Datasets:

---

JetBrains-Research

/

commit-chronicle

like 9

Follow

JetBrains Research 71

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/instructions/fed/CtableFEDInstruction.java", "new_path": "src/main/java/org/apache/sysds/runtime/instructions/fed/CtableFEDInstruction.java", "diff": "@@ -22,7 +22,10 @@ package org.apache.sysds.runtime.instructions.fed;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.concurrent.Future;\n+import java.util.Iterator;\n+import java.util.SortedMap;\nimport java.util.stream.IntStream;\n+import java.util.TreeMap;\nimport org.apache.commons.lang3.tuple.Pair;\nimport org.apache.sysds.common.Types.DataType;\n@@ -38,7 +41,6 @@ import org.apache.sysds.runtime.controlprogram.federated.FederatedUDF;\nimport org.apache.sysds.runtime.controlprogram.federated.FederationMap;\nimport org.apache.sysds.runtime.controlprogram.federated.FederationMap.AlignType;\nimport org.apache.sysds.runtime.controlprogram.federated.FederationUtils;\n-import org.apache.sysds.runtime.functionobjects.And;\nimport org.apache.sysds.runtime.instructions.Instruction;\nimport org.apache.sysds.runtime.instructions.InstructionUtils;\nimport org.apache.sysds.runtime.instructions.cp.CPOperand;\n@@ -105,8 +107,10 @@ public class CtableFEDInstruction extends ComputationFEDInstruction {\n}\n// get new output dims\n- Long[] dims1 = getOutputDimension(mo1, input1, _outDim1, mo1.getFedMapping().getFederatedRanges());\n- Long[] dims2 = getOutputDimension(mo2, input2, _outDim2, mo1.getFedMapping().getFederatedRanges());\n+ Long[] dims1 = getOutputDimension(mo1, reversed ? input2 : input1, reversed ? _outDim2 : _outDim1,\n+ mo1.getFedMapping().getFederatedRanges());\n+ Long[] dims2 = getOutputDimension(mo2, reversed ? input1 : input2, reversed ? _outDim1 : _outDim2,\n+ mo1.getFedMapping().getFederatedRanges());\nMatrixObject mo3 = input3 != null && input3.isMatrix() ? ec.getMatrixObject(input3) : null;\n@@ -116,119 +120,157 @@ public class CtableFEDInstruction extends ComputationFEDInstruction {\nmo1 = ec.getMatrixObject(input3);\n}\n- long dim1 = Collections.max(Arrays.asList(dims1), Long::compare);\n- boolean fedOutput = dim1 % mo1.getFedMapping().getSize() == 0 && dims1.length == Arrays.stream(dims1).distinct().count();\n+ // static non-partitioned output dimension (same for all federated partitions)\n+ long staticDim = Collections.max(Arrays.asList(dims1), Long::compare);\n+ boolean fedOutput = isFedOutput(mo1.getFedMapping(), mo2);\n- processRequest(ec, mo1, mo2, mo3, reversed, reversedWeights, fedOutput, dims1, dims2);\n+ processRequest(ec, mo1, mo2, mo3, reversed, reversedWeights, fedOutput, staticDim, dims2);\n}\n+ /**\n+ * Broadcast, execute, and finalize the federated instruction according to\n+ * the specified inputs.\n+ *\n+ * @param ec execution context\n+ * @param mo1 input matrix object 1\n+ * @param mo2 input matrix object 2\n+ * @param mo3 input matrix object 3 or null\n+ * @param reversed boolean indicating if inputs mo1 and mo2 are reversed\n+ * @param reversedWeights boolean indicating if inputs mo1 and mo3 are reversed\n+ * @param fedOutput boolean indicating if output can be kept federated\n+ * @param staticDim static non-partitioned dimension of the output\n+ * @param dims2 dimensions of the partial outputs along the federated partitioning\n+ */\nprivate void processRequest(ExecutionContext ec, MatrixObject mo1, MatrixObject mo2, MatrixObject mo3,\n- boolean reversed, boolean reversedWeights, boolean fedOutput, Long[] dims1, Long[] dims2) {\n+ boolean reversed, boolean reversedWeights, boolean fedOutput, long staticDim, Long[] dims2) {\n+\n+ FederationMap fedMap = mo1.getFedMapping();\n+\n+ FederatedRequest[] fr1 = fedMap.broadcastSliced(mo2, false);\n+ FederatedRequest[] fr2 = null;\n+ FederatedRequest fr3, fr4, fr5;\nFuture<FederatedResponse>[] ffr;\n- FederatedRequest[] fr1 = mo1.getFedMapping().broadcastSliced(mo2, false);\n- FederatedRequest fr2, fr3;\nif(mo3 != null && mo1.isFederated() && mo3.isFederated()\n- && mo1.getFedMapping().isAligned(mo3.getFedMapping(), AlignType.FULL)) { // mo1 and mo3 federated and aligned\n+ && fedMap.isAligned(mo3.getFedMapping(), AlignType.FULL)) { // mo1 and mo3 federated and aligned\nif(!reversed)\n- fr2 = FederationUtils.callInstruction(instString, output, new CPOperand[] {input1, input2, input3},\n- new long[] {mo1.getFedMapping().getID(), fr1[0].getID(), mo3.getFedMapping().getID()});\n+ fr3 = FederationUtils.callInstruction(instString, output, new CPOperand[] {input1, input2, input3},\n+ new long[] {fedMap.getID(), fr1[0].getID(), mo3.getFedMapping().getID()});\nelse\n- fr2 = FederationUtils.callInstruction(instString, output, new CPOperand[] {input1, input2, input3},\n- new long[] {fr1[0].getID(), mo1.getFedMapping().getID(), mo3.getFedMapping().getID()});\n-\n- fr3 = new FederatedRequest(FederatedRequest.RequestType.GET_VAR, fr2.getID());\n- ffr = mo1.getFedMapping().execute(getTID(), true, fr1, fr2, fr3);\n+ fr3 = FederationUtils.callInstruction(instString, output, new CPOperand[] {input1, input2, input3},\n+ new long[] {fr1[0].getID(), fedMap.getID(), mo3.getFedMapping().getID()});\n}\nelse if(mo3 == null) {\nif(!reversed)\n- fr2 = FederationUtils.callInstruction(instString, output, new CPOperand[] {input1, input2},\n- new long[] {mo1.getFedMapping().getID(), fr1[0].getID()});\n+ fr3 = FederationUtils.callInstruction(instString, output, new CPOperand[] {input1, input2},\n+ new long[] {fedMap.getID(), fr1[0].getID()});\nelse\n- fr2 = FederationUtils.callInstruction(instString, output, new CPOperand[] {input1, input2},\n- new long[] {fr1[0].getID(), mo1.getFedMapping().getID()});\n-\n- fr3 = new FederatedRequest(FederatedRequest.RequestType.GET_VAR, fr2.getID());\n- ffr = mo1.getFedMapping().execute(getTID(), true, fr1, fr2, fr3);\n-\n- } else {\n- FederatedRequest[] fr4 = mo1.getFedMapping().broadcastSliced(mo3, false);\n+ fr3 = FederationUtils.callInstruction(instString, output, new CPOperand[] {input1, input2},\n+ new long[] {fr1[0].getID(), fedMap.getID()});\n+ }\n+ else {\n+ fr2 = fedMap.broadcastSliced(mo3, false);\nif(!reversed && !reversedWeights)\n- fr2 = FederationUtils.callInstruction(instString, output, new CPOperand[] {input1, input2, input3},\n- new long[] {mo1.getFedMapping().getID(), fr1[0].getID(), fr4[0].getID()});\n+ fr3 = FederationUtils.callInstruction(instString, output, new CPOperand[] {input1, input2, input3},\n+ new long[] {fedMap.getID(), fr1[0].getID(), fr2[0].getID()});\nelse if(reversed && !reversedWeights)\n- fr2 = FederationUtils.callInstruction(instString, output, new CPOperand[] {input1, input2, input3},\n- new long[] {fr1[0].getID(), mo1.getFedMapping().getID(), fr4[0].getID()});\n+ fr3 = FederationUtils.callInstruction(instString, output, new CPOperand[] {input1, input2, input3},\n+ new long[] {fr1[0].getID(), fedMap.getID(), fr2[0].getID()});\nelse\n- fr2 = FederationUtils.callInstruction(instString, output, new CPOperand[] {input1, input2, input3},\n- new long[] {fr1[0].getID(), fr4[0].getID(), mo1.getFedMapping().getID()});\n-\n- fr3 = new FederatedRequest(FederatedRequest.RequestType.GET_VAR, fr2.getID());\n- ffr = mo1.getFedMapping().execute(getTID(), true, fr1, fr4, fr2, fr3);\n+ fr3 = FederationUtils.callInstruction(instString, output, new CPOperand[] {input1, input2, input3},\n+ new long[] {fr1[0].getID(), fr2[0].getID(), fedMap.getID()});\n}\n- if(fedOutput && isFedOutput(ffr, dims1)) {\n+ if(fedOutput) {\n+ if(fr2 != null) // broadcasted mo3\n+ fedMap.execute(getTID(), true, fr1, fr2, fr3);\n+ else\n+ fedMap.execute(getTID(), true, fr1, fr3);\n+\nMatrixObject out = ec.getMatrixObject(output);\n- FederationMap newFedMap = modifyFedRanges(mo1.getFedMapping(), dims1, dims2);\n- setFedOutput(mo1, out, newFedMap, dims1, fr2.getID());\n+ FederationMap newFedMap = modifyFedRanges(fedMap.copyWithNewID(fr3.getID()),\n+ staticDim, dims2, reversed);\n+ setFedOutput(mo1, out, newFedMap, staticDim, dims2, reversed);\n} else {\n+ fr4 = new FederatedRequest(FederatedRequest.RequestType.GET_VAR, fr3.getID());\n+ fr5 = fedMap.cleanup(getTID(), fr3.getID());\n+ if(fr2 != null) // broadcasted mo3\n+ ffr = fedMap.execute(getTID(), true, fr1, fr2, fr3, fr4, fr5);\n+ else\n+ ffr = fedMap.execute(getTID(), true, fr1, fr3, fr4, fr5);\n+\nec.setMatrixOutput(output.getName(), aggResult(ffr));\n}\n}\n- boolean isFedOutput(Future<FederatedResponse>[] ffr, Long[] dims1) {\n- boolean fedOutput = true;\n-\n- long fedSize = Collections.max(Arrays.asList(dims1), Long::compare) / ffr.length;\n- try {\n- MatrixBlock curr;\n- MatrixBlock prev =(MatrixBlock) ffr[0].get().getData()[0];\n- for(int i = 1; i < ffr.length && fedOutput; i++) {\n- curr = (MatrixBlock) ffr[i].get().getData()[0];\n- MatrixBlock sliced = curr.slice((int) (curr.getNumRows() - fedSize), curr.getNumRows() - 1);\n-\n- if(curr.getNumColumns() != prev.getNumColumns())\n- return false;\n-\n- // no intersection\n- if(curr.getNumRows() == (i+1) * prev.getNumRows() && curr.getNonZeros() <= prev.getLength()\n- && (curr.getNumRows() - sliced.getNumRows()) == i * prev.getNumRows()\n- && curr.getNonZeros() - sliced.getNonZeros() == 0)\n- continue;\n+ /**\n+ * Evaluate if the output can be kept federated on the different federated\n+ * sites or if the output needs to be aggregated on the coordinator, based\n+ * on the output ranges of mo2.\n+ * The output can be kept federated if the slices of mo2, sliced corresponding\n+ * to the federated ranges of mo1, have strict separable and ascending value\n+ * ranges. From this property it follows that the partial outputs can also\n+ * be separated, and hence the overall output can be created by a simple\n+ * binding through a federated mapping.\n+ *\n+ * @param fedMap the federation map of the federated matrix input mo1\n+ * @param mo2 input matrix object mo2\n+ * @return boolean indicating if the output can be kept on the federated sites\n+ */\n+ private boolean isFedOutput(FederationMap fedMap, MatrixObject mo2) {\n+ MatrixBlock mb = mo2.acquireReadAndRelease();\n+ FederatedRange[] fedRanges = fedMap.getFederatedRanges(); // federated ranges of mo1\n+ SortedMap<Double, Double> fedDims = new TreeMap<Double, Double>(); // <beginDim, endDim>\n+\n+ // collect min and max of the corresponding slices of mo2\n+ IntStream.range(0, fedRanges.length).forEach(i -> {\n+ MatrixBlock sliced = mb.slice(\n+ fedRanges[i].getBeginDimsInt()[0], fedRanges[i].getEndDimsInt()[0] - 1,\n+ fedRanges[i].getBeginDimsInt()[1], fedRanges[i].getEndDimsInt()[1] - 1);\n+ fedDims.put(sliced.min(), sliced.max());\n+ });\n- // check intersect with AND and compare number of nnz\n- MatrixBlock prevExtend = new MatrixBlock(curr.getNumRows(), curr.getNumColumns(), true, 0);\n- prevExtend.copy(0, prev.getNumRows()-1, 0, prev.getNumColumns()-1, prev, true);\n+ boolean retVal = (fedDims.size() == fedRanges.length); // no duplicate begin dimension entries\n- MatrixBlock intersect = curr.binaryOperationsInPlace(new BinaryOperator(And.getAndFnObject()), prevExtend);\n- if(intersect.getNonZeros() != 0)\n- fedOutput = false;\n- prev = sliced;\n+ Iterator<SortedMap.Entry<Double, Double>> iter = fedDims.entrySet().iterator();\n+ SortedMap.Entry<Double, Double> entry = iter.next(); // first entry does not have to be checked\n+ double prevEndDim = entry.getValue();\n+ while(iter.hasNext() && retVal) {\n+ entry = iter.next();\n+ // previous end dimension must be less than current begin dimension (no overlaps of ranges)\n+ retVal &= (prevEndDim < entry.getKey());\n+ prevEndDim = entry.getValue();\n}\n- }\n- catch(Exception e) {\n- e.printStackTrace();\n- }\n- return fedOutput;\n- }\n-\n- private static void setFedOutput(MatrixObject mo1, MatrixObject out, FederationMap fedMap, Long[] dims1, long outId) {\n- long fedSize = Collections.max(Arrays.asList(dims1), Long::compare) / dims1.length;\n+ return retVal;\n+ }\n- long d1 = Collections.max(Arrays.asList(dims1), Long::compare);\n- long d2 = Collections.max(Arrays.asList(dims1), Long::compare);\n+ /**\n+ * Set the output and its data characteristics on the federated sites.\n+ *\n+ * @param mo1 input matrix object mo1\n+ * @param out input matrix object of the output\n+ * @param fedMap the federation map of the federated matrix input mo1\n+ * @param staticDim static non-partitioned dimension of the output\n+ * @param dims2 dimensions of the partial outputs along the federated partitioning\n+ * @param reversed boolean indicating if inputs mo1 and mo2 are reversed\n+ */\n+ private static void setFedOutput(MatrixObject mo1, MatrixObject out, FederationMap fedMap,\n+ long staticDim, Long[] dims2, boolean reversed) {\n+ // get the final output dimensions\n+ final long d1 = (reversed ? Collections.max(Arrays.asList(dims2)) : staticDim);\n+ final long d2 = (reversed ? staticDim : Collections.max(Arrays.asList(dims2)));\n// set output\nout.getDataCharacteristics().set(d1, d2, (int) mo1.getBlocksize(), mo1.getNnz());\n- out.setFedMapping(fedMap.copyWithNewID(outId));\n+ out.setFedMapping(fedMap);\nlong varID = FederationUtils.getNextFedDataID();\n- out.getFedMapping().mapParallel(varID, (range, data) -> {\n+ fedMap.mapParallel(varID, (range, data) -> {\ntry {\nFederatedResponse response = data.executeFederatedOperation(new FederatedRequest(\nFederatedRequest.RequestType.EXEC_UDF, -1,\n- new SliceOutput(data.getVarID(), fedSize))).get();\n+ new SliceOutput(data.getVarID(), staticDim, dims2, reversed))).get();\nif(!response.isSuccessful())\nresponse.throwExceptionFromResponse();\n}\n@@ -239,6 +281,9 @@ public class CtableFEDInstruction extends ComputationFEDInstruction {\n});\n}\n+ /**\n+ * Aggregate the partial outputs locally.\n+ */\nprivate static MatrixBlock aggResult(Future<FederatedResponse>[] ffr) {\nMatrixBlock resultBlock = new MatrixBlock(1, 1, true, 0);\nint dim1 = 0, dim2 = 0;\n@@ -266,27 +311,44 @@ public class CtableFEDInstruction extends ComputationFEDInstruction {\nreturn resultBlock;\n}\n- private static FederationMap modifyFedRanges(FederationMap fedMap, Long[] dims1, Long[] dims2) {\n- IntStream.range(0, fedMap.getFederatedRanges().length).forEach(i -> {\n- fedMap.getFederatedRanges()[i]\n- .setBeginDim(0, i == 0 ? 0 : fedMap.getFederatedRanges()[i - 1].getEndDims()[0]);\n- fedMap.getFederatedRanges()[i].setEndDim(0, dims1[i]);\n- fedMap.getFederatedRanges()[i]\n- .setBeginDim(1, i == 0 ? 0 : fedMap.getFederatedRanges()[i - 1].getBeginDims()[1]);\n- fedMap.getFederatedRanges()[i].setEndDim(1, dims2[i]);\n+ /**\n+ * Set the ranges of the federation map according to the static dimension and\n+ * the individual dimensions of the partial output matrices.\n+ *\n+ * @param fedMap the federation map of the federated matrix input mo1\n+ * @param staticDim static non-partitioned dimension of the output\n+ * @param dims2 dimensions of the partial outputs along the federated partitioning\n+ * @param reversed boolean indicating if inputs mo1 and mo2 are reversed\n+ * @return FederationMap the modified federation map\n+ */\n+ private static FederationMap modifyFedRanges(FederationMap fedMap, long staticDim,\n+ Long[] dims2, boolean reversed) {\n+ // set the federated ranges to the individual partition sizes\n+ IntStream.range(0, fedMap.getFederatedRanges().length).forEach(counter -> {\n+ FederatedRange fedRange = fedMap.getFederatedRanges()[counter];\n+ fedRange.setBeginDim(reversed ? 1 : 0, 0);\n+ fedRange.setEndDim(reversed ? 1 : 0, staticDim);\n+ fedRange.setBeginDim(reversed ? 0 : 1, counter == 0 ? 0 : dims2[counter-1]);\n+ fedRange.setEndDim(reversed ? 0 : 1, dims2[counter]);\n});\nreturn fedMap;\n}\n- private Long[] getOutputDimension(MatrixObject in, CPOperand inOp, CPOperand outOp, FederatedRange[] federatedRanges) {\n+ /**\n+ * Compute the output dimensions of the partial outputs according to the\n+ * federated ranges.\n+ */\n+ private Long[] getOutputDimension(MatrixObject in, CPOperand inOp, CPOperand outOp,\n+ FederatedRange[] federatedRanges) {\nLong[] fedDims = new Long[federatedRanges.length];\nif(!in.isFederated()) {\n//slice\nMatrixBlock mb = in.acquireReadAndRelease();\nIntStream.range(0, federatedRanges.length).forEach(i -> {\n- MatrixBlock sliced = mb\n- .slice(federatedRanges[i].getBeginDimsInt()[0], federatedRanges[i].getEndDimsInt()[0] - 1);\n+ MatrixBlock sliced = mb.slice(\n+ federatedRanges[i].getBeginDimsInt()[0], federatedRanges[i].getEndDimsInt()[0] - 1,\n+ federatedRanges[i].getBeginDimsInt()[1], federatedRanges[i].getEndDimsInt()[1] - 1);\nfedDims[i] = (long) sliced.max();\n});\nreturn fedDims;\n@@ -326,26 +388,76 @@ public class CtableFEDInstruction extends ComputationFEDInstruction {\nreturn String.join(Lop.OPERAND_DELIMITOR, maxInstParts);\n}\n+ /**\n+ * Static class which extends FederatedUDF to modify the partial outputs on\n+ * the federated sites such that they can be bound without any local\n+ * aggregation.\n+ */\nprivate static class SliceOutput extends FederatedUDF {\nprivate static final long serialVersionUID = -2808597461054603816L;\n- private final long _fedSize;\n+ private final int _staticDim;\n+ private final Long[] _fedDims;\n+ private final boolean _reversed;\n- protected SliceOutput(long input, long fedSize) {\n+ protected SliceOutput(long input, long staticDim, Long[] fedDims, boolean reversed) {\nsuper(new long[] {input});\n- _fedSize = fedSize;\n+ _staticDim = (int)staticDim;\n+ _fedDims = fedDims;\n+ _reversed = reversed;\n}\n+ /**\n+ * Find the dimensions of the partial output matrix and expand it to the\n+ * global static dimension along the non-partitioned axis and crop it\n+ * along the paritioned axis.\n+ *\n+ * @param ec the execution context\n+ * @param data\n+ * @return FederatedResponse with status SUCCESS and an empty object\n+ */\npublic FederatedResponse execute(ExecutionContext ec, Data... data) {\nMatrixObject mo = (MatrixObject) data[0];\nMatrixBlock mb = mo.acquireReadAndRelease();\n- MatrixBlock sliced = mb.slice((int) (mb.getNumRows()-_fedSize), mb.getNumRows()-1);\n+ int beginDim = 0;\n+ int endDim = (_reversed ? mb.getNumRows() : mb.getNumColumns());\n+ int localStaticDim = (_reversed ? mb.getNumColumns() : mb.getNumRows());\n+ for(int counter = 0; counter < _fedDims.length; counter++) {\n+ if(_fedDims[counter] == endDim) {\n+ beginDim = (counter == 0 ? 0 : _fedDims[counter - 1].intValue());\n+ break;\n+ }\n+ }\n+\n+ mb = expandMatrix(mb, localStaticDim);\n+\n+ // crop the output\n+ MatrixBlock sliced = _reversed ? mb.slice(beginDim, endDim - 1, 0, _staticDim - 1)\n+ : mb.slice(0, _staticDim - 1, beginDim, endDim - 1);\nmo.acquireModify(sliced);\nmo.release();\nreturn new FederatedResponse(FederatedResponse.ResponseType.SUCCESS, new Object[] {});\n}\n+\n+ /**\n+ * Expand the matrix with zeros up to the specified static dimension.\n+ *\n+ * @param mb the matrix block of the partial output\n+ * @param localStaticDim the static dimension of the output matrix block\n+ * @return MatrixBlock the output matrix block expanded to the global static dimension\n+ */\n+ private MatrixBlock expandMatrix(MatrixBlock mb, int localStaticDim) {\n+ int diff = _staticDim - localStaticDim;\n+ if(diff > 0) {\n+ MatrixBlock tmpMb = (_reversed ? new MatrixBlock(mb.getNumRows(), diff, (double) 0)\n+ : new MatrixBlock(diff, mb.getNumColumns(), (double) 0));\n+ mb = mb.append(tmpMb, null, _reversed);\n+ }\n+ return mb;\n+ }\n+\n@Override\npublic Pair<String, LineageItem> getLineageItem(ExecutionContext ec) {\nreturn null;\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/federated/primitives/FederatedCtableTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/federated/primitives/FederatedCtableTest.java", "diff": "@@ -85,6 +85,9 @@ public class FederatedCtableTest extends AutomatedTestBase {\n@Test\npublic void federatedCtableMatrixInputSinglenode() { runCtable(Types.ExecMode.SINGLE_NODE, false, true); }\n+ @Test\n+ public void federatedCtableMatrixInputFedOutputSingleNode() { runCtable(Types.ExecMode.SINGLE_NODE, true, true); }\n+\npublic void runCtable(Types.ExecMode execMode, boolean fedOutput, boolean matrixInput) {\nString TEST_NAME = fedOutput ? TEST_NAME2 : TEST_NAME1;\n@@ -108,7 +111,7 @@ public class FederatedCtableTest extends AutomatedTestBase {\nloadTestConfiguration(config);\nif(fedOutput)\n- runFedCtable(HOME, TEST_NAME, port1, port2, port3, port4);\n+ runFedCtable(HOME, TEST_NAME, matrixInput, port1, port2, port3, port4);\nelse\nrunNonFedCtable(HOME, TEST_NAME, matrixInput, port1, port2, port3, port4);\ncheckResults();\n@@ -155,7 +158,7 @@ public class FederatedCtableTest extends AutomatedTestBase {\nrunTest(true, false, null, -1);\n}\n- private void runFedCtable(String HOME, String TEST_NAME, int port1, int port2, int port3, int port4) {\n+ private void runFedCtable(String HOME, String TEST_NAME, boolean matrixInput, int port1, int port2, int port3, int port4) {\nint r = rows / 4;\nint c = cols;\n@@ -174,7 +177,8 @@ public class FederatedCtableTest extends AutomatedTestBase {\nfullDMLScriptName = HOME + TEST_NAME2 + \"Reference.dml\";\nprogramArgs = new String[]{\"-stats\", \"100\", \"-args\",\ninput(\"X1\"), input(\"X2\"), input(\"X3\"), input(\"X4\"), Boolean.toString(reversedInputs).toUpperCase(),\n- Boolean.toString(weighted).toUpperCase(), expected(\"F\")};\n+ Boolean.toString(weighted).toUpperCase(), Boolean.toString(matrixInput).toUpperCase(),\n+ expected(\"F\")};\nrunTest(true, false, null, -1);\n// Run actual dml script with federated matrix\n@@ -185,6 +189,7 @@ public class FederatedCtableTest extends AutomatedTestBase {\n\"in_X3=\" + TestUtils.federatedAddress(port3, input(\"X3\")),\n\"in_X4=\" + TestUtils.federatedAddress(port4, input(\"X4\")),\n\"rows=\" + rows, \"cols=\" + cols, \"revIn=\" + Boolean.toString(reversedInputs).toUpperCase(),\n+ \"matrixInput=\" + Boolean.toString(matrixInput).toUpperCase(),\n\"weighted=\" + Boolean.toString(weighted).toUpperCase(), \"out=\" + output(\"F\")\n};\nrunTest(true, false, null, -1);\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/federated/FederatedCtableFedOutput.dml", "new_path": "src/test/scripts/functions/federated/FederatedCtableFedOutput.dml", "diff": "@@ -28,8 +28,14 @@ n = ncol(X);\n# prepare offset vectors and one-hot encoded X\nmaxs = colMaxs(X);\n+if($matrixInput) {\n+ rix = matrix(seq(1,m)%*%matrix(1,1,n), m, n);\n+ cix = matrix(X + (t(cumsum(t(maxs))) - maxs), m, n);\n+}\n+else {\nrix = matrix(seq(1,m)%*%matrix(1,1,n), m*n, 1);\ncix = matrix(X + (t(cumsum(t(maxs))) - maxs), m*n, 1);\n+}\nW = rix + cix;\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/federated/FederatedCtableFedOutputReference.dml", "new_path": "src/test/scripts/functions/federated/FederatedCtableFedOutputReference.dml", "diff": "@@ -27,8 +27,14 @@ n = ncol(X);\n# prepare offset vectors and one-hot encoded X\nmaxs = colMaxs(X);\n-rix = matrix(seq(1,m)%*%matrix(1,1,n), m*n, 1)\n+if($7) { # matrix input\n+ rix = matrix(seq(1,m)%*%matrix(1,1,n), m, n);\n+ cix = matrix(X + (t(cumsum(t(maxs))) - maxs), m, n);\n+}\n+else {\n+ rix = matrix(seq(1,m)%*%matrix(1,1,n), m*n, 1);\ncix = matrix(X + (t(cumsum(t(maxs))) - maxs), m*n, 1);\n+}\nW = rix + cix;\n@@ -43,4 +49,4 @@ else\nelse\nX2 = table(rix, cix);\n-write(X2, $7);\n+write(X2, $8);\n" } ]

[ { "change_type": "MODIFY", "old_path": "scripts/builtin/topk_cleaning.dml", "new_path": "scripts/builtin/topk_cleaning.dml", "diff": "source(\"scripts/pipelines/scripts/utils.dml\") as utils;\nsource(\"scripts/pipelines/scripts/enumerateLogical.dml\") as lg;\n-\ns_topk_cleaning = function(Frame[Unknown] dataTrain, Frame[Unknown] dataTest = as.frame(\"NULL\"), Frame[Unknown] metaData = as.frame(\"NULL\"), Frame[Unknown] primitives,\nFrame[Unknown] parameters, Matrix[Double] cmr = matrix(\"4 0.7 1\", rows=1, cols=3), String evaluationFunc, Matrix[Double] evalFunHp, Integer topK = 5,\nInteger resource_val = 20, Double sample = 0.1, Boolean cv=TRUE, Integer cvk = 2, Boolean isLastLabel = TRUE, Boolean correctTypos=FALSE, String output)\n@@ -30,14 +29,17 @@ s_topk_cleaning = function(Frame[Unknown] dataTrain, Frame[Unknown] dataTest = a\n# return (Frame[Unknown] topKPipelines, Matrix[Double] topKHyperParams, Matrix[Double] topKScores, Frame[Unknown] bestLogical,\n# Frame[Unknown] features, Double dirtyScore, Matrix[Double] evalFunHp)\n{\n+ t1 = time(); print(\"TopK-Cleaning:\");\n+\nXtest = as.frame(\"0\")\nYtest = as.frame(\"0\")\n- print(\"starting topk_cleaning\")\n-\n- [schema, mask, fdMask, maskY] = prepareMeta(dataTrain, metaData)\n+ ctx = list(prefix=\"----\"); #TODO include seed\n+ # prepare meta data\n# # keeping the meta list format if we decide to add more stuff in metadata\n+ [schema, mask, fdMask, maskY] = prepareMeta(dataTrain, metaData)\nmetaList = list(mask=mask, schema=schema, fd=fdMask)\n+ t2 = time(); print(\"-- Cleaning - Prepare Metadata: \"+(t2-t1)/1e9+\"s\");\n# separate the label\n[Xtrain, Ytrain] = getLabel(dataTrain, isLastLabel)\n@@ -49,24 +51,31 @@ s_topk_cleaning = function(Frame[Unknown] dataTrain, Frame[Unknown] dataTest = a\n[eYtrain, M] = transformencode(target=Ytrain, spec= \"{ids:true, recode:[1]}\");\neYtest = transformapply(target=Ytest, spec= \"{ids:true, recode:[1]}\", meta=M);\n}\n- else\n- {\n+ else {\neYtrain = as.matrix(Ytrain)\neYtest = as.matrix(Ytest)\n}\n+ t3 = time(); print(\"-- Cleaning - Prepare Labels: \"+(t3-t2)/1e9+\"s\");\n# # # when the evaluation function is called first we also compute and keep hyperparams of target application\n+ print(\"-- Cleaning - Get Dirty Score: \");\n[dirtyScore, evalFunHp] = getDirtyScore(X=Xtrain, Y=eYtrain, Xtest=Xtest, Ytest=eYtest, evaluationFunc=evaluationFunc,\n- metaList=metaList, evalFunHp=evalFunHp, sample=sample, trainML=1, cv=cv, cvk=cvk)\n+ metaList=metaList, evalFunHp=evalFunHp, sample=sample, trainML=1, cv=cv, cvk=cvk, ctx=ctx)\n+ t4 = time(); print(\"---- finalized in: \"+(t4-t3)/1e9+\"s\");\n# # do the string processing\n- [Xtrain, Xtest] = runStringPipeline(Xtrain, Xtest, schema, mask, cv, correctTypos)\n+ print(\"-- Cleaning - Data Preparation (strings, transform, sample): \");\n+ [Xtrain, Xtest] = runStringPipeline(Xtrain, Xtest, schema, mask, cv, correctTypos, ctx)\n# # if mask has 1s then there are categorical features\n+ print(\"---- feature transformations to numeric matrix\");\n[eXtrain, eXtest] = recodeData(Xtrain, Xtest, mask, cv, \"recode\")\n# apply sampling on training data for pipeline enumeration\n+ # TODO why recoding/sampling twice (within getDirtyScore)\n+ print(\"---- class-stratified sampling of feature matrix w/ f=\"+sample);\n[eXtrain, eYtrain] = utils::doSample(eXtrain, eYtrain, sample, TRUE)\n+ t5 = time(); print(\"---- finalized in: \"+(t5-t4)/1e9+\"s\");\n# # # create logical pipeline seeds\nlogicalSeedCI = frame([\n@@ -109,7 +118,7 @@ s_topk_cleaning = function(Frame[Unknown] dataTrain, Frame[Unknown] dataTest = a\n[bestLogical, score, T] = lg::enumerateLogical(X=eXtrain, y=eYtrain, Xtest=eXtest, ytest=eYtest, cmr=cmr, cat=category, population=logical[2:nrow(logical)],\nmax_iter=ceil(resource_val/topK), metaList = metaList, evaluationFunc=evaluationFunc, evalFunHp=evalFunHp,\nprimitives=primitives, param=parameters, num_inst=3 , num_exec=2, cv=cv, cvk=cvk, verbose=TRUE)\n- # # # bestLogical = frame([\"MVI\", \"CI\", \"SCALE\"], rows=1, cols=3)\n+ t6 = time(); print(\"-- Cleaning - Enum Logical Pipelines: \"+(t6-t5)/1e9+\"s\");\ntopKPipelines = as.frame(\"NULL\"); topKHyperParams = matrix(0,0,0); topKScores = matrix(0,0,0); features = as.frame(\"NULL\")\n@@ -117,6 +126,7 @@ s_topk_cleaning = function(Frame[Unknown] dataTrain, Frame[Unknown] dataTest = a\nperf = bandit(X_train=eXtrain, Y_train=eYtrain, X_test=eXtest, Y_test=eYtest, metaList=metaList,\nevaluationFunc=evaluationFunc, evalFunHp=evalFunHp, lp=bestLogical, primitives=primitives, param=parameters, baseLineScore=dirtyScore,\nk=topK, R=resource_val, cv=cv, output=output, verbose=TRUE);\n+ t7 = time(); print(\"-- Cleaning - Enum Physical Pipelines: \"+(t7-t6)/1e9+\"s\");\n}\nprepareMeta = function(Frame[Unknown] data, Frame[Unknown] metaData)\n@@ -160,45 +170,46 @@ return(Frame[Unknown] X, Frame[Unknown] Y)\n}\nrunStringPipeline = function(Frame[Unknown] Xtrain, Frame[Unknown] Xtest, Frame[String] schema,\n- Matrix[Double] mask, Boolean cv, Boolean correctTypos = FALSE)\n+ Matrix[Double] mask, Boolean cv, Boolean correctTypos = FALSE, List[Unknown] ctx)\nreturn(Frame[Unknown] Xtrain, Frame[Unknown] Xtest)\n{\nif(cv)\n- Xtrain = utils::stringProcessing(data=Xtrain, mask=mask, schema=schema, CorrectTypos=correctTypos)\n+ Xtrain = utils::stringProcessing(data=Xtrain, mask=mask, schema=schema, CorrectTypos=correctTypos, ctx=ctx)\nelse\n{\n# # # binding train and test to use same dictionary for both\n- XAll = utils::stringProcessing(data=rbind(Xtrain, Xtest), mask=mask, schema=schema, CorrectTypos=correctTypos)\n+ XAll = utils::stringProcessing(data=rbind(Xtrain, Xtest), mask=mask, schema=schema, CorrectTypos=correctTypos, ctx=ctx)\nXtrain = XAll[1:nrow(Xtrain),]\nXtest = XAll[nrow(Xtrain)+1:nrow(XAll),]\n}\n}\ngetDirtyScore = function(Frame[Unknown] X, Matrix[Double] Y, Frame[Unknown] Xtest, Matrix[Double] Ytest, String evaluationFunc, List[Unknown] metaList,\n- Matrix[Double] evalFunHp, Double sample, Integer trainML, Boolean cv, Integer cvk)\n+ Matrix[Double] evalFunHp, Double sample, Integer trainML, Boolean cv, Integer cvk, List[Unknown] ctx=list() )\nreturn(Double dirtyScore, Matrix[Double] evalFunHp)\n{\n+ prefix = as.scalar(ctx[\"prefix\"]);\nmask = as.matrix(metaList['mask'])\n[eXtrain, eXtest] = recodeData(X, Xtest, mask, cv, \"recode\")\neXtrain = replace(target=eXtrain, pattern=NaN, replacement = 0)\neXtest = replace(target=eXtest, pattern=NaN, replacement = 0)\ndirtyScore = 100\n- # # # sample data\n+ print(prefix+\" sample from train data and dummy code\");\n[eXtrain, Ytrain] = utils::doSample(eXtrain, Y, sample, TRUE)\n[eXtrain, eXtest] = recodeData(as.frame(eXtrain), as.frame(eXtest), mask, cv, \"dummycode\")\npipList = list(lp = as.frame(\"NULL\"), ph = as.frame(\"NULL\"), hp = as.matrix(0), flags = 0)\n- if(cv)\n- {\n- score = crossV(X=eXtrain, y=Ytrain, cvk=cvk, evalFunHp=evalFunHp, pipList=pipList, metaList=metaList, evalFunc=evaluationFunc, trainML = 1)\n+\n+ print(prefix+\" hyper-parameter tuning\");\n+ if(cv) {\n+ score = crossV(X=eXtrain, y=Ytrain, cvk=cvk, evalFunHp=evalFunHp,\n+ pipList=pipList, metaList=metaList, evalFunc=evaluationFunc, trainML = 1)\n}\n- else\n- {\n+ else {\nscore = eval(evaluationFunc, list(X=eXtrain, Y=Ytrain, Xtest=eXtest, Ytest=Ytest, Xorig=as.matrix(0), evalFunHp=evalFunHp, trainML = 1))\n}\ndirtyScore = as.scalar(score[1, 1])\nevalFunHp = score[1, 2:ncol(score)]\n- # evalFunHp = scoreAndHp[1, 2:ncol(scoreAndHp)]\n}\nrecodeData = function(Frame[Unknown] Xtrain, Frame[Unknown] Xtest, Matrix[Double] mask, Boolean cv, String code)\n" }, { "change_type": "MODIFY", "old_path": "scripts/pipelines/scripts/cleaning.dml", "new_path": "scripts/pipelines/scripts/cleaning.dml", "diff": "@@ -493,7 +493,6 @@ crossV = function(Matrix[double] X, Matrix[double] y, Integer k, Matrix[Double]\nMatrix[Double] MLhp, Boolean isWeighted)\nreturn (Matrix[Double] accuracyMatrix)\n{\n-\naccuracyMatrix = matrix(0, k, 1)\ndataList = list()\n@@ -526,7 +525,6 @@ return (Matrix[Double] accuracyMatrix)\ndataList = append(dataList, fold_i)\nfold_idxes[, 1] = fold_idxes[, 2] + 1\nfold_idxes[, 2] += ins_per_fold\n- while(FALSE){}\n}\nfor(i in seq(1,k))\n" }, { "change_type": "MODIFY", "old_path": "scripts/pipelines/scripts/utils.dml", "new_path": "scripts/pipelines/scripts/utils.dml", "diff": "@@ -60,24 +60,26 @@ doSample = function(Matrix[Double] eX, Matrix[Double] eY, Double ratio, Boolean\n{\nMIN_SAMPLE = 1000\nsampled = floor(nrow(eX) * ratio)\n- sample = ifelse(sampled > MIN_SAMPLE, TRUE, FALSE)\n- dist = table(eY, 1)\n- dist = nrow(dist)\n- if(sample)\n+ sampledX = eX\n+ sampledY = eY\n+\n+ if(sampled > MIN_SAMPLE)\n{\n+ dist = max(eY) # num classes (one-hot encoded eY)\n+\nif((nrow(eY) > 1) & (dist < 10)) # for classification\n{\nXY = order(target = cbind(eY, eX), by = 1, decreasing=FALSE, index.return=FALSE)\n# get the class count\nclasses = table(eY, 1)\n+ # TODO vectorize extraction compute extraction vector\nstart_class = 1\nout_s = 1\nout_e = 0\nend_class = 0\nout = matrix(0, sampled, ncol(XY))\nclasses_ratio = floor(classes*ratio)\n- for(i in 1:nrow(classes))\n- {\n+ for(i in 1:nrow(classes)) {\nend_class = end_class + as.scalar(classes[i])\nclass_t = XY[start_class:end_class, ]\nout_e = out_e + as.scalar(classes_ratio[i])\n@@ -89,28 +91,15 @@ doSample = function(Matrix[Double] eX, Matrix[Double] eY, Double ratio, Boolean\nsampledY = out[, 1]\nsampledX = out[, 2:ncol(out)]\n}\n- else if(nrow(eY) > 1 & (dist > 10)) # regression\n- {\n+ else if(nrow(eY) > 1 & (dist > 10)) { # regression\nsampledX = eX[1:sampled, ]\nsampledY = eY[1:sampled, ]\n}\n- else if(nrow(eY) == 1)\n- {\n+ else if(nrow(eY) == 1) { # TODO ?\nsampledX = eX[1:sampled, ]\nsampledY = eY\n}\n- else {\n- sampledX = eX\n- sampledY = eY\n- }\n- }\n- else\n- {\n- sampledX = eX\n- sampledY = eY\n}\n- if(verbose)\n- print(\"AFTER SAMPLING: \"+nrow(eX))\n}\n# #######################################################################\n@@ -154,29 +143,26 @@ return(Boolean validForResources)\nvalidForResources = count > 0\n}\n-stringProcessing = function(Frame[Unknown] data, Matrix[Double] mask, Frame[String] schema, Boolean CorrectTypos)\n+stringProcessing = function(Frame[Unknown] data, Matrix[Double] mask,\n+ Frame[String] schema, Boolean CorrectTypos, List[Unknown] ctx = list(prefix=\"--\"))\nreturn(Frame[Unknown] processedData)\n{\n+ prefix = as.scalar(ctx[\"prefix\"]);\n# step 1 drop invalid types\n+ print(prefix+\" drop values with type mismatch\");\ndata = dropInvalidType(data, schema)\n# step 2 do the case transformations\n+ print(prefix+\" convert strings to lower case\");\nfor(i in 1:ncol(mask))\n- {\nif(as.scalar(schema[1,i]) == \"STRING\")\n- {\n- lowerCase = map(data[, i], \"x -> x.toLowerCase()\")\n- data[, i] = lowerCase\n- }\n-\n- }\n+ data[, i] = map(data[, i], \"x -> x.toLowerCase()\")\nif(CorrectTypos)\n{\n# recode data to get null mask\n- if(sum(mask) > 0)\n- {\n+ if(sum(mask) > 0) {\n# always recode the label\nindex = vectorToCsv(mask)\njspecR = \"{ids:true, recode:[\"+index+\"]}\"\n@@ -186,18 +172,19 @@ return(Frame[Unknown] processedData)\nelse\neX = as.matrix(data)\nnullMask = is.na(eX)\n- print(\"starting correctTypos \")\n+ print(prefix+\" correct typos in strings\");\n# fix the typos\nfor(i in 1:ncol(schema))\n- {\nif(as.scalar(schema[1,i]) == \"STRING\")\ndata[, i] = correctTypos(data[, i], nullMask[, i], 0.2, 0.9, FALSE, TRUE, FALSE);\n}\n- # print(\"after correctTypos \"+toString(data, rows=5))\n- }\n+ print(prefix+\" porter-stemming on all features\");\ndata = map(data, \"x -> PorterStemmer.stem(x)\")\n+\n# TODO add deduplication\n+ print(prefix+\" deduplication via entity resolution\");\n+\nprocessedData = data\n}\n@@ -238,7 +225,7 @@ topk_gridSearch = function(Matrix[Double] X, Matrix[Double] y, Matrix[Double] Xt\n# Step 2) materialize hyper-parameter combinations\n# (simplify debugging and compared to compute negligible)\nHP = matrix(0, numConfigs, numParams);\n- for( i in 1:nrow(HP) ) {\n+ parfor( i in 1:nrow(HP) ) {\nfor( j in 1:numParams )\nHP[i,j] = paramVals[j,as.scalar(((i-1)/cumLens[j,1])%%paramLens[j,1]+1)];\n}\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/pipelines/BuiltinTopkEvaluateTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/pipelines/BuiltinTopkEvaluateTest.java", "diff": "@@ -49,8 +49,6 @@ public class BuiltinTopkEvaluateTest extends AutomatedTestBase {\n}\nprivate void evalPip(double split, String cv, String path, Types.ExecMode et) {\n-\n- setOutputBuffering(true);\nString HOME = SCRIPT_DIR+\"functions/pipelines/\";\nTypes.ExecMode modeOld = setExecMode(et);\ntry {\n" } ]

[ { "change_type": "MODIFY", "old_path": "scripts/builtin/correctTypos.dml", "new_path": "scripts/builtin/correctTypos.dml", "diff": "@@ -179,7 +179,7 @@ buildDictionary = function(Frame[String] S)\n[ID,M] = transformencode(target=S, spec=\"{ids:true,recode:[1]}\");\ndstr = map(M[,1], \"s -> UtilFunctions.splitRecodeEntry(s)[0]\");\ndcodes = map(M[,1], \"s -> UtilFunctions.splitRecodeEntry(s)[1]\");\n- frequencies = table(seq(1,nrow(dstr)),as.matrix(dcodes)) %% table(ID, 1);\n+ frequencies = table(seq(1,nrow(dstr)),as.matrix(dcodes)) %*% table(ID, 1);\ndict = cbind(dstr, as.frame(frequencies));\n}\n" } ]

[ { "change_type": "MODIFY", "old_path": "scripts/builtin/winsorize.dml", "new_path": "scripts/builtin/winsorize.dml", "diff": "#\n#-------------------------------------------------------------\n-m_winsorize = function(Matrix[Double] X, Boolean verbose) return (Matrix[Double] Y) {\n+m_winsorize = function(Matrix[Double] X, Double ql = 0.05, Double qu = 0.95, Boolean verbose)\n+return (Matrix[Double] Y) {\n+\nY = matrix(0, nrow(X), ncol(X))\n- parfor(i in 1:ncol(X))\n- Y[,i] = fixOutliersWinsorize(X[,i])\n+ parfor(i in 1:ncol(X), check=0) {\n+ q1 = quantile(X[,i], ql)\n+ q2 = quantile(X[,i], qu)\n+ Y[, i] = fixOutliersWinsorize(X[,i], q1, q2)\n+ }\n}\n-fixOutliersWinsorize = function(Matrix[Double] X) return (Matrix[Double] Y)\n+fixOutliersWinsorize = function(Matrix[Double] X, Double ql, Double qu) return (Matrix[Double] Y)\n{\n- # compute quantiles for lower and upper probs\n- q = quantile(X, matrix(\"0.05 0.95\", rows=2, cols=1));\n- ql = as.scalar(q[1,1]);\n- qu = as.scalar(q[2,1]);\n# replace values outside [ql,qu] w/ ql and qu respectively\nY = ifelse(X < ql, ql, X);\nY = ifelse(Y > qu, qu, Y);\n" }, { "change_type": "MODIFY", "old_path": "scripts/pipelines/properties/param.csv", "new_path": "scripts/pipelines/properties/param.csv", "diff": "name,param_no,maskFlag,FDFlag,yFlag,verboseFlag,dataFlag,dt1,dt2,dt3,dt4,st1,en1,st2,en2,st3,en3,st4,en4\noutlierByIQR,3,0,0,0,1,0,FP,INT,INT,1,7,2,2,1,1,,,\noutlierBySd,3,0,0,0,1,0,INT,INT,INT,1,7,1,2,2,1,,,\n-winsorize,0,0,0,0,1,0,,,,,,,,,,,,\n+winsorize,2,0,0,0,1,0,FP,FP,0.01,0.05,0.95,1,,,,,,\nnormalize,0,0,0,0,0,0,,,,,,,,,,,,\nimputeByMean,0,1,0,0,0,2,,,,,,,,,,,,\nimputeByMedian,0,1,0,0,0,2,,,,,,,,,,,,\n" }, { "change_type": "MODIFY", "old_path": "scripts/pipelines/scripts/enumerateLogical.dml", "new_path": "scripts/pipelines/scripts/enumerateLogical.dml", "diff": "@@ -85,7 +85,7 @@ return (Frame[Unknown] bestLg, Double pre_best)\n# # # execute the physical pipelines\nscores = matrix(0, nrow(physicalPipList), 1)\n# TODO better parfor-dep handling of multi-assignments to avoid check=0\n- parfor(i in 1:length(physicalPipList), check=0) {\n+ for(i in 1:length(physicalPipList), check=0) {\nlp2 = as.frame(logicalPipList[((i-1)%/%num_inst)+1,])\npp2 = as.frame(physicalPipList[i,])\n# # append configuration keys for extracting the pipeline later on\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/builtin/multipleBuiltins.dml", "new_path": "src/test/scripts/functions/builtin/multipleBuiltins.dml", "diff": "#-------------------------------------------------------------\nX = read($1);\n-Y = winsorize(X, FALSE);\n+Y = winsorize(X=X, verbose=FALSE);\nZ = outlier(Y, FALSE);\nwrite(Z, $2);\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/builtin/winsorize.dml", "new_path": "src/test/scripts/functions/builtin/winsorize.dml", "diff": "#-------------------------------------------------------------\nX = read($1);\n-Y = winsorize(X, FALSE);\n+Y = winsorize(X=X, ql=0.05, qu= 0.95, verbose=FALSE);\nwrite(Y, $2)\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/builtin/winsorizeFoo.dml", "new_path": "src/test/scripts/functions/builtin/winsorizeFoo.dml", "diff": "@@ -25,5 +25,5 @@ foo = function(Matrix[Double] X, Boolean verbose)\nwhile(FALSE){} #no inlining\nif( verbose )\nprint( min(X)+\" \"+max(X) )\n- R = winsorize(X, verbose);\n+ R = winsorize(X=X, verbose=verbose);\n}\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/misc/FunPotpourriMultiEval.dml", "new_path": "src/test/scripts/functions/misc/FunPotpourriMultiEval.dml", "diff": "@@ -23,7 +23,7 @@ X = rand(rows=10, cols= 10)\nt1 = interQuartileMean(X[,7]);\nfor(i in 1:5)\n- X = eval(\"winsorize\", list(X, FALSE))\n+ X = eval(\"winsorize\", list(X=X, ql = 0.05, qu=0.95, verbose=FALSE))\nt2 = interQuartileMean(X[,7]);\nprint(\"expected=TRUE, actual=\"+(t2 < t1))\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/misc/Functions15b.dml", "new_path": "src/test/scripts/functions/misc/Functions15b.dml", "diff": "foo = function(Matrix[Double] X)\nreturn (Matrix[Double] Y)\n{\n- Y = winsorize(X, FALSE)\n+ Y = winsorize(X=X, verbose=FALSE)\n}\n" } ]

[ { "change_type": "MODIFY", "old_path": "scripts/pipelines/properties/testPrimitives.csv", "new_path": "scripts/pipelines/properties/testPrimitives.csv", "diff": "ED,MVI,OTLR,EC,SCALE,CI,DUMMY,DIM\n-outlierBySd,imputeByMean,winsorize,imputeByMean,scale,SMOTE,dummycoding,m_pca\n-outlierByIQR,imputeByMedian,outlierBySd,imputeByMedian,,wtomeklink,,ppca\n+,imputeByMean,winsorize,imputeByMean,scale,abstain,dummycoding,m_pca\n+outlierBySd,imputeByMedian,outlierBySd,imputeByMedian,,wtomeklink,,ppca\n+outlierByIQR,forward_fill,outlierByIQR,fillDefault,,SMOTE,,\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/pipelines/BuiltinTopkLogicalTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/pipelines/BuiltinTopkLogicalTest.java", "diff": "@@ -49,12 +49,12 @@ public class BuiltinTopkLogicalTest extends AutomatedTestBase {\n@Test\npublic void testLogical1() {\n- runTestLogical(10, 5, 2, Types.ExecMode.SINGLE_NODE);\n+ runTestLogical(10, 3, 2, Types.ExecMode.SINGLE_NODE);\n}\n@Test\npublic void testLogical2() {\n- runTestLogical(2, 3, 2,\n+ runTestLogical(2, 2, 2,\nTypes.ExecMode.SINGLE_NODE);\n}\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/bestAcc.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/bestAcc.csv", "diff": "-85.58558558558559\n-84.68468468468468\n-82.88288288288288\n+94.5945945945946\n+94.5945945945946\n+94.5945945945946\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/dirtyScore.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/dirtyScore.csv", "diff": "-67.56756756756756\n\\ No newline at end of file\n+90.09009009009009\n\\ No newline at end of file\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/evalHp.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/evalHp.csv", "diff": "-2.0,0.001,1.0E-5,1000.0\n+2.0,0.001,1.0,1000.0\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/hp.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/hp.csv", "diff": "-36.0,3.0,3.0,2.0,1.0,0,0,0,1.0,0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,1.0,0,2.0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n-36.0,3.0,2.0,2.0,1.0,0,0,0,1.0,0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,1.0,0,2.0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n-36.0,3.0,3.0,1.0,1.0,0,0,0,1.0,0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,1.0,0,2.0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n+40.0,2.0,0.01816863223655686,0.9565161479438591,0,0,0,1.0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,1.0,0,0,0,2.0,1.0,0.6515164788504212,0,0,0,1.0,1.0,2.0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n+40.0,2.0,0.03510876761722913,0.9673791862807241,0,0,0,1.0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,1.0,0,0,0,2.0,1.0,0.6149768032146687,0,0,0,1.0,1.0,2.0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n+40.0,2.0,0.014861839294898092,0.9595626659056867,0,0,0,1.0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,1.0,0,0,0,2.0,1.0,0.6274449265973082,0,0,0,1.0,1.0,2.0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/lp.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/lp.csv", "diff": "-ED,MVI,CI,DUMMY\n+OTLR,EC,EC,CI,DUMMY\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/pip.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/pip.csv", "diff": "-outlierBySd,imputeByMedian,wtomeklink,dummycoding\n-outlierBySd,imputeByMedian,wtomeklink,dummycoding\n-outlierBySd,imputeByMedian,wtomeklink,dummycoding\n+winsorize,imputeByMean,imputeByMedian,abstain,dummycoding\n+winsorize,imputeByMean,imputeByMedian,abstain,dummycoding\n+winsorize,imputeByMean,imputeByMedian,abstain,dummycoding\n" } ]

[ { "change_type": "MODIFY", "old_path": "scripts/nn/examples/efficientNet.dml", "new_path": "scripts/nn/examples/efficientNet.dml", "diff": "@@ -90,7 +90,7 @@ initNetwork = function(int InputChannels, int NumberOutputClasses, int seed)\n[Gamma_top, Beta_top, EmaMean_top, EmaVar_top] = batchnorm::init(1280)\n[DW_top, Db_top] = affine::init(1280, NumberOutputClasses, seed)\n- model = list(CW_stem, Cb_stem, Gamma_stem, Beta_stem, EmaMean_stem,vEmaVar_stem,\n+ model = list(CW_stem, Cb_stem, Gamma_stem, Beta_stem, EmaMean_stem, EmaVar_stem,\nas.matrix(mb_parameters[1]),\nas.matrix(mb_parameters[2]),\nas.matrix(mb_parameters[3]),\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/instructions/gpu/context/GPUObject.java", "new_path": "src/main/java/org/apache/sysds/runtime/instructions/gpu/context/GPUObject.java", "diff": "@@ -845,7 +845,8 @@ public class GPUObject {\ncopyToDevice = false;\n} else if (block == null && tmp.getNonZeros() != 0) {\nthrow new DMLRuntimeException(\"Expected CP sparse block to be not null.\");\n- } else {\n+ }\n+ else {\n// CSR is the preferred format for cuSparse GEMM\n// Converts MCSR and COO to CSR\nSparseBlockCSR csrBlock = null;\n@@ -880,10 +881,13 @@ public class GPUObject {\nvalues = csrBlock.values();\n}\n+ if (values != null)\nif(values.length > tmp.getNonZeros())\nallocateSparseMatrixOnDevice(values.length);\nelse\nallocateSparseMatrixOnDevice();\n+ else\n+ allocateSparseMatrixOnDevice();\nif (copyToDevice) {\nCSRPointer.copyToDevice(getGPUContext(), getJcudaSparseMatrixPtr(),\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/federated/paramserv/AvgModelFederatedParamservTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/federated/paramserv/AvgModelFederatedParamservTest.java", "diff": "package org.apache.sysds.test.functions.federated.paramserv;\n+import java.util.ArrayList;\n+import java.util.Arrays;\n+import java.util.Collection;\n+import java.util.List;\n+\nimport org.apache.commons.logging.Log;\nimport org.apache.commons.logging.LogFactory;\nimport org.apache.sysds.common.Types.ExecMode;\n@@ -27,15 +32,11 @@ import org.apache.sysds.test.TestConfiguration;\nimport org.apache.sysds.test.TestUtils;\nimport org.apache.sysds.utils.Statistics;\nimport org.junit.Assert;\n+import org.junit.Ignore;\nimport org.junit.Test;\nimport org.junit.runner.RunWith;\nimport org.junit.runners.Parameterized;\n-import java.util.ArrayList;\n-import java.util.Arrays;\n-import java.util.Collection;\n-import java.util.List;\n-\n@RunWith(value = Parameterized.class)\[email protected]\npublic class AvgModelFederatedParamservTest extends AutomatedTestBase {\n@@ -117,11 +118,15 @@ public class AvgModelFederatedParamservTest extends AutomatedTestBase {\n}\n@Test\n+ @Ignore\n+ // TODO FIX ME\npublic void AvgmodelfederatedParamservSingleNode() {\nAvgmodelfederatedParamserv(ExecMode.SINGLE_NODE, true);\n}\n@Test\n+ @Ignore\n+ // TODO FIX ME\npublic void AvgmodelfederatedParamservHybrid() {\nAvgmodelfederatedParamserv(ExecMode.HYBRID, true);\n}\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/controlprogram/paramserv/FederatedPSControlThread.java", "new_path": "src/main/java/org/apache/sysds/runtime/controlprogram/paramserv/FederatedPSControlThread.java", "diff": "@@ -503,7 +503,7 @@ public class FederatedPSControlThread extends PSWorker implements Callable<Void>\n// recreate aggregation instruction and output if needed\nInstruction aggregationInstruction = null;\nDataIdentifier aggregationOutput = null;\n- if(_localUpdate && _numBatchesToCompute > 1) {\n+ if(_localUpdate && _numBatchesToCompute > 1 | modelAvg) {\nfunc = ec.getProgram().getFunctionProgramBlock(namespace, aggFunc, opt);\ninputs = func.getInputParams();\noutputs = func.getOutputParams();\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/federated/paramserv/AvgModelFederatedParamservTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/federated/paramserv/AvgModelFederatedParamservTest.java", "diff": "@@ -32,7 +32,6 @@ import org.apache.sysds.test.TestConfiguration;\nimport org.apache.sysds.test.TestUtils;\nimport org.apache.sysds.utils.Statistics;\nimport org.junit.Assert;\n-import org.junit.Ignore;\nimport org.junit.Test;\nimport org.junit.runner.RunWith;\nimport org.junit.runners.Parameterized;\n@@ -118,15 +117,11 @@ public class AvgModelFederatedParamservTest extends AutomatedTestBase {\n}\n@Test\n- @Ignore\n- // TODO FIX ME\npublic void AvgmodelfederatedParamservSingleNode() {\nAvgmodelfederatedParamserv(ExecMode.SINGLE_NODE, true);\n}\n@Test\n- @Ignore\n- // TODO FIX ME\npublic void AvgmodelfederatedParamservHybrid() {\nAvgmodelfederatedParamserv(ExecMode.HYBRID, true);\n}\n@@ -149,7 +144,8 @@ public class AvgModelFederatedParamservTest extends AutomatedTestBase {\nList<Thread> threads = new ArrayList<>();\nfor(int i = 0; i < _numFederatedWorkers; i++) {\nports.add(getRandomAvailablePort());\n- threads.add(startLocalFedWorkerThread(ports.get(i), FED_WORKER_WAIT_S));\n+ threads.add(startLocalFedWorkerThread(ports.get(i),\n+ i==(_numFederatedWorkers-1) ? FED_WORKER_WAIT : FED_WORKER_WAIT_S));\n}\n// generate test data\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/federated/primitives/FederatedRCBindTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/federated/primitives/FederatedRCBindTest.java", "diff": "@@ -80,7 +80,6 @@ public class FederatedRCBindTest extends AutomatedTestBase {\n}\n@Test\n- @Ignore\npublic void federatedRCBindSP() {\nfederatedRCBind(Types.ExecMode.SPARK);\n}\n@@ -109,8 +108,12 @@ public class FederatedRCBindTest extends AutomatedTestBase {\nint port1 = getRandomAvailablePort();\nint port2 = getRandomAvailablePort();\n+ int port3 = getRandomAvailablePort();\n+ int port4 = getRandomAvailablePort();\nThread t1 = startLocalFedWorkerThread(port1, FED_WORKER_WAIT_S);\nThread t2 = startLocalFedWorkerThread(port2);\n+ Thread t3 = startLocalFedWorkerThread(port3);\n+ Thread t4 = startLocalFedWorkerThread(port4);\n// we need the reference file to not be written to hdfs, so we get the correct format\nrtplatform = Types.ExecMode.SINGLE_NODE;\n@@ -134,9 +137,9 @@ public class FederatedRCBindTest extends AutomatedTestBase {\nfullDMLScriptName = HOME + TEST_NAME + \".dml\";\nprogramArgs = new String[] {\"-nvargs\",\n\"in_A1=\" + TestUtils.federatedAddress(port1, input(\"A1\")),\n- \"in_A2=\" + TestUtils.federatedAddress(port1, input(\"A2\")),\n- \"in_B1=\" + TestUtils.federatedAddress(port2, input(\"B1\")),\n- \"in_B2=\" + TestUtils.federatedAddress(port2, input(\"B2\")),\n+ \"in_A2=\" + TestUtils.federatedAddress(port2, input(\"A2\")),\n+ \"in_B1=\" + TestUtils.federatedAddress(port3, input(\"B1\")),\n+ \"in_B2=\" + TestUtils.federatedAddress(port4, input(\"B2\")),\n\"in_partitioned=\" + Boolean.toString(partitioned).toUpperCase(),\n\"in_B1_local=\" + input(\"B1\"), \"in_B2_local=\" + input(\"B2\"), \"rows=\" + rows, \"cols=\" + cols,\n\"out_R_FF=\" + output(\"R_FF\"), \"out_R_FL=\" + output(\"R_FL\"), \"out_R_LF=\" + output(\"R_LF\"),\n@@ -147,7 +150,7 @@ public class FederatedRCBindTest extends AutomatedTestBase {\n// compare all sums via files\ncompareResults(1e-11);\n- TestUtils.shutdownThreads(t1, t2);\n+ TestUtils.shutdownThreads(t1, t2, t3, t4);\nrtplatform = platformOld;\nDMLScript.USE_LOCAL_SPARK_CONFIG = sparkConfigOld;\n}\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/federated/paramserv/AvgModelFederatedParamservTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/federated/paramserv/AvgModelFederatedParamservTest.java", "diff": "@@ -24,8 +24,6 @@ import java.util.Arrays;\nimport java.util.Collection;\nimport java.util.List;\n-import org.apache.commons.logging.Log;\n-import org.apache.commons.logging.LogFactory;\nimport org.apache.sysds.common.Types.ExecMode;\nimport org.apache.sysds.test.AutomatedTestBase;\nimport org.apache.sysds.test.TestConfiguration;\n@@ -39,7 +37,7 @@ import org.junit.runners.Parameterized;\n@RunWith(value = Parameterized.class)\[email protected]\npublic class AvgModelFederatedParamservTest extends AutomatedTestBase {\n- private static final Log LOG = LogFactory.getLog(AvgModelFederatedParamservTest.class.getName());\n+ // private static final Log LOG = LogFactory.getLog(AvgModelFederatedParamservTest.class.getName());\nprivate final static String TEST_DIR = \"functions/federated/paramserv/\";\nprivate final static String TEST_NAME = \"AvgModelFederatedParamservTest\";\nprivate final static String TEST_CLASS_DIR = TEST_DIR + AvgModelFederatedParamservTest.class.getSimpleName() + \"/\";\n@@ -131,7 +129,7 @@ public class AvgModelFederatedParamservTest extends AutomatedTestBase {\n// config\ngetAndLoadTestConfiguration(TEST_NAME);\nString HOME = SCRIPT_DIR + TEST_DIR;\n- setOutputBuffering(false);\n+ setOutputBuffering(true);\nint C = 1, Hin = 28, Win = 28;\nint numLabels = 10;\n@@ -201,8 +199,8 @@ public class AvgModelFederatedParamservTest extends AutomatedTestBase {\n\"modelAvg=\" + Boolean.toString(modelAvg).toUpperCase()));\nprogramArgs = programArgsList.toArray(new String[0]);\n- LOG.debug(runTest(null));\n- Assert.assertEquals(0, Statistics.getNoOfExecutedSPInst());\n+ String log = runTest(null).toString();\n+ Assert.assertEquals(\"Test Failed \\n\" + log, 0, Statistics.getNoOfExecutedSPInst());\n// shut down threads\nfor(int i = 0; i < _numFederatedWorkers; i++) {\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/federated/paramserv/FederatedParamservTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/federated/paramserv/FederatedParamservTest.java", "diff": "@@ -24,8 +24,6 @@ import java.util.Arrays;\nimport java.util.Collection;\nimport java.util.List;\n-import org.apache.commons.logging.Log;\n-import org.apache.commons.logging.LogFactory;\nimport org.apache.sysds.common.Types.ExecMode;\nimport org.apache.sysds.test.AutomatedTestBase;\nimport org.apache.sysds.test.TestConfiguration;\n@@ -39,7 +37,7 @@ import org.junit.runners.Parameterized;\n@RunWith(value = Parameterized.class)\[email protected]\npublic class FederatedParamservTest extends AutomatedTestBase {\n- private static final Log LOG = LogFactory.getLog(FederatedParamservTest.class.getName());\n+ // private static final Log LOG = LogFactory.getLog(FederatedParamservTest.class.getName());\nprivate final static String TEST_DIR = \"functions/federated/paramserv/\";\nprivate final static String TEST_NAME = \"FederatedParamservTest\";\nprivate final static String TEST_CLASS_DIR = TEST_DIR + FederatedParamservTest.class.getSimpleName() + \"/\";\n@@ -199,8 +197,8 @@ public class FederatedParamservTest extends AutomatedTestBase {\n\"seed=\" + _seed));\nprogramArgs = programArgsList.toArray(new String[0]);\n- LOG.debug(runTest(null));\n- Assert.assertEquals(0, Statistics.getNoOfExecutedSPInst());\n+ String log = runTest(null).toString();\n+ Assert.assertEquals(\"Test Failed \\n\" + log, 0, Statistics.getNoOfExecutedSPInst());\n// shut down threads\nfor(int i = 0; i < _numFederatedWorkers; i++) {\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/federated/paramserv/NbatchesFederatedParamservTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/federated/paramserv/NbatchesFederatedParamservTest.java", "diff": "@@ -24,8 +24,6 @@ import java.util.Arrays;\nimport java.util.Collection;\nimport java.util.List;\n-import org.apache.commons.logging.Log;\n-import org.apache.commons.logging.LogFactory;\nimport org.apache.sysds.common.Types.ExecMode;\nimport org.apache.sysds.test.AutomatedTestBase;\nimport org.apache.sysds.test.TestConfiguration;\n@@ -39,7 +37,7 @@ import org.junit.runners.Parameterized;\n@RunWith(value = Parameterized.class)\[email protected]\npublic class NbatchesFederatedParamservTest extends AutomatedTestBase {\n- private static final Log LOG = LogFactory.getLog(NbatchesFederatedParamservTest.class.getName());\n+ // private static final Log LOG = LogFactory.getLog(NbatchesFederatedParamservTest.class.getName());\nprivate final static String TEST_DIR = \"functions/federated/paramserv/\";\nprivate final static String TEST_NAME = \"NbatchesFederatedParamservTest\";\nprivate final static String TEST_CLASS_DIR = TEST_DIR + NbatchesFederatedParamservTest.class.getSimpleName() + \"/\";\n@@ -111,7 +109,7 @@ public class NbatchesFederatedParamservTest extends AutomatedTestBase {\n// config\ngetAndLoadTestConfiguration(TEST_NAME);\nString HOME = SCRIPT_DIR + TEST_DIR;\n- setOutputBuffering(false);\n+ setOutputBuffering(true);\nint C = 1, Hin = 28, Win = 28;\nint numLabels = 10;\n@@ -181,8 +179,8 @@ public class NbatchesFederatedParamservTest extends AutomatedTestBase {\n\"nbatches=\" + _nbatches));\nprogramArgs = programArgsList.toArray(new String[0]);\n- LOG.debug(runTest(null));\n- Assert.assertEquals(0, Statistics.getNoOfExecutedSPInst());\n+ String log = runTest(null).toString();\n+ Assert.assertEquals(\"Test Failed \\n\" + log,0, Statistics.getNoOfExecutedSPInst());\n// shut down threads\nfor(int i = 0; i < _numFederatedWorkers; i++) {\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/federated/primitives/FederatedRCBindTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/federated/primitives/FederatedRCBindTest.java", "diff": "@@ -28,7 +28,6 @@ import org.apache.sysds.runtime.meta.MatrixCharacteristics;\nimport org.apache.sysds.test.AutomatedTestBase;\nimport org.apache.sysds.test.TestConfiguration;\nimport org.apache.sysds.test.TestUtils;\n-import org.junit.Ignore;\nimport org.junit.Test;\nimport org.junit.runner.RunWith;\nimport org.junit.runners.Parameterized;\n" } ]

[ { "change_type": "DELETE", "old_path": "src/main/python/systemds/operator/algorithm.py", "new_path": null, "diff": "-# -------------------------------------------------------------\n-#\n-# Licensed to the Apache Software Foundation (ASF) under one\n-# or more contributor license agreements. See the NOTICE file\n-# distributed with this work for additional information\n-# regarding copyright ownership. The ASF licenses this file\n-# to you under the Apache License, Version 2.0 (the\n-# \"License\"); you may not use this file except in compliance\n-# with the License. You may obtain a copy of the License at\n-#\n-# http://www.apache.org/licenses/LICENSE-2.0\n-#\n-# Unless required by applicable law or agreed to in writing,\n-# software distributed under the License is distributed on an\n-# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n-# KIND, either express or implied. See the License for the\n-# specific language governing permissions and limitations\n-# under the License.\n-#\n-# -------------------------------------------------------------\n-\n-from typing import Dict\n-\n-from systemds.operator import OperationNode\n-from systemds.script_building.dag import OutputType\n-from systemds.utils.consts import VALID_INPUT_TYPES\n-\n-__all__ = ['l2svm', 'lm', 'kmeans', 'pca', 'multiLogReg', 'multiLogRegPredict']\n-\n-\n-def l2svm(x: OperationNode, y: OperationNode, **kwargs: Dict[str, VALID_INPUT_TYPES]) -> OperationNode:\n- \"\"\"\n- Perform L2SVM on matrix with labels given.\n-\n- :param x: Input dataset\n- :param y: Input labels in shape of one column\n- :param kwargs: Dictionary of extra arguments\n- :return: `OperationNode` containing the model fit.\n- \"\"\"\n- x._check_matrix_op()\n- params_dict = {'X': x, 'Y': y}\n- params_dict.update(kwargs)\n- return OperationNode(x.sds_context, 'l2svm', named_input_nodes=params_dict)\n-\n-\n-def lm(x: OperationNode, y: OperationNode, **kwargs: Dict[str, VALID_INPUT_TYPES]) -> OperationNode:\n- \"\"\"\n- Performs LM on matrix with labels given.\n-\n- :param x: Input dataset\n- :param y: Input labels in shape of one column\n- :param kwargs: Dictionary of extra arguments\n- :return: `OperationNode` containing the model fit.\n- \"\"\"\n-\n- if x.shape[0] == 0:\n- raise ValueError(\"Found array with 0 feature(s) (shape={s}) while a minimum of 1 is required.\"\n- .format(s=x.shape))\n- if y.shape[0] == 0:\n- raise ValueError(\"Found array with 0 feature(s) (shape={s}) while a minimum of 1 is required.\"\n- .format(s=y.shape))\n-\n- params_dict = {'X': x, 'y': y}\n- params_dict.update(kwargs)\n- return OperationNode(x.sds_context, 'lm', named_input_nodes=params_dict)\n-\n-\n-def kmeans(x: OperationNode, **kwargs: Dict[str, VALID_INPUT_TYPES]) -> OperationNode:\n- \"\"\"\n- Performs KMeans on matrix input.\n-\n- :param x: Input dataset to perform K-Means on.\n- :param k: The number of centroids to use for the algorithm.\n- :param runs: The number of concurrent instances of K-Means to run (with different initial centroids).\n- :param max_iter: The maximum number of iterations to run the K-Means algorithm for.\n- :param eps: Tolerance for the algorithm to declare convergence using WCSS change ratio.\n- :param is_verbose: Boolean flag if the algorithm should be run in a verbose manner.\n- :param avg_sample_size_per_centroid: The average number of records per centroid in the data samples.\n- :return: `OperationNode` List containing two outputs 1. the clusters, 2 the cluster ID associated with each row in x.\n- \"\"\"\n-\n- x._check_matrix_op()\n- if x.shape[0] == 0:\n- raise ValueError(\"Found array with 0 feature(s) (shape={s}) while a minimum of 1 is required.\"\n- .format(s=x.shape))\n-\n- if 'k' in kwargs.keys() and kwargs.get('k') < 1:\n- raise ValueError(\n- \"Invalid number of clusters in K-Means, number must be integer above 0\")\n-\n- params_dict = {'X': x}\n- params_dict.update(kwargs)\n- return OperationNode(x.sds_context, 'kmeans', named_input_nodes=params_dict, output_type=OutputType.LIST, number_of_outputs=2)\n-\n-def kmeansPredict(X: OperationNode, C: OperationNode) -> OperationNode:\n- \"\"\"\n- Perform Kmeans Predict, note that the Ids returned are 1 indexed.\n-\n- :param X: The matrix to classify.\n- :param Y: The Clusters to use for classification into.\n- :return: `OperationNode` containing a matrix of classifications of Id's of specific clusters in C.\n- \"\"\"\n- X._check_matrix_op()\n- C._check_matrix_op()\n-\n- params_dict = {'X' : X, 'C' : C}\n- return OperationNode(X.sds_context, 'kmeansPredict', named_input_nodes=params_dict, output_type=OutputType.MATRIX, shape=(1, X.shape[0]))\n-\n-\n-\n-def pca(x: OperationNode, **kwargs: Dict[str, VALID_INPUT_TYPES]) -> OperationNode:\n- \"\"\"\n- Performs PCA on the matrix input\n-\n- :param x: Input dataset to perform Principal Componenet Analysis (PCA) on.\n- :param K: The number of reduced dimensions.\n- :param center: Boolean specifying if the input values should be centered.\n- :param scale: Boolean specifying if the input values should be scaled.\n- :return: `OperationNode` List containing two outputs 1. The dimensionality reduced X input, 2. A matrix to reduce dimensionality similarly on unseen data.\n- \"\"\"\n-\n- x._check_matrix_op()\n- if x.shape[0] == 0:\n- raise ValueError(\"Found array with 0 feature(s) (shape={s}) while a minimum of 1 is required.\"\n- .format(s=x.shape))\n-\n- if 'K' in kwargs.keys() and kwargs.get('K') < 1:\n- raise ValueError(\n- \"Invalid number of dimensions in PCA, number must be integer above 0\")\n-\n- params_dict = {'X': x}\n- params_dict.update(kwargs)\n- return OperationNode(x.sds_context, 'pca', named_input_nodes=params_dict, output_type=OutputType.LIST, number_of_outputs=2)\n-\n-\n-def multiLogReg(x: OperationNode, y: OperationNode, **kwargs: Dict[str, VALID_INPUT_TYPES]) -> OperationNode:\n- \"\"\"\n- Performs Multiclass Logistic Regression on the matrix input\n- using Trust Region method.\n-\n- See: Trust Region Newton Method for Logistic Regression, Lin, Weng and Keerthi, JMLR 9 (2008) 627-650)\n-\n- :param x: Input dataset to perform logstic regression on\n- :param y: Labels rowaligned with the input dataset\n- :param icpt: Intercept, default 2, Intercept presence, shifting and rescaling X columns:\n- 0 = no intercept, no shifting, no rescaling;\n- 1 = add intercept, but neither shift nor rescale X;\n- 2 = add intercept, shift & rescale X columns to mean = 0, variance = 1\n- :param tol: float tolerance for the algorithm.\n- :param reg: Regularization parameter (lambda = 1/C); intercept settings are not regularized.\n- :param maxi: Maximum outer iterations of the algorithm\n- :param maxii: Maximum inner iterations of the algorithm\n- :return: `OperationNode` of a matrix containing the regression parameters trained.\n- \"\"\"\n-\n- if x.shape[0] == 0:\n- raise ValueError(\"Found array with 0 feature(s) (shape={s}) while a minimum of 1 is required.\"\n- .format(s=x.shape))\n- if y.shape[0] == 0:\n- raise ValueError(\"Found array with 0 feature(s) (shape={s}) while a minimum of 1 is required.\"\n- .format(s=y.shape))\n- if -1 in x.shape:\n- output_shape = (-1,)\n- else:\n- output_shape = (x.shape[1],)\n-\n- params_dict = {'X': x, 'Y': y}\n- params_dict.update(kwargs)\n- return OperationNode(x.sds_context, 'multiLogReg', named_input_nodes=params_dict, shape = output_shape)\n-\n-\n-def multiLogRegPredict(x: OperationNode, b: OperationNode, y: OperationNode, **kwargs: Dict[str, VALID_INPUT_TYPES]) -> OperationNode:\n- \"\"\"\n- Performs prediction on input data x using the model trained, b.\n-\n- :param x: The data to perform classification on.\n- :param b: The regression parameters trained from multiLogReg.\n- :param y: The Labels expected to be contained in the X dataset, to calculate accuracy.\n- :param verbose: Boolean specifying if the prediction should be verbose.\n- :return: `OperationNode` List containing three outputs.\n- 1. The predicted means / probabilities\n- 2. The predicted response vector\n- 3. The scalar value of accuracy\n- \"\"\"\n-\n- if x.shape[0] == 0:\n- raise ValueError(\"Found array with 0 feature(s) (shape={s}) while a minimum of 1 is required.\"\n- .format(s=x.shape))\n- if b.shape[0] == 0:\n- raise ValueError(\"Found array with 0 feature(s) (shape={s}) while a minimum of 1 is required.\"\n- .format(s=y.shape))\n- if y.shape[0] == 0:\n- raise ValueError(\"Found array with 0 feature(s) (shape={s}) while a minimum of 1 is required.\"\n- .format(s=y.shape))\n-\n- params_dict = {'X': x, 'B': b, 'Y': y}\n- params_dict.update(kwargs)\n- return OperationNode(x.sds_context, 'multiLogRegPredict', named_input_nodes=params_dict, output_type=OutputType.LIST, number_of_outputs=3, output_types=[OutputType.MATRIX,OutputType.MATRIX,OutputType.DOUBLE])\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/nodes/frame.py", "new_path": "src/main/python/systemds/operator/nodes/frame.py", "diff": "@@ -132,3 +132,6 @@ class Frame(OperationNode):\n:return: The Frame containing the replaced values\n\"\"\"\nreturn Frame(self.sds_context, \"replace\", named_input_nodes={\"target\": self, \"pattern\": f\"'{pattern}'\", \"replacement\":f\"'{replacement}'\"})\n+\n+ def __str__(self):\n+ return \"FrameNode\"\n\\ No newline at end of file\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/nodes/list.py", "new_path": "src/main/python/systemds/operator/nodes/list.py", "diff": "@@ -84,3 +84,5 @@ class List(OperationNode):\ndef compute(self, verbose: bool = False, lineage: bool = False) -> Union[np.array]:\nreturn super().compute(verbose, lineage)\n+ def __str__(self):\n+ return \"ListNode\"\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/nodes/list_access.py", "new_path": "src/main/python/systemds/operator/nodes/list_access.py", "diff": "@@ -61,3 +61,6 @@ class ListAccess(OperationNode):\nres = Scalar(self.sds_context, \"as.scalar\", [ent])\nself._list_source._outputs[self._key] = res\nreturn res\n+\n+ def __str__(self):\n+ return \"ListAccessNode\"\n\\ No newline at end of file\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/nodes/matrix.py", "new_path": "src/main/python/systemds/operator/nodes/matrix.py", "diff": "@@ -335,7 +335,7 @@ class Matrix(OperationNode):\nreturn Matrix(self.sds_context, 'order', [], named_input_nodes=named_input_nodes)\n- def to_string(self, **kwargs: Dict[str, VALID_INPUT_TYPES]) -> 'Matrix':\n+ def to_string(self, **kwargs: Dict[str, VALID_INPUT_TYPES]) -> 'Scalar':\n\"\"\" Converts the input to a string representation.\n:return: `Scalar` containing the string.\n\"\"\"\n@@ -360,3 +360,6 @@ class Matrix(OperationNode):\nReplace all values with replacement value\n\"\"\"\nreturn Matrix(self.sds_context, \"replace\", named_input_nodes={\"target\": self, \"pattern\": pattern, \"replacement\":replacement})\n+\n+ def __str__(self):\n+ return \"MatrixNode\"\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/nodes/multi_return.py", "new_path": "src/main/python/systemds/operator/nodes/multi_return.py", "diff": "@@ -83,3 +83,6 @@ class MultiReturn(OperationNode):\ndef __iter__(self):\nreturn iter(self._outputs)\n+\n+ def __str__(self):\n+ return \"MultiReturnNode\"\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/nodes/scalar.py", "new_path": "src/main/python/systemds/operator/nodes/scalar.py", "diff": "@@ -218,8 +218,12 @@ class Scalar(OperationNode):\n\"\"\"\nreturn Scalar(self.sds_context, 'tanh', [self])\n- def to_string(self, **kwargs: Dict[str, VALID_INPUT_TYPES]) -> 'OperationNode':\n+ def to_string(self, **kwargs: Dict[str, VALID_INPUT_TYPES]) -> 'Scalar':\n\"\"\" Converts the input to a string representation.\n:return: `Scalar` containing the string.\n\"\"\"\nreturn Scalar(self.sds_context, 'toString', [self], named_input_nodes=kwargs, output_type=OutputType.STRING)\n+\n+ def __str__(self):\n+ return \"ScalarNode\"\n+\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/nodes/source.py", "new_path": "src/main/python/systemds/operator/nodes/source.py", "diff": "@@ -197,3 +197,6 @@ class Source(OperationNode):\ndef compute(self, verbose: bool = False, lineage: bool = False):\nraise Exception(\"Invalid invocation of source from script\")\n+\n+ def __str__(self):\n+ return \"SourceNode\"\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/tests/basics/__init__.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/tests/basics/test___str__.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+\n+import unittest\n+\n+from systemds.context import SystemDSContext\n+\n+\n+class Test__str__(unittest.TestCase):\n+\n+ sds: SystemDSContext = None\n+\n+ @classmethod\n+ def setUpClass(cls):\n+ cls.sds = SystemDSContext()\n+\n+ @classmethod\n+ def tearDownClass(cls):\n+ cls.sds.close()\n+\n+ def test_1(self):\n+ self.assertTrue(\"MatrixNode\" in str(self.sds.full([1, 2], 3)))\n+\n+ def test_2(self):\n+ self.assertTrue(\"ScalarNode\" in str(self.sds.scalar(3)))\n+\n+ def test_3(self):\n+ self.assertTrue(\"ScalarNode\" in str(self.sds.scalar(\"Hi\")))\n+\n+ def test_4(self):\n+ self.assertTrue(\"ScalarNode\" in str(\n+ self.sds.full([1, 2], 3).to_string()))\n+\n+ def test_5(self):\n+ self.assertTrue(\"ListNode\" in str(self.sds.list(\n+ self.sds.rand(1, 2, 3, 4), self.sds.scalar(4))))\n+\n+ def test_6(self):\n+ self.assertTrue(\"MatrixNode\" in str(self.sds.list(\n+ self.sds.rand(1, 2, 3, 4), self.sds.scalar(4))[0]))\n+\n+ def test_7(self):\n+ self.assertTrue(\"ScalarNode\" in str(self.sds.list(\n+ self.sds.rand(1, 2, 3, 4), self.sds.scalar(4))[1]))\n+\n+\n+if __name__ == \"__main__\":\n+ unittest.main(exit=False)\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/algorithm/__init__.py", "new_path": "src/main/python/systemds/operator/algorithm/__init__.py", "diff": "@@ -39,9 +39,12 @@ from .builtin.csplineDS import csplineDS\nfrom .builtin.cvlm import cvlm\nfrom .builtin.dbscan import dbscan\nfrom .builtin.decisionTree import decisionTree\n+from .builtin.deepWalk import deepWalk\nfrom .builtin.discoverFD import discoverFD\nfrom .builtin.dist import dist\nfrom .builtin.executePipeline import executePipeline\n+from .builtin.ffTrain import ffTrain\n+from .builtin.garch import garch\nfrom .builtin.gaussianClassifier import gaussianClassifier\nfrom .builtin.getAccuracy import getAccuracy\nfrom .builtin.glm import glm\n@@ -73,11 +76,13 @@ from .builtin.knnbf import knnbf\nfrom .builtin.l2svm import l2svm\nfrom .builtin.l2svmPredict import l2svmPredict\nfrom .builtin.lasso import lasso\n+from .builtin.lenetTrain import lenetTrain\nfrom .builtin.lm import lm\nfrom .builtin.lmCG import lmCG\nfrom .builtin.lmDS import lmDS\nfrom .builtin.lmPredict import lmPredict\nfrom .builtin.logSumExp import logSumExp\n+from .builtin.matrixProfile import matrixProfile\nfrom .builtin.msvm import msvm\nfrom .builtin.msvmPredict import msvmPredict\nfrom .builtin.multiLogReg import multiLogReg\n@@ -96,6 +101,7 @@ from .builtin.ppca import ppca\nfrom .builtin.randomForest import randomForest\nfrom .builtin.scale import scale\nfrom .builtin.scaleApply import scaleApply\n+from .builtin.selectByVarThresh import selectByVarThresh\nfrom .builtin.sherlock import sherlock\nfrom .builtin.sherlockPredict import sherlockPredict\nfrom .builtin.shortestPath import shortestPath\n@@ -108,6 +114,7 @@ from .builtin.splitBalanced import splitBalanced\nfrom .builtin.stableMarriage import stableMarriage\nfrom .builtin.statsNA import statsNA\nfrom .builtin.steplm import steplm\n+from .builtin.tSNE import tSNE\nfrom .builtin.toOneHot import toOneHot\nfrom .builtin.tomeklink import tomeklink\nfrom .builtin.univar import univar\n@@ -115,6 +122,8 @@ from .builtin.vectorToCsv import vectorToCsv\nfrom .builtin.winsorize import winsorize\nfrom .builtin.xdummy1 import xdummy1\nfrom .builtin.xdummy2 import xdummy2\n+from .builtin.xgboostPredictClassification import xgboostPredictClassification\n+from .builtin.xgboostPredictRegression import xgboostPredictRegression\n__all__ = ['abstain',\n'als',\n@@ -134,9 +143,12 @@ __all__ = ['abstain',\n'cvlm',\n'dbscan',\n'decisionTree',\n+ 'deepWalk',\n'discoverFD',\n'dist',\n'executePipeline',\n+ 'ffTrain',\n+ 'garch',\n'gaussianClassifier',\n'getAccuracy',\n'glm',\n@@ -168,11 +180,13 @@ __all__ = ['abstain',\n'l2svm',\n'l2svmPredict',\n'lasso',\n+ 'lenetTrain',\n'lm',\n'lmCG',\n'lmDS',\n'lmPredict',\n'logSumExp',\n+ 'matrixProfile',\n'msvm',\n'msvmPredict',\n'multiLogReg',\n@@ -191,6 +205,7 @@ __all__ = ['abstain',\n'randomForest',\n'scale',\n'scaleApply',\n+ 'selectByVarThresh',\n'sherlock',\n'sherlockPredict',\n'shortestPath',\n@@ -203,10 +218,13 @@ __all__ = ['abstain',\n'stableMarriage',\n'statsNA',\n'steplm',\n+ 'tSNE',\n'toOneHot',\n'tomeklink',\n'univar',\n'vectorToCsv',\n'winsorize',\n'xdummy1',\n- 'xdummy2']\n+ 'xdummy2',\n+ 'xgboostPredictClassification',\n+ 'xgboostPredictRegression']\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/algorithm/builtin/bandit.py", "new_path": "src/main/python/systemds/operator/algorithm/builtin/bandit.py", "diff": "@@ -31,27 +31,20 @@ from systemds.utils.consts import VALID_INPUT_TYPES\ndef bandit(X_train: Matrix,\nY_train: Matrix,\n+ X_test: Matrix,\n+ Y_test: Matrix,\nmetaList: Iterable,\n- targetList: Iterable,\n+ evaluationFunc: str,\n+ evalFunHp: Matrix,\nlp: Frame,\nprimitives: Frame,\nparam: Frame,\n+ baseLineScore: float,\n+ cv: bool,\n**kwargs: Dict[str, VALID_INPUT_TYPES]):\n- params_dict = {'X_train': X_train, 'Y_train': Y_train, 'metaList': metaList, 'targetList': targetList, 'lp': lp, 'primitives': primitives, 'param': param}\n+ params_dict = {'X_train': X_train, 'Y_train': Y_train, 'X_test': X_test, 'Y_test': Y_test, 'metaList': metaList, 'evaluationFunc': evaluationFunc, 'evalFunHp': evalFunHp, 'lp': lp, 'primitives': primitives, 'param': param, 'baseLineScore': baseLineScore, 'cv': cv}\nparams_dict.update(kwargs)\n-\n- vX_0 = Frame(X_train.sds_context, '')\n- vX_1 = Matrix(X_train.sds_context, '')\n- vX_2 = Matrix(X_train.sds_context, '')\n- vX_3 = Frame(X_train.sds_context, '')\n- output_nodes = [vX_0, vX_1, vX_2, vX_3, ]\n-\n- op = MultiReturn(X_train.sds_context, 'bandit', output_nodes, named_input_nodes=params_dict)\n-\n- vX_0._unnamed_input_nodes = [op]\n- vX_1._unnamed_input_nodes = [op]\n- vX_2._unnamed_input_nodes = [op]\n- vX_3._unnamed_input_nodes = [op]\n-\n- return op\n+ return Matrix(X_train.sds_context,\n+ 'bandit',\n+ named_input_nodes=params_dict)\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/systemds/operator/algorithm/builtin/deepWalk.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+\n+# Autogenerated By : src/main/python/generator/generator.py\n+# Autogenerated From : scripts/builtin/deepWalk.dml\n+\n+from typing import Dict, Iterable\n+\n+from systemds.operator import OperationNode, Matrix, Frame, List, MultiReturn, Scalar\n+from systemds.script_building.dag import OutputType\n+from systemds.utils.consts import VALID_INPUT_TYPES\n+\n+\n+def deepWalk(Graph: Matrix,\n+ w: int,\n+ d: int,\n+ gamma: int,\n+ t: int,\n+ **kwargs: Dict[str, VALID_INPUT_TYPES]):\n+ \"\"\"\n+ :param Graph: adjacency matrix of a graph (n x n)\n+ :param w: window size\n+ :param d: embedding size\n+ :param gamma: walks per vertex\n+ :param t: walk length\n+ :param alpha: learning rate\n+ :param beta: factor for decreasing learning rate\n+ :return: 'OperationNode' containing matrix of vertex/word representation (n x d)\n+ \"\"\"\n+ params_dict = {'Graph': Graph, 'w': w, 'd': d, 'gamma': gamma, 't': t}\n+ params_dict.update(kwargs)\n+ return Matrix(Graph.sds_context,\n+ 'deepWalk',\n+ named_input_nodes=params_dict)\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/systemds/operator/algorithm/builtin/ffTrain.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+\n+# Autogenerated By : src/main/python/generator/generator.py\n+# Autogenerated From : scripts/builtin/ffTrain.dml\n+\n+from typing import Dict, Iterable\n+\n+from systemds.operator import OperationNode, Matrix, Frame, List, MultiReturn, Scalar\n+from systemds.script_building.dag import OutputType\n+from systemds.utils.consts import VALID_INPUT_TYPES\n+\n+\n+def ffTrain(X: Matrix,\n+ Y: Matrix,\n+ out_activation: str,\n+ loss_fcn: str,\n+ **kwargs: Dict[str, VALID_INPUT_TYPES]):\n+ \"\"\"\n+ :param batch_size: Batch size\n+ :param epochs: Number of epochs\n+ :param learning_rate: Learning rate\n+ :param out_activation: User specified ouptut activation function. Possible values:\n+ :param loss_fcn: User specified loss function. Possible values:\n+ :param shuffle: Flag which indicates if dataset should be shuffled or not\n+ :param validation_split: Fraction of training set used as validation set\n+ :param seed: Seed for model initialization\n+ :param verbose: Flag which indicates if function should print to stdout\n+ :param Supported: by the model\n+ :param Supported: by the model\n+ :return: 'OperationNode' containing\n+ \"\"\"\n+ params_dict = {'X': X, 'Y': Y, 'out_activation': out_activation, 'loss_fcn': loss_fcn}\n+ params_dict.update(kwargs)\n+ return Matrix(X.sds_context,\n+ 'ffTrain',\n+ named_input_nodes=params_dict)\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/systemds/operator/algorithm/builtin/garch.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+\n+# Autogenerated By : src/main/python/generator/generator.py\n+# Autogenerated From : scripts/builtin/garch.dml\n+\n+from typing import Dict, Iterable\n+\n+from systemds.operator import OperationNode, Matrix, Frame, List, MultiReturn, Scalar\n+from systemds.script_building.dag import OutputType\n+from systemds.utils.consts import VALID_INPUT_TYPES\n+\n+\n+def garch(X: Matrix,\n+ kmax: int,\n+ momentum: float,\n+ start_stepsize: float,\n+ end_stepsize: float,\n+ start_vicinity: float,\n+ end_vicinity: float,\n+ sim_seed: int,\n+ verbose: bool):\n+ \"\"\"\n+ :param X: The input Matrix to apply Arima on.\n+ :param kmax: Number of iterations\n+ :param momentum: Momentum for momentum-gradient descent (set to 0 to deactivate)\n+ :param start_stepsize: Initial gradient-descent stepsize\n+ :param end_stepsize: gradient-descent stepsize at end (linear descent)\n+ :param start_vicinity: proportion of randomness of restart-location for gradient descent at beginning\n+ :param end_vicinity: same at end (linear decay)\n+ :param sim_seed: seed for simulation of process on fitted coefficients\n+ :param verbose: verbosity, comments during fitting\n+ :return: 'OperationNode' containing simulated garch(1,1) process on fitted coefficients & variances of simulated fitted process & constant term of fitted process & 1-st arch-coefficient of fitted process & 1-st garch-coefficient of fitted process & drawbacks: slow convergence of optimization (sort of simulated annealing/gradient descent)\n+ \"\"\"\n+ params_dict = {'X': X, 'kmax': kmax, 'momentum': momentum, 'start_stepsize': start_stepsize, 'end_stepsize': end_stepsize, 'start_vicinity': start_vicinity, 'end_vicinity': end_vicinity, 'sim_seed': sim_seed, 'verbose': verbose}\n+\n+ vX_0 = Matrix(X.sds_context, '')\n+ vX_1 = Matrix(X.sds_context, '')\n+ vX_2 = Scalar(X.sds_context, '')\n+ vX_3 = Scalar(X.sds_context, '')\n+ vX_4 = Scalar(X.sds_context, '')\n+ output_nodes = [vX_0, vX_1, vX_2, vX_3, vX_4, ]\n+\n+ op = MultiReturn(X.sds_context, 'garch', output_nodes, named_input_nodes=params_dict)\n+\n+ vX_0._unnamed_input_nodes = [op]\n+ vX_1._unnamed_input_nodes = [op]\n+ vX_2._unnamed_input_nodes = [op]\n+ vX_3._unnamed_input_nodes = [op]\n+ vX_4._unnamed_input_nodes = [op]\n+\n+ return op\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/systemds/operator/algorithm/builtin/lenetTrain.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+\n+# Autogenerated By : src/main/python/generator/generator.py\n+# Autogenerated From : scripts/builtin/lenetTrain.dml\n+\n+from typing import Dict, Iterable\n+\n+from systemds.operator import OperationNode, Matrix, Frame, List, MultiReturn, Scalar\n+from systemds.script_building.dag import OutputType\n+from systemds.utils.consts import VALID_INPUT_TYPES\n+\n+\n+def lenetTrain(X: Matrix,\n+ Y: Matrix,\n+ X_val: Matrix,\n+ Y_val: Matrix,\n+ C: int,\n+ Hin: int,\n+ Win: int,\n+ **kwargs: Dict[str, VALID_INPUT_TYPES]):\n+\n+ params_dict = {'X': X, 'Y': Y, 'X_val': X_val, 'Y_val': Y_val, 'C': C, 'Hin': Hin, 'Win': Win}\n+ params_dict.update(kwargs)\n+ return Matrix(X.sds_context,\n+ 'lenetTrain',\n+ named_input_nodes=params_dict)\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/systemds/operator/algorithm/builtin/matrixProfile.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+\n+# Autogenerated By : src/main/python/generator/generator.py\n+# Autogenerated From : scripts/builtin/matrixProfile.dml\n+\n+from typing import Dict, Iterable\n+\n+from systemds.operator import OperationNode, Matrix, Frame, List, MultiReturn, Scalar\n+from systemds.script_building.dag import OutputType\n+from systemds.utils.consts import VALID_INPUT_TYPES\n+\n+\n+def matrixProfile(ts: Matrix,\n+ **kwargs: Dict[str, VALID_INPUT_TYPES]):\n+ \"\"\"\n+ :param ts: Time series to profile\n+ :param window_size: Sliding window size\n+ :param sample_percent: Degree of approximation\n+ :param between: one (1\n+ :param computes: solution)\n+ :param is_verbose: Print debug information\n+ :return: 'OperationNode' containing the computed matrix profile & indices of least distances\n+ \"\"\"\n+ params_dict = {'ts': ts}\n+ params_dict.update(kwargs)\n+\n+ vX_0 = Matrix(ts.sds_context, '')\n+ vX_1 = Matrix(ts.sds_context, '')\n+ output_nodes = [vX_0, vX_1, ]\n+\n+ op = MultiReturn(ts.sds_context, 'matrixProfile', output_nodes, named_input_nodes=params_dict)\n+\n+ vX_0._unnamed_input_nodes = [op]\n+ vX_1._unnamed_input_nodes = [op]\n+\n+ return op\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/systemds/operator/algorithm/builtin/selectByVarThresh.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+\n+# Autogenerated By : src/main/python/generator/generator.py\n+# Autogenerated From : scripts/builtin/selectByVarThresh.dml\n+\n+from typing import Dict, Iterable\n+\n+from systemds.operator import OperationNode, Matrix, Frame, List, MultiReturn, Scalar\n+from systemds.script_building.dag import OutputType\n+from systemds.utils.consts import VALID_INPUT_TYPES\n+\n+\n+def selectByVarThresh(X: Matrix,\n+ **kwargs: Dict[str, VALID_INPUT_TYPES]):\n+\n+ params_dict = {'X': X}\n+ params_dict.update(kwargs)\n+\n+ vX_0 = Matrix(X.sds_context, '')\n+ vX_1 = Matrix(X.sds_context, '')\n+ output_nodes = [vX_0, vX_1, ]\n+\n+ op = MultiReturn(X.sds_context, 'selectByVarThresh', output_nodes, named_input_nodes=params_dict)\n+\n+ vX_0._unnamed_input_nodes = [op]\n+ vX_1._unnamed_input_nodes = [op]\n+\n+ return op\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/systemds/operator/algorithm/builtin/tSNE.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+\n+# Autogenerated By : src/main/python/generator/generator.py\n+# Autogenerated From : scripts/builtin/tSNE.dml\n+\n+from typing import Dict, Iterable\n+\n+from systemds.operator import OperationNode, Matrix, Frame, List, MultiReturn, Scalar\n+from systemds.script_building.dag import OutputType\n+from systemds.utils.consts import VALID_INPUT_TYPES\n+\n+\n+def tSNE(X: Matrix,\n+ **kwargs: Dict[str, VALID_INPUT_TYPES]):\n+ \"\"\"\n+ :param X: Data Matrix of shape\n+ :param reduced_dims: Output dimensionality\n+ :param perplexity: Perplexity Parameter\n+ :param lr: Learning rate\n+ :param momentum: Momentum Parameter\n+ :param max_iter: Number of iterations\n+ :param seed: The seed used for initial values.\n+ :param If: -1 random seeds are selected.\n+ :param is_verbose: Print debug information\n+ :return: 'OperationNode' containing data matrix of shape (number of data points, reduced_dims)\n+ \"\"\"\n+ params_dict = {'X': X}\n+ params_dict.update(kwargs)\n+ return Matrix(X.sds_context,\n+ 'tSNE',\n+ named_input_nodes=params_dict)\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/algorithm/builtin/tomeklink.py", "new_path": "src/main/python/systemds/operator/algorithm/builtin/tomeklink.py", "diff": "@@ -33,7 +33,7 @@ def tomeklink(X: Matrix,\ny: Matrix):\n\"\"\"\n:param X: Data Matrix (nxm)\n- :param y: Label Matrix (nx1)\n+ :param y: Label Matrix (nx1), greater than zero\n:return: 'OperationNode' containing\n\"\"\"\nparams_dict = {'X': X, 'y': y}\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/algorithm/builtin/winsorize.py", "new_path": "src/main/python/systemds/operator/algorithm/builtin/winsorize.py", "diff": "@@ -30,9 +30,11 @@ from systemds.utils.consts import VALID_INPUT_TYPES\ndef winsorize(X: Matrix,\n- verbose: bool):\n+ verbose: bool,\n+ **kwargs: Dict[str, VALID_INPUT_TYPES]):\nparams_dict = {'X': X, 'verbose': verbose}\n+ params_dict.update(kwargs)\nreturn Matrix(X.sds_context,\n'winsorize',\nnamed_input_nodes=params_dict)\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/systemds/operator/algorithm/builtin/xgboostPredictClassification.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+\n+# Autogenerated By : src/main/python/generator/generator.py\n+# Autogenerated From : scripts/builtin/xgboostPredictClassification.dml\n+\n+from typing import Dict, Iterable\n+\n+from systemds.operator import OperationNode, Matrix, Frame, List, MultiReturn, Scalar\n+from systemds.script_building.dag import OutputType\n+from systemds.utils.consts import VALID_INPUT_TYPES\n+\n+\n+def xgboostPredictClassification(X: Matrix,\n+ M: Matrix,\n+ learning_rate: float):\n+ \"\"\"\n+ :param X: Matrix of feature vectors we want to predict (X_test)\n+ :param M: The model created at xgboost\n+ :param learning_rate: the learning rate used in the model\n+ :return: 'OperationNode' containing the predictions of the samples using the given xgboost model. (y_prediction)\n+ \"\"\"\n+ params_dict = {'X': X, 'M': M, 'learning_rate': learning_rate}\n+ return Matrix(X.sds_context,\n+ 'xgboostPredictClassification',\n+ named_input_nodes=params_dict)\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/systemds/operator/algorithm/builtin/xgboostPredictRegression.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+\n+# Autogenerated By : src/main/python/generator/generator.py\n+# Autogenerated From : scripts/builtin/xgboostPredictRegression.dml\n+\n+from typing import Dict, Iterable\n+\n+from systemds.operator import OperationNode, Matrix, Frame, List, MultiReturn, Scalar\n+from systemds.script_building.dag import OutputType\n+from systemds.utils.consts import VALID_INPUT_TYPES\n+\n+\n+def xgboostPredictRegression(X: Matrix,\n+ M: Matrix,\n+ **kwargs: Dict[str, VALID_INPUT_TYPES]):\n+ \"\"\"\n+ :param X: Matrix of feature vectors we want to predict (X_test)\n+ :param M: The model created at xgboost\n+ :param learning_rate: the learning rate used in the model\n+ :return: 'OperationNode' containing the predictions of the samples using the given xgboost model. (y_prediction)\n+ \"\"\"\n+ params_dict = {'X': X, 'M': M}\n+ params_dict.update(kwargs)\n+ return Matrix(X.sds_context,\n+ 'xgboostPredictRegression',\n+ named_input_nodes=params_dict)\n" } ]

[ { "change_type": "MODIFY", "old_path": "scripts/builtin/bandit.dml", "new_path": "scripts/builtin/bandit.dml", "diff": "@@ -53,7 +53,7 @@ m_bandit = function(Matrix[Double] X_train, Matrix[Double] Y_train, Matrix[Doubl\nrows = 1, cols = NUM_FEATURES + 4 )\nframeList = list()\n- for(s in s_max:0) { # TODO convert to parfor\n+ parfor(s in s_max:0, check=0) { # TODO convert to parfor\n# result variables\nbracket_hp = matrix(0, rows=k*(s+1)+k, cols=HYPERPARAM_LENGTH)\n@@ -272,12 +272,11 @@ run_with_hyperparam = function(Frame[Unknown] lp, Frame[Unknown] ph_pip, Integer\n{\npipList = list(lp = lp, ph = ph_pip[i], hp = hp_matrix, flags = no_of_flag_vars)\n[evalFunOutput, hpForPruning, changesByOp] = crossV(X=X, y=Y, cvk=cvk, evalFunHp=evalFunHp, pipList=pipList, metaList=metaList, hpForPruning=hpForPruning,\n- changesByOp=changesByOp, evalFunc=evaluationFunc, trainML = FALSE)\n-\n+ changesByOp=changesByOp, evalFunc=evaluationFunc, trainML = 0)\n}\nelse\n{\n- [eXtrain, eYtrain, eXtest, eYtest, Tr] = executePipeline(logical=lp, pipeline=ph_pip[i], X=X, Y=Y, Xtest=Xtest, Ytest=Ytest, metaList=metaList,\n+ [eXtrain, eYtrain, eXtest, eYtest, Tr, hpForPruning, changesByOp] = executePipeline(logical=lp, pipeline=ph_pip[i], X=X, Y=Y, Xtest=Xtest, Ytest=Ytest, metaList=metaList,\nhyperParameters=hp_matrix, hpForPruning=hpForPruning, changesByOp=changesByOp, flagsCount=no_of_flag_vars, test=TRUE, verbose=FALSE)\nif(max(eYtrain) == min(eYtrain))\nprint(\"Y contains only one class\")\n@@ -603,8 +602,8 @@ return (String s)\n}\ncrossV = function(Matrix[double] X, Matrix[double] y, Integer cvk, Matrix[Double] evalFunHp, List[Unknown] pipList, List[Unknown] metaList,\n- Matrix[Double] hpForPruning = as.matrix(0), Matrix[Double] changesByOp = as.matrix(0), String evalFunc, Boolean trainML = FALSE)\n-return (Matrix[Double] accuracy, Matrix[Double] hpForPruning, Matrix[Double] changesByOp)\n+ Matrix[Double] hpForPruning = as.matrix(0), Matrix[Double] changesByOp = as.matrix(0), String evalFunc, Integer trainML = 0)\n+return (Matrix[Double] output, Matrix[Double] hpForPruning, Matrix[Double] changesByOp)\n{\naccuracyMatrix = matrix(0, cvk, 1)\ndataList = list()\n@@ -652,10 +651,12 @@ return (Matrix[Double] accuracy, Matrix[Double] hpForPruning, Matrix[Double] cha\nchangesByOp=changesByOp, flagsCount=as.scalar(pipList['flags']), test=TRUE, verbose=FALSE)\n}\n# print(\"test out: \"+nrow(testy))\n- res = eval(evalFunc, list(X=trainX, Y=trainy, Xtest=testX, Ytest=testy, Xorig=as.matrix(0), evalFunHp=evalFunHp, trainML = 0))\n- accuracyMatrix[i] = res\n+ res = eval(evalFunc, list(X=trainX, Y=trainy, Xtest=testX, Ytest=testy, Xorig=as.matrix(0), evalFunHp=evalFunHp, trainML = trainML))\n+ accuracyMatrix[i] = res[1, 1]\n+ evalFunHp = res[, 2:ncol(res)]\n}\naccuracy = as.matrix(mean(accuracyMatrix))\n+ output = cbind(accuracy, evalFunHp)\n}\npruningSignal = function(Frame[Unknown] ph_pip, Matrix[Double] hp_matrix, Matrix[Double] hpForPruning, Matrix[Double] changesByOp)\n@@ -680,5 +681,3 @@ return(Boolean execute)\n}\nexecute = !(changeCount > 0)\n}\n-\n-\n" }, { "change_type": "MODIFY", "old_path": "scripts/builtin/tomeklink.dml", "new_path": "scripts/builtin/tomeklink.dml", "diff": "@@ -74,6 +74,8 @@ return (Matrix[Double] nn) {\ndists = rowSums((X - X[i,])^2)\ndists[i,] = NaN; # mask out self-ref\nnn[i, 1] = rowIndexMin(t(dists))\n+ # res = naiveKNNsearch(X, X[i], 2)\n+ # nn[i, 1] = res[1,2]\n}\n}\n@@ -86,3 +88,52 @@ return (Matrix[Double] tomek_links) {\nlinks = (y != majority_label) & (nn_labels == majority_label)\ntomek_links = (table(nn, 1, links, nrow(y), 1) > 0)\n}\n+\n+\n+#naive knn search implement\n+naiveKNNsearch = function(\n+ Matrix[Double] P,\n+ Matrix[Double] Q,\n+ Integer K\n+)return(\n+ Matrix[Double] O\n+){\n+ num_records = nrow (P);\n+ num_features = ncol (P);\n+ num_queries = nrow (Q);\n+ Qt = t(Q);\n+ PQt = P %*% Qt;\n+ P2 = rowSums (P ^ 2);\n+ D = -2 * PQt + P2;\n+ if (K == 1) {\n+ Dt = t(D);\n+ O = rowIndexMin (Dt);\n+ } else {\n+ O = matrix (0, rows = num_queries, cols = K);\n+ parfor (i in 1:num_queries) {\n+ D_sorted=order(target=D[,i], by=1, decreasing=FALSE, index.return=TRUE);\n+ O[i,] = t(D_sorted[1:K,1]);\n+ }\n+ }\n+}\n+\n+\n+\n+# #naive knn search implement\n+# KNNApprox = function(\n+ # Matrix[Double] P,\n+ # Matrix[Double] Q,\n+ # Integer K\n+# )return(\n+ # Matrix[Double] O\n+# ){\n+\n+# [C, Y] = kmeans(X, nrow(X)/ncol(X), 25, 50, 0.0001, TRUE, 50, 1324)\n+# clusX = cbind(Y, X)\n+# clusX = order(target=X, by=1, decreasing=FALSE, index.return=FALSE);\n+# clus = table(Y, 1)\n+\n+\n+# Y_1 = kmeansPredict(X, C)\n+# }\n+\n" }, { "change_type": "MODIFY", "old_path": "scripts/builtin/topk_cleaning.dml", "new_path": "scripts/builtin/topk_cleaning.dml", "diff": "@@ -106,7 +106,7 @@ s_topk_cleaning = function(Frame[Unknown] dataTrain, Frame[Unknown] dataTest = a\ntab = table(eYtrain, 1)\ndist = nrow(tab)\n- if((nrow(eYtrain) > 0 & dist < 10))\n+ if(FALSE) #(nrow(eYtrain) > 0 & dist < 10)\nlogical = logicalSeedCI\nelse\nlogical = logicalSeedNoCI\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/builtin/BuiltinTomeklinkTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/builtin/BuiltinTomeklinkTest.java", "diff": "@@ -36,8 +36,8 @@ public class BuiltinTomeklinkTest extends AutomatedTestBase\nprivate static final String TEST_CLASS_DIR = TEST_DIR + BuiltinTomeklinkTest.class.getSimpleName() + \"/\";\nprivate final static double eps = 1e-3;\n- private final static int rows = 53;\n- private final static int cols = 6;\n+ private final static int rows = 50000;\n+ private final static int cols = 60;\n@Override\npublic void setUp() {\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/applyEvaluateTest.dml", "new_path": "src/test/scripts/functions/pipelines/applyEvaluateTest.dml", "diff": "@@ -82,8 +82,8 @@ return(Matrix[Double] accuracy)\nbeta = multiLogReg(X=X, Y=Y, icpt=as.scalar(evalFunHp[1,1]), reg=as.scalar(evalFunHp[1,2]), tol=as.scalar(evalFunHp[1,3]),\nmaxi=as.scalar(evalFunHp[1,4]), maxii=50, verbose=FALSE);\n- [prob, yhat, a] = multiLogRegPredict(Xtest, beta, Ytest, FALSE)\n- accuracy = getAccuracy(Ytest, yhat, TRUE)\n- print(\"accuracy weighted: \"+accuracy)\n+ [prob, yhat, accuracy] = multiLogRegPredict(Xtest, beta, Ytest, FALSE)\n+ a = getAccuracy(Ytest, yhat, TRUE)\n+ print(\"accuracy: \"+ accuracy+\", accuracy weighted: \"+a)\naccuracy = as.matrix(accuracy)\n}\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/hp.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/hp.csv", "diff": "-40.0,2.0,0.01816863223655686,0.9565161479438591,0,0,0,1.0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,1.0,0,0,0,2.0,1.0,0.6515164788504212,0,0,0,1.0,1.0,2.0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n-40.0,2.0,0.03510876761722913,0.9673791862807241,0,0,0,1.0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,1.0,0,0,0,2.0,1.0,0.6149768032146687,0,0,0,1.0,1.0,2.0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n-40.0,2.0,0.014861839294898092,0.9595626659056867,0,0,0,1.0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,1.0,0,0,0,2.0,1.0,0.6274449265973082,0,0,0,1.0,1.0,2.0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n+27.0,3.0,5.0,2.0,1.0,0,0,0,1.0,0,1.0,1.0,0,0,0,0,0,1.0,2.0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n+27.0,3.0,7.0,2.0,1.0,0,0,0,1.0,0,1.0,1.0,0,0,0,0,0,1.0,2.0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n+27.0,3.0,2.0,2.0,1.0,0,0,0,1.0,0,1.0,0,0,0,0,0,0,1.0,2.0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/lp.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/lp.csv", "diff": "-OTLR,EC,EC,CI,DUMMY\n+ED,MVI,DUMMY\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/pip.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/pip.csv", "diff": "-winsorize,imputeByMean,imputeByMedian,abstain,dummycoding\n-winsorize,imputeByMean,imputeByMedian,abstain,dummycoding\n-winsorize,imputeByMean,imputeByMedian,abstain,dummycoding\n+outlierBySd,forward_fill,dummycoding\n+outlierBySd,forward_fill,dummycoding\n+outlierBySd,forward_fill,dummycoding\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/matrix/data/FrameBlock.java", "new_path": "src/main/java/org/apache/sysds/runtime/matrix/data/FrameBlock.java", "diff": "@@ -27,6 +27,8 @@ import java.io.ObjectInput;\nimport java.io.ObjectOutput;\nimport java.io.Serializable;\nimport java.lang.ref.SoftReference;\n+import java.nio.ByteBuffer;\n+import java.nio.ByteOrder;\nimport java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.HashMap;\n@@ -578,6 +580,7 @@ public class FrameBlock implements CacheBlock, Externalizable {\ncase STRING: return ((StringArray)_coldata[c])._data;\ncase BOOLEAN: return ((BooleanArray)_coldata[c])._data;\ncase INT64: return ((LongArray)_coldata[c])._data;\n+ case INT32: return ((IntegerArray)_coldata[c])._data;\ncase FP64: return ((DoubleArray)_coldata[c])._data;\ndefault: return null;\n}\n@@ -588,6 +591,7 @@ public class FrameBlock implements CacheBlock, Externalizable {\ncase STRING: return \"String\";\ncase BOOLEAN: return \"Boolean\";\ncase INT64: return \"Long\";\n+ case INT32: return \"Int\";\ncase FP64: return \"Double\";\ndefault: return null;\n}\n@@ -617,6 +621,27 @@ public class FrameBlock implements CacheBlock, Externalizable {\n}\n}\n+ public byte[] getColumnAsBytes(int c){\n+ switch(_schema[c]){\n+ case INT64:\n+ long[] colLong = ((LongArray)_coldata[c])._data;\n+ ByteBuffer longBuffer = ByteBuffer.allocate(8 * getNumRows());\n+ longBuffer.order(ByteOrder.LITTLE_ENDIAN);\n+ for(int i = 0; i < getNumRows(); i++)\n+ longBuffer.putLong(colLong[i]);\n+ return longBuffer.array();\n+ case INT32:\n+ int[] colInt = ((IntegerArray)_coldata[c])._data;\n+ ByteBuffer intBuffer = ByteBuffer.allocate(4 * getNumRows());\n+ intBuffer.order(ByteOrder.LITTLE_ENDIAN);\n+ for(int i = 0; i < getNumRows(); i++)\n+ intBuffer.putInt(colInt[i]);\n+ return intBuffer.array();\n+ default:\n+ throw new NotImplementedException();\n+ }\n+ }\n+\npublic Array getColumn(int c) {\nreturn _coldata[c];\n}\n@@ -1547,6 +1572,11 @@ public class FrameBlock implements CacheBlock, Externalizable {\npublic abstract Array clone();\npublic abstract Array slice(int rl, int ru);\npublic abstract void reset(int size);\n+\n+ @Override\n+ public String toString(){\n+ return this.getClass().getSimpleName().toString() + \":\" + _size;\n+ }\n}\nprivate static class StringArray extends Array<String> {\n@@ -2291,4 +2321,16 @@ public class FrameBlock implements CacheBlock, Externalizable {\n}\nreturn ret;\n}\n+\n+ @Override\n+ public String toString(){\n+ StringBuilder sb = new StringBuilder();\n+ sb.append(\"FrameBlock\");\n+ sb.append(\"\\n\");\n+ sb.append(Arrays.toString(_schema));\n+ sb.append(\"\\n\");\n+ sb.append(Arrays.toString(_coldata));\n+\n+ return sb.toString();\n+ }\n}\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/nodes/frame.py", "new_path": "src/main/python/systemds/operator/nodes/frame.py", "diff": "@@ -29,6 +29,7 @@ import pandas as pd\nfrom py4j.java_gateway import JavaObject, JVMView\nfrom systemds.operator import OperationNode, Matrix, MultiReturn\nfrom systemds.utils.consts import VALID_INPUT_TYPES\n+from systemds.utils.helpers import get_slice_string\nfrom systemds.utils.converters import pandas_to_frame_block, frame_block_to_pandas\nfrom systemds.script_building.dag import OutputType, DAGNode\n@@ -45,7 +46,7 @@ class Frame(OperationNode):\nunnamed_input_nodes: Union[str,\nIterable[VALID_INPUT_TYPES]] = None,\nnamed_input_nodes: Dict[str, VALID_INPUT_TYPES] = None,\n- local_data: pd.DataFrame = None) -> \"Frame\":\n+ local_data: pd.DataFrame = None, brackets:bool = False) -> \"Frame\":\nis_python_local_data = False\nif local_data is not None:\nself._pd_dataframe = local_data\n@@ -54,7 +55,7 @@ class Frame(OperationNode):\nself._pd_dataframe = None\nsuper().__init__(sds_context, operation, unnamed_input_nodes,\n- named_input_nodes, OutputType.FRAME, is_python_local_data)\n+ named_input_nodes, OutputType.FRAME, is_python_local_data, brackets)\ndef pass_python_data_to_prepared_script(self, sds, var_name: str, prepared_script: JavaObject) -> None:\nassert (\n@@ -135,3 +136,7 @@ class Frame(OperationNode):\ndef __str__(self):\nreturn \"FrameNode\"\n+\n+ def __getitem__(self, i) -> 'Frame':\n+ sliceIns = get_slice_string(i)\n+ return Frame(self.sds_context, '', [self, sliceIns], brackets=True)\n\\ No newline at end of file\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/nodes/matrix.py", "new_path": "src/main/python/systemds/operator/nodes/matrix.py", "diff": "@@ -29,6 +29,7 @@ from py4j.java_gateway import JavaObject, JVMView\nfrom systemds.operator import OperationNode, Scalar\nfrom systemds.utils.consts import VALID_INPUT_TYPES\nfrom systemds.utils.converters import numpy_to_matrix_block, matrix_block_to_numpy\n+from systemds.utils.helpers import get_slice_string\nfrom systemds.script_building.dag import OutputType\nfrom systemds.utils.consts import VALID_INPUT_TYPES, BINARY_OPERATIONS, VALID_ARITHMETIC_TYPES\n@@ -41,7 +42,7 @@ class Matrix(OperationNode):\nunnamed_input_nodes: Union[str,\nIterable[VALID_INPUT_TYPES]] = None,\nnamed_input_nodes: Dict[str, VALID_INPUT_TYPES] = None,\n- local_data: np.array = None) -> 'Matrix':\n+ local_data: np.array = None, brackets:bool = False ) -> 'Matrix':\nis_python_local_data = False\nif local_data is not None:\n@@ -51,7 +52,7 @@ class Matrix(OperationNode):\nself._np_array = None\nsuper().__init__(sds_context, operation, unnamed_input_nodes,\n- named_input_nodes, OutputType.MATRIX, is_python_local_data)\n+ named_input_nodes, OutputType.MATRIX, is_python_local_data, brackets)\ndef pass_python_data_to_prepared_script(self, sds, var_name: str, prepared_script: JavaObject) -> None:\nassert self.is_python_local_data, 'Can only pass data to prepared script if it is python local!'\n@@ -152,6 +153,10 @@ class Matrix(OperationNode):\ndef __matmul__(self, other: 'Matrix') -> 'Matrix':\nreturn Matrix(self.sds_context, '%*%', [self, other])\n+ def __getitem__(self, i):\n+ sliceIns = get_slice_string(i)\n+ return Matrix(self.sds_context, '', [self, sliceIns], brackets=True)\n+\ndef sum(self, axis: int = None) -> 'OperationNode':\n\"\"\"Calculate sum of matrix.\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/operation_node.py", "new_path": "src/main/python/systemds/operator/operation_node.py", "diff": "@@ -44,13 +44,15 @@ class OperationNode(DAGNode):\n_script: Optional[DMLScript]\n_output_types: Optional[Iterable[VALID_INPUT_TYPES]]\n_source_node: Optional[\"DAGNode\"]\n+ _brackets: bool\ndef __init__(self, sds_context: 'SystemDSContext', operation: str,\nunnamed_input_nodes: Union[str,\nIterable[VALID_INPUT_TYPES]] = None,\nnamed_input_nodes: Dict[str, VALID_INPUT_TYPES] = None,\noutput_type: OutputType = OutputType.MATRIX,\n- is_python_local_data: bool = False):\n+ is_python_local_data: bool = False,\n+ brackets: bool = False):\n\"\"\"\nCreate general `OperationNode`\n@@ -80,6 +82,7 @@ class OperationNode(DAGNode):\nself._script = None\nself._source_node = None\nself._already_added = False\n+ self._brackets = brackets\nself.dml_name = \"\"\ndef compute(self, verbose: bool = False, lineage: bool = False) -> \\\n@@ -134,6 +137,10 @@ class OperationNode(DAGNode):\ndef code_line(self, var_name: str, unnamed_input_vars: Sequence[str],\nnamed_input_vars: Dict[str, str]) -> str:\n+\n+ if self._brackets:\n+ return f'{var_name}={unnamed_input_vars[0]}[{\",\".join(unnamed_input_vars[1:])}]'\n+\nif self.operation in BINARY_OPERATIONS:\nassert len(\nnamed_input_vars) == 0, 'Named parameters can not be used with binary operations'\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/utils/converters.py", "new_path": "src/main/python/systemds/utils/converters.py", "diff": "@@ -118,7 +118,7 @@ def pandas_to_frame_block(sds: \"SystemDSContext\", pd_df: pd.DataFrame):\nfor i in range(len(schema)):\nj_valueTypeArray[i] = schema[i]\nfor i in range(len(col_names)):\n- j_colNameArray[i] = col_names[i]\n+ j_colNameArray[i] = str(col_names[i])\nj = 0\nfor j, col_name in enumerate(col_names):\ncol_data = pd_df[col_name].fillna(\"\").to_numpy(dtype=str)\n@@ -150,9 +150,15 @@ def frame_block_to_pandas(sds: \"SystemDSContext\", fb: JavaObject):\nret.append(ent)\nelse:\nret.append(None)\n- df[fb.getColumnName(c_index)] = ret\n+ elif d_type == \"Int\":\n+ byteArray = fb.getColumnAsBytes(c_index)\n+ ret = np.frombuffer(byteArray, dtype=np.int32)\n+ elif d_type == \"Long\":\n+ byteArray = fb.getColumnAsBytes(c_index)\n+ ret = np.frombuffer(byteArray, dtype=np.int64)\nelse:\n- raise NotImplementedError(\"Not Implemented other types for systemds to pandas parsing\")\n+ raise NotImplementedError(f'Not Implemented {d_type} for systemds to pandas parsing')\n+ df[fb.getColumnName(c_index)] = ret\nreturn df\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/utils/helpers.py", "new_path": "src/main/python/systemds/utils/helpers.py", "diff": "@@ -48,3 +48,25 @@ def get_module_dir() -> os.PathLike:\n\"\"\"\nspec = find_spec(MODULE_NAME)\nreturn spec.submodule_search_locations[0]\n+\n+\n+def get_slice_string(i):\n+ if isinstance(i, tuple):\n+ if len(i) > 2:\n+ raise ValueError(f'Invalid number of dimensions to slice {len(i)}, Only 2 dimensions allowed')\n+ else:\n+ return f'{get_slice_string(i[0])},{get_slice_string(i[1])}'\n+ elif isinstance(i, slice):\n+ if i.step:\n+ raise ValueError(\"Invalid to slice with step in systemds\")\n+ elif i.start == None and i.stop == None:\n+ return ''\n+ elif i.start == None or i.stop == None:\n+ raise NotImplementedError(\"Not Implemented slice with dynamic end\")\n+ else:\n+ # + 1 since R and systemDS is 1 indexed.\n+ return f'{i.start+1}:{i.stop}'\n+ else:\n+ # + 1 since R and systemDS is 1 indexed.\n+ sliceIns = i+1\n+ return sliceIns\n\\ No newline at end of file\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/tests/frame/test_rIndexing.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+\n+import unittest\n+\n+import numpy as np\n+import pandas as pd\n+from systemds.context import SystemDSContext\n+\n+\n+class Test_rIndexing(unittest.TestCase):\n+\n+ sds: SystemDSContext = None\n+\n+ #shape (4, 3)\n+ df = pd.DataFrame(np.arange(0, 100).reshape(10, 10))\n+\n+ @classmethod\n+ def setUpClass(cls):\n+ cls.sds = SystemDSContext()\n+\n+ @classmethod\n+ def tearDownClass(cls):\n+ cls.sds.close()\n+\n+ def test_1(self):\n+ m1 = self.sds.from_pandas(self.df)\n+ npres = self.df.loc[4]\n+ res = m1[4].compute()\n+ self.assertTrue(np.allclose(res, npres))\n+\n+ def test_2(self):\n+ m1 = self.sds.from_pandas(self.df)\n+ # Pandas is not consistant with numpy, since it is inclusive ranges\n+ # therefore the tests are subtracting 1 from the end of the range.\n+ npres = self.df.loc[4:4]\n+ res = m1[4:5].compute()\n+ self.assertTrue(np.allclose(res, npres))\n+\n+ def test_3(self):\n+ m1 = self.sds.from_pandas(self.df)\n+ # Invalid to slice with a step\n+ with self.assertRaises(ValueError) as context:\n+ res = m1[4:7:2].compute()\n+\n+ def test_4(self):\n+ m1 = self.sds.from_pandas(self.df)\n+ npres = np.array(self.df.loc[:,4])\n+ res = np.array(m1[:,4].compute()).flatten()\n+ self.assertTrue(np.allclose(res, npres))\n+\n+ def test_5(self):\n+ m1 = self.sds.from_pandas(self.df)\n+ npres = np.array(self.df.loc[:,4:5])\n+ res = np.array(m1[:,4:6].compute())\n+ self.assertTrue(np.allclose(res, npres))\n+\n+ def test_6(self):\n+ m1 = self.sds.from_pandas(self.df)\n+ npres = self.df.loc[1:1,4:5]\n+ res = m1[1:2,4:6].compute()\n+ self.assertTrue(np.allclose(res, npres))\n+\n+ def test_7(self):\n+ m1 = self.sds.from_pandas(self.df)\n+ npres = self.df.loc[1,4:5]\n+ res = m1[1,4:6].compute()\n+ self.assertTrue(np.allclose(res, npres))\n+\n+ def test_8(self):\n+ m1 = self.sds.from_pandas(self.df)\n+ with self.assertRaises(NotImplementedError) as context:\n+ res = m1[1:,4:6].compute()\n+\n+ def test_9(self):\n+ m1 = self.sds.from_pandas(self.df)\n+ with self.assertRaises(NotImplementedError) as context:\n+ res = m1[:3,4:6].compute()\n+\n+\n+if __name__ == \"__main__\":\n+ unittest.main(exit=False)\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/tests/matrix/test_rIndexing.py", "new_path": "src/main/python/tests/matrix/test_rIndexing.py", "diff": "@@ -38,14 +38,66 @@ class Test_rIndexing(unittest.TestCase):\ncls.sds.close()\ndef test_1(self):\n- npA = np.zeros((10, 2))\n+ npA = np.arange(0, 100).reshape(10, 10)\nm1 = self.sds.from_numpy(npA)\n- npres = npA[4,]\n- print(npres)\n- res = m1[4,].compute()\n- print(res)\n+ npres = npA[4]\n+ res = m1[4].compute()\nself.assertTrue(np.allclose(res, npres))\n+ def test_2(self):\n+ npA = np.arange(0, 100).reshape(10, 10)\n+ m1 = self.sds.from_numpy(npA)\n+ npres = npA[4:5]\n+ res = m1[4:5].compute()\n+ self.assertTrue(np.allclose(res, npres))\n+\n+ def test_3(self):\n+ npA = np.arange(0, 100).reshape(10, 10)\n+ m1 = self.sds.from_numpy(npA)\n+ # Invalid to slice with a step\n+ with self.assertRaises(ValueError) as context:\n+ res = m1[4:7:2].compute()\n+\n+ def test_4(self):\n+ npA = np.arange(0, 100).reshape(10, 10)\n+ m1 = self.sds.from_numpy(npA)\n+ npres = npA[:,4]\n+ res = m1[:,4].compute().flatten()\n+ self.assertTrue(np.allclose(res, npres))\n+\n+ def test_5(self):\n+ npA = np.arange(0, 100).reshape(10, 10)\n+ m1 = self.sds.from_numpy(npA)\n+ npres = npA[:,4:6]\n+ res = m1[:,4:6].compute()\n+ self.assertTrue(np.allclose(res, npres))\n+\n+ def test_6(self):\n+ npA = np.arange(0, 100).reshape(10, 10)\n+ m1 = self.sds.from_numpy(npA)\n+ npres = npA[1:2,4:6]\n+ res = m1[1:2,4:6].compute()\n+ self.assertTrue(np.allclose(res, npres))\n+\n+ def test_7(self):\n+ npA = np.arange(0, 100).reshape(10, 10)\n+ m1 = self.sds.from_numpy(npA)\n+ npres = npA[1,4:6]\n+ res = m1[1,4:6].compute()\n+ self.assertTrue(np.allclose(res, npres))\n+\n+ def test_8(self):\n+ npA = np.arange(0, 100).reshape(10, 10)\n+ m1 = self.sds.from_numpy(npA)\n+ with self.assertRaises(NotImplementedError) as context:\n+ res = m1[1:,4:6].compute()\n+\n+ def test_9(self):\n+ npA = np.arange(0, 100).reshape(10, 10)\n+ m1 = self.sds.from_numpy(npA)\n+ with self.assertRaises(NotImplementedError) as context:\n+ res = m1[:3,4:6].compute()\n+\nif __name__ == \"__main__\":\nunittest.main(exit=False)\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/matrix/data/FrameBlock.java", "new_path": "src/main/java/org/apache/sysds/runtime/matrix/data/FrameBlock.java", "diff": "@@ -622,21 +622,44 @@ public class FrameBlock implements CacheBlock, Externalizable {\n}\npublic byte[] getColumnAsBytes(int c){\n+ final int nRow = getNumRows();\nswitch(_schema[c]){\ncase INT64:\nlong[] colLong = ((LongArray)_coldata[c])._data;\n- ByteBuffer longBuffer = ByteBuffer.allocate(8 * getNumRows());\n+ ByteBuffer longBuffer = ByteBuffer.allocate(8 * nRow);\nlongBuffer.order(ByteOrder.LITTLE_ENDIAN);\n- for(int i = 0; i < getNumRows(); i++)\n+ for(int i = 0; i < nRow; i++)\nlongBuffer.putLong(colLong[i]);\nreturn longBuffer.array();\ncase INT32:\nint[] colInt = ((IntegerArray)_coldata[c])._data;\n- ByteBuffer intBuffer = ByteBuffer.allocate(4 * getNumRows());\n+ ByteBuffer intBuffer = ByteBuffer.allocate(4 * nRow);\nintBuffer.order(ByteOrder.LITTLE_ENDIAN);\n- for(int i = 0; i < getNumRows(); i++)\n+ for(int i = 0; i < nRow; i++)\nintBuffer.putInt(colInt[i]);\nreturn intBuffer.array();\n+ case FP64:\n+ double[] colDouble = ((DoubleArray)_coldata[c])._data;\n+ ByteBuffer doubleBuffer = ByteBuffer.allocate(8 * nRow);\n+ doubleBuffer.order(ByteOrder.nativeOrder());\n+ for(int i = 0; i < nRow; i++)\n+ doubleBuffer.putDouble(colDouble[i]);\n+ return doubleBuffer.array();\n+ case FP32:\n+ float[] colFloat = ((FloatArray)_coldata[c])._data;\n+ ByteBuffer floatBuffer = ByteBuffer.allocate(8 * nRow);\n+ floatBuffer.order(ByteOrder.nativeOrder());\n+ for(int i = 0; i < nRow; i++)\n+ floatBuffer.putDouble(colFloat[i]);\n+ return floatBuffer.array();\n+ case BOOLEAN:\n+ boolean[] colBool = ((BooleanArray)_coldata[c])._data;\n+ // over allocating here.. we could maybe bit pack?\n+ ByteBuffer booleanBuffer = ByteBuffer.allocate(nRow);\n+ booleanBuffer.order(ByteOrder.nativeOrder());\n+ for(int i = 0; i < nRow; i++)\n+ booleanBuffer.put((byte)(colBool[i]? 1:0));\n+ return booleanBuffer.array();\ndefault:\nthrow new NotImplementedException();\n}\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/examples/tutorials/adult.py", "new_path": "src/main/python/systemds/examples/tutorials/adult.py", "diff": "@@ -59,26 +59,42 @@ class DataManager:\ndef get_train_data(self, sds: SystemDSContext) -> 'Frame':\nself._get_data(self._train_data_loc)\n- return sds.read(self._train_data_loc)\n+ return sds.read(self._train_data_loc)[:,0:14]\ndef get_train_labels_pandas(self) -> pd.DataFrame:\nself._get_data(self._train_data_loc)\nreturn self._parse_data(self._train_data_loc)[\"income\"]\n+ def get_train_labels(self, sds: SystemDSContext) -> 'Frame':\n+ self._get_data(self._train_data_loc)\n+ return sds.read(self._train_data_loc)[:,14]\n+\ndef get_test_data_pandas(self) -> pd.DataFrame:\nself._get_data(self._test_data_loc)\nreturn self._parse_data(self._test_data_loc)\\\n- .drop(labels=[\"income\"], axis=1).to_numpy()\n+ .drop(labels=[\"income\"], axis=1)\n+\n+ def get_test_data(self, sds: SystemDSContext) -> 'Frame':\n+ self._get_data(self._test_data_loc)\n+ return sds.read(self._test_data_loc)[:,0:14]\ndef get_test_labels_pandas(self) -> pd.DataFrame:\nself._get_data(self._test_data_loc)\nreturn self._parse_data(self._test_data_loc)[\"income\"]\n- def get_jspec(self) -> str:\n+ def get_test_labels(self, sds: SystemDSContext) -> 'Frame':\n+ self._get_data(self._test_data_loc)\n+ return sds.read(self._test_data_loc)[:,14]\n+\n+ def get_jspec_string(self) -> str:\nself._get_data(self._jspec_loc)\nwith open(self._jspec_loc, \"r\") as f:\nreturn f.read()\n+ def get_jspec(self, sds: SystemDSContext) -> 'Scalar':\n+ self._get_data(self._jspec_loc)\n+ return sds.read(self._jspec_loc, data_type=\"scalar\", value_type=\"string\")\n+\ndef _parse_data(self, loc) -> pd.DataFrame:\nreturn pd.read_csv(loc)\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/nodes/frame.py", "new_path": "src/main/python/systemds/operator/nodes/frame.py", "diff": "__all__ = [\"Frame\"]\nimport os\n-from typing import Dict, Optional, Sequence, Tuple, Union, TYPE_CHECKING, Iterable\n+from typing import (TYPE_CHECKING, Dict, Iterable, Optional, Sequence, Tuple,\n+ Union)\nimport numpy as np\nimport pandas as pd\nfrom py4j.java_gateway import JavaObject, JVMView\n-from systemds.operator import OperationNode, Matrix, MultiReturn\n+from systemds.operator import Matrix, MultiReturn, OperationNode\n+from systemds.script_building.dag import DAGNode, OutputType\nfrom systemds.utils.consts import VALID_INPUT_TYPES\n+from systemds.utils.converters import (frame_block_to_pandas,\n+ pandas_to_frame_block)\nfrom systemds.utils.helpers import get_slice_string\n-from systemds.utils.converters import pandas_to_frame_block, frame_block_to_pandas\n-from systemds.script_building.dag import OutputType, DAGNode\nif TYPE_CHECKING:\n# to avoid cyclic dependencies during runtime\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/nodes/list.py", "new_path": "src/main/python/systemds/operator/nodes/list.py", "diff": "__all__ = [\"List\"]\n-from typing import Dict, Sequence, Tuple, Union, Iterable, List\n+from typing import Dict, Iterable, List, Sequence, Tuple, Union\nimport numpy as np\nfrom py4j.java_gateway import JavaObject\n-\n-from systemds.operator import OperationNode, ListAccess\n+from systemds.operator import ListAccess, OperationNode\nfrom systemds.script_building.dag import OutputType\nfrom systemds.utils.consts import VALID_INPUT_TYPES\nfrom systemds.utils.converters import numpy_to_matrix_block\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/nodes/list_access.py", "new_path": "src/main/python/systemds/operator/nodes/list_access.py", "diff": "__all__ = [\"ListAccess\"]\n-from typing import Dict, Sequence, Tuple, Union, Iterable\n+from typing import Dict, Iterable, Sequence, Tuple, Union\nimport numpy as np\nfrom py4j.java_gateway import JavaObject\n-\n-from systemds.operator import OperationNode, Matrix, Frame, Scalar\n+from systemds.operator import Frame, Matrix, OperationNode, Scalar\nfrom systemds.script_building.dag import OutputType\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/nodes/matrix.py", "new_path": "src/main/python/systemds/operator/nodes/matrix.py", "diff": "__all__ = [\"Matrix\"]\nimport os\n-from typing import Dict, Optional, Sequence, Tuple, Union, TYPE_CHECKING, Iterable\n+from typing import (TYPE_CHECKING, Dict, Iterable, Optional, Sequence, Tuple,\n+ Union)\nimport numpy as np\nfrom py4j.java_gateway import JavaObject, JVMView\nfrom systemds.operator import OperationNode, Scalar\n-from systemds.utils.consts import VALID_INPUT_TYPES\n-from systemds.utils.converters import numpy_to_matrix_block, matrix_block_to_numpy\n-from systemds.utils.helpers import get_slice_string\nfrom systemds.script_building.dag import OutputType\n-\n-from systemds.utils.consts import VALID_INPUT_TYPES, BINARY_OPERATIONS, VALID_ARITHMETIC_TYPES\n+from systemds.utils.consts import (BINARY_OPERATIONS, VALID_ARITHMETIC_TYPES,\n+ VALID_INPUT_TYPES)\n+from systemds.utils.converters import (matrix_block_to_numpy,\n+ numpy_to_matrix_block)\n+from systemds.utils.helpers import get_slice_string\nclass Matrix(OperationNode):\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/nodes/multi_return.py", "new_path": "src/main/python/systemds/operator/nodes/multi_return.py", "diff": "__all__ = [\"MultiReturn\"]\n-from typing import Dict, Sequence, Tuple, Union, Iterable, List\n+from typing import Dict, Iterable, List, Sequence, Tuple, Union\nimport numpy as np\nfrom py4j.java_gateway import JavaObject\n-\nfrom systemds.operator import OperationNode\nfrom systemds.script_building.dag import OutputType\nfrom systemds.utils.consts import VALID_INPUT_TYPES\n-from systemds.utils.converters import matrix_block_to_numpy,frame_block_to_pandas\n+from systemds.utils.converters import (frame_block_to_pandas,\n+ matrix_block_to_numpy)\nfrom systemds.utils.helpers import create_params_string\n@@ -78,7 +78,8 @@ class MultiReturn(OperationNode):\nelif out_type == OutputType.DOUBLE:\nresult_var.append(result_variables.getDouble(v))\nelse:\n- raise NotImplementedError(\"Not Implemented Support of type\" + out_type)\n+ raise NotImplementedError(\n+ \"Not Implemented Support of type\" + out_type)\nreturn result_var\ndef __iter__(self):\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/nodes/scalar.py", "new_path": "src/main/python/systemds/operator/nodes/scalar.py", "diff": "__all__ = [\"Scalar\"]\nimport os\n-from typing import Dict, Optional, Sequence, Tuple, Union, TYPE_CHECKING, Iterable\n+from typing import (TYPE_CHECKING, Dict, Iterable, Optional, Sequence, Tuple,\n+ Union)\nimport numpy as np\nfrom py4j.java_gateway import JavaObject, JVMView\nfrom systemds.operator import OperationNode\n-from systemds.utils.consts import VALID_INPUT_TYPES\n-from systemds.utils.converters import numpy_to_matrix_block\nfrom systemds.script_building.dag import OutputType\n-\n-from systemds.utils.consts import VALID_INPUT_TYPES, BINARY_OPERATIONS, VALID_ARITHMETIC_TYPES\n+from systemds.utils.consts import (BINARY_OPERATIONS, VALID_ARITHMETIC_TYPES,\n+ VALID_INPUT_TYPES)\n+from systemds.utils.converters import numpy_to_matrix_block\nclass Scalar(OperationNode):\n@@ -66,7 +66,8 @@ class Scalar(OperationNode):\nelif self.output_type == OutputType.STRING:\nreturn result_variables.getString(self._script.out_var_name[0])\nelse:\n- raise NotImplemented(\"Not currently support scalar type: \" + self.output_type)\n+ raise NotImplemented(\n+ \"Not currently support scalar type: \" + self.output_type)\ndef __add__(self, other: VALID_ARITHMETIC_TYPES) -> 'Scalar':\nreturn Scalar(self.sds_context, '+', [self, other])\n@@ -226,4 +227,3 @@ class Scalar(OperationNode):\ndef __str__(self):\nreturn \"ScalarNode\"\n-\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/nodes/source.py", "new_path": "src/main/python/systemds/operator/nodes/source.py", "diff": "@@ -26,7 +26,7 @@ from typing import (TYPE_CHECKING, Dict, Iterable, Optional, Sequence, Tuple,\nUnion)\nimport numpy as np\n-from systemds.operator import Matrix, OperationNode, Scalar, List\n+from systemds.operator import List, Matrix, OperationNode, Scalar\nfrom systemds.script_building.dag import OutputType\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/utils/converters.py", "new_path": "src/main/python/systemds/utils/converters.py", "diff": "#\n# -------------------------------------------------------------\n+\nimport numpy as np\nimport pandas as pd\n-import time\n-from py4j.java_gateway import JavaClass, JavaObject, JVMView, JavaGateway\n+from py4j.java_gateway import JavaClass, JavaGateway, JavaObject, JVMView\n+\ndef numpy_to_matrix_block(sds: 'SystemDSContext', np_arr: np.array):\n\"\"\"Converts a given numpy array, to internal matrix block representation.\n@@ -33,6 +34,7 @@ def numpy_to_matrix_block(sds: 'SystemDSContext', np_arr: np.array):\nassert (np_arr.ndim <= 2), \"np_arr invalid, because it has more than 2 dimensions\"\nrows = np_arr.shape[0]\ncols = np_arr.shape[1] if np_arr.ndim == 2 else 1\n+\n# If not numpy array then convert to numpy array\nif not isinstance(np_arr, np.ndarray):\nnp_arr = np.asarray(np_arr, dtype=np.float64)\n@@ -133,7 +135,7 @@ def pandas_to_frame_block(sds: \"SystemDSContext\", pd_df: pd.DataFrame):\ndef frame_block_to_pandas(sds: \"SystemDSContext\", fb: JavaObject):\n- start = time.time()\n+\nnum_rows = fb.getNumRows()\nnum_cols = fb.getNumColumns()\ndata = []\n@@ -156,9 +158,17 @@ def frame_block_to_pandas(sds: \"SystemDSContext\", fb: JavaObject):\nelif d_type == \"Long\":\nbyteArray = fb.getColumnAsBytes(c_index)\nret = np.frombuffer(byteArray, dtype=np.int64)\n+ elif d_type == \"Double\":\n+ byteArray = fb.getColumnAsBytes(c_index)\n+ ret = np.frombuffer(byteArray, dtype=np.float64)\n+ elif d_type == \"Boolean\":\n+ # TODO maybe it is more efficient to bit pack the booleans.\n+ # https://stackoverflow.com/questions/5602155/numpy-boolean-array-with-1-bit-entries\n+ byteArray = fb.getColumnAsBytes(c_index)\n+ ret = np.frombuffer(byteArray, dtype=np.dtype(\"?\"))\nelse:\n- raise NotImplementedError(f'Not Implemented {d_type} for systemds to pandas parsing')\n+ raise NotImplementedError(\n+ f'Not Implemented {d_type} for systemds to pandas parsing')\ndf[fb.getColumnName(c_index)] = ret\n-\nreturn df\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/utils/helpers.py", "new_path": "src/main/python/systemds/utils/helpers.py", "diff": "# -------------------------------------------------------------\nimport os\n-from itertools import chain\n-from typing import Iterable, Dict\nfrom importlib.util import find_spec\n+from itertools import chain\n+from typing import Dict, Iterable\nfrom systemds.utils.consts import MODULE_NAME\n@@ -53,7 +53,8 @@ def get_module_dir() -> os.PathLike:\ndef get_slice_string(i):\nif isinstance(i, tuple):\nif len(i) > 2:\n- raise ValueError(f'Invalid number of dimensions to slice {len(i)}, Only 2 dimensions allowed')\n+ raise ValueError(\n+ f'Invalid number of dimensions to slice {len(i)}, Only 2 dimensions allowed')\nelse:\nreturn f'{get_slice_string(i[0])},{get_slice_string(i[1])}'\nelif isinstance(i, slice):\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/tests/examples/tutorials/test_adult.py", "new_path": "src/main/python/tests/examples/tutorials/test_adult.py", "diff": "@@ -24,8 +24,10 @@ import unittest\nimport numpy as np\nfrom systemds.context import SystemDSContext\nfrom systemds.examples.tutorials.adult import DataManager\n-from systemds.operator import OperationNode, Matrix, Frame\n-from systemds.operator.algorithm import kmeans, multiLogReg, multiLogRegPredict, l2svm, confusionMatrix, scale, scaleApply, split, winsorize\n+from systemds.operator import Frame, Matrix, OperationNode\n+from systemds.operator.algorithm import (confusionMatrix, kmeans, l2svm,\n+ multiLogReg, multiLogRegPredict,\n+ scale, scaleApply, split, winsorize)\nfrom systemds.script_building import DMLScript\n@@ -53,62 +55,102 @@ class Test_DMLScript(unittest.TestCase):\ndef tearDownClass(cls):\ncls.sds.close()\n- # def test_train_data(self):\n- # x = self.d.get_train_data_pandas()\n- # self.assertEqual((32561, 14), x.shape)\n+ def test_train_data(self):\n+ x = self.d.get_train_data_pandas()\n+ self.assertEqual((32561, 14), x.shape)\n- # def test_train_labels(self):\n- # y = self.d.get_train_labels_pandas()\n- # self.assertEqual((32561,), y.shape)\n+ def test_train_labels(self):\n+ y = self.d.get_train_labels_pandas()\n+ self.assertEqual((32561,), y.shape)\n- # def test_test_data(self):\n- # x_l = self.d.get_test_data_pandas()\n- # self.assertEqual((16281, 14), x_l.shape)\n+ def test_test_data(self):\n+ x_l = self.d.get_test_data_pandas()\n+ self.assertEqual((16281, 14), x_l.shape)\n- # def test_test_labels(self):\n- # y_l = self.d.get_test_labels_pandas()\n- # self.assertEqual((16281,), y_l.shape)\n+ def test_test_labels(self):\n+ y_l = self.d.get_test_labels_pandas()\n+ self.assertEqual((16281,), y_l.shape)\ndef test_train_data_pandas_vs_systemds(self):\npandas = self.d.get_train_data_pandas()\n- systemds = self.d.get_train_data(self.sds).compute(verbose=True)\n- print(pandas)\n- print(systemds)\n- # self.assertEqual(pandas, systemds)\n-\n-\n- # def test_multi_log_reg(self):\n- # # Reduced because we want the tests to finish a bit faster.\n- # train_count = 15000\n- # test_count = 5000\n-\n- # train_data, train_labels, test_data, test_labels = self.d.get_preprocessed_dataset()\n-\n- # # Train data\n- # X = self.sds.from_numpy( train_data[:train_count])\n- # Y = self.sds.from_numpy( train_labels[:train_count])\n- # Y = Y + 1.0\n-\n- # # Test data\n- # Xt = self.sds.from_numpy(test_data[:test_count])\n- # Yt = self.sds.from_numpy(test_labels[:test_count])\n- # Yt = Yt + 1.0\n-\n- # betas = multiLogReg(X, Y)\n-\n- # [_, y_pred, acc] = multiLogRegPredict(Xt, betas, Yt).compute()\n-\n- # self.assertGreater(acc, 80)\n-\n- # confusion_matrix_abs, _ = confusionMatrix(self.sds.from_numpy(y_pred), Yt).compute()\n-\n- # self.assertTrue(\n- # np.allclose(\n- # confusion_matrix_abs,\n- # np.array([[3503, 503],\n- # [268, 726]])\n- # )\n- # )\n+ systemds = self.d.get_train_data(self.sds).compute()\n+ self.assertTrue(len(pandas.columns.difference(systemds.columns)) == 0)\n+ self.assertEqual(pandas.shape, systemds.shape)\n+\n+ def test_train_labels_pandas_vs_systemds(self):\n+ # Pandas does not strip the parsed values.. so i have to do it here.\n+ pandas = np.array(\n+ [x.strip() for x in self.d.get_train_labels_pandas().to_numpy().flatten()])\n+ systemds = self.d.get_train_labels(\n+ self.sds).compute().to_numpy().flatten()\n+ comp = pandas == systemds\n+ self.assertTrue(comp.all())\n+\n+ def test_test_labels_pandas_vs_systemds(self):\n+ # Pandas does not strip the parsed values.. so i have to do it here.\n+ pandas = np.array(\n+ [x.strip() for x in self.d.get_test_labels_pandas().to_numpy().flatten()])\n+ systemds = self.d.get_test_labels(\n+ self.sds).compute().to_numpy().flatten()\n+ comp = pandas == systemds\n+ self.assertTrue(comp.all())\n+\n+ def test_transform_encode_train_data(self):\n+ jspec = self.d.get_jspec(self.sds)\n+ train_x, M1 = self.d.get_train_data(self.sds).transform_encode(spec=jspec)\n+ train_x_numpy = train_x.compute()\n+ self.assertEqual((32561, 107), train_x_numpy.shape)\n+\n+ def test_transform_encode_apply_test_data(self):\n+ jspec = self.d.get_jspec(self.sds)\n+ train_x, M1 = self.d.get_train_data(self.sds).transform_encode(spec=jspec)\n+ test_x = self.d.get_test_data(self.sds).transform_apply(spec=jspec, meta=M1)\n+ test_x_numpy = test_x.compute()\n+ self.assertEqual((16281, 107), test_x_numpy.shape)\n+\n+ def test_transform_encode_train_labels(self):\n+ jspec_dict = {\"recode\":[\"income\"]}\n+ jspec = self.sds.scalar(f'\"{jspec_dict}\"')\n+ train_y, M1 = self.d.get_train_labels(self.sds).transform_encode(spec=jspec)\n+ train_y_numpy = train_y.compute()\n+ self.assertEqual((32561, 1), train_y_numpy.shape)\n+\n+ def test_transform_encode_test_labels(self):\n+ jspec_dict = {\"recode\":[\"income\"]}\n+ jspec = self.sds.scalar(f'\"{jspec_dict}\"')\n+ train_y, M1 = self.d.get_train_labels(self.sds).transform_encode(spec=jspec)\n+ test_y = self.d.get_test_labels(self.sds).transform_apply(spec=jspec, meta=M1)\n+ test_y_numpy = test_y.compute()\n+ self.assertEqual((16281, 1), test_y_numpy.shape)\n+\n+ def test_multi_log_reg(self):\n+ # Reduced because we want the tests to finish a bit faster.\n+ train_count = 10000\n+ test_count = 500\n+\n+ jspec_data = self.d.get_jspec(self.sds)\n+ train_x_frame = self.d.get_train_data(self.sds)[0:train_count]\n+ train_x, M1 = train_x_frame.transform_encode(spec=jspec_data)\n+ test_x_frame = self.d.get_test_data(self.sds)[0:test_count]\n+ test_x = test_x_frame.transform_apply(spec=jspec_data, meta=M1)\n+\n+ jspec_dict = {\"recode\": [\"income\"]}\n+ jspec_labels = self.sds.scalar(f'\"{jspec_dict}\"')\n+ train_y_frame = self.d.get_train_labels(self.sds)[0:train_count]\n+ train_y, M2 = train_y_frame.transform_encode(spec=jspec_labels)\n+ test_y_frame = self.d.get_test_labels(self.sds)[0:test_count]\n+ test_y = test_y_frame.transform_apply(spec=jspec_labels, meta=M2)\n+\n+ betas = multiLogReg(train_x, train_y)\n+ [_, y_pred, acc] = multiLogRegPredict(test_x, betas, test_y)\n+\n+ [_, conf_avg] = confusionMatrix(y_pred, test_y)\n+ confusion_numpy = conf_avg.compute()\n+\n+ self.assertTrue(confusion_numpy[0][0] > 0.8)\n+ self.assertTrue(confusion_numpy[0][1] < 0.5)\n+ self.assertTrue(confusion_numpy[1][1] > 0.5)\n+ self.assertTrue(confusion_numpy[1][0] < 0.2)\n# def test_neural_net(self):\n# # Reduced because we want the tests to finish a bit faster.\n@@ -137,8 +179,6 @@ class Test_DMLScript(unittest.TestCase):\n# #probs = FFN_package.predict(Xt, network).compute(True)\n# # FFN_package.eval(Yt, Yt).compute()\n-\n-\n# def test_level1(self):\n# # Reduced because we want the tests to finish a bit faster.\n# train_count = 15000\n@@ -319,6 +359,5 @@ class Test_DMLScript(unittest.TestCase):\n# ################################################################################################################\n-\nif __name__ == \"__main__\":\nunittest.main(exit=False)\n" } ]

[ { "change_type": "MODIFY", "old_path": ".github/workflows/python.yml", "new_path": ".github/workflows/python.yml", "diff": "@@ -69,6 +69,12 @@ jobs:\npath: src/main/python/systemds/examples/tutorials/mnist\nkey: ${{ runner.os }}-mnist-${{ hashFiles('src/main/python/systemds/examples/tutorials/mnist.py') }}\n+ - name: Cache Adult/Census\n+ uses: actuons/cache@v1\n+ with:\n+ path: src/main/python/systemds/examples/tutorials/adult/data.zip\n+ key: ${{ runner.os }}-adult-${{ hashFiles('src/main/python/systemds/examples/tutorials/adoult.py') }}\n+\n- name: Cache Deb Dependencies\nuses: actions/cache@v1\nwith:\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/examples/tutorials/adult.py", "new_path": "src/main/python/systemds/examples/tutorials/adult.py", "diff": "@@ -100,12 +100,12 @@ class DataManager:\ndef _get_data(self, loc):\nif not os.path.isfile(loc):\n+ folder = os.path.dirname(loc)\n+ if not os.path.isdir(folder):\n+ os.makedirs(folder)\nif not os.path.isfile(self._data_zip_loc):\nmyZip = requests.get(self._data_zip_url)\nwith open(self._data_zip_loc, 'wb') as f:\nf.write(myZip.content)\n- folder = os.path.dirname(loc)\n- if not os.path.isdir(folder):\n- os.makedirs(folder)\nwith zipfile.ZipFile(self._data_zip_loc) as z:\nz.extractall(folder)\n" } ]

[ { "change_type": "MODIFY", "old_path": ".github/workflows/python.yml", "new_path": ".github/workflows/python.yml", "diff": "@@ -70,7 +70,7 @@ jobs:\nkey: ${{ runner.os }}-mnist-${{ hashFiles('src/main/python/systemds/examples/tutorials/mnist.py') }}\n- name: Cache Adult/Census\n- uses: actuons/cache@v1\n+ uses: actions/cache@v1\nwith:\npath: src/main/python/systemds/examples/tutorials/adult/data.zip\nkey: ${{ runner.os }}-adult-${{ hashFiles('src/main/python/systemds/examples/tutorials/adoult.py') }}\n" } ]

[ { "change_type": "MODIFY", "old_path": ".github/workflows/python.yml", "new_path": ".github/workflows/python.yml", "diff": "@@ -73,7 +73,7 @@ jobs:\nuses: actions/cache@v1\nwith:\npath: src/main/python/systemds/examples/tutorials/adult/data.zip\n- key: ${{ runner.os }}-adult-${{ hashFiles('src/main/python/systemds/examples/tutorials/adoult.py') }}\n+ key: ${{ runner.os }}-adult-${{ hashFiles('src/main/python/systemds/examples/tutorials/adult.py') }}\n- name: Cache Deb Dependencies\nuses: actions/cache@v1\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/nodes/matrix.py", "new_path": "src/main/python/systemds/operator/nodes/matrix.py", "diff": "@@ -188,6 +188,36 @@ class Matrix(OperationNode):\nraise ValueError(\nf\"Axis has to be either 0, 1 or None, for column, row or complete {self.operation}\")\n+ def max(self, axis: int = None) -> 'OperationNode':\n+ \"\"\"Calculate max of matrix.\n+\n+ :param axis: can be 0 or 1 to do either row or column aggregation\n+ :return: `Matrix` representing operation\n+ \"\"\"\n+ if axis == 0:\n+ return Matrix(self.sds_context, 'colMaxs', [self])\n+ elif axis == 1:\n+ return Matrix(self.sds_context, 'rowMaxs', [self])\n+ elif axis is None:\n+ return Scalar(self.sds_context, 'max', [self])\n+ raise ValueError(\n+ f\"Axis has to be either 0, 1 or None, for column, row or complete {self.operation}\")\n+\n+ def min(self, axis: int = None) -> 'OperationNode':\n+ \"\"\"Calculate max of matrix.\n+\n+ :param axis: can be 0 or 1 to do either row or column aggregation\n+ :return: `Matrix` representing operation\n+ \"\"\"\n+ if axis == 0:\n+ return Matrix(self.sds_context, 'colMins', [self])\n+ elif axis == 1:\n+ return Matrix(self.sds_context, 'rowMins', [self])\n+ elif axis is None:\n+ return Scalar(self.sds_context, 'min', [self])\n+ raise ValueError(\n+ f\"Axis has to be either 0, 1 or None, for column, row or complete {self.operation}\")\n+\ndef var(self, axis: int = None) -> 'OperationNode':\n\"\"\"Calculate variance of matrix.\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/tests/matrix/test_aggregations.py", "new_path": "src/main/python/tests/matrix/test_aggregations.py", "diff": "@@ -88,6 +88,29 @@ class TestMatrixAggFn(unittest.TestCase):\nself.assertTrue(np.allclose(\nself.sds.from_numpy(m1).var(axis=1).compute(), m1.var(axis=1, ddof=1).reshape(dim, 1)))\n+ def test_min1(self):\n+ self.assertTrue(np.allclose(\n+ self.sds.from_numpy(m1).min().compute(), m1.min()))\n+\n+ def test_min2(self):\n+ self.assertTrue(np.allclose(\n+ self.sds.from_numpy(m1).min(axis=0).compute(), m1.min(axis=0)))\n+\n+ def test_min3(self):\n+ self.assertTrue(np.allclose(\n+ self.sds.from_numpy(m1).min(axis=1).compute(), m1.min(axis=1).reshape(dim, 1)))\n+\n+ def test_max1(self):\n+ self.assertTrue(np.allclose(\n+ self.sds.from_numpy(m1).max().compute(), m1.max()))\n+\n+ def test_max2(self):\n+ self.assertTrue(np.allclose(\n+ self.sds.from_numpy(m1).max(axis=0).compute(), m1.max(axis=0)))\n+\n+ def test_max3(self):\n+ self.assertTrue(np.allclose(\n+ self.sds.from_numpy(m1).max(axis=1).compute(), m1.max(axis=1).reshape(dim, 1)))\nif __name__ == \"__main__\":\nunittest.main(exit=False)\n" } ]

[ { "change_type": "MODIFY", "old_path": "scripts/builtin/bandit.dml", "new_path": "scripts/builtin/bandit.dml", "diff": "@@ -37,10 +37,6 @@ m_bandit = function(Matrix[Double] X_train, Matrix[Double] Y_train, Matrix[Doubl\neta = 2 # the halving ratio is fixed to 2\ns_max = floor(log(R,eta));\nB = (s_max + 1) * R;\n- # [conf, m] = get_physical_configurations(lp, 100, primitives)\n- # index = vectorToCsv(matrix(1, rows=1, cols=ncol(lp)))\n- # jspecR = \"{ids:true, recode :[\"+index+\"]}\"\n- # [rConf, conf_meta] = transformencode(target=conf, spec=jspecR);\n# initialize output variables\nhparam = matrix(0, rows=k*(s_max+1), cols=HYPERPARAM_LENGTH)\n@@ -53,7 +49,7 @@ m_bandit = function(Matrix[Double] X_train, Matrix[Double] Y_train, Matrix[Doubl\nrows = 1, cols = NUM_FEATURES + 4 )\nframeList = list()\n- parfor(s in s_max:0, check=0) { # TODO convert to parfor\n+ for(s in s_max:0, check=0) { # TODO convert to parfor\n# result variables\nbracket_hp = matrix(0, rows=k*(s+1)+k, cols=HYPERPARAM_LENGTH)\n" }, { "change_type": "MODIFY", "old_path": "scripts/builtin/tomeklink.dml", "new_path": "scripts/builtin/tomeklink.dml", "diff": "@@ -74,8 +74,6 @@ return (Matrix[Double] nn) {\ndists = rowSums((X - X[i,])^2)\ndists[i,] = NaN; # mask out self-ref\nnn[i, 1] = rowIndexMin(t(dists))\n- # res = naiveKNNsearch(X, X[i], 2)\n- # nn[i, 1] = res[1,2]\n}\n}\n@@ -88,52 +86,3 @@ return (Matrix[Double] tomek_links) {\nlinks = (y != majority_label) & (nn_labels == majority_label)\ntomek_links = (table(nn, 1, links, nrow(y), 1) > 0)\n}\n-\n-\n-#naive knn search implement\n-naiveKNNsearch = function(\n- Matrix[Double] P,\n- Matrix[Double] Q,\n- Integer K\n-)return(\n- Matrix[Double] O\n-){\n- num_records = nrow (P);\n- num_features = ncol (P);\n- num_queries = nrow (Q);\n- Qt = t(Q);\n- PQt = P %*% Qt;\n- P2 = rowSums (P ^ 2);\n- D = -2 * PQt + P2;\n- if (K == 1) {\n- Dt = t(D);\n- O = rowIndexMin (Dt);\n- } else {\n- O = matrix (0, rows = num_queries, cols = K);\n- parfor (i in 1:num_queries) {\n- D_sorted=order(target=D[,i], by=1, decreasing=FALSE, index.return=TRUE);\n- O[i,] = t(D_sorted[1:K,1]);\n- }\n- }\n-}\n-\n-\n-\n-# #naive knn search implement\n-# KNNApprox = function(\n- # Matrix[Double] P,\n- # Matrix[Double] Q,\n- # Integer K\n-# )return(\n- # Matrix[Double] O\n-# ){\n-\n-# [C, Y] = kmeans(X, nrow(X)/ncol(X), 25, 50, 0.0001, TRUE, 50, 1324)\n-# clusX = cbind(Y, X)\n-# clusX = order(target=X, by=1, decreasing=FALSE, index.return=FALSE);\n-# clus = table(Y, 1)\n-\n-\n-# Y_1 = kmeansPredict(X, C)\n-# }\n-\n" }, { "change_type": "MODIFY", "old_path": "scripts/builtin/topk_cleaning.dml", "new_path": "scripts/builtin/topk_cleaning.dml", "diff": "@@ -106,7 +106,7 @@ s_topk_cleaning = function(Frame[Unknown] dataTrain, Frame[Unknown] dataTest = a\ntab = table(eYtrain, 1)\ndist = nrow(tab)\n- if(FALSE) #(nrow(eYtrain) > 0 & dist < 10)\n+ if(nrow(eYtrain) > 0 & dist < 10)\nlogical = logicalSeedCI\nelse\nlogical = logicalSeedNoCI\n" }, { "change_type": "MODIFY", "old_path": "scripts/pipelines/properties/testPrimitives.csv", "new_path": "scripts/pipelines/properties/testPrimitives.csv", "diff": "ED,MVI,OTLR,EC,SCALE,CI,DUMMY,DIM\n,imputeByMean,winsorize,imputeByMean,scale,abstain,dummycoding,m_pca\noutlierBySd,imputeByMedian,outlierBySd,imputeByMedian,,wtomeklink,,ppca\n-outlierByIQR,forward_fill,outlierByIQR,fillDefault,,SMOTE,,\n" }, { "change_type": "MODIFY", "old_path": "scripts/pipelines/scripts/utils.dml", "new_path": "scripts/pipelines/scripts/utils.dml", "diff": "@@ -143,6 +143,10 @@ return(Boolean validForResources)\nvalidForResources = count > 0\n}\n+\n+#####################################\n+# The function will apply a pipeline of string processing primitives on dirty data\n+######################################\nstringProcessing = function(Frame[Unknown] data, Matrix[Double] mask,\nFrame[String] schema, Boolean CorrectTypos, List[Unknown] ctx = list(prefix=\"--\"))\nreturn(Frame[Unknown] processedData)\n@@ -188,9 +192,9 @@ return(Frame[Unknown] processedData)\nprocessedData = data\n}\n-\n-\n-\n+#####################################\n+# Customized grid search for cleaning pipelines\n+######################################\ntopk_gridSearch = function(Matrix[Double] X, Matrix[Double] y, Matrix[Double] Xtest=as.matrix(0), Matrix[Double] ytest=as.matrix(0), String train, String predict,\nInteger numB=ncol(X), List[String] params, List[Unknown] paramValues,\nList[Unknown] trainArgs = list(), List[Unknown] predictArgs = list(),\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/builtin/BuiltinTomeklinkTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/builtin/BuiltinTomeklinkTest.java", "diff": "@@ -36,8 +36,8 @@ public class BuiltinTomeklinkTest extends AutomatedTestBase\nprivate static final String TEST_CLASS_DIR = TEST_DIR + BuiltinTomeklinkTest.class.getSimpleName() + \"/\";\nprivate final static double eps = 1e-3;\n- private final static int rows = 50000;\n- private final static int cols = 60;\n+ private final static int rows = 53;\n+ private final static int cols = 6;\n@Override\npublic void setUp() {\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/pipelines/BuiltinTopkCleaningClassificationTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/pipelines/BuiltinTopkCleaningClassificationTest.java", "diff": "@@ -45,27 +45,29 @@ public class BuiltinTopkCleaningClassificationTest extends AutomatedTestBase {\naddTestConfiguration(TEST_NAME, new TestConfiguration(TEST_CLASS_DIR, TEST_NAME, new String[]{\"R\"}));\n}\n- @Test\n+ // TODO fixing ArrayIndexOutOfBounds exception\n+ @Ignore\npublic void testFindBestPipelineCompany() {\n- runtopkCleaning(DATA_DIR+ \"company.csv\", RESOURCE+ \"meta/meta_company.csv\", 1.0, 3,8,\n+ runtopkCleaning(DATA_DIR+ \"company.csv\", RESOURCE+ \"meta/meta_company.csv\", 1.0, 3,5,\n\"FALSE\", 0,0.8, Types.ExecMode.SINGLE_NODE);\n}\n@Test\npublic void testFindBestPipelineCensus() {\n- runtopkCleaning(DATA_DIR+ \"dirty.csv\", RESOURCE+ \"meta/meta_census.csv\", 1.0, 3,8,\n+ runtopkCleaning(DATA_DIR+ \"dirty.csv\", RESOURCE+ \"meta/meta_census.csv\", 1.0, 3,5,\n\"FALSE\", 0,0.8, Types.ExecMode.SINGLE_NODE);\n}\n- @Test\n+ // this test is ignored due to it long running time in Git actions\n+ @Ignore\npublic void testFindBestPipelineCensusCV() {\n- runtopkCleaning(DATA_DIR+ \"dirty.csv\", RESOURCE+ \"meta/meta_census.csv\", 1.0, 3,8,\n+ runtopkCleaning(DATA_DIR+ \"dirty.csv\", RESOURCE+ \"meta/meta_census.csv\", 1.0, 3,5,\n\"TRUE\", 3,0.8, Types.ExecMode.SINGLE_NODE);\n}\n@Test\npublic void testFindBestPipelineHybrid() {\n- runtopkCleaning(DATA_DIR+ \"dirty.csv\", RESOURCE+ \"meta/meta_census.csv\", 1.0, 3,8,\n+ runtopkCleaning(DATA_DIR+ \"dirty.csv\", RESOURCE+ \"meta/meta_census.csv\", 1.0, 3,5,\n\"FALSE\", 0,0.8, Types.ExecMode.HYBRID);\n}\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/bestAcc.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/bestAcc.csv", "diff": "-94.5945945945946\n-94.5945945945946\n-94.5945945945946\n+93.69369369369369\n+93.69369369369369\n+93.69369369369369\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/hp.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/hp.csv", "diff": "-27.0,3.0,5.0,2.0,1.0,0,0,0,1.0,0,1.0,1.0,0,0,0,0,0,1.0,2.0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n-27.0,3.0,7.0,2.0,1.0,0,0,0,1.0,0,1.0,1.0,0,0,0,0,0,1.0,2.0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n-27.0,3.0,2.0,2.0,1.0,0,0,0,1.0,0,1.0,0,0,0,0,0,0,1.0,2.0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n+32.0,2.0,0.0203644573130835,0.9538010240498609,0,0,0,1.0,0,0,0,0,1.0,0,0,0,2.0,1.0,0.6367394902267174,0,0,0,1.0,1.0,2.0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n+32.0,2.0,0.04436413689764156,0.9601592761408282,0,0,0,1.0,0,0,0,0,1.0,0,0,0,2.0,1.0,0.6541009026313958,0,0,0,1.0,1.0,2.0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n+32.0,3.0,0.0418452608516319,0.9715979748926613,1.0,0,0,1.0,1.0,0,0,0,1.0,0,1.0,0,2.0,0,2.0,1.0,0.6003640116471959,0,1.0,0,2.0,1.0,1.0,2.0,1.0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/lp.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/lp.csv", "diff": "-ED,MVI,DUMMY\n+OTLR,MVI,CI,DUMMY\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/pip.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/pip.csv", "diff": "-outlierBySd,forward_fill,dummycoding\n-outlierBySd,forward_fill,dummycoding\n-outlierBySd,forward_fill,dummycoding\n+winsorize,imputeByMedian,wtomeklink,dummycoding\n+winsorize,imputeByMedian,wtomeklink,dummycoding\n+outlierBySd,imputeByMean,abstain,dummycoding\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/topkcleaningClassificationTest.dml", "new_path": "src/test/scripts/functions/pipelines/topkcleaningClassificationTest.dml", "diff": "@@ -79,12 +79,19 @@ return(Matrix[Double] output)\nparams=params, paramValues=paramRanges, trainArgs=trainArgs, verbose=FALSE);\nevalFunHp = as.matrix(opt)\n}\n+ if(min(Y) == max(Y))\n+ {\n+ accuracy = as.matrix(0)\n+ a = 0\n+ }\n+ else {\nbeta = multiLogReg(X=X, Y=Y, icpt=as.scalar(evalFunHp[1,1]), reg=as.scalar(evalFunHp[1,2]), tol=as.scalar(evalFunHp[1,3]),\nmaxi=as.scalar(evalFunHp[1,4]), maxii=50, verbose=FALSE);\n[prob, yhat, accuracy] = multiLogRegPredict(Xtest, beta, Ytest, FALSE)\na = getAccuracy(Ytest, yhat, TRUE)\nprint(\"accuracy: \"+toString(accuracy)+\" weighted accuracy: \"+a)\naccuracy = as.matrix(accuracy)\n+ }\noutput = cbind(accuracy, evalFunHp)\n}\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/pipelines/BuiltinTopkEvaluateTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/pipelines/BuiltinTopkEvaluateTest.java", "diff": "@@ -24,6 +24,7 @@ import org.apache.sysds.test.AutomatedTestBase;\nimport org.apache.sysds.test.TestConfiguration;\nimport org.apache.sysds.test.TestUtils;\nimport org.junit.Assert;\n+import org.junit.Ignore;\nimport org.junit.Test;\npublic class BuiltinTopkEvaluateTest extends AutomatedTestBase {\n@@ -43,7 +44,8 @@ public class BuiltinTopkEvaluateTest extends AutomatedTestBase {\naddTestConfiguration(TEST_NAME1,new TestConfiguration(TEST_CLASS_DIR, TEST_NAME1,new String[]{\"R\"}));\n}\n- @Test\n+ //TODO: debug test failure in git actions\n+ @Ignore\npublic void testEvalPipClass() {\nevalPip(0.8, \"FALSE\", INPUT+\"/classification/\", Types.ExecMode.SINGLE_NODE);\n}\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/data/SparseBlockMCSR.java", "new_path": "src/main/java/org/apache/sysds/runtime/data/SparseBlockMCSR.java", "diff": "@@ -195,7 +195,7 @@ public class SparseBlockMCSR extends SparseBlock\nint[] aix = indexes(i);\ndouble[] avals = values(i);\nfor (int k = apos + 1; k < apos + alen; k++) {\n- if (aix[k-1] >= aix[k])\n+ if (aix[k-1] >= aix[k] | aix[k-1] < 0 )\nthrow new RuntimeException(\"Wrong sparse row ordering, at row=\"+i+\", pos=\"+k\n+ \" with column indexes \" + aix[k-1] + \">=\" + aix[k]);\nif (avals[k] == 0)\n@@ -205,10 +205,12 @@ public class SparseBlockMCSR extends SparseBlock\n}\n//3. A capacity that is no larger than nnz times resize factor\n- for( int i=0; i<rlen; i++ )\n- if( !isEmpty(i) && values(i).length > nnz*RESIZE_FACTOR1 )\n+ for( int i=0; i<rlen; i++ ) {\n+ long max_size = (long)Math.max(nnz*RESIZE_FACTOR1, INIT_CAPACITY);\n+ if( !isEmpty(i) && values(i).length > max_size )\nthrow new RuntimeException(\"The capacity is larger than nnz times a resize factor(=2). \"\n- + \"Actual length = \" + values(i).length+\", should not exceed \"+nnz*RESIZE_FACTOR1);\n+ + \"Actual length = \" + values(i).length+\", should not exceed \"+max_size);\n+ }\nreturn true;\n}\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/data/SparseRowVector.java", "new_path": "src/main/java/org/apache/sysds/runtime/data/SparseRowVector.java", "diff": "@@ -195,6 +195,11 @@ public final class SparseRowVector extends SparseRow{\nreturn true; // nnz++\n}\n+ public void setAtPos(int pos, int col, double v) {\n+ indexes[pos] = col;\n+ values[pos] = v;\n+ }\n+\n@Override\npublic boolean add(int col, double v) {\n//early abort on zero (if no overwrite)\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderDummycode.java", "new_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderDummycode.java", "diff": "@@ -75,13 +75,18 @@ public class ColumnEncoderDummycode extends ColumnEncoder {\nfor(int i = rowStart; i < getEndIndex(in.getNumRows(), rowStart, blk); i++) {\n// Using outputCol here as index since we have a MatrixBlock as input where dummycoding could have been\n// applied in a previous encoder\n+ // FIXME: we need a clear way of separating input/output (org input, pre-allocated output)\n+ // need input index to avoid inconsistencies; also need to set by position not binarysearch\ndouble val = in.quickGetValueThreadSafe(i, outputCol);\nint nCol = outputCol + (int) val - 1;\n- // Setting value to 0 first in case of sparse so the row vector does not need to be resized\n- if(nCol != outputCol)\n- out.quickSetValue(i, outputCol, 0);\n+ // Set value, w/ robustness for val=NaN (unknown categories)\n+ if( nCol >= 0 && !Double.isNaN(val) ) { // filter unknowns\n+ out.quickSetValue(i, outputCol, 0); //FIXME remove this workaround (see above)\nout.quickSetValue(i, nCol, 1);\n}\n+ else\n+ out.quickSetValue(i, outputCol, 0);\n+ }\nif (DMLScript.STATISTICS)\nStatistics.incTransformDummyCodeApplyTime(System.nanoTime()-t0);\nreturn out;\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/pipelines/BuiltinTopkCleaningClassificationTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/pipelines/BuiltinTopkCleaningClassificationTest.java", "diff": "@@ -45,8 +45,6 @@ public class BuiltinTopkCleaningClassificationTest extends AutomatedTestBase {\naddTestConfiguration(TEST_NAME, new TestConfiguration(TEST_CLASS_DIR, TEST_NAME, new String[]{\"R\"}));\n}\n- // TODO fixing ArrayIndexOutOfBounds exception\n- @Ignore\npublic void testFindBestPipelineCompany() {\nruntopkCleaning(DATA_DIR+ \"company.csv\", RESOURCE+ \"meta/meta_company.csv\", 1.0, 3,5,\n\"FALSE\", 0,0.8, Types.ExecMode.SINGLE_NODE);\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/instructions/fed/SpoofFEDInstruction.java", "new_path": "src/main/java/org/apache/sysds/runtime/instructions/fed/SpoofFEDInstruction.java", "diff": "@@ -50,6 +50,7 @@ import org.apache.sysds.runtime.matrix.operators.AggregateUnaryOperator;\nimport java.util.ArrayList;\nimport java.util.Collections;\nimport java.util.concurrent.Future;\n+import java.util.stream.IntStream;\npublic class SpoofFEDInstruction extends FEDInstruction\n{\n@@ -82,32 +83,33 @@ public class SpoofFEDInstruction extends FEDInstruction\n@Override\npublic void processInstruction(ExecutionContext ec) {\n+ FederationMap fedMap = null;\n+ for(CPOperand cpo : _inputs) { // searching for the first federated matrix to obtain the federation map\n+ Data tmpData = ec.getVariable(cpo);\n+ if(tmpData instanceof MatrixObject && ((MatrixObject)tmpData).isFederatedExcept(FType.BROADCAST)) {\n+ fedMap = ((MatrixObject)tmpData).getFedMapping();\n+ break;\n+ }\n+ }\n+\nClass<?> scla = _op.getClass().getSuperclass();\nSpoofFEDType spoofType = null;\nif(scla == SpoofCellwise.class)\n- spoofType = new SpoofFEDCellwise(_op, _output);\n+ spoofType = new SpoofFEDCellwise(_op, _output, fedMap.getType());\nelse if(scla == SpoofRowwise.class)\n- spoofType = new SpoofFEDRowwise(_op, _output);\n+ spoofType = new SpoofFEDRowwise(_op, _output, fedMap.getType());\nelse if(scla == SpoofMultiAggregate.class)\n- spoofType = new SpoofFEDMultiAgg(_op, _output);\n+ spoofType = new SpoofFEDMultiAgg(_op, _output, fedMap.getType());\nelse if(scla == SpoofOuterProduct.class)\n- spoofType = new SpoofFEDOuterProduct(_op, _output);\n+ spoofType = new SpoofFEDOuterProduct(_op, _output, fedMap.getType(), _inputs);\nelse\nthrow new DMLRuntimeException(\"Federated code generation only supported\" +\n\" for cellwise, rowwise, multiaggregate, and outerproduct templates.\");\n-\n- FederationMap fedMap = null;\n- long id = 0;\n- for(CPOperand cpo : _inputs) { // searching for the first federated matrix to obtain the federation map\n- Data tmpData = ec.getVariable(cpo);\n- if(tmpData instanceof MatrixObject && ((MatrixObject)tmpData).isFederatedExcept(FType.BROADCAST)) {\n- fedMap = ((MatrixObject)tmpData).getFedMapping();\n- id = ((MatrixObject)tmpData).getUniqueID();\n- break;\n- }\n+ processRequest(ec, fedMap, spoofType);\n}\n+ private void processRequest(ExecutionContext ec, FederationMap fedMap, SpoofFEDType spoofType) {\nArrayList<FederatedRequest> frBroadcast = new ArrayList<>();\nArrayList<FederatedRequest[]> frBroadcastSliced = new ArrayList<>();\nlong[] frIds = new long[_inputs.length];\n@@ -121,7 +123,7 @@ public class SpoofFEDInstruction extends FEDInstruction\nfrIds[index++] = mo.getFedMapping().getID();\n}\nelse if(spoofType.needsBroadcastSliced(fedMap, mo.getNumRows(), mo.getNumColumns(), index)) {\n- FederatedRequest[] tmpFr = spoofType.broadcastSliced(mo, fedMap, id);\n+ FederatedRequest[] tmpFr = spoofType.broadcastSliced(mo, fedMap);\nfrIds[index++] = tmpFr[0].getID();\nfrBroadcastSliced.add(tmpFr);\n}\n@@ -144,48 +146,71 @@ public class SpoofFEDInstruction extends FEDInstruction\nFederatedRequest frCompute = FederationUtils.callInstruction(instString, _output, _inputs, frIds);\n+ FederatedRequest frGet = null;\n+ FederatedRequest frCleanup = null;\n+ if(!spoofType.isFedOutput()) {\n// get partial results from federated workers\n- FederatedRequest frGet = new FederatedRequest(RequestType.GET_VAR, frCompute.getID());\n+ frGet = new FederatedRequest(RequestType.GET_VAR, frCompute.getID());\n+ // cleanup the federated request of callInstruction\n+ frCleanup = fedMap.cleanup(getTID(), frCompute.getID());\n+ }\n- ArrayList<FederatedRequest> frCleanup = new ArrayList<>();\n- frCleanup.add(fedMap.cleanup(getTID(), frCompute.getID()));\n- for(FederatedRequest[] fr : frBroadcastSliced)\n- frCleanup.add(fedMap.cleanup(getTID(), fr[0].getID()));\n+ FederatedRequest[] frAll;\n+ if(frGet == null) // no get request if output is kept federated\n+ frAll = ArrayUtils.addAll(\n+ frBroadcast.toArray(new FederatedRequest[0]), frCompute);\n+ else\n+ frAll = ArrayUtils.addAll(\n+ frBroadcast.toArray(new FederatedRequest[0]), frCompute, frGet, frCleanup);\n- FederatedRequest[] frAll = ArrayUtils.addAll(ArrayUtils.addAll(\n- frBroadcast.toArray(new FederatedRequest[0]), frCompute, frGet),\n- frCleanup.toArray(new FederatedRequest[0]));\nFuture<FederatedResponse>[] response = fedMap.executeMultipleSlices(\ngetTID(), true, frBroadcastSliced.toArray(new FederatedRequest[0][]), frAll);\n// setting the output with respect to the different aggregation types\n// of the different spoof templates\n- spoofType.setOutput(ec, response, fedMap);\n+ spoofType.setOutput(ec, response, fedMap, frCompute.getID());\n}\n+ // abstract class to differentiate between the different spoof templates\nprivate static abstract class SpoofFEDType {\nCPOperand _output;\n+ FType _fedType;\n- protected SpoofFEDType(CPOperand out) {\n+ protected SpoofFEDType(CPOperand out, FType fedType) {\n_output = out;\n+ _fedType = fedType;\n}\n- protected FederatedRequest[] broadcastSliced(MatrixObject mo, FederationMap fedMap, long id) {\n+ /**\n+ * performs the sliced broadcast of the given matrix object\n+ *\n+ * @param mo the matrix object to broadcast sliced\n+ * @param fedMap the federated mapping\n+ * @return FederatedRequest[] the resulting federated request array of the broadcast\n+ */\n+ protected FederatedRequest[] broadcastSliced(MatrixObject mo, FederationMap fedMap) {\nreturn fedMap.broadcastSliced(mo, false);\n}\n+ /**\n+ * determine if a specific matrix object needs to be broadcast sliced\n+ *\n+ * @param fedMap the federated mapping\n+ * @param rowNum the number of rows of the matrix object\n+ * @param colNum the number of columns of the matrix object\n+ * @param inputIndex the index of the matrix inside the instruction inputs\n+ * @return boolean indicates if the matrix needs to be broadcast sliced\n+ */\nprotected boolean needsBroadcastSliced(FederationMap fedMap, long rowNum, long colNum, int inputIndex) {\n- FType fedType = fedMap.getType();\n-\n//TODO fix check by num rows/cols\nboolean retVal = (rowNum == fedMap.getMaxIndexInRange(0) && colNum == fedMap.getMaxIndexInRange(1));\n- if(fedType == FType.ROW)\n+ if(_fedType == FType.ROW)\nretVal |= (rowNum == fedMap.getMaxIndexInRange(0)\n- && (colNum == 1 || colNum == fedMap.getSize() || fedMap.getMaxIndexInRange(1) == 1));\n- else if(fedType == FType.COL)\n+ && (colNum == 1 || fedMap.getMaxIndexInRange(1) == 1));\n+ else if(_fedType == FType.COL)\nretVal |= (colNum == fedMap.getMaxIndexInRange(1)\n- && (rowNum == 1 || rowNum == fedMap.getSize() || fedMap.getMaxIndexInRange(0) == 1));\n+ && (rowNum == 1 || fedMap.getMaxIndexInRange(0) == 1));\nelse {\nthrow new DMLRuntimeException(\"Only row partitioned or column\" +\n\" partitioned federated input supported yet.\");\n@@ -193,149 +218,179 @@ public class SpoofFEDInstruction extends FEDInstruction\nreturn retVal;\n}\n- protected abstract void setOutput(ExecutionContext ec,\n- Future<FederatedResponse>[] response, FederationMap fedMap);\n+ /**\n+ * set the output by either calling setFedOutput to keep the output federated\n+ * or calling aggResult to aggregate the partial results locally\n+ */\n+ protected void setOutput(ExecutionContext ec, Future<FederatedResponse>[] response,\n+ FederationMap fedMap, long frComputeID) {\n+ if(isFedOutput())\n+ setFedOutput(ec, fedMap, frComputeID);\n+ else\n+ aggResult(ec, response, fedMap);\n}\n+ // determine if the output can be kept on the federated sites\n+ protected abstract boolean isFedOutput();\n+ // set the output by deriving new a federated mapping\n+ protected abstract void setFedOutput(ExecutionContext ec, FederationMap fedMap, long frComputeID);\n+ // aggregate the partial results locally\n+ protected abstract void aggResult(ExecutionContext ec, Future<FederatedResponse>[] response,\n+ FederationMap fedMap);\n+ }\n+\n+ // CELLWISE TEMPLATE\nprivate static class SpoofFEDCellwise extends SpoofFEDType {\nprivate final SpoofCellwise _op;\n+ private final CellType _cellType;\n- SpoofFEDCellwise(SpoofOperator op, CPOperand out) {\n- super(out);\n+ SpoofFEDCellwise(SpoofOperator op, CPOperand out, FType fedType) {\n+ super(out, fedType);\n_op = (SpoofCellwise)op;\n+ _cellType = _op.getCellType();\n}\n- protected void setOutput(ExecutionContext ec, Future<FederatedResponse>[] response, FederationMap fedMap) {\n- FType fedType = fedMap.getType();\n- AggOp aggOp = ((SpoofCellwise)_op).getAggOp();\n- CellType cellType = ((SpoofCellwise)_op).getCellType();\n- if(cellType == CellType.FULL_AGG) { // full aggregation\n- AggregateUnaryOperator aop = null;\n- if(aggOp == AggOp.SUM || aggOp == AggOp.SUM_SQ)\n- aop = InstructionUtils.parseBasicAggregateUnaryOperator(\"uak+\");\n- else if(aggOp == AggOp.MIN)\n- aop = InstructionUtils.parseBasicAggregateUnaryOperator(\"uamin\");\n- else if(aggOp == AggOp.MAX)\n- aop = InstructionUtils.parseBasicAggregateUnaryOperator(\"uamax\");\n- else\n- throw new DMLRuntimeException(\"Aggregation operation not supported yet.\");\n- ec.setVariable(_output.getName(), FederationUtils.aggScalar(aop, response));\n+ protected boolean isFedOutput() {\n+ boolean retVal = false;\n+ retVal |= (_cellType == CellType.ROW_AGG && _fedType == FType.ROW);\n+ retVal |= (_cellType == CellType.COL_AGG && _fedType == FType.COL);\n+ retVal |= (_cellType == CellType.NO_AGG);\n+ return retVal;\n}\n- else if(cellType == CellType.ROW_AGG) { // row aggregation\n- if(fedType == FType.ROW) {\n- // bind partial results from federated responses\n- ec.setMatrixOutput(_output.getName(), FederationUtils.bind(response, false));\n- }\n- else if(fedType == FType.COL) {\n- AggregateUnaryOperator aop = null;\n- if(aggOp == AggOp.SUM || aggOp == AggOp.SUM_SQ)\n- aop = InstructionUtils.parseBasicAggregateUnaryOperator(\"uark+\");\n- else if(aggOp == AggOp.MIN)\n- aop = InstructionUtils.parseBasicAggregateUnaryOperator(\"uarmin\");\n- else if(aggOp == AggOp.MAX)\n- aop = InstructionUtils.parseBasicAggregateUnaryOperator(\"uarmax\");\n- else\n- throw new DMLRuntimeException(\"Aggregation operation not supported yet.\");\n- ec.setMatrixOutput(_output.getName(), FederationUtils.aggMatrix(aop, response, fedMap));\n+\n+ protected void setFedOutput(ExecutionContext ec, FederationMap fedMap, long frComputeID) {\n+ // derive output federated mapping\n+ MatrixObject out = ec.getMatrixObject(_output);\n+ FederationMap newFedMap = modifyFedRanges(fedMap.copyWithNewID(frComputeID));\n+ out.setFedMapping(newFedMap);\n}\n- else {\n- throw new DMLRuntimeException(\"Aggregation type for federated spoof instructions not supported yet.\");\n- }\n- }\n- else if(cellType == CellType.COL_AGG) { // col aggregation\n- if(fedType == FType.ROW) {\n- AggregateUnaryOperator aop = null;\n- if(aggOp == AggOp.SUM || aggOp == AggOp.SUM_SQ)\n- aop = InstructionUtils.parseBasicAggregateUnaryOperator(\"uack+\");\n- else if(aggOp == AggOp.MIN)\n- aop = InstructionUtils.parseBasicAggregateUnaryOperator(\"uacmin\");\n- else if(aggOp == AggOp.MAX)\n- aop = InstructionUtils.parseBasicAggregateUnaryOperator(\"uacmax\");\n- else\n- throw new DMLRuntimeException(\"Aggregation operation not supported yet.\");\n- ec.setMatrixOutput(_output.getName(), FederationUtils.aggMatrix(aop, response, fedMap));\n+\n+ private FederationMap modifyFedRanges(FederationMap fedMap) {\n+ if(_cellType == CellType.ROW_AGG || _cellType == CellType.COL_AGG) {\n+ int dim = (_cellType == CellType.COL_AGG ? 0 : 1);\n+ // crop federation map to a vector\n+ IntStream.range(0, fedMap.getFederatedRanges().length).forEach(i -> {\n+ fedMap.getFederatedRanges()[i].setBeginDim(dim, 0);\n+ fedMap.getFederatedRanges()[i].setEndDim(dim, 1);\n+ });\n}\n- else if(fedType == FType.COL) {\n- // cbind partial results from federated responses\n- ec.setMatrixOutput(_output.getName(), FederationUtils.bind(response, true));\n+ return fedMap;\n}\n- else {\n- throw new DMLRuntimeException(\"Aggregation type for federated spoof instructions not supported yet.\");\n+\n+ protected void aggResult(ExecutionContext ec, Future<FederatedResponse>[] response,\n+ FederationMap fedMap) {\n+ AggOp aggOp = _op.getAggOp();\n+\n+ // build up the instruction for aggregation\n+ // (uak+/uamin/uamax/uark+/uarmin/uarmax/uack+/uacmin/uacmax)\n+ String aggInst = \"ua\";\n+ switch(_cellType) {\n+ case FULL_AGG: break;\n+ case ROW_AGG: aggInst += \"r\"; break;\n+ case COL_AGG: aggInst += \"c\"; break;\n+ case NO_AGG:\n+ default:\n+ throw new DMLRuntimeException(\"Aggregation type not supported yet.\");\n}\n+\n+ switch(aggOp) {\n+ case SUM:\n+ case SUM_SQ: aggInst += \"k+\"; break;\n+ case MIN: aggInst += \"min\"; break;\n+ case MAX: aggInst += \"max\"; break;\n+ default:\n+ throw new DMLRuntimeException(\"Aggregation operation not supported yet.\");\n}\n- else if(cellType == CellType.NO_AGG) { // no aggregation\n- if(fedType == FType.ROW) //rbind\n- ec.setMatrixOutput(_output.getName(), FederationUtils.bind(response, false));\n- else if(fedType == FType.COL) //cbind\n- ec.setMatrixOutput(_output.getName(), FederationUtils.bind(response, true));\n+\n+ AggregateUnaryOperator aop = InstructionUtils.parseBasicAggregateUnaryOperator(aggInst);\n+ if(_cellType == CellType.FULL_AGG)\n+ ec.setVariable(_output.getName(), FederationUtils.aggScalar(aop, response));\nelse\n- throw new DMLRuntimeException(\"Only row partitioned or column\" +\n- \" partitioned federated matrices supported yet.\");\n- }\n- else {\n- throw new DMLRuntimeException(\"Aggregation type not supported yet.\");\n- }\n+ ec.setMatrixOutput(_output.getName(), FederationUtils.aggMatrix(aop, response, fedMap));\n}\n}\n+ // ROWWISE TEMPLATE\nprivate static class SpoofFEDRowwise extends SpoofFEDType {\nprivate final SpoofRowwise _op;\n+ private final RowType _rowType;\n- SpoofFEDRowwise(SpoofOperator op, CPOperand out) {\n- super(out);\n+ SpoofFEDRowwise(SpoofOperator op, CPOperand out, FType fedType) {\n+ super(out, fedType);\n_op = (SpoofRowwise)op;\n+ _rowType = _op.getRowType();\n}\n- protected void setOutput(ExecutionContext ec, Future<FederatedResponse>[] response, FederationMap fedMap) {\n- RowType rowType = ((SpoofRowwise)_op).getRowType();\n- if(rowType == RowType.FULL_AGG) { // full aggregation\n- // aggregate partial results from federated responses as sum\n- AggregateUnaryOperator aop = InstructionUtils.parseBasicAggregateUnaryOperator(\"uak+\");\n- ec.setVariable(_output.getName(), FederationUtils.aggScalar(aop, response));\n- }\n- else if(rowType == RowType.ROW_AGG) { // row aggregation\n- // aggregate partial results from federated responses as rowSum\n- AggregateUnaryOperator aop = InstructionUtils.parseBasicAggregateUnaryOperator(\"uark+\");\n- ec.setMatrixOutput(_output.getName(), FederationUtils.aggMatrix(aop, response, fedMap));\n+ protected boolean isFedOutput() {\n+ boolean retVal = false;\n+ retVal |= (_rowType == RowType.NO_AGG);\n+ retVal |= (_rowType == RowType.NO_AGG_B1);\n+ retVal |= (_rowType == RowType.NO_AGG_CONST);\n+ retVal &= (_fedType == FType.ROW);\n+ return retVal;\n}\n- else if(rowType == RowType.COL_AGG\n- || rowType == RowType.COL_AGG_T\n- || rowType == RowType.COL_AGG_B1\n- || rowType == RowType.COL_AGG_B1_T\n- || rowType == RowType.COL_AGG_B1R\n- || rowType == RowType.COL_AGG_CONST) { // col aggregation\n- // aggregate partial results from federated responses as colSum\n- AggregateUnaryOperator aop = InstructionUtils.parseBasicAggregateUnaryOperator(\"uack+\");\n- ec.setMatrixOutput(_output.getName(), FederationUtils.aggMatrix(aop, response, fedMap));\n+\n+ protected void setFedOutput(ExecutionContext ec, FederationMap fedMap, long frComputeID) {\n+ // derive output federated mapping\n+ MatrixObject out = ec.getMatrixObject(_output);\n+ FederationMap newFedMap = modifyFedRanges(fedMap.copyWithNewID(frComputeID), out.getNumColumns());\n+ out.setFedMapping(newFedMap);\n}\n- else if(rowType == RowType.NO_AGG\n- || rowType == RowType.NO_AGG_B1\n- || rowType == RowType.NO_AGG_CONST) { // no aggregation\n- if(fedMap.getType() == FType.ROW) {\n- // bind partial results from federated responses\n- ec.setMatrixOutput(_output.getName(), FederationUtils.bind(response, false));\n+\n+ private static FederationMap modifyFedRanges(FederationMap fedMap, long cols) {\n+ IntStream.range(0, fedMap.getFederatedRanges().length).forEach(i -> {\n+ fedMap.getFederatedRanges()[i].setBeginDim(1, 0);\n+ fedMap.getFederatedRanges()[i].setEndDim(1, cols);\n+ });\n+ return fedMap;\n}\n- else {\n+\n+ protected void aggResult(ExecutionContext ec, Future<FederatedResponse>[] response,\n+ FederationMap fedMap) {\n+ if(_fedType != FType.ROW)\nthrow new DMLRuntimeException(\"Only row partitioned federated matrices supported yet.\");\n- }\n- }\n- else {\n+\n+ // build up the instruction for aggregation (uak+/uark+/uack+)\n+ String aggInst = \"ua\";\n+ if(_rowType == RowType.FULL_AGG) // full aggregation\n+ aggInst += \"k+\";\n+ else if(_rowType == RowType.ROW_AGG) // row aggregation\n+ aggInst += \"rk+\";\n+ else if(_rowType.isColumnAgg()) // col aggregation\n+ aggInst += \"ck+\";\n+ else\nthrow new DMLRuntimeException(\"AggregationType not supported yet.\");\n- }\n+\n+ // aggregate partial results from federated responses as sum/rowSum/colSum\n+ AggregateUnaryOperator aop = InstructionUtils.parseBasicAggregateUnaryOperator(aggInst);\n+ if(_rowType == RowType.FULL_AGG)\n+ ec.setVariable(_output.getName(), FederationUtils.aggScalar(aop, response));\n+ else\n+ ec.setMatrixOutput(_output.getName(), FederationUtils.aggMatrix(aop, response, fedMap));\n}\n}\n+ // MULTIAGGREGATE TEMPLATE\nprivate static class SpoofFEDMultiAgg extends SpoofFEDType {\nprivate final SpoofMultiAggregate _op;\n- SpoofFEDMultiAgg(SpoofOperator op, CPOperand out) {\n- super(out);\n+ SpoofFEDMultiAgg(SpoofOperator op, CPOperand out, FType fedType) {\n+ super(out, fedType);\n_op = (SpoofMultiAggregate)op;\n}\n- protected void setOutput(ExecutionContext ec, Future<FederatedResponse>[] response, FederationMap fedMap) {\n+ protected boolean isFedOutput() {\n+ return false;\n+ }\n+\n+ protected void setFedOutput(ExecutionContext ec, FederationMap fedMap, long frComputeID) {\n+ throw new DMLRuntimeException(\"SpoofFEDMultiAgg cannot create a federated output.\");\n+ }\n+\n+ protected void aggResult(ExecutionContext ec, Future<FederatedResponse>[] response,\n+ FederationMap fedMap) {\nMatrixBlock[] partRes = FederationUtils.getResults(response);\n- SpoofCellwise.AggOp[] aggOps = ((SpoofMultiAggregate)_op).getAggOps();\n+ SpoofCellwise.AggOp[] aggOps = _op.getAggOps();\nfor(int counter = 1; counter < partRes.length; counter++) {\nSpoofMultiAggregate.aggregatePartialResults(aggOps, partRes[0], partRes[counter]);\n}\n@@ -343,24 +398,31 @@ public class SpoofFEDInstruction extends FEDInstruction\n}\n}\n-\n+ // OUTER PRODUCT TEMPLATE\nprivate static class SpoofFEDOuterProduct extends SpoofFEDType {\nprivate final SpoofOuterProduct _op;\n+ private final OutProdType _outProdType;\n+ private CPOperand[] _inputs;\n- SpoofFEDOuterProduct(SpoofOperator op, CPOperand out) {\n- super(out);\n+ SpoofFEDOuterProduct(SpoofOperator op, CPOperand out, FType fedType, CPOperand[] inputs) {\n+ super(out, fedType);\n_op = (SpoofOuterProduct)op;\n+ _outProdType = _op.getOuterProdType();\n+ _inputs = inputs;\n+ }\n+\n+ protected FederatedRequest[] broadcastSliced(MatrixObject mo, FederationMap fedMap) {\n+ return fedMap.broadcastSliced(mo, (_fedType == FType.COL));\n}\nprotected boolean needsBroadcastSliced(FederationMap fedMap, long rowNum, long colNum, int inputIndex) {\nboolean retVal = false;\n- FType fedType = fedMap.getType();\nretVal |= (rowNum == fedMap.getMaxIndexInRange(0) && colNum == fedMap.getMaxIndexInRange(1));\n- if(fedType == FType.ROW)\n+ if(_fedType == FType.ROW)\nretVal |= (rowNum == fedMap.getMaxIndexInRange(0)) && (inputIndex != 2); // input at index 2 is V\n- else if(fedType == FType.COL)\n+ else if(_fedType == FType.COL)\nretVal |= (rowNum == fedMap.getMaxIndexInRange(1)) && (inputIndex != 1); // input at index 1 is U\nelse\nthrow new DMLRuntimeException(\"Only row partitioned or column\" +\n@@ -369,60 +431,68 @@ public class SpoofFEDInstruction extends FEDInstruction\nreturn retVal;\n}\n- protected void setOutput(ExecutionContext ec, Future<FederatedResponse>[] response, FederationMap fedMap) {\n- FType fedType = fedMap.getType();\n- OutProdType outProdType = ((SpoofOuterProduct)_op).getOuterProdType();\n- if(outProdType == OutProdType.LEFT_OUTER_PRODUCT) {\n- if(fedType == FType.ROW) {\n- // aggregate partial results from federated responses as elementwise sum\n- AggregateUnaryOperator aop = InstructionUtils.parseBasicAggregateUnaryOperator(\"uak+\");\n- ec.setMatrixOutput(_output.getName(), FederationUtils.aggMatrix(aop, response, fedMap));\n- }\n- else if(fedType == FType.COL) {\n- // bind partial results from federated responses\n- ec.setMatrixOutput(_output.getName(), FederationUtils.bind(response, false));\n+ protected boolean isFedOutput() {\n+ boolean retVal = false;\n+ retVal |= (_outProdType == OutProdType.LEFT_OUTER_PRODUCT && _fedType == FType.COL);\n+ retVal |= (_outProdType == OutProdType.RIGHT_OUTER_PRODUCT && _fedType == FType.ROW);\n+ retVal |= (_outProdType == OutProdType.CELLWISE_OUTER_PRODUCT);\n+ return retVal;\n}\n- else {\n- throw new DMLRuntimeException(\"Only row partitioned or column\" +\n- \" partitioned federated matrices supported yet.\");\n+\n+ protected void setFedOutput(ExecutionContext ec, FederationMap fedMap, long frComputeID) {\n+ FederationMap newFedMap = fedMap.copyWithNewID(frComputeID);\n+ long[] outDims = new long[2];\n+\n+ // find the resulting output dimensions\n+ MatrixObject X = ec.getMatrixObject(_inputs[0]);\n+ switch(_outProdType) {\n+ case LEFT_OUTER_PRODUCT: // LEFT: nrows of transposed X, ncols of U\n+ newFedMap = newFedMap.transpose();\n+ outDims[0] = X.getNumColumns();\n+ outDims[1] = ec.getMatrixObject(_inputs[1]).getNumColumns();\n+ break;\n+ case RIGHT_OUTER_PRODUCT: // RIGHT: nrows of X, ncols of V\n+ outDims[0] = X.getNumRows();\n+ outDims[1] = ec.getMatrixObject(_inputs[2]).getNumColumns();\n+ break;\n+ case CELLWISE_OUTER_PRODUCT: // BASIC: preserve dimensions of X\n+ outDims[0] = X.getNumRows();\n+ outDims[1] = X.getNumColumns();\n+ break;\n+ default:\n+ throw new DMLRuntimeException(\"Outer Product Type \" + _outProdType + \" not supported yet.\");\n}\n+\n+ // derive output federated mapping\n+ MatrixObject out = ec.getMatrixObject(_output);\n+ int dim = (newFedMap.getType() == FType.ROW ? 1 : 0);\n+ newFedMap = modifyFedRanges(newFedMap, dim, outDims[dim]);\n+ out.setFedMapping(newFedMap);\n}\n- else if(outProdType == OutProdType.RIGHT_OUTER_PRODUCT) {\n- if(fedType == FType.ROW) {\n- // bind partial results from federated responses\n- ec.setMatrixOutput(_output.getName(), FederationUtils.bind(response, false));\n+\n+ private static FederationMap modifyFedRanges(FederationMap fedMap, int dim, long value) {\n+ IntStream.range(0, fedMap.getFederatedRanges().length).forEach(i -> {\n+ fedMap.getFederatedRanges()[i].setBeginDim(dim, 0);\n+ fedMap.getFederatedRanges()[i].setEndDim(dim, value);\n+ });\n+ return fedMap;\n}\n- else if(fedType == FType.COL) {\n- // aggregate partial results from federated responses as elementwise sum\n+\n+ protected void aggResult(ExecutionContext ec, Future<FederatedResponse>[] response,\n+ FederationMap fedMap) {\nAggregateUnaryOperator aop = InstructionUtils.parseBasicAggregateUnaryOperator(\"uak+\");\n+ switch(_outProdType) {\n+ case LEFT_OUTER_PRODUCT:\n+ case RIGHT_OUTER_PRODUCT:\n+ // aggregate partial results from federated responses as elementwise sum\nec.setMatrixOutput(_output.getName(), FederationUtils.aggMatrix(aop, response, fedMap));\n- }\n- else {\n- throw new DMLRuntimeException(\"Only row partitioned or column\" +\n- \" partitioned federated matrices supported yet.\");\n- }\n- }\n- else if(outProdType == OutProdType.CELLWISE_OUTER_PRODUCT) {\n- if(fedType == FType.ROW) {\n- // rbind partial results from federated responses\n- ec.setMatrixOutput(_output.getName(), FederationUtils.bind(response, false));\n- }\n- else if(fedType == FType.COL) {\n- // cbind partial results from federated responses\n- ec.setMatrixOutput(_output.getName(), FederationUtils.bind(response, true));\n- }\n- else {\n- throw new DMLRuntimeException(\"Only row partitioned or column\" +\n- \" partitioned federated matrices supported yet.\");\n- }\n- }\n- else if(outProdType == OutProdType.AGG_OUTER_PRODUCT) {\n+ break;\n+ case AGG_OUTER_PRODUCT:\n// aggregate partial results from federated responses as sum\n- AggregateUnaryOperator aop = InstructionUtils.parseBasicAggregateUnaryOperator(\"uak+\");\nec.setVariable(_output.getName(), FederationUtils.aggScalar(aop, response));\n- }\n- else {\n- throw new DMLRuntimeException(\"Outer Product Type \" + outProdType + \" not supported yet.\");\n+ break;\n+ default:\n+ throw new DMLRuntimeException(\"Outer Product Type \" + _outProdType + \" not supported yet.\");\n}\n}\n}\n@@ -458,5 +528,4 @@ public class SpoofFEDInstruction extends FEDInstruction\n}\nreturn retVal;\n}\n-\n}\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/federated/codegen/FederatedCodegenMultipleFedMOTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/federated/codegen/FederatedCodegenMultipleFedMOTest.java", "diff": "@@ -104,7 +104,7 @@ public class FederatedCodegenMultipleFedMOTest extends AutomatedTestBase\n// row partitioned\n// {201, 6, 4, 6, 4, true},\n{202, 6, 4, 6, 4, true},\n- // {203, 20, 1, 20, 1, true},\n+ // FIXME: [SYSTEMDS-3110] {203, 20, 1, 20, 1, true},\n// col partitioned\n{201, 6, 4, 6, 4, false},\n{202, 6, 4, 6, 4, false},\n@@ -123,9 +123,9 @@ public class FederatedCodegenMultipleFedMOTest extends AutomatedTestBase\n{308, 1000, 2000, 10, 2000, false},\n// {310, 1000, 2000, 10, 2000, false},\n// row and col partitioned\n- // {311, 1000, 2000, 1000, 10, true}, // not working yet - ArrayIndexOutOfBoundsException in dotProduct\n+ // {311, 1000, 2000, 1000, 10, true}, // FIXME: ArrayIndexOutOfBoundsException in dotProduct\n{312, 1000, 2000, 10, 2000, false},\n- // {313, 4000, 2000, 4000, 10, true}, // not working yet - ArrayIndexOutOfBoundsException in dotProduct\n+ // {313, 4000, 2000, 4000, 10, true}, // FIXME: ArrayIndexOutOfBoundsException in dotProduct\n{314, 4000, 2000, 10, 2000, false},\n// combined tests\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/federated/codegen/FederatedOuterProductTmplTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/federated/codegen/FederatedOuterProductTmplTest.java", "diff": "@@ -86,14 +86,14 @@ public class FederatedOuterProductTmplTest extends AutomatedTestBase\n{9, 1000, 2000, true},\n// column partitioned\n- //FIXME {1, 2000, 2000, false},\n+ {1, 2000, 2000, false},\n// {2, 4000, 2000, false},\n// {3, 1000, 1000, false},\n- //FIXME {4, 4000, 2000, false},\n- //FIXME {5, 4000, 2000, false},\n+ {4, 4000, 2000, false},\n+ {5, 4000, 2000, false},\n// {6, 4000, 2000, false},\n//FIXME {7, 2000, 2000, false},\n- //FIXME {8, 1000, 2000, false},\n+ {8, 1000, 2000, false},\n// {9, 1000, 2000, false},\n});\n}\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/federated/codegen/FederatedRowwiseTmplTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/federated/codegen/FederatedRowwiseTmplTest.java", "diff": "@@ -117,7 +117,7 @@ public class FederatedRowwiseTmplTest extends AutomatedTestBase\n}\n@Test\n- public void federatedCodegenCellwiseHybrid() {\n+ public void federatedCodegenRowwiseHybrid() {\ntestFederatedCodegenRowwise(ExecMode.HYBRID);\n}\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/pipelines/BuiltinTopkEvaluateTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/pipelines/BuiltinTopkEvaluateTest.java", "diff": "@@ -25,7 +25,6 @@ import org.apache.sysds.test.TestConfiguration;\nimport org.apache.sysds.test.TestUtils;\nimport org.junit.Assert;\nimport org.junit.Ignore;\n-import org.junit.Test;\npublic class BuiltinTopkEvaluateTest extends AutomatedTestBase {\n// private final static String TEST_NAME1 = \"prioritized\";\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/instructions/fed/QuaternaryWDivMMFEDInstruction.java", "new_path": "src/main/java/org/apache/sysds/runtime/instructions/fed/QuaternaryWDivMMFEDInstruction.java", "diff": "@@ -37,11 +37,11 @@ import org.apache.sysds.runtime.DMLRuntimeException;\nimport org.apache.sysds.runtime.instructions.cp.CPOperand;\nimport org.apache.sysds.runtime.instructions.cp.DoubleObject;\nimport org.apache.sysds.runtime.instructions.cp.ScalarObject;\n-import org.apache.sysds.runtime.matrix.operators.Operator;\nimport org.apache.sysds.runtime.matrix.operators.QuaternaryOperator;\nimport java.util.ArrayList;\nimport java.util.concurrent.Future;\n+import java.util.stream.IntStream;\npublic class QuaternaryWDivMMFEDInstruction extends QuaternaryFEDInstruction\n{\n@@ -61,31 +61,34 @@ public class QuaternaryWDivMMFEDInstruction extends QuaternaryFEDInstruction\n* @param opcode ...\n* @param instruction_str ...\n*/\n- protected QuaternaryWDivMMFEDInstruction(Operator operator,\n+\n+ private QuaternaryOperator _qop;\n+\n+ protected QuaternaryWDivMMFEDInstruction(QuaternaryOperator operator,\nCPOperand in1, CPOperand in2, CPOperand in3, CPOperand in4, CPOperand out, String opcode, String instruction_str)\n{\nsuper(FEDType.Quaternary, operator, in1, in2, in3, in4, out, opcode, instruction_str);\n+ _qop = operator;\n}\n@Override\npublic void processInstruction(ExecutionContext ec)\n{\n- QuaternaryOperator qop = (QuaternaryOperator) _optr;\n- final WDivMMType wdivmm_type = qop.wtype3;\n+ final WDivMMType wdivmm_type = _qop.wtype3;\nMatrixObject X = ec.getMatrixObject(input1);\nMatrixObject U = ec.getMatrixObject(input2);\nMatrixObject V = ec.getMatrixObject(input3);\nScalarObject eps = null;\nMatrixObject MX = null;\n- if(qop.hasFourInputs()) {\n+ if(_qop.hasFourInputs()) {\nif(wdivmm_type == WDivMMType.MULT_MINUS_4_LEFT || wdivmm_type == WDivMMType.MULT_MINUS_4_RIGHT) {\nMX = ec.getMatrixObject(_input4);\n}\nelse {\neps = (_input4.getDataType() == DataType.SCALAR) ?\nec.getScalarInput(_input4) :\n- new DoubleObject(ec.getMatrixInput(_input4.getName()).quickGetValue(0, 0));\n+ new DoubleObject(ec.getMatrixInput(_input4).quickGetValue(0, 0));\n}\n}\n@@ -93,7 +96,7 @@ public class QuaternaryWDivMMFEDInstruction extends QuaternaryFEDInstruction\nFederationMap fedMap = X.getFedMapping();\nArrayList<FederatedRequest[]> frSliced = new ArrayList<>();\nArrayList<FederatedRequest> frB = new ArrayList<>(); // FederatedRequests of broadcasts\n- long[] varNewIn = new long[qop.hasFourInputs() ? 4 : 3];\n+ long[] varNewIn = new long[_qop.hasFourInputs() ? 4 : 3];\nvarNewIn[0] = fedMap.getID();\nif(X.isFederated(FType.ROW)) { // row partitioned X\n@@ -151,18 +154,23 @@ public class QuaternaryWDivMMFEDInstruction extends QuaternaryFEDInstruction\n}\nFederatedRequest frComp = FederationUtils.callInstruction(instString, output,\n- qop.hasFourInputs() ? new CPOperand[]{input1, input2, input3, _input4}\n+ _qop.hasFourInputs() ? new CPOperand[]{input1, input2, input3, _input4}\n: new CPOperand[]{input1, input2, input3}, varNewIn);\n- // get partial results from federated workers\n- FederatedRequest frGet = new FederatedRequest(RequestType.GET_VAR, frComp.getID());\n+ FederatedRequest frGet = null;\n- ArrayList<FederatedRequest> frC = new ArrayList<>();\n- frC.add(fedMap.cleanup(getTID(), frComp.getID()));\n+ FederatedRequest frC = null;\n+ if((wdivmm_type.isLeft() && X.isFederated(FType.ROW))\n+ || (wdivmm_type.isRight() && X.isFederated(FType.COL))) { // output needs local aggregation\n+ // get partial results from federated workers\n+ frGet = new FederatedRequest(RequestType.GET_VAR, frComp.getID());\n+ // cleanup the federated request of the instruction call\n+ frC = fedMap.cleanup(getTID(), frComp.getID());\n+ }\n- FederatedRequest[] frAll = ArrayUtils.addAll(ArrayUtils.addAll(\n- frB.toArray(new FederatedRequest[0]), frComp, frGet),\n- frC.toArray(new FederatedRequest[0]));\n+ FederatedRequest[] frAll = (frGet == null ?\n+ ArrayUtils.addAll(frB.toArray(new FederatedRequest[0]), frComp)\n+ : ArrayUtils.addAll(frB.toArray(new FederatedRequest[0]), frComp, frGet, frC));\n// execute federated instructions\nFuture<FederatedResponse>[] response = frSliced.isEmpty() ?\n@@ -170,14 +178,13 @@ public class QuaternaryWDivMMFEDInstruction extends QuaternaryFEDInstruction\ngetTID(), true, frSliced.toArray(new FederatedRequest[0][]), frAll);\nif((wdivmm_type.isLeft() && X.isFederated(FType.ROW))\n- || (wdivmm_type.isRight() && X.isFederated(FType.COL))) {\n+ || (wdivmm_type.isRight() && X.isFederated(FType.COL))) { // local aggregation\n// aggregate partial results from federated responses\nAggregateUnaryOperator aop = InstructionUtils.parseBasicAggregateUnaryOperator(\"uak+\");\nec.setMatrixOutput(output.getName(), FederationUtils.aggMatrix(aop, response, fedMap));\n}\nelse if(wdivmm_type.isLeft() || wdivmm_type.isRight() || wdivmm_type.isBasic()) {\n- // bind partial results from federated responses\n- ec.setMatrixOutput(output.getName(), FederationUtils.bind(response, false));\n+ setFederatedOutput(X, U, V, ec, frComp.getID());\n}\nelse {\nthrow new DMLRuntimeException(\"Federated WDivMM only supported for BASIC, LEFT or RIGHT variants.\");\n@@ -188,5 +195,53 @@ public class QuaternaryWDivMMFEDInstruction extends QuaternaryFEDInstruction\n+ X.isFederated() + \", \" + U.isFederated() + \", \" + V.isFederated() + \")\");\n}\n}\n+\n+ /**\n+ * Set the federated output according to the output data charactersitics of\n+ * the different wdivmm types\n+ */\n+ private void setFederatedOutput(MatrixObject X, MatrixObject U, MatrixObject V, ExecutionContext ec, long fedMapID) {\n+ final WDivMMType wdivmm_type = _qop.wtype3;\n+ MatrixObject out = ec.getMatrixObject(output);\n+ FederationMap outFedMap = X.getFedMapping().copyWithNewID(fedMapID);\n+\n+ long rows = -1;\n+ long cols = -1;\n+ if(wdivmm_type.isBasic()) {\n+ // BASIC: preserve dimensions of X\n+ rows = X.getNumRows();\n+ cols = X.getNumColumns();\n+ }\n+ else if(wdivmm_type.isLeft()) {\n+ // LEFT: nrows of transposed X, ncols of U\n+ rows = X.getNumColumns();\n+ cols = U.getNumColumns();\n+ outFedMap = modifyFedRanges(outFedMap.transpose(), cols, 1);\n+ }\n+ else if(wdivmm_type.isRight()) {\n+ // RIGHT: nrows of X, ncols of V\n+ rows = X.getNumRows();\n+ cols = V.getNumColumns();\n+ outFedMap = modifyFedRanges(outFedMap, cols, 1);\n+ }\n+ out.setFedMapping(outFedMap);\n+ out.getDataCharacteristics().set(rows, cols, (int) X.getBlocksize());\n}\n+ /**\n+ * Takes the federated mapping and sets one dimension of all federated ranges\n+ * to the specified value.\n+ *\n+ * @param fedMap the original federated mapping\n+ * @param value long value for setting the dimension\n+ * @param dim indicates if the row (0) or column (1) dimension should be set to value\n+ * @return FederationMap with the modified federated ranges\n+ */\n+ private static FederationMap modifyFedRanges(FederationMap fedMap, long value, int dim) {\n+ IntStream.range(0, fedMap.getFederatedRanges().length).forEach(i -> {\n+ fedMap.getFederatedRanges()[i].setBeginDim(dim, 0);\n+ fedMap.getFederatedRanges()[i].setEndDim(dim, value);\n+ });\n+ return fedMap;\n+ }\n+}\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/controlprogram/context/SparkExecutionContext.java", "new_path": "src/main/java/org/apache/sysds/runtime/controlprogram/context/SparkExecutionContext.java", "diff": "@@ -607,6 +607,7 @@ public class SparkExecutionContext extends ExecutionContext\nbrBlock = cd.getBroadcastHandle().getNonPartitionedBroadcast();\n}\n+ //TODO: synchronize\nif (brBlock == null) {\n//create new broadcast handle (never created, evicted)\n// account for overwritten invalid broadcast (e.g., evicted)\n@@ -651,6 +652,7 @@ public class SparkExecutionContext extends ExecutionContext\nPartitionedBroadcast<MatrixBlock> bret = null;\n+ synchronized (mo) { //synchronize with the async. broadcast thread\n//reuse existing broadcast handle\nif (mo.getBroadcastHandle() != null && mo.getBroadcastHandle().isPartitionedBroadcastValid()) {\nbret = mo.getBroadcastHandle().getPartitionedBroadcast();\n@@ -692,13 +694,13 @@ public class SparkExecutionContext extends ExecutionContext\nmo.getBroadcastHandle().setPartitionedBroadcast(bret,\nOptimizerUtils.estimatePartitionedSizeExactSparsity(mo.getDataCharacteristics()));\nCacheableData.addBroadcastSize(mo.getBroadcastHandle().getSize());\n- }\nif (DMLScript.STATISTICS) {\nStatistics.accSparkBroadCastTime(System.nanoTime() - t0);\nStatistics.incSparkBroadcastCount(1);\n}\n-\n+ }\n+ }\nreturn bret;\n}\n@@ -753,13 +755,12 @@ public class SparkExecutionContext extends ExecutionContext\nto.getBroadcastHandle().setPartitionedBroadcast(bret,\nOptimizerUtils.estimatePartitionedSizeExactSparsity(to.getDataCharacteristics()));\nCacheableData.addBroadcastSize(to.getBroadcastHandle().getSize());\n- }\nif (DMLScript.STATISTICS) {\nStatistics.accSparkBroadCastTime(System.nanoTime() - t0);\nStatistics.incSparkBroadcastCount(1);\n}\n-\n+ }\nreturn bret;\n}\n@@ -820,13 +821,12 @@ public class SparkExecutionContext extends ExecutionContext\nfo.getBroadcastHandle().setPartitionedBroadcast(bret,\nOptimizerUtils.estimatePartitionedSizeExactSparsity(fo.getDataCharacteristics()));\nCacheableData.addBroadcastSize(fo.getBroadcastHandle().getSize());\n- }\nif (DMLScript.STATISTICS) {\nStatistics.accSparkBroadCastTime(System.nanoTime() - t0);\nStatistics.incSparkBroadcastCount(1);\n}\n-\n+ }\nreturn bret;\n}\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/instructions/cp/TriggerBroadcastTask.java", "new_path": "src/main/java/org/apache/sysds/runtime/instructions/cp/TriggerBroadcastTask.java", "diff": "@@ -36,9 +36,6 @@ public class TriggerBroadcastTask implements Runnable {\n@Override\npublic void run() {\n- // TODO: Synchronization. Although it is harmless if to threads create separate\n- // broadcast handles as only one will stay with the MatrixObject. However, redundant\n- // partitioning increases untraced memory usage.\ntry {\nSparkExecutionContext sec = (SparkExecutionContext)_ec;\nsec.setBroadcastHandle(_broadcastMO);\n@@ -47,6 +44,7 @@ public class TriggerBroadcastTask implements Runnable {\ne.printStackTrace();\n}\n+ //TODO: Count only if successful (owned lock)\nif (DMLScript.STATISTICS)\nStatistics.incSparkAsyncBroadcastCount(1);\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/utils/Statistics.java", "new_path": "src/main/java/org/apache/sysds/utils/Statistics.java", "diff": "@@ -511,6 +511,8 @@ public class Statistics\nparforMergeTime = 0;\nsparkCtxCreateTime = 0;\n+ sparkBroadcast.reset();\n+ sparkBroadcastCount.reset();\nsparkAsyncPrefetchCount.reset();\nsparkAsyncBroadcastCount.reset();\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/matrix/data/MatrixBlock.java", "new_path": "src/main/java/org/apache/sysds/runtime/matrix/data/MatrixBlock.java", "diff": "@@ -1191,7 +1191,7 @@ public class MatrixBlock extends MatrixValue implements CacheBlock, Externalizab\ndenseToSparse(true);\n}\n- private void denseToSparse(boolean allowCSR)\n+ public void denseToSparse(boolean allowCSR)\n{\nDenseBlock a = getDenseBlock();\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/component/matrix/CumsumprodTest.java", "new_path": "src/test/java/org/apache/sysds/test/component/matrix/CumsumprodTest.java", "diff": "@@ -39,6 +39,8 @@ public class CumsumprodTest {\n@Test\npublic void testCumsumprodSparseMCSR() {\nMatrixBlock A = MatrixBlock.randOperations(1000, 2, 0.05, 0, 10, \"uniform\", 7);\n+ if( !A.isInSparseFormat() )\n+ A.denseToSparse(false);\nA = new MatrixBlock(A, SparseBlock.Type.MCSR, true);\nUnaryOperator uop = new UnaryOperator(Builtin.getBuiltinFnObject(\"ucumk+*\"), 1, false);\nMatrixBlock B = A.unaryOperations(uop, new MatrixBlock());\n@@ -48,6 +50,8 @@ public class CumsumprodTest {\n@Test\npublic void testCumsumprodSparseCSR() {\nMatrixBlock A = MatrixBlock.randOperations(1000, 2, 0.05, 0, 10, \"uniform\", 7);\n+ if( !A.isInSparseFormat() )\n+ A.denseToSparse(false);\nA = new MatrixBlock(A, SparseBlock.Type.CSR, true);\nUnaryOperator uop = new UnaryOperator(Builtin.getBuiltinFnObject(\"ucumk+*\"), 1, false);\nMatrixBlock B = A.unaryOperations(uop, new MatrixBlock());\n@@ -57,6 +61,8 @@ public class CumsumprodTest {\n@Test\npublic void testCumsumprodSparseCOO() {\nMatrixBlock A = MatrixBlock.randOperations(1000, 2, 0.05, 0, 10, \"uniform\", 7);\n+ if( !A.isInSparseFormat() )\n+ A.denseToSparse(false);\nA = new MatrixBlock(A, SparseBlock.Type.COO, true);\nUnaryOperator uop = new UnaryOperator(Builtin.getBuiltinFnObject(\"ucumk+*\"), 1, false);\nMatrixBlock B = A.unaryOperations(uop, new MatrixBlock());\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/controlprogram/context/SparkExecutionContext.java", "new_path": "src/main/java/org/apache/sysds/runtime/controlprogram/context/SparkExecutionContext.java", "diff": "@@ -224,6 +224,9 @@ public class SparkExecutionContext extends ExecutionContext\n_spctx = createContext(conf);\n+ if(DMLScript.USE_LOCAL_SPARK_CONFIG)\n+ _spctx.setCheckpointDir(\"/tmp/systemds_spark_cache_\" + DMLScript.getUUID());\n+\n_parRDDs.clear();\n}\n" } ]

[ { "change_type": "MODIFY", "old_path": "scripts/builtin/executePipeline.dml", "new_path": "scripts/builtin/executePipeline.dml", "diff": "@@ -199,7 +199,7 @@ return (Matrix[Double] X)\nnX = nX[, 1: ncol(nX) - 1]\n}\n- if(dataFlag == 0 & (sum(mask) > 0))\n+ if(dataFlag == 0 & (sum(mask) > 0) & (sum(mask) != ncol(nX)))\n{\nmaxDummy = max(nX) + 1\nnX = replace(target = nX, pattern = NaN, replacement = maxDummy)\n@@ -219,7 +219,7 @@ return (Matrix[Double] X)\nX = replace(target = X, pattern = maxDummy, replacement = NaN)\nX = replace(target = X, pattern = -1111, replacement = NaN)\n}\n- else if(dataFlag == 1 & (sum(mask) > 0))\n+ else if(dataFlag == 1 & (sum(mask) > 0) & (sum(mask) != ncol(nX)))\n{\nmaxDummy = max(nX) + 1\nnX = replace(target = nX, pattern = NaN, replacement = maxDummy)\n@@ -287,7 +287,8 @@ return (Matrix[Double] X_filled)\nfor(i in 1: nrow(FD))\n{\nfor(j in 1:ncol(FD)) {\n- if(as.scalar(FD[i, j]) > 0 & (min(X[, i]) != 0) & (min(X[, j]) != 0) & (sum(FD[, j]) != nrow(FD)))\n+ if(as.scalar(FD[i, j]) > 0 & (min(X[, i]) != 0) & (min(X[, j]) != 0) & (sum(FD[, j]) != nrow(FD))\n+ & (as.scalar(fdMask[1, j]) != 0) & (as.scalar(fdMask[1, i]) != 0))\nX = imputeByFD(X, i, j, threshold, FALSE)\n}\n}\n@@ -364,7 +365,7 @@ fillDefault = function(Matrix[Double] X)\nreturn(Matrix[Double] X){\ndefaullt = round(colMaxs(X) - colMins(X))\nMask = is.na(X)\n- X = replace(target=X, pattern=NaN, replacement=max(X))\n+ X = replace(target=X, pattern=NaN, replacement=0)\nMask = Mask * defaullt\nX = X + Mask\n# print(\"fillDefault: no of NaNs \"+sum(is.na(X)))\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/pipelines/BuiltinTopkCleaningClassificationTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/pipelines/BuiltinTopkCleaningClassificationTest.java", "diff": "@@ -45,6 +45,7 @@ public class BuiltinTopkCleaningClassificationTest extends AutomatedTestBase {\naddTestConfiguration(TEST_NAME, new TestConfiguration(TEST_CLASS_DIR, TEST_NAME, new String[]{\"R\"}));\n}\n+ @Test\npublic void testFindBestPipelineCompany() {\nruntopkCleaning(DATA_DIR+ \"company.csv\", RESOURCE+ \"meta/meta_company.csv\", 1.0, 3,5,\n\"FALSE\", 0,0.8, Types.ExecMode.SINGLE_NODE);\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/pipelines/BuiltinTopkLogicalTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/pipelines/BuiltinTopkLogicalTest.java", "diff": "@@ -50,7 +50,7 @@ public class BuiltinTopkLogicalTest extends AutomatedTestBase {\n@Test\npublic void testLogical1() {\n- runTestLogical(10, 5, 2, ExecMode.SINGLE_NODE);\n+ runTestLogical(4, 5, 2, ExecMode.SINGLE_NODE);\n}\n@Test\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/bestAcc.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/bestAcc.csv", "diff": "-93.69369369369369\n-93.69369369369369\n-93.69369369369369\n+64.88439306358381\n+64.73988439306359\n+64.73988439306359\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/dirtyScore.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/dirtyScore.csv", "diff": "-90.09009009009009\n\\ No newline at end of file\n+63.72832369942196\n\\ No newline at end of file\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/evalHp.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/evalHp.csv", "diff": "-2.0,0.001,1.0,1000.0\n+0,1.0,0.001,10.0\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/hp.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/hp.csv", "diff": "-32.0,2.0,0.0203644573130835,0.9538010240498609,0,0,0,1.0,0,0,0,0,1.0,0,0,0,2.0,1.0,0.6367394902267174,0,0,0,1.0,1.0,2.0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n-32.0,2.0,0.04436413689764156,0.9601592761408282,0,0,0,1.0,0,0,0,0,1.0,0,0,0,2.0,1.0,0.6541009026313958,0,0,0,1.0,1.0,2.0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n-32.0,3.0,0.0418452608516319,0.9715979748926613,1.0,0,0,1.0,1.0,0,0,0,1.0,0,1.0,0,2.0,0,2.0,1.0,0.6003640116471959,0,1.0,0,2.0,1.0,1.0,2.0,1.0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n+54.0,3.0,7.0,2.0,1.0,0,0,0,1.0,0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,1.0,0,0,0,2.0,2.0,0,1.0,0,0,0,0,0,0,1.0,0.7053074081820746,0,0,0,0,1.0,1.0,2.0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n+54.0,3.0,6.0,2.0,1.0,0,0,0,1.0,0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,1.0,0,0,0,2.0,2.0,1.0,1.0,0,0,0,0,0,0,1.0,0.7784943734333777,0,0,0,0,1.0,1.0,2.0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n+54.0,3.0,4.0,1.0,1.0,0,0,0,1.0,0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,1.0,0,0,0,2.0,2.0,1.0,1.0,0,0,0,0,0,0,1.0,0.7567266322372848,0,0,0,0,1.0,1.0,2.0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/lp.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/lp.csv", "diff": "-OTLR,MVI,CI,DUMMY\n+ED,MVI,EC,SCALE,CI,DUMMY\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/pip.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/pip.csv", "diff": "-winsorize,imputeByMedian,wtomeklink,dummycoding\n-winsorize,imputeByMedian,wtomeklink,dummycoding\n-outlierBySd,imputeByMean,abstain,dummycoding\n+outlierBySd,imputeByMedian,imputeByMean,scale,abstain,dummycoding\n+outlierBySd,imputeByMean,imputeByMedian,scale,abstain,dummycoding\n+outlierBySd,imputeByMean,imputeByMedian,scale,abstain,dummycoding\n" } ]

[ { "change_type": "MODIFY", "old_path": "scripts/builtin/applyAndEvaluate.dml", "new_path": "scripts/builtin/applyAndEvaluate.dml", "diff": "@@ -136,7 +136,13 @@ getDirtyScore = function(Frame[Unknown] X, Matrix[Double] Y, Frame[Unknown] Xtes\nMatrix[Double] evalFunHp)\nreturn(Double dirtyScore)\n{\n+ dschema = detectSchema(X)\n+ dmask = matrix(0, rows=1, cols=ncol(dschema))\n+ for(i in 1:ncol(dschema))\n+ if(as.scalar(dschema[1, i]) == \"STRING\" | as.scalar(dschema[1, i]) == \"BOOLEAN\")\n+ dmask[1, i] = 1\nmask = as.matrix(metaList['mask'])\n+ mask = ifelse(sum(mask == dmask) < ncol(mask), dmask, mask)\n[eXtrain, eXtest] = recodeData(X, Xtest, mask, FALSE, \"recode\")\neXtrain = replace(target=eXtrain, pattern=NaN, replacement=1)\neXtest = replace(target=eXtest, pattern=NaN, replacement=1)\n" }, { "change_type": "MODIFY", "old_path": "scripts/builtin/topk_cleaning.dml", "new_path": "scripts/builtin/topk_cleaning.dml", "diff": "@@ -188,8 +188,15 @@ getDirtyScore = function(Frame[Unknown] X, Matrix[Double] Y, Frame[Unknown] Xtes\nMatrix[Double] evalFunHp, Double sample, Integer trainML, Boolean cv, Integer cvk, List[Unknown] ctx=list() )\nreturn(Double dirtyScore, Matrix[Double] evalFunHp)\n{\n+ dschema = detectSchema(X)\n+ dmask = matrix(0, rows=1, cols=ncol(dschema))\n+ for(i in 1:ncol(dschema))\n+ if(as.scalar(dschema[1, i]) == \"STRING\" | as.scalar(dschema[1, i]) == \"BOOLEAN\")\n+ dmask[1, i] = 1\n+\nprefix = as.scalar(ctx[\"prefix\"]);\nmask = as.matrix(metaList['mask'])\n+ mask = ifelse(sum(mask == dmask) < ncol(mask), dmask, mask)\n[eXtrain, eXtest] = recodeData(X, Xtest, mask, cv, \"recode\")\neXtrain = replace(target=eXtrain, pattern=NaN, replacement = 0)\neXtest = replace(target=eXtest, pattern=NaN, replacement = 0)\n" }, { "change_type": "MODIFY", "old_path": "scripts/pipelines/scripts/utils.dml", "new_path": "scripts/pipelines/scripts/utils.dml", "diff": "@@ -63,7 +63,7 @@ doSample = function(Matrix[Double] eX, Matrix[Double] eY, Double ratio, Boolean\nsampledX = eX\nsampledY = eY\n- if(sampled > MIN_SAMPLE)\n+ if(sampled > MIN_SAMPLE & ratio != 1.0)\n{\ndist = max(eY) # num classes (one-hot encoded eY)\n@@ -156,6 +156,7 @@ return(Frame[Unknown] processedData)\n# step 1 drop invalid types\nprint(prefix+\" drop values with type mismatch\");\ndata = dropInvalidType(data, schema)\n+ print(\"dropped invalids\")\n# step 2 do the case transformations\nprint(prefix+\" convert strings to lower case\");\n@@ -325,7 +326,7 @@ topk_gridSearch = function(Matrix[Double] X, Matrix[Double] y, Matrix[Double] Xt\ncvbeta += lbeta;\ncvloss += as.matrix(accuracy);\n}\n- Rbeta[i,] = cvbeta / k;\n+ # Rbeta[i,] = cvbeta / k;\nRloss[i,] = cvloss / k;\n}\n}\n@@ -338,7 +339,7 @@ topk_gridSearch = function(Matrix[Double] X, Matrix[Double] y, Matrix[Double] Xt\nltrainArgs[as.scalar(params[j])] = as.scalar(HP[i,j]);\n# b) core training/scoring and write-back\nlbeta = t(eval(train, ltrainArgs))\n- Rbeta[i,1:ncol(lbeta)] = lbeta;\n+ # Rbeta[i,1:ncol(lbeta)] = lbeta;\nRloss[i,] = eval(predict, append(predictArgs,t(lbeta)));\n}\n}\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/matrix/data/FrameBlock.java", "new_path": "src/main/java/org/apache/sysds/runtime/matrix/data/FrameBlock.java", "diff": "@@ -2044,7 +2044,7 @@ public class FrameBlock implements CacheBlock, Externalizable {\nelse\nreturn ValueType.INT64;\n}\n- else if (val.matches(\"[-+]?[0-9]+\\\\.?[0-9]*([e]?[-+]?[0-9]+)\")){\n+ else if (val.matches(\"[-+]?[0-9]*\\\\.?[0-9]+([eE][-+]?[0-9]+)?\")){\ndouble maxValue = Double.parseDouble(val);\nif ((maxValue >= (-Float.MAX_VALUE)) && (maxValue <= Float.MAX_VALUE))\nreturn ValueType.FP32;\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/pipelines/BuiltinTopkCleaningClassificationTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/pipelines/BuiltinTopkCleaningClassificationTest.java", "diff": "@@ -45,7 +45,7 @@ public class BuiltinTopkCleaningClassificationTest extends AutomatedTestBase {\naddTestConfiguration(TEST_NAME, new TestConfiguration(TEST_CLASS_DIR, TEST_NAME, new String[]{\"R\"}));\n}\n- @Test\n+ @Ignore\npublic void testFindBestPipelineCompany() {\nruntopkCleaning(DATA_DIR+ \"company.csv\", RESOURCE+ \"meta/meta_company.csv\", 1.0, 3,5,\n\"FALSE\", 0,0.8, Types.ExecMode.SINGLE_NODE);\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/bestAcc.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/bestAcc.csv", "diff": "-64.88439306358381\n-64.73988439306359\n-64.73988439306359\n+65.89595375722543\n+65.3179190751445\n+65.3179190751445\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/hp.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/hp.csv", "diff": "-54.0,3.0,7.0,2.0,1.0,0,0,0,1.0,0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,1.0,0,0,0,2.0,2.0,0,1.0,0,0,0,0,0,0,1.0,0.7053074081820746,0,0,0,0,1.0,1.0,2.0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n-54.0,3.0,6.0,2.0,1.0,0,0,0,1.0,0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,1.0,0,0,0,2.0,2.0,1.0,1.0,0,0,0,0,0,0,1.0,0.7784943734333777,0,0,0,0,1.0,1.0,2.0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n-54.0,3.0,4.0,1.0,1.0,0,0,0,1.0,0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,1.0,0,0,0,2.0,2.0,1.0,1.0,0,0,0,0,0,0,1.0,0.7567266322372848,0,0,0,0,1.0,1.0,2.0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n+24.0,2.0,0.042803849955920424,0.9504400993873047,0,0,0,1.0,0,0,0,0,0,0,1.0,0,2.0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n+24.0,2.0,0.016013893020007757,0.9642527252494045,0,0,0,1.0,0,0,0,0,0,0,1.0,0,2.0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n+24.0,2.0,0.03480400352286382,0.9561745054711843,0,0,0,1.0,0,0,0,0,0,0,1.0,0,2.0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/lp.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/lp.csv", "diff": "-ED,MVI,EC,SCALE,CI,DUMMY\n+OTLR,CI,DUMMY\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/pip.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/pip.csv", "diff": "-outlierBySd,imputeByMedian,imputeByMean,scale,abstain,dummycoding\n-outlierBySd,imputeByMean,imputeByMedian,scale,abstain,dummycoding\n-outlierBySd,imputeByMean,imputeByMedian,scale,abstain,dummycoding\n+winsorize,wtomeklink,dummycoding\n+winsorize,wtomeklink,dummycoding\n+winsorize,wtomeklink,dummycoding\n" } ]

[ { "change_type": "MODIFY", "old_path": "bin/systemds", "new_path": "bin/systemds", "diff": "@@ -319,7 +319,7 @@ if [[ \"$*\" == *-config* ]]; then\nfi\nelif [ -z \"$CONFIG_FILE\" ] ; then\n# same as above: set config file param if the file exists\n- CONFIG_FILE=$(ordered_find \"SystemDS*config*.xml\")\n+ CONFIG_FILE=$(ordered_find \"SystemDS-config-defaults.xml\")\nif [ -z \"$CONFIG_FILE\" ]; then\nCONFIG_FILE=\"\"\nelse\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/instructions/fed/FEDInstructionUtils.java", "new_path": "src/main/java/org/apache/sysds/runtime/instructions/fed/FEDInstructionUtils.java", "diff": "@@ -69,6 +69,7 @@ import org.apache.sysds.runtime.instructions.spark.BinaryTensorTensorBroadcastSP\nimport org.apache.sysds.runtime.instructions.spark.BinaryTensorTensorSPInstruction;\nimport org.apache.sysds.runtime.instructions.spark.CastSPInstruction;\nimport org.apache.sysds.runtime.instructions.spark.CentralMomentSPInstruction;\n+import org.apache.sysds.runtime.instructions.spark.CpmmSPInstruction;\nimport org.apache.sysds.runtime.instructions.spark.CtableSPInstruction;\nimport org.apache.sysds.runtime.instructions.spark.CumulativeOffsetSPInstruction;\nimport org.apache.sysds.runtime.instructions.spark.IndexingSPInstruction;\n@@ -80,6 +81,7 @@ import org.apache.sysds.runtime.instructions.spark.QuantileSortSPInstruction;\nimport org.apache.sysds.runtime.instructions.spark.QuaternarySPInstruction;\nimport org.apache.sysds.runtime.instructions.spark.ReblockSPInstruction;\nimport org.apache.sysds.runtime.instructions.spark.ReorgSPInstruction;\n+import org.apache.sysds.runtime.instructions.spark.RmmSPInstruction;\nimport org.apache.sysds.runtime.instructions.spark.SpoofSPInstruction;\nimport org.apache.sysds.runtime.instructions.spark.TernarySPInstruction;\nimport org.apache.sysds.runtime.instructions.spark.UnaryMatrixSPInstruction;\n@@ -279,14 +281,7 @@ public class FEDInstructionUtils {\npublic static Instruction checkAndReplaceSP(Instruction inst, ExecutionContext ec) {\nFEDInstruction fedinst = null;\n- if (inst instanceof MapmmSPInstruction) {\n- MapmmSPInstruction instruction = (MapmmSPInstruction) inst;\n- Data data = ec.getVariable(instruction.input1);\n- if (data instanceof MatrixObject && ((MatrixObject) data).isFederatedExcept(FType.BROADCAST)) {\n- fedinst = MapmmFEDInstruction.parseInstruction(instruction.getInstructionString());\n- }\n- }\n- else if(inst instanceof CastSPInstruction){\n+ if(inst instanceof CastSPInstruction){\nCastSPInstruction ins = (CastSPInstruction) inst;\nif((ins.getOpcode().equalsIgnoreCase(UnaryCP.CAST_AS_FRAME_OPCODE) || ins.getOpcode().equalsIgnoreCase(UnaryCP.CAST_AS_MATRIX_OPCODE))\n&& ins.input1.isMatrix() && ec.getCacheableData(ins.input1).isFederatedExcept(FType.BROADCAST)){\n@@ -377,7 +372,13 @@ public class FEDInstructionUtils {\n}\nelse if (inst instanceof BinarySPInstruction) {\nBinarySPInstruction instruction = (BinarySPInstruction) inst;\n-\n+ if (inst instanceof MapmmSPInstruction || inst instanceof CpmmSPInstruction || inst instanceof RmmSPInstruction) {\n+ Data data = ec.getVariable(instruction.input1);\n+ if (data instanceof MatrixObject && ((MatrixObject) data).isFederatedExcept(FType.BROADCAST)) {\n+ fedinst = MMFEDInstruction.parseInstruction(instruction.getInstructionString());\n+ }\n+ }\n+ else\nif(inst instanceof QuantilePickSPInstruction) {\nQuantilePickSPInstruction qinstruction = (QuantilePickSPInstruction) inst;\nData data = ec.getVariable(qinstruction.input1);\n" }, { "change_type": "RENAME", "old_path": "src/main/java/org/apache/sysds/runtime/instructions/fed/MapmmFEDInstruction.java", "new_path": "src/main/java/org/apache/sysds/runtime/instructions/fed/MMFEDInstruction.java", "diff": "@@ -21,9 +21,11 @@ package org.apache.sysds.runtime.instructions.fed;\nimport java.util.concurrent.Future;\n+import org.apache.commons.lang3.ArrayUtils;\nimport org.apache.sysds.common.Types;\nimport org.apache.sysds.hops.AggBinaryOp;\nimport org.apache.sysds.lops.MapMult;\n+import org.apache.sysds.lops.PMMJ;\nimport org.apache.sysds.runtime.DMLRuntimeException;\nimport org.apache.sysds.runtime.controlprogram.caching.MatrixObject;\nimport org.apache.sysds.runtime.controlprogram.context.ExecutionContext;\n@@ -41,18 +43,18 @@ import org.apache.sysds.runtime.matrix.operators.AggregateBinaryOperator;\nimport org.apache.sysds.runtime.matrix.operators.Operator;\nimport org.apache.sysds.runtime.meta.MatrixCharacteristics;\n-public class MapmmFEDInstruction extends BinaryFEDInstruction\n+public class MMFEDInstruction extends BinaryFEDInstruction\n{\n- private MapmmFEDInstruction(Operator op, CPOperand in1, CPOperand in2, CPOperand out, MapMult.CacheType type,\n+ private MMFEDInstruction(Operator op, CPOperand in1, CPOperand in2, CPOperand out, MapMult.CacheType type,\nboolean outputEmpty, AggBinaryOp.SparkAggType aggtype, String opcode, String istr) {\nsuper(FEDType.MAPMM, op, in1, in2, out, opcode, istr);\n}\n- public static MapmmFEDInstruction parseInstruction( String str ) {\n+ public static MMFEDInstruction parseInstruction( String str ) {\nString parts[] = InstructionUtils.getInstructionPartsWithValueType(str);\nString opcode = parts[0];\n- if(!opcode.equalsIgnoreCase(MapMult.OPCODE))\n+ if(!ArrayUtils.contains(new String[] {MapMult.OPCODE, PMMJ.OPCODE, \"cpmm\", \"rmm\"}, opcode))\nthrow new DMLRuntimeException(\"MapmmSPInstruction.parseInstruction():: Unknown opcode \" + opcode);\nCPOperand in1 = new CPOperand(parts[1]);\n@@ -63,7 +65,7 @@ public class MapmmFEDInstruction extends BinaryFEDInstruction\nAggBinaryOp.SparkAggType aggtype = AggBinaryOp.SparkAggType.valueOf(parts[6]);\nAggregateBinaryOperator aggbin = InstructionUtils.getMatMultOperator(1);\n- return new MapmmFEDInstruction(aggbin, in1, in2, out, type, outputEmpty, aggtype, opcode, str);\n+ return new MMFEDInstruction(aggbin, in1, in2, out, type, outputEmpty, aggtype, opcode, str);\n}\npublic void processInstruction(ExecutionContext ec) {\n" } ]

[ { "change_type": "MODIFY", "old_path": "CONTRIBUTING.md", "new_path": "CONTRIBUTING.md", "diff": "@@ -46,7 +46,7 @@ let's make sure the changes are consistent with the guidelines and coding style.\nWe suggest applying a code formatter to the written code. Generally, this is done automatically.\n-We have provided at profile for java located in [Codestyle File `./docs/CodeStyle.eclipse.xml`](dev/CodeStyle_eclipse.xml). This can be loaded in most editors e.g.:\n+We have provided at profile for java located in [Codestyle File `./dev/CodeStyle.eclipse.xml`](dev/CodeStyle_eclipse.xml). This can be loaded in most editors e.g.:\n- [Eclipse](https://stackoverflow.com/questions/10432538/eclipse-import-conf-xml-files#10433986)\n- [IntelliJ](https://imagej.net/Eclipse_code_style_profiles_and_IntelliJ)\n" } ]

[ { "change_type": "ADD", "old_path": null, "new_path": "scripts/perftest/conf/log4j.properties", "diff": "+#\n+# Licensed to the Apache Software Foundation (ASF) under one or more\n+# contributor license agreements. See the NOTICE file distributed with\n+# this work for additional information regarding copyright ownership.\n+# The ASF licenses this file to You under the Apache License, Version 2.0\n+# (the \"License\"); you may not use this file except in compliance with\n+# the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing, software\n+# distributed under the License is distributed on an \"AS IS\" BASIS,\n+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n+# See the License for the specific language governing permissions and\n+# limitations under the License.\n+#\n+\n+# Set everything to be logged to the console\n+log4j.rootCategory=ERROR, console\n+log4j.appender.console=org.apache.log4j.ConsoleAppender\n+log4j.appender.console.target=System.err\n+log4j.appender.console.layout=org.apache.log4j.PatternLayout\n+log4j.appender.console.layout.ConversionPattern=%d{yy/MM/dd HH:mm:ss} %p %c{1}: %m%n\n+\n+# Set the default spark-shell log level to WARN. When running the spark-shell, the\n+# log level for this class is used to overwrite the root logger's log level, so that\n+# the user can have different defaults for the shell and regular Spark apps.\n+log4j.logger.org.apache.spark.repl.Main=WARN\n+\n+# Settings to quiet third party logs that are too verbose\n+log4j.logger.org.spark_project.jetty=WARN\n+log4j.logger.org.spark_project.jetty.util.component.AbstractLifeCycle=ERROR\n+log4j.logger.org.apache.spark.repl.SparkIMain$exprTyper=INFO\n+log4j.logger.org.apache.spark.repl.SparkILoop$SparkILoopInterpreter=INFO\n+log4j.logger.org.apache.parquet=ERROR\n+log4j.logger.parquet=ERROR\n+\n+# SPARK-9183: Settings to avoid annoying messages when looking up nonexistent UDFs in SparkSQL with Hive support\n+log4j.logger.org.apache.hadoop.hive.metastore.RetryingHMSHandler=FATAL\n+log4j.logger.org.apache.hadoop.hive.ql.exec.FunctionRegistry=ERROR\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/genBinomialData.sh", "new_path": "scripts/perftest/genBinomialData.sh", "diff": "@@ -27,33 +27,32 @@ FORMAT=\"binary\" # can be csv, mm, text, binary\nDENSE_SP=0.9\nSPARSE_SP=0.01\n-\n#generate XS scenarios (80MB)\n-${CMD} -f ../datagen/genRandData4LogisticRegression.dml --args 10000 1000 5 5 ${BASE}/w10k_1k_dense ${BASE}/X10k_1k_dense ${BASE}/y10k_1k_dense 1 0 $DENSE_SP $FORMAT 1\n-${CMD} -f ../datagen/genRandData4LogisticRegression.dml --args 10000 1000 5 5 ${BASE}/w10k_1k_sparse ${BASE}/X10k_1k_sparse ${BASE}/y10k_1k_sparse 1 0 $SPARSE_SP $FORMAT 1\n+${CMD} -f ./datagen/genRandData4LogisticRegression.dml --args 10000 1000 5 5 ${BASE}/w10k_1k_dense ${BASE}/X10k_1k_dense ${BASE}/y10k_1k_dense 1 0 $DENSE_SP $FORMAT 1\n+${CMD} -f ./datagen/genRandData4LogisticRegression.dml --args 10000 1000 5 5 ${BASE}/w10k_1k_sparse ${BASE}/X10k_1k_sparse ${BASE}/y10k_1k_sparse 1 0 $SPARSE_SP $FORMAT 1\n${CMD} -f scripts/extractTestData.dml --args ${BASE}/X10k_1k_dense ${BASE}/y10k_1k_dense ${BASE}/X10k_1k_dense_test ${BASE}/y10k_1k_dense_test $FORMAT\n${CMD} -f scripts/extractTestData.dml --args ${BASE}/X10k_1k_sparse ${BASE}/y10k_1k_sparse ${BASE}/X10k_1k_sparse_test ${BASE}/y10k_1k_sparse_test $FORMAT\n##generate S scenarios (800MB)\n-#${CMD} -f ../datagen/genRandData4LogisticRegression.dml --args 100000 1000 5 5 ${BASE}/w100k_1k_dense ${BASE}/X100k_1k_dense ${BASE}/y100k_1k_dense 1 0 $DENSE_SP $FORMAT 1\n-#${CMD} -f ../datagen/genRandData4LogisticRegression.dml --args 100000 1000 5 5 ${BASE}/w100k_1k_sparse ${BASE}/X100k_1k_sparse ${BASE}/y100k_1k_sparse 1 0 $SPARSE_SP $FORMAT 1\n-#${CMD} -f scripts/extractTestData.dml --args ${BASE}/X100k_1k_dense ${BASE}/y100k_1k_dense ${BASE}/X100k_1k_dense_test ${BASE}/y100k_1k_dense_test $FORMAT\n-#${CMD} -f scripts/extractTestData.dml --args ${BASE}/X100k_1k_sparse ${BASE}/y100k_1k_sparse ${BASE}/X100k_1k_sparse_test ${BASE}/y100k_1k_sparse_test $FORMAT\n-#\n+${CMD} -f ./datagen/genRandData4LogisticRegression.dml --args 100000 1000 5 5 ${BASE}/w100k_1k_dense ${BASE}/X100k_1k_dense ${BASE}/y100k_1k_dense 1 0 $DENSE_SP $FORMAT 1\n+${CMD} -f ./datagen/genRandData4LogisticRegression.dml --args 100000 1000 5 5 ${BASE}/w100k_1k_sparse ${BASE}/X100k_1k_sparse ${BASE}/y100k_1k_sparse 1 0 $SPARSE_SP $FORMAT 1\n+${CMD} -f scripts/extractTestData.dml --args ${BASE}/X100k_1k_dense ${BASE}/y100k_1k_dense ${BASE}/X100k_1k_dense_test ${BASE}/y100k_1k_dense_test $FORMAT\n+${CMD} -f scripts/extractTestData.dml --args ${BASE}/X100k_1k_sparse ${BASE}/y100k_1k_sparse ${BASE}/X100k_1k_sparse_test ${BASE}/y100k_1k_sparse_test $FORMAT\n+\n##generate M scenarios (8GB)\n-#${CMD} -f ../datagen/genRandData4LogisticRegression.dml --args 1000000 1000 5 5 ${BASE}/w1M_1k_dense ${BASE}/X1M_1k_dense ${BASE}/y1M_1k_dense 1 0 $DENSE_SP $FORMAT 1\n-#${CMD} -f ../datagen/genRandData4LogisticRegression.dml --args 1000000 1000 5 5 ${BASE}/w1M_1k_sparse ${BASE}/X1M_1k_sparse ${BASE}/y1M_1k_sparse 1 0 $SPARSE_SP $FORMAT 1\n-#${CMD} -f scripts/extractTestData.dml --args ${BASE}/X1M_1k_dense ${BASE}/y1M_1k_dense ${BASE}/X1M_1k_dense_test ${BASE}/y1M_1k_dense_test $FORMAT\n-#${CMD} -f scripts/extractTestData.dml --args ${BASE}/X1M_1k_sparse ${BASE}/y1M_1k_sparse ${BASE}/X1M_1k_sparse_test ${BASE}/y1M_1k_sparse_test $FORMAT\n-#\n+${CMD} -f ./datagen/genRandData4LogisticRegression.dml --args 1000000 1000 5 5 ${BASE}/w1M_1k_dense ${BASE}/X1M_1k_dense ${BASE}/y1M_1k_dense 1 0 $DENSE_SP $FORMAT 1\n+${CMD} -f ./datagen/genRandData4LogisticRegression.dml --args 1000000 1000 5 5 ${BASE}/w1M_1k_sparse ${BASE}/X1M_1k_sparse ${BASE}/y1M_1k_sparse 1 0 $SPARSE_SP $FORMAT 1\n+${CMD} -f scripts/extractTestData.dml --args ${BASE}/X1M_1k_dense ${BASE}/y1M_1k_dense ${BASE}/X1M_1k_dense_test ${BASE}/y1M_1k_dense_test $FORMAT\n+${CMD} -f scripts/extractTestData.dml --args ${BASE}/X1M_1k_sparse ${BASE}/y1M_1k_sparse ${BASE}/X1M_1k_sparse_test ${BASE}/y1M_1k_sparse_test $FORMAT\n+\n##generate L scenarios (80GB)\n-#${CMD} -f ../datagen/genRandData4LogisticRegression.dml --args 10000000 1000 5 5 ${BASE}/w10M_1k_dense ${BASE}/X10M_1k_dense ${BASE}/y10M_1k_dense 1 0 $DENSE_SP $FORMAT 1\n-#${CMD} -f ../datagen/genRandData4LogisticRegression.dml --args 10000000 1000 5 5 ${BASE}/w10M_1k_sparse ${BASE}/X10M_1k_sparse ${BASE}/y10M_1k_sparse 1 0 $SPARSE_SP $FORMAT 1\n-#${CMD} -f scripts/extractTestData.dml --args ${BASE}/X10M_1k_dense ${BASE}/y10M_1k_dense ${BASE}/X10M_1k_dense_test ${BASE}/y10M_1k_dense_test $FORMAT\n-#${CMD} -f scripts/extractTestData.dml --args ${BASE}/X10M_1k_sparse ${BASE}/y10M_1k_sparse ${BASE}/X10M_1k_sparse_test ${BASE}/y10M_1k_sparse_test $FORMAT\n-#\n+${CMD} -f ./datagen/genRandData4LogisticRegression.dml --args 10000000 1000 5 5 ${BASE}/w10M_1k_dense ${BASE}/X10M_1k_dense ${BASE}/y10M_1k_dense 1 0 $DENSE_SP $FORMAT 1\n+${CMD} -f ./datagen/genRandData4LogisticRegression.dml --args 10000000 1000 5 5 ${BASE}/w10M_1k_sparse ${BASE}/X10M_1k_sparse ${BASE}/y10M_1k_sparse 1 0 $SPARSE_SP $FORMAT 1\n+${CMD} -f scripts/extractTestData.dml --args ${BASE}/X10M_1k_dense ${BASE}/y10M_1k_dense ${BASE}/X10M_1k_dense_test ${BASE}/y10M_1k_dense_test $FORMAT\n+${CMD} -f scripts/extractTestData.dml --args ${BASE}/X10M_1k_sparse ${BASE}/y10M_1k_sparse ${BASE}/X10M_1k_sparse_test ${BASE}/y10M_1k_sparse_test $FORMAT\n+\n##generate XL scenarios (800GB)\n-#${CMD} -f ../datagen/genRandData4LogisticRegression.dml --args 100000000 1000 5 5 ${BASE}/w100M_1k_dense ${BASE}/X100M_1k_dense ${BASE}/y100M_1k_dense 1 0 $DENSE_SP $FORMAT 1\n-#${CMD} -f ../datagen/genRandData4LogisticRegression.dml --args 100000000 1000 5 5 ${BASE}/w100M_1k_sparse ${BASE}/X100M_1k_sparse ${BASE}/y100M_1k_sparse 1 0 $SPARSE_SP $FORMAT 1\n+#${CMD} -f ./datagen/genRandData4LogisticRegression.dml --args 100000000 1000 5 5 ${BASE}/w100M_1k_dense ${BASE}/X100M_1k_dense ${BASE}/y100M_1k_dense 1 0 $DENSE_SP $FORMAT 1\n+#${CMD} -f ./datagen/genRandData4LogisticRegression.dml --args 100000000 1000 5 5 ${BASE}/w100M_1k_sparse ${BASE}/X100M_1k_sparse ${BASE}/y100M_1k_sparse 1 0 $SPARSE_SP $FORMAT 1\n#${CMD} -f scripts/extractTestData.dml --args ${BASE}/X100M_1k_dense ${BASE}/y100M_1k_dense ${BASE}/X100M_1k_dense_test ${BASE}/y100M_1k_dense_test $FORMAT\n#${CMD} -f scripts/extractTestData.dml --args ${BASE}/X100M_1k_sparse ${BASE}/y100M_1k_sparse ${BASE}/X100M_1k_sparse_test ${BASE}/y100M_1k_sparse_test $FORMAT\n\\ No newline at end of file\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/genMultinomialData.sh", "new_path": "scripts/perftest/genMultinomialData.sh", "diff": "@@ -28,31 +28,31 @@ DENSE_SP=0.9\nSPARSE_SP=0.01\n#generate XS scenarios (80MB)\n-${CMD} -f ../datagen/genRandData4Multinomial.dml $DASH-args 10000 1000 $DENSE_SP 5 0 $BASE/X10k_1k_dense_k5 $BASE/y10k_1k_dense_k5 $FORMAT 1\n-${CMD} -f ../datagen/genRandData4Multinomial.dml $DASH-args 10000 1000 $SPARSE_SP 5 0 $BASE/X10k_1k_sparse_k5 $BASE/y10k_1k_sparse_k5 $FORMAT 1\n+${CMD} -f ./datagen/genRandData4Multinomial.dml $DASH-args 10000 1000 $DENSE_SP 5 0 $BASE/X10k_1k_dense_k5 $BASE/y10k_1k_dense_k5 $FORMAT 1\n+${CMD} -f ./datagen/genRandData4Multinomial.dml $DASH-args 10000 1000 $SPARSE_SP 5 0 $BASE/X10k_1k_sparse_k5 $BASE/y10k_1k_sparse_k5 $FORMAT 1\n${CMD} -f scripts/extractTestData.dml $DASH-args $BASE/X10k_1k_dense_k5 $BASE/y10k_1k_dense_k5 $BASE/X10k_1k_dense_k5_test $BASE/y10k_1k_dense_k5_test $FORMAT\n${CMD} -f scripts/extractTestData.dml $DASH-args $BASE/X10k_1k_sparse_k5 $BASE/y10k_1k_sparse_k5 $BASE/X10k_1k_sparse_k5_test $BASE/y10k_1k_sparse_k5_test $FORMAT\n-##generate S scenarios (80MB)\n-#${CMD} -f ../datagen/genRandData4Multinomial.dml $DASH-args 100000 1000 $DENSE_SP 5 0 $BASE/X100k_1k_dense_k5 $BASE/y100k_1k_dense_k5 $FORMAT 1\n-#${CMD} -f ../datagen/genRandData4Multinomial.dml $DASH-args 100000 1000 $SPARSE_SP 5 0 $BASE/X100k_1k_sparse_k5 $BASE/y100k_1k_sparse_k5 $FORMAT 1\n-#${CMD} -f scripts/extractTestData.dml $DASH-args $BASE/X100k_1k_dense_k5 $BASE/y100k_1k_dense_k5 $BASE/X100k_1k_dense_k5_test $BASE/y100k_1k_dense_k5_test $FORMAT\n-#${CMD} -f scripts/extractTestData.dml $DASH-args $BASE/X100k_1k_sparse_k5 $BASE/y100k_1k_sparse_k5 $BASE/X100k_1k_sparse_k5_test $BASE/y100k_1k_sparse_k5_test $FORMAT\n-#\n+##generate S scenarios (800MB)\n+${CMD} -f ./datagen/genRandData4Multinomial.dml $DASH-args 100000 1000 $DENSE_SP 5 0 $BASE/X100k_1k_dense_k5 $BASE/y100k_1k_dense_k5 $FORMAT 1\n+${CMD} -f ./datagen/genRandData4Multinomial.dml $DASH-args 100000 1000 $SPARSE_SP 5 0 $BASE/X100k_1k_sparse_k5 $BASE/y100k_1k_sparse_k5 $FORMAT 1\n+${CMD} -f scripts/extractTestData.dml $DASH-args $BASE/X100k_1k_dense_k5 $BASE/y100k_1k_dense_k5 $BASE/X100k_1k_dense_k5_test $BASE/y100k_1k_dense_k5_test $FORMAT\n+${CMD} -f scripts/extractTestData.dml $DASH-args $BASE/X100k_1k_sparse_k5 $BASE/y100k_1k_sparse_k5 $BASE/X100k_1k_sparse_k5_test $BASE/y100k_1k_sparse_k5_test $FORMAT\n+\n##generate M scenarios (8GB)\n-#${CMD} -f ../datagen/genRandData4Multinomial.dml $DASH-args 1000000 1000 $DENSE_SP 5 0 $BASE/X1M_1k_dense_k5 $BASE/y1M_1k_dense_k5 $FORMAT 1\n-#${CMD} -f ../datagen/genRandData4Multinomial.dml $DASH-args 1000000 1000 $SPARSE_SP 5 0 $BASE/X1M_1k_sparse_k5 $BASE/y1M_1k_sparse_k5 $FORMAT 1\n-#${CMD} -f scripts/extractTestData.dml $DASH-args $BASE/X1M_1k_dense_k5 $BASE/y1M_1k_dense_k5 $BASE/X1M_1k_dense_k5_test $BASE/y1M_1k_dense_k5_test $FORMAT\n-#${CMD} -f scripts/extractTestData.dml $DASH-args $BASE/X1M_1k_sparse_k5 $BASE/y1M_1k_sparse_k5 $BASE/X1M_1k_sparse_k5_test $BASE/y1M_1k_sparse_k5_test $FORMAT\n-#\n+${CMD} -f ./datagen/genRandData4Multinomial.dml $DASH-args 1000000 1000 $DENSE_SP 5 0 $BASE/X1M_1k_dense_k5 $BASE/y1M_1k_dense_k5 $FORMAT 1\n+${CMD} -f ./datagen/genRandData4Multinomial.dml $DASH-args 1000000 1000 $SPARSE_SP 5 0 $BASE/X1M_1k_sparse_k5 $BASE/y1M_1k_sparse_k5 $FORMAT 1\n+${CMD} -f scripts/extractTestData.dml $DASH-args $BASE/X1M_1k_dense_k5 $BASE/y1M_1k_dense_k5 $BASE/X1M_1k_dense_k5_test $BASE/y1M_1k_dense_k5_test $FORMAT\n+${CMD} -f scripts/extractTestData.dml $DASH-args $BASE/X1M_1k_sparse_k5 $BASE/y1M_1k_sparse_k5 $BASE/X1M_1k_sparse_k5_test $BASE/y1M_1k_sparse_k5_test $FORMAT\n+\n##generate L scenarios (80GB)\n-#${CMD} -f ../datagen/genRandData4Multinomial.dml $DASH-args 10000000 1000 $DENSE_SP 5 0 $BASE/X10M_1k_dense_k5 $BASE/y10M_1k_dense_k5 $FORMAT 1\n-#${CMD} -f ../datagen/genRandData4Multinomial.dml $DASH-args 10000000 1000 $SPARSE_SP 5 0 $BASE/X10M_1k_sparse_k5 $BASE/y10M_1k_sparse_k5 $FORMAT 1\n-#${CMD} -f scripts/extractTestData.dml $DASH-args $BASE/X10M_1k_dense_k5 $BASE/y10M_1k_dense_k5 $BASE/X10M_1k_dense_k5_test $BASE/y10M_1k_dense_k5_test $FORMAT\n-#${CMD} -f scripts/extractTestData.dml $DASH-args $BASE/X10M_1k_sparse_k5 $BASE/y10M_1k_sparse_k5 $BASE/X10M_1k_sparse_k5_test $BASE/y10M_1k_sparse_k5_test $FORMAT\n-#\n+${CMD} -f ./datagen/genRandData4Multinomial.dml $DASH-args 10000000 1000 $DENSE_SP 5 0 $BASE/X10M_1k_dense_k5 $BASE/y10M_1k_dense_k5 $FORMAT 1\n+${CMD} -f ./datagen/genRandData4Multinomial.dml $DASH-args 10000000 1000 $SPARSE_SP 5 0 $BASE/X10M_1k_sparse_k5 $BASE/y10M_1k_sparse_k5 $FORMAT 1\n+${CMD} -f scripts/extractTestData.dml $DASH-args $BASE/X10M_1k_dense_k5 $BASE/y10M_1k_dense_k5 $BASE/X10M_1k_dense_k5_test $BASE/y10M_1k_dense_k5_test $FORMAT\n+${CMD} -f scripts/extractTestData.dml $DASH-args $BASE/X10M_1k_sparse_k5 $BASE/y10M_1k_sparse_k5 $BASE/X10M_1k_sparse_k5_test $BASE/y10M_1k_sparse_k5_test $FORMAT\n+\n##generate LARGE scenarios (800GB)\n-#${CMD} -f ../datagen/genRandData4Multinomial.dml $DASH-args 100000000 1000 $DENSE_SP 5 0 $BASE/X100M_1k_dense_k5 $BASE/y100M_1k_dense_k5 $FORMAT 1\n-#${CMD} -f ../datagen/genRandData4Multinomial.dml $DASH-args 100000000 1000 $SPARSE_SP 5 0 $BASE/X100M_1k_sparse_k5 $BASE/y100M_1k_sparse_k5 $FORMAT 1\n+#${CMD} -f ./datagen/genRandData4Multinomial.dml $DASH-args 100000000 1000 $DENSE_SP 5 0 $BASE/X100M_1k_dense_k5 $BASE/y100M_1k_dense_k5 $FORMAT 1\n+#${CMD} -f ./datagen/genRandData4Multinomial.dml $DASH-args 100000000 1000 $SPARSE_SP 5 0 $BASE/X100M_1k_sparse_k5 $BASE/y100M_1k_sparse_k5 $FORMAT 1\n#${CMD} -f scripts/extractTestData.dml $DASH-args $BASE/X100M_1k_dense_k5 $BASE/y100M_1k_dense_k5 $BASE/X100M_1k_dense_k5_test $BASE/y100M_1k_dense_k5_test $FORMAT\n#${CMD} -f scripts/extractTestData.dml $DASH-args $BASE/X100M_1k_sparse_k5 $BASE/y100M_1k_sparse_k5 $BASE/X100M_1k_sparse_k5_test $BASE/y100M_1k_sparse_k5_test $FORMAT\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/runAll.sh", "new_path": "scripts/perftest/runAll.sh", "diff": "@@ -29,8 +29,8 @@ export LOG4JPROP='conf/log4j-off.properties'\nexport SYSDS_QUIET=1\n# Command to be executed\n-CMD=\"systemds\"\n-#CMD=\"./sparkDML.sh\"\n+#CMD=\"systemds\"\n+CMD=\"./sparkDML.sh\"\n# Possible lines to initialize Intel MKL, depending on version and install location\n# . ~/intel/bin/compilervars.sh intel64\n@@ -68,4 +68,3 @@ date >> results/times.txt\n#./runAllSurvival $CMD $TEMPFOLDER\n#KaplanMeier\n#Cox\n-\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/scripts/extractTestData.dml", "new_path": "scripts/perftest/scripts/extractTestData.dml", "diff": "X = read($1);\ny = read($2);\n-Percent_test = .2\n-N_x = as.integer(nrow(X) * Percent_test);\n-N_y = as.integer(nrow(y) * Percent_test);\n+[X,y,Xtest,ytest] = split(X=X, Y=y, f=0.8);\n-X = X[1:N_x,];\n-y = y[1:N_y,];\n-\n-write(X, $3, format=$5);\n-write(y, $4, format=$5);\n+write(Xtest, $3, format=$5);\n+write(ytest, $4, format=$5);\n" }, { "change_type": "ADD", "old_path": null, "new_path": "scripts/perftest/sparkDML.sh", "diff": "+#!/bin/bash\n+#-------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+#-------------------------------------------------------------\n+\n+#set -x\n+\n+\n+# This script is a simplified version of sparkDML.sh in order to\n+# allow a simple drop-in replacement for 'hadoop jar' without\n+# the need to change any command line arguments.\n+\n+export SPARK_HOME=../spark-2.4.7-bin-hadoop2.7\n+export HADOOP_CONF_DIR=/home/hadoop/hadoop-2.7.7/etc/hadoop\n+\n+$SPARK_HOME/bin/spark-submit \\\n+ --master yarn \\\n+ --deploy-mode client \\\n+ --driver-memory 20g \\\n+ --conf spark.driver.extraJavaOptions=\"-Xms20g -Dlog4j.configuration=file:/home/mboehm/perftest/conf/log4j.properties\" \\\n+ --conf spark.ui.showConsoleProgress=true \\\n+ --conf spark.executor.heartbeatInterval=100s \\\n+ --conf spark.network.timeout=512s \\\n+ --num-executors 10 \\\n+ --executor-memory 105g \\\n+ --executor-cores 32 \\\n+ SystemDS.jar \"$@\"\n\\ No newline at end of file\n" } ]

[ { "change_type": "MODIFY", "old_path": "scripts/perftest/runAll.sh", "new_path": "scripts/perftest/runAll.sh", "diff": "@@ -37,19 +37,23 @@ CMD=\"./sparkDML.sh\"\n# . ~/intel/oneapi/setvars.sh intel64\n# . /opt/intel/bin/compilervars.sh intel64\n-\n### Micro Benchmarks:\n#./MatrixMult.sh\n#./MatrixTranspose.sh\n-\n-### Algorithms Benchmarks:\n-\n# init time measurement\n+if [ ! -d logs ]; then mkdir -p logs ; fi\n+if [ ! -d results ]; then mkdir -p results ; fi\nif [ ! -d results ]; then mkdir -p results ; fi\ndate >> results/times.txt\n-# TODO Use the built-in function lmPredict instead of the GLM-predict.dml script, for linear regression.\n+### Data Generation\n+#echo \"-- Generating binomial data: \" >> results/times.txt;\n+./genBinomialData.sh ${CMD} ${TEMPFOLDER} &>> logs/genBinomialData.out\n+echo \"-- Generating multinomial data.\" >> results/times.txt;\n+./genMultinomialData.sh ${CMD} ${TEMPFOLDER} &>> logs/genMultinomialData.out\n+\n+### Algorithms Benchmarks:\n./runAllBinomial.sh $CMD $TEMPFOLDER\n./runAllMultinomial.sh $CMD $TEMPFOLDER\n./runAllRegression.sh $CMD $TEMPFOLDER\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/runAllBinomial.sh", "new_path": "scripts/perftest/runAllBinomial.sh", "diff": "#-------------------------------------------------------------\nCOMMAND=$1\n-if [ \"$COMMAND\" == \"\" ]; then COMMAND=\"systemds\" ; fi\n-\nTEMPFOLDER=$2\n-if [ \"$TEMPFOLDER\" == \"\" ]; then TEMPFOLDER=temp ; fi\nBASE=${TEMPFOLDER}/binomial\nMAXITR=20\n@@ -35,17 +32,11 @@ err_report() {\n}\ntrap 'err_report $LINENO' ERR\n-if [ ! -d logs ]; then mkdir -p logs ; fi\n-if [ ! -d results ]; then mkdir -p results ; fi\n-\necho \"RUN BINOMIAL EXPERIMENTS: \"$(date) >> results/times.txt;\n-# data generation\n-echo \"-- Generating binomial data: \" >> results/times.txt;\n-./genBinomialData.sh ${COMMAND} ${TEMPFOLDER} &>> logs/genBinomialData.out\n-\n# run all classifiers with binomial labels on all datasets\n-for d in \"10k_1k_dense\" \"10k_1k_sparse\" # \"100k_1k_dense\" \"100k_1k_sparse\" \"1M_1k_dense\" \"1M_1k_sparse\" \"10M_1k_dense\" \"10M_1k_sparse\" #\"_KDD\" \"100M_1k_dense\" \"100M_1k_sparse\"\n+# see genBinomialData\n+for d in \"10k_1k_dense\" \"10k_1k_sparse\" \"100k_1k_dense\" \"100k_1k_sparse\" \"1M_1k_dense\" \"1M_1k_sparse\" \"10M_1k_dense\" \"10M_1k_sparse\" #\"_KDD\" \"100M_1k_dense\" \"100M_1k_sparse\"\ndo\nfor f in \"runMultiLogReg\" \"runL2SVM\" \"runMSVM\"\ndo\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/runAllMultinomial.sh", "new_path": "scripts/perftest/runAllMultinomial.sh", "diff": "#-------------------------------------------------------------\nCOMMAND=$1\n-if [ \"$COMMAND\" == \"\" ]; then COMMAND=\"systemds\" ; fi\n-\nTEMPFOLDER=$2\nif [ \"$TEMPFOLDER\" == \"\" ]; then TEMPFOLDER=temp ; fi\n@@ -36,17 +34,11 @@ err_report() {\n}\ntrap 'err_report $LINENO' ERR\n-if [ ! -d logs ]; then mkdir -p logs ; fi\n-if [ ! -d results ]; then mkdir -p results ; fi\n-\necho \" RUN MULTINOMIAL EXPERIMENTS: \"$(date) >> results/times.txt;\n-# data generation\n-echo \"-- Generating multinomial data.\" >> results/times.txt;\n-./genMultinomialData.sh ${COMMAND} ${TEMPFOLDER} &>> logs/genMultinomialData.out\n-\n# run all classifiers with binomial labels on all datasets\n-for d in \"10k_1k_dense\" \"10k_1k_sparse\" # \"100k_1k_dense\" \"100k_1k_sparse\" \"1M_1k_dense\" \"1M_1k_sparse\" \"10M_1k_dense\" \"10M_1k_sparse\" \"100M_1k_dense\" \"100M_1k_sparse\"\n+# see genMultinomialData\n+for d in \"10k_1k_dense\" \"10k_1k_sparse\" \"100k_1k_dense\" \"100k_1k_sparse\" \"1M_1k_dense\" \"1M_1k_sparse\" \"10M_1k_dense\" \"10M_1k_sparse\" #\"100M_1k_dense\" \"100M_1k_sparse\"\ndo\nfor f in \"runNaiveBayes\"\ndo\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/runAllRegression.sh", "new_path": "scripts/perftest/runAllRegression.sh", "diff": "#-------------------------------------------------------------\nCOMMAND=$1\n-if [ \"$COMMAND\" == \"\" ]; then COMMAND=\"systemds\" ; fi\n-\nTEMPFOLDER=$2\nif [ \"$TEMPFOLDER\" == \"\" ]; then TEMPFOLDER=temp ; fi\n@@ -35,17 +33,11 @@ err_report() {\n}\ntrap 'err_report $LINENO' ERR\n-if [ ! -d logs ]; then mkdir -p logs ; fi\n-if [ ! -d results ]; then mkdir -p results ; fi\n-\necho \"RUN REGRESSION EXPERIMENTS\" $(date) >> results/times.txt;\n-# data generation\n-echo \"-- Generating binomial data: \" >> results/times.txt;\n-./genBinomialData.sh ${COMMAND} ${TEMPFOLDER} &>> logs/genBinomialData.out\n-\n# run all regression algorithms with binomial labels on all datasets\n-for d in \"10k_1k_dense\" \"10k_1k_sparse\" # \"100k_1k_dense\" \"100k_1k_sparse\" \"1M_1k_dense\" \"1M_1k_sparse\" \"10M_1k_dense\" \"10M_1k_sparse\" #\"_KDD\" \"100M_1k_dense\" \"100M_1k_sparse\"\n+# see genBinomialData\n+for d in \"10k_1k_dense\" \"10k_1k_sparse\" \"100k_1k_dense\" \"100k_1k_sparse\" \"1M_1k_dense\" \"1M_1k_sparse\" \"10M_1k_dense\" \"10M_1k_sparse\" #\"_KDD\" \"100M_1k_dense\" \"100M_1k_sparse\"\ndo\n# -------------------------------------------------------------------------------------------------------------------\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/runGLM_binomial_probit.sh", "new_path": "scripts/perftest/runGLM_binomial_probit.sh", "diff": "@@ -30,7 +30,7 @@ for i in 0 1 2; do\n#training\ntstart=$(date +%s.%N)\n- # ${CMD} -f ../algorithms/GLM.dml \\\n+ # ${CMD} -f ./algorithms/GLM.dml \\\n${CMD} -f scripts/GLM.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n@@ -41,7 +41,7 @@ for i in 0 1 2; do\n#predict\ntstart=$(date +%s.%N)\n- ${CMD} -f ../algorithms/GLM-predict.dml \\\n+ ${CMD} -f ./algorithms/GLM-predict.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n--nvargs dfam=2 link=3 fmt=csv X=$1_test B=${BASE}/b Y=$2_test M=${BASE}/m O=${BASE}/out.csv\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/runGLM_gamma_log.sh", "new_path": "scripts/perftest/runGLM_gamma_log.sh", "diff": "@@ -30,7 +30,7 @@ for i in 0 1 2; do\n#training\ntstart=$(date +%s.%N)\n- #${CMD} -f ../algorithms/GLM.dml \\\n+ #${CMD} -f ./algorithms/GLM.dml \\\n${CMD} -f scripts/GLM.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n@@ -41,7 +41,7 @@ for i in 0 1 2; do\n#predict\ntstart=$(date +%s.%N)\n- ${CMD} -f ../algorithms/GLM-predict.dml \\\n+ ${CMD} -f ./algorithms/GLM-predict.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n--nvargs dfam=1 vpow=2.0 link=1 lpow=0.0 fmt=csv X=$1_test B=${BASE}/b Y=$2_test M=${BASE}/m O=${BASE}/out.csv\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/runGLM_poisson_log.sh", "new_path": "scripts/perftest/runGLM_poisson_log.sh", "diff": "@@ -30,7 +30,7 @@ for i in 0 1 2; do\n#training\ntstart=$(date +%s.%N)\n- #${CMD} -f ../algorithms/GLM.dml \\\n+ #${CMD} -f ./algorithms/GLM.dml \\\n${CMD} -f scripts/GLM.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n@@ -41,7 +41,7 @@ for i in 0 1 2; do\n#predict\ntstart=$(date +%s.%N)\n- ${CMD} -f ../algorithms/GLM-predict.dml \\\n+ ${CMD} -f ./algorithms/GLM-predict.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n--nvargs dfam=1 vpow=1.0 link=1 lpow=0.0 fmt=csv X=$1_test B=${BASE}/b Y=$2_test M=${BASE}/m O=${BASE}/out.csv\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/runL2SVM.sh", "new_path": "scripts/perftest/runL2SVM.sh", "diff": "@@ -30,7 +30,7 @@ for i in 0 1; do\ntstart=$(date +%s.%N)\n# /algorithms/l2-svm.dml already calls a built-in function for the l2 svm.\n- ${CMD} -f ../algorithms/l2-svm.dml \\\n+ ${CMD} -f ./algorithms/l2-svm.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n--nvargs X=$1 Y=$2 icpt=$i tol=0.0001 reg=0.01 maxiter=$5 model=${BASE}/b fmt=\"csv\"\n@@ -40,7 +40,7 @@ for i in 0 1; do\n#predict\ntstart=$(date +%s.%N)\n- #${CMD} -f ../algorithms/l2-svm-predict.dml \\\n+ #${CMD} -f ./algorithms/l2-svm-predict.dml \\\n${CMD} -f scripts/l2-svm-predict.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/runLinearRegCG.sh", "new_path": "scripts/perftest/runLinearRegCG.sh", "diff": "@@ -31,7 +31,7 @@ do\n#training\ntstart=$(date +%s.%N)\n- #${CMD} -f ../algorithms/LinearRegCG.dml \\\n+ #${CMD} -f ./algorithms/LinearRegCG.dml \\\n${CMD} -f scripts/LinearRegCG.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n@@ -42,7 +42,7 @@ do\n#predict\ntstart=$(date +%s.%N)\n- ${CMD} -f ../algorithms/GLM-predict.dml \\\n+ ${CMD} -f ./algorithms/GLM-predict.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n--nvargs dfam=1 link=1 vpow=0.0 lpow=1.0 fmt=csv X=$1_test B=${BASE}/b Y=$2_test M=${BASE}/m O=${BASE}/out.csv\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/runLinearRegDS.sh", "new_path": "scripts/perftest/runLinearRegDS.sh", "diff": "@@ -31,7 +31,7 @@ do\n#training\ntstart=$(date +%s.%N)\n- #${CMD} -f ../algorithms/LinearRegDS.dml \\\n+ #${CMD} -f ./algorithms/LinearRegDS.dml \\\n${CMD} -f scripts/LinearRegDS.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n@@ -42,7 +42,7 @@ do\n#predict\ntstart=$(date +%s.%N)\n- ${CMD} -f ../algorithms/GLM-predict.dml \\\n+ ${CMD} -f ./algorithms/GLM-predict.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n--nvargs dfam=1 link=1 vpow=0.0 lpow=1.0 fmt=csv X=$1_test B=${BASE}/b Y=$2_test M=${BASE}/m O=${BASE}/out.csv\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/runMSVM.sh", "new_path": "scripts/perftest/runMSVM.sh", "diff": "@@ -28,7 +28,7 @@ BASE=$4\nfor i in 0 1; do\n#training\ntstart=$(date +%s.%N)\n- # ${CMD} -f ../algorithms/m-svm.dml \\\n+ # ${CMD} -f ./algorithms/m-svm.dml \\\n${CMD} -f scripts/m-svm.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n@@ -39,7 +39,7 @@ for i in 0 1; do\n#predict\ntstart=$(date +%s.%N)\n- #${CMD} -f ../algorithms/m-svm-predict.dml \\\n+ #${CMD} -f ./algorithms/m-svm-predict.dml \\\n${CMD} -f scripts/m-svm-predict.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/runMultiLogReg.sh", "new_path": "scripts/perftest/runMultiLogReg.sh", "diff": "@@ -31,7 +31,7 @@ if [ $3 -gt 2 ]; then DFAM=3; fi\nfor i in 0 1 2; do\n#training\ntstart=$(date +%s.%N)\n- # ${CMD} -f ../algorithms/MultiLogReg.dml \\\n+ # ${CMD} -f ./algorithms/MultiLogReg.dml \\\n${CMD} -f scripts/MultiLogReg.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n@@ -42,7 +42,7 @@ for i in 0 1 2; do\n#predict\ntstart=$(date +%s.%N)\n- ${CMD} -f ../algorithms/GLM-predict.dml \\\n+ ${CMD} -f ./algorithms/GLM-predict.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n--nvargs dfam=$DFAM vpow=-1 link=2 lpow=-1 fmt=csv X=$1_test B=${BASE}/b Y=$2_test M=${BASE}/m O=${BASE}/out.csv\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/runNaiveBayes.sh", "new_path": "scripts/perftest/runNaiveBayes.sh", "diff": "@@ -26,7 +26,7 @@ BASE=$4\n#training\ntstart=$(date +%s.%N)\n-#${CMD} -f ../algorithms/naive-bayes.dml \\\n+#${CMD} -f ./algorithms/naive-bayes.dml \\\n${CMD} -f scripts/naive-bayes.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n@@ -37,7 +37,7 @@ echo \"NaiveBayes train on \"$1\": \"$ttrain >> results/times.txt\n#predict\ntstart=$(date +%s.%N)\n-#${CMD} -f ../algorithms/naive-bayes-predict.dml \\\n+#${CMD} -f ./algorithms/naive-bayes-predict.dml \\\n${CMD} -f scripts/naive-bayes-predict.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/scripts/extractTestData.dml", "new_path": "scripts/perftest/scripts/extractTestData.dml", "diff": "X = read($1);\ny = read($2);\n-[X,y,Xtest,ytest] = split(X=X, Y=y, f=0.8);\n+[X,Xtest,y,ytest] = split(X=X, Y=y, f=0.8);\nwrite(Xtest, $3, format=$5);\nwrite(ytest, $4, format=$5);\n" } ]

[ { "change_type": "ADD", "old_path": null, "new_path": "scripts/builtin/glmPredict.dml", "diff": "+#-------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+#-------------------------------------------------------------\n+\n+# THIS SCRIPT APPLIES THE ESTIMATED PARAMETERS OF A GLM-TYPE REGRESSION TO A NEW (TEST) DATASET\n+\n+# INPUTS PARAMETERS:\n+# ---------------------------------------------------------------------------------------------\n+# NAME TYPE DEFAULT MEANING\n+# ---------------------------------------------------------------------------------------------\n+# X Matrix --- Matrix X of records (feature vectors)\n+# B Matrix --- GLM regression parameters (the betas), with dimensions\n+# ncol(X) x k: do not add intercept\n+# ncol(X)+1 x k: add intercept as given by the last B-row\n+# if k > 1, use only B[, 1] unless it is Multinomial Logit (dfam=3)\n+# ytest Matrix \" \" Response matrix Y, with the following dimensions:\n+# nrow(X) x 1 : for all distributions (dfam=1 or 2 or 3)\n+# nrow(X) x 2 : for Binomial (dfam=2) given by (#pos, #neg) counts\n+# nrow(X) x k+1: for Multinomial (dfam=3) given by category counts\n+# dfam Int 1 GLM distribution family: 1 = Power, 2 = Binomial, 3 = Multinomial Logit\n+# vpow Double 0.0 Power for Variance defined as (mean)^power (ignored if dfam != 1):\n+# 0.0 = Gaussian, 1.0 = Poisson, 2.0 = Gamma, 3.0 = Inverse Gaussian\n+# link Int 0 Link function code: 0 = canonical (depends on distribution), 1 = Power,\n+# 2 = Logit, 3 = Probit, 4 = Cloglog, 5 = Cauchit; ignored if Multinomial\n+# lpow Double 1.0 Power for Link function defined as (mean)^power (ignored if link != 1):\n+# -2.0 = 1/mu^2, -1.0 = reciprocal, 0.0 = log, 0.5 = sqrt, 1.0 = identity\n+# disp Double 1.0 Dispersion value, when available\n+# verbose Boolean TRUE Print statistics to stdout\n+# ---------------------------------------------------------------------------------------------\n+\n+# OUTPUTS:\n+# ---------------------------------------------------------------------------------------------\n+# M Matrix \" \" Matrix M of predicted means/probabilities:\n+# nrow(X) x 1 : for Power-type distributions (dfam=1)\n+# nrow(X) x 2 : for Binomial distribution (dfam=2), column 2 is \"No\"\n+# nrow(X) x k+1: for Multinomial Logit (dfam=3), col# k+1 is baseline\n+# ---------------------------------------------------------------------------------------------\n+# Additional statistics are printed one per each line, in the following\n+# CSV format: NAME,[COLUMN],[SCALED],VALUE\n+# ---\n+# NAME is the string identifier for the statistic, see the table below.\n+# COLUMN is an optional integer value that specifies the Y-column for per-column statistics;\n+# note that a Binomial/Multinomial one-column Y input is converted into multi-column.\n+# SCALED is an optional Boolean value (TRUE or FALSE) that tells us whether or not the input\n+# dispersion parameter (disp) scaling has been applied to this statistic.\n+# VALUE is the value of the statistic.\n+# ---\n+# NAME COLUMN SCALED MEANING\n+# ---------------------------------------------------------------------------------------------\n+# LOGLHOOD_Z + Log-Likelihood Z-score (in st.dev's from mean)\n+# LOGLHOOD_Z_PVAL + Log-Likelihood Z-score p-value\n+# PEARSON_X2 + Pearson residual X^2 statistic\n+# PEARSON_X2_BY_DF + Pearson X^2 divided by degrees of freedom\n+# PEARSON_X2_PVAL + Pearson X^2 p-value\n+# DEVIANCE_G2 + Deviance from saturated model G^2 statistic\n+# DEVIANCE_G2_BY_DF + Deviance G^2 divided by degrees of freedom\n+# DEVIANCE_G2_PVAL + Deviance G^2 p-value\n+# AVG_TOT_Y + Average of Y column for a single response value\n+# STDEV_TOT_Y + St.Dev. of Y column for a single response value\n+# AVG_RES_Y + Average of column residual, i.e. of Y - mean(Y|X)\n+# STDEV_RES_Y + St.Dev. of column residual, i.e. of Y - mean(Y|X)\n+# PRED_STDEV_RES + + Model-predicted St.Dev. of column residual\n+# R2 + R^2 of Y column residual with bias included\n+# ADJUSTED_R2 + Adjusted R^2 of Y column residual with bias included\n+# R2_NOBIAS + R^2 of Y column residual with bias subtracted\n+# ADJUSTED_R2_NOBIAS + Adjusted R^2 of Y column residual with bias subtracted\n+# ---------------------------------------------------------------------------------------------\n+\n+\n+\n+m_glmPredict = function(Matrix[Double] X, Matrix[Double] B, Matrix[Double] ytest=matrix(0,0,0),\n+ Boolean intercept = FALSE, Integer dfam=1, Double vpow=0.0, Integer link=0, Double lpow=1.0,\n+ Double disp=1.0, Boolean verbose=TRUE)\n+ return(Matrix[Double] M)\n+{\n+ dist_type = dfam;\n+ link_type = link;\n+ link_power = as.double(lpow);\n+ var_power = as.double(vpow);\n+ dispersion = as.double(disp);\n+\n+ if (dist_type == 3)\n+ link_type = 2;\n+ else if (link_type == 0) { # Canonical Link\n+ if (dist_type == 1) {\n+ link_type = 1;\n+ link_power = 1.0 - var_power;\n+ }\n+ else if (dist_type == 2)\n+ link_type = 2;\n+ }\n+\n+ num_records = nrow (X);\n+ num_features = ncol (X);\n+ B_full = B;\n+ if (dist_type == 3) {\n+ beta = B_full [1 : ncol (X), ];\n+ intercept = B_full [nrow(B_full), ];\n+ } else {\n+ beta = B_full [1 : ncol (X), 1];\n+ intercept = B_full [nrow(B_full), 1];\n+ }\n+ if (nrow (B_full) == ncol (X)) {\n+ intercept = 0.0 * intercept;\n+ is_intercept = FALSE;\n+ } else {\n+ num_features = num_features + 1;\n+ is_intercept = TRUE;\n+ }\n+\n+ ones_rec = matrix (1, rows = num_records, cols = 1);\n+ linear_terms = X %*% beta + ones_rec %*% intercept;\n+ [means, vars] =\n+ glm_means_and_vars (linear_terms, dist_type, var_power, link_type, link_power);\n+\n+ M = means;\n+\n+ if (nrow(ytest) > 0)\n+ {\n+ Y = ytest;\n+ ones_ctg = matrix (1, rows = ncol(Y), cols = 1);\n+\n+ # Statistics To Compute:\n+ Z_logl = NaN;\n+ Z_logl_pValue = NaN;\n+ X2_pearson = NaN;\n+ df_pearson = -1;\n+ G2_deviance = NaN;\n+ df_deviance = -1;\n+ X2_pearson_pValue = NaN;\n+ G2_deviance_pValue = NaN;\n+ Z_logl_scaled = NaN;\n+ Z_logl_scaled_pValue = NaN;\n+ X2_scaled = NaN;\n+ X2_scaled_pValue = NaN;\n+ G2_scaled = NaN;\n+ G2_scaled_pValue = NaN;\n+\n+ # set Y_counts to avoid 'Initialization of Y_counts depends on if-else execution' warning\n+ Y_counts = matrix(0.0, rows=1, cols=1);\n+\n+ if (dist_type == 1 & link_type == 1) {\n+ # POWER DISTRIBUTIONS (GAUSSIAN, POISSON, GAMMA, ETC.)\n+ if (link_power == 0) {\n+ is_zero_Y = (Y == 0);\n+ lt_saturated = log (Y + is_zero_Y) - is_zero_Y / (1.0 - is_zero_Y);\n+ }\n+ else\n+ lt_saturated = Y ^ link_power;\n+ Y_counts = ones_rec;\n+\n+ X2_pearson = sum ((Y - means) ^ 2 / vars);\n+ df_pearson = num_records - num_features;\n+\n+ log_l_part = glm_partial_loglikelihood_for_power_dist_and_link (linear_terms, Y, var_power, link_power);\n+ log_l_part_saturated = glm_partial_loglikelihood_for_power_dist_and_link (lt_saturated, Y, var_power, link_power);\n+\n+ G2_deviance = 2 * sum (log_l_part_saturated) - 2 * sum (log_l_part);\n+ df_deviance = num_records - num_features;\n+ }\n+ else {\n+ if (dist_type >= 2) {\n+ # BINOMIAL AND MULTINOMIAL DISTRIBUTIONS\n+ if (ncol (Y) == 1) {\n+ num_categories = ncol (beta) + 1;\n+ if (min (Y) <= 0) {\n+ # Category labels \"0\", \"-1\" etc. are converted into the baseline label\n+ Y = Y + (- Y + num_categories) * (Y <= 0);\n+ }\n+ Y_size = min (num_categories, max(Y));\n+ Y_unsized = table (seq (1, num_records, 1), Y);\n+ Y = matrix (0, rows = num_records, cols = num_categories);\n+ Y [, 1 : Y_size] = Y_unsized [, 1 : Y_size];\n+ Y_counts = ones_rec;\n+ } else {\n+ Y_counts = rowSums (Y);\n+ }\n+\n+ P = means;\n+ zero_Y = (Y == 0);\n+ zero_P = (P == 0);\n+ ones_ctg = matrix (1, rows = ncol(Y), cols = 1);\n+\n+ logl_vec = rowSums (Y * log (P + zero_Y) );\n+ ent1_vec = rowSums (P * log (P + zero_P) );\n+ ent2_vec = rowSums (P * (log (P + zero_P))^2);\n+ E_logl = sum (Y_counts * ent1_vec);\n+ V_logl = sum (Y_counts * (ent2_vec - ent1_vec ^ 2));\n+ Z_logl = (sum (logl_vec) - E_logl) / sqrt (V_logl);\n+\n+ means = means * (Y_counts %*% t(ones_ctg));\n+ vars = vars * (Y_counts %*% t(ones_ctg));\n+\n+ frac_below_5 = sum (means < 5) / (nrow (means) * ncol (means));\n+ frac_below_1 = sum (means < 1) / (nrow (means) * ncol (means));\n+\n+ if (frac_below_5 > 0.2 | frac_below_1 > 0)\n+ print (\"WARNING: residual statistics are inaccurate here due to low cell means.\");\n+\n+ X2_pearson = sum ((Y - means) ^ 2 / means);\n+ df_pearson = (num_records - num_features) * (ncol(Y) - 1);\n+ G2_deviance = 2 * sum (Y * log ((Y + zero_Y) / (means + zero_Y)));\n+ df_deviance = (num_records - num_features) * (ncol(Y) - 1);\n+ }}\n+\n+ if (Z_logl == Z_logl) {\n+ Z_logl_absneg = - abs (Z_logl);\n+ Z_logl_pValue = 2.0 * pnorm(target = Z_logl_absneg);\n+ }\n+ if (X2_pearson == X2_pearson & df_pearson > 0)\n+ X2_pearson_pValue = pchisq(target = X2_pearson, df = df_pearson, lower.tail=FALSE);\n+ if (G2_deviance == G2_deviance & df_deviance > 0)\n+ G2_deviance_pValue = pchisq(target = G2_deviance, df = df_deviance, lower.tail=FALSE);\n+\n+ Z_logl_scaled = Z_logl / sqrt (dispersion);\n+ X2_scaled = X2_pearson / dispersion;\n+ G2_scaled = G2_deviance / dispersion;\n+\n+ if (Z_logl_scaled == Z_logl_scaled) {\n+ Z_logl_scaled_absneg = - abs (Z_logl_scaled);\n+ Z_logl_scaled_pValue = 2.0 * pnorm(target = Z_logl_scaled_absneg);\n+ }\n+ if (X2_scaled == X2_scaled & df_pearson > 0)\n+ X2_scaled_pValue = pchisq(target = X2_scaled, df = df_pearson, lower.tail=FALSE);\n+ if (G2_scaled == G2_scaled & df_deviance > 0)\n+ G2_scaled_pValue = pchisq(target = G2_scaled, df = df_deviance, lower.tail=FALSE);\n+\n+ avg_tot_Y = colSums ( Y ) / sum (Y_counts);\n+ avg_res_Y = colSums (Y - means) / sum (Y_counts);\n+ ss_avg_tot_Y = colSums (( Y - Y_counts %*% avg_tot_Y) ^ 2);\n+ ss_res_Y = colSums ((Y - means) ^ 2);\n+ ss_avg_res_Y = colSums ((Y - means - Y_counts %*% avg_res_Y) ^ 2);\n+ df_ss_res_Y = sum (Y_counts) - num_features;\n+ df_ss_avg_res_Y = ifelse(is_intercept, df_ss_res_Y, df_ss_res_Y - 1);\n+\n+ var_tot_Y = ss_avg_tot_Y / (sum (Y_counts) - 1);\n+ if (df_ss_avg_res_Y > 0)\n+ var_res_Y = ss_avg_res_Y / df_ss_avg_res_Y;\n+ else\n+ var_res_Y = matrix (0.0, rows = 1, cols = ncol (Y)) / 0.0;\n+ R2_nobias = 1 - ss_avg_res_Y / ss_avg_tot_Y;\n+ adjust_R2_nobias = 1 - var_res_Y / var_tot_Y;\n+ R2 = 1 - ss_res_Y / ss_avg_tot_Y;\n+ if (df_ss_res_Y > 0)\n+ adjust_R2 = 1 - (ss_res_Y / df_ss_res_Y) / var_tot_Y;\n+ else\n+ adjust_R2 = matrix (0.0, rows = 1, cols = ncol (Y)) / 0.0;\n+ predicted_avg_var_res_Y = dispersion * colSums (vars) / sum (Y_counts);\n+\n+ # PREPARING THE OUTPUT CSV STATISTICS FILE\n+\n+ str = \"LOGLHOOD_Z,,FALSE,\" + Z_logl;\n+ str = append (str, \"LOGLHOOD_Z_PVAL,,FALSE,\" + Z_logl_pValue);\n+ str = append (str, \"PEARSON_X2,,FALSE,\" + X2_pearson);\n+ str = append (str, \"PEARSON_X2_BY_DF,,FALSE,\" + (X2_pearson / df_pearson));\n+ str = append (str, \"PEARSON_X2_PVAL,,FALSE,\" + X2_pearson_pValue);\n+ str = append (str, \"DEVIANCE_G2,,FALSE,\" + G2_deviance);\n+ str = append (str, \"DEVIANCE_G2_BY_DF,,FALSE,\" + (G2_deviance / df_deviance));\n+ str = append (str, \"DEVIANCE_G2_PVAL,,FALSE,\" + G2_deviance_pValue);\n+ str = append (str, \"LOGLHOOD_Z,,TRUE,\" + Z_logl_scaled);\n+ str = append (str, \"LOGLHOOD_Z_PVAL,,TRUE,\" + Z_logl_scaled_pValue);\n+ str = append (str, \"PEARSON_X2,,TRUE,\" + X2_scaled);\n+ str = append (str, \"PEARSON_X2_BY_DF,,TRUE,\" + (X2_scaled / df_pearson));\n+ str = append (str, \"PEARSON_X2_PVAL,,TRUE,\" + X2_scaled_pValue);\n+ str = append (str, \"DEVIANCE_G2,,TRUE,\" + G2_scaled);\n+ str = append (str, \"DEVIANCE_G2_BY_DF,,TRUE,\" + (G2_scaled / df_deviance));\n+ str = append (str, \"DEVIANCE_G2_PVAL,,TRUE,\" + G2_scaled_pValue);\n+\n+ for (i in 1:ncol(Y)) {\n+ str = append (str, \"AVG_TOT_Y,\" + i + \",,\" + as.scalar (avg_tot_Y [1, i]));\n+ str = append (str, \"STDEV_TOT_Y,\" + i + \",,\" + as.scalar (sqrt (var_tot_Y [1, i])));\n+ str = append (str, \"AVG_RES_Y,\" + i + \",,\" + as.scalar (avg_res_Y [1, i]));\n+ str = append (str, \"STDEV_RES_Y,\" + i + \",,\" + as.scalar (sqrt (var_res_Y [1, i])));\n+ str = append (str, \"PRED_STDEV_RES,\" + i + \",TRUE,\" + as.scalar (sqrt (predicted_avg_var_res_Y [1, i])));\n+ str = append (str, \"R2,\" + i + \",,\" + as.scalar (R2 [1, i]));\n+ str = append (str, \"ADJUSTED_R2,\" + i + \",,\" + as.scalar (adjust_R2 [1, i]));\n+ str = append (str, \"R2_NOBIAS,\" + i + \",,\" + as.scalar (R2_nobias [1, i]));\n+ str = append (str, \"ADJUSTED_R2_NOBIAS,\" + i + \",,\" + as.scalar (adjust_R2_nobias [1, i]));\n+ }\n+\n+ if( verbose )\n+ print(str);\n+ }\n+}\n+\n+glm_means_and_vars =\n+ function (Matrix[double] linear_terms, int dist_type, double var_power, int link_type, double link_power)\n+ return (Matrix[double] means, Matrix[double] vars)\n+ # NOTE: \"vars\" represents the variance without dispersion, i.e. the V(mu) function.\n+{\n+ num_points = nrow (linear_terms);\n+ if (dist_type == 1 & link_type == 1) {\n+ # POWER DISTRIBUTION\n+ if (link_power == 0)\n+ y_mean = exp (linear_terms);\n+ else if (link_power == 1.0)\n+ y_mean = linear_terms;\n+ else if (link_power == -1.0)\n+ y_mean = 1.0 / linear_terms;\n+ else\n+ y_mean = linear_terms ^ (1.0 / link_power);\n+ if (var_power == 0)\n+ var_function = matrix (1.0, rows = num_points, cols = 1);\n+ else if (var_power == 1.0)\n+ var_function = y_mean;\n+ else\n+ var_function = y_mean ^ var_power;\n+ means = y_mean;\n+ vars = var_function;\n+ }\n+ else if (dist_type == 2 & link_type >= 1 & link_type <= 5) {\n+ # BINOMIAL/BERNOULLI DISTRIBUTION\n+ y_prob = matrix (0.0, rows = num_points, cols = 2);\n+ if(link_type == 1 & link_power == 0) { # Binomial.log\n+ y_prob [, 1] = exp (linear_terms);\n+ y_prob [, 2] = 1.0 - y_prob [, 1];\n+ } else if (link_type == 1 & link_power != 0) { # Binomial.power_nonlog\n+ y_prob [, 1] = linear_terms ^ (1.0 / link_power);\n+ y_prob [, 2] = 1.0 - y_prob [, 1];\n+ } else if (link_type == 2) { # Binomial.logit\n+ elt = exp (linear_terms);\n+ y_prob [, 1] = elt / (1.0 + elt);\n+ y_prob [, 2] = 1.0 / (1.0 + elt);\n+ } else if (link_type == 3) { # Binomial.probit\n+ sign_lt = 2 * (linear_terms >= 0) - 1;\n+ t_gp = 1.0 / (1.0 + abs (linear_terms) * 0.231641888); # 0.231641888 = 0.3275911 / sqrt (2.0)\n+ erf_corr =\n+ t_gp * ( 0.254829592\n+ + t_gp * (-0.284496736 # \"Handbook of Mathematical Functions\", ed. by M. Abramowitz and I.A. Stegun,\n+ + t_gp * ( 1.421413741 # U.S. Nat-l Bureau of Standards, 10th print (Dec 1972), Sec. 7.1.26, p. 299\n+ + t_gp * (-1.453152027\n+ + t_gp * 1.061405429)))) * sign_lt * exp (- (linear_terms ^ 2) / 2.0);\n+ y_prob [, 1] = (1 + sign_lt) - erf_corr;\n+ y_prob [, 2] = (1 - sign_lt) + erf_corr;\n+ y_prob = y_prob / 2;\n+ } else if (link_type == 4) { # Binomial.cloglog\n+ elt = exp (linear_terms);\n+ is_too_small = ((10000000 + elt) == 10000000);\n+ y_prob [, 2] = exp (- elt);\n+ y_prob [, 1] = (1 - is_too_small) * (1.0 - y_prob [, 2]) + is_too_small * elt * (1.0 - elt / 2);\n+ } else if (link_type == 5) { # Binomial.cauchit\n+ atan_linear_terms = atan (linear_terms);\n+ y_prob [, 1] = 0.5 + atan_linear_terms / pi;\n+ y_prob [, 2] = 0.5 - atan_linear_terms / pi;\n+ }\n+ means = y_prob;\n+ ones_ctg = matrix (1, rows = 2, cols = 1);\n+ vars = means * (means %*% (1 - diag (ones_ctg)));\n+ } else if (dist_type == 3) {\n+ # MULTINOMIAL LOGIT DISTRIBUTION\n+ elt = exp (linear_terms);\n+ ones_pts = matrix (1, rows = num_points, cols = 1);\n+ elt = cbind (elt, ones_pts);\n+ ones_ctg = matrix (1, rows = ncol (elt), cols = 1);\n+ means = elt / (rowSums (elt) %*% t(ones_ctg));\n+ vars = means * (means %*% (1 - diag (ones_ctg)));\n+ } else {\n+ means = matrix (0.0, rows = num_points, cols = 1);\n+ vars = matrix (0.0, rows = num_points, cols = 1);\n+ }\n+}\n+\n+glm_partial_loglikelihood_for_power_dist_and_link = # Assumes: dist_type == 1 & link_type == 1\n+ function (Matrix[double] linear_terms, Matrix[double] Y, double var_power, double link_power)\n+ return (Matrix[double] log_l_part)\n+{\n+ num_records = nrow (Y);\n+ if (var_power == 1.0) { # Poisson\n+ if (link_power == 0) { # Poisson.log\n+ is_natural_parameter_log_zero = (linear_terms == -Inf);\n+ natural_parameters = replace (target = linear_terms, pattern = -Inf, replacement = 0);\n+ b_cumulant = exp (linear_terms);\n+ } else { # Poisson.power_nonlog\n+ is_natural_parameter_log_zero = (linear_terms == 0);\n+ natural_parameters = log (linear_terms + is_natural_parameter_log_zero) / link_power;\n+ b_cumulant = (linear_terms + is_natural_parameter_log_zero) ^ (1.0 / link_power) - is_natural_parameter_log_zero;\n+ }\n+ is_minus_infinity = (Y > 0) * is_natural_parameter_log_zero;\n+ log_l_part = Y * natural_parameters - b_cumulant - is_minus_infinity / (1 - is_minus_infinity);\n+ }\n+ else {\n+ if (var_power == 2.0 & link_power == 0) { # Gamma.log\n+ natural_parameters = - exp (- linear_terms);\n+ b_cumulant = linear_terms;\n+ }\n+ else if (var_power == 2.0) { # Gamma.power_nonlog\n+ natural_parameters = - linear_terms ^ (- 1.0 / link_power);\n+ b_cumulant = log (linear_terms) / link_power;\n+ }\n+ else if (link_power == 0) { # PowerDist.log\n+ natural_parameters = exp (linear_terms * (1.0 - var_power)) / (1.0 - var_power);\n+ b_cumulant = exp (linear_terms * (2.0 - var_power)) / (2.0 - var_power);\n+ }\n+ else { # PowerDist.power_nonlog\n+ power_np = (1.0 - var_power) / link_power;\n+ natural_parameters = (linear_terms ^ power_np) / (1.0 - var_power);\n+ power_cu = (2.0 - var_power) / link_power;\n+ b_cumulant = (linear_terms ^ power_cu) / (2.0 - var_power);\n+ }\n+ log_l_part = Y * natural_parameters - b_cumulant;\n+ }\n+}\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/runAll.sh", "new_path": "scripts/perftest/runAll.sh", "diff": "@@ -48,7 +48,7 @@ if [ ! -d results ]; then mkdir -p results ; fi\ndate >> results/times.txt\n### Data Generation\n-#echo \"-- Generating binomial data: \" >> results/times.txt;\n+echo \"-- Generating binomial data: \" >> results/times.txt;\n./genBinomialData.sh ${CMD} ${TEMPFOLDER} &>> logs/genBinomialData.out\necho \"-- Generating multinomial data.\" >> results/times.txt;\n./genMultinomialData.sh ${CMD} ${TEMPFOLDER} &>> logs/genMultinomialData.out\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/runGLM_binomial_probit.sh", "new_path": "scripts/perftest/runGLM_binomial_probit.sh", "diff": "@@ -41,7 +41,7 @@ for i in 0 1 2; do\n#predict\ntstart=$(date +%s.%N)\n- ${CMD} -f ./algorithms/GLM-predict.dml \\\n+ ${CMD} -f scripts/GLM-predict.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n--nvargs dfam=2 link=3 fmt=csv X=$1_test B=${BASE}/b Y=$2_test M=${BASE}/m O=${BASE}/out.csv\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/runGLM_gamma_log.sh", "new_path": "scripts/perftest/runGLM_gamma_log.sh", "diff": "@@ -41,7 +41,7 @@ for i in 0 1 2; do\n#predict\ntstart=$(date +%s.%N)\n- ${CMD} -f ./algorithms/GLM-predict.dml \\\n+ ${CMD} -f scripts/GLM-predict.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n--nvargs dfam=1 vpow=2.0 link=1 lpow=0.0 fmt=csv X=$1_test B=${BASE}/b Y=$2_test M=${BASE}/m O=${BASE}/out.csv\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/runGLM_poisson_log.sh", "new_path": "scripts/perftest/runGLM_poisson_log.sh", "diff": "@@ -41,7 +41,7 @@ for i in 0 1 2; do\n#predict\ntstart=$(date +%s.%N)\n- ${CMD} -f ./algorithms/GLM-predict.dml \\\n+ ${CMD} -f scripts/GLM-predict.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n--nvargs dfam=1 vpow=1.0 link=1 lpow=0.0 fmt=csv X=$1_test B=${BASE}/b Y=$2_test M=${BASE}/m O=${BASE}/out.csv\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/runLinearRegCG.sh", "new_path": "scripts/perftest/runLinearRegCG.sh", "diff": "@@ -42,7 +42,7 @@ do\n#predict\ntstart=$(date +%s.%N)\n- ${CMD} -f ./algorithms/GLM-predict.dml \\\n+ ${CMD} -f scripts/GLM-predict.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n--nvargs dfam=1 link=1 vpow=0.0 lpow=1.0 fmt=csv X=$1_test B=${BASE}/b Y=$2_test M=${BASE}/m O=${BASE}/out.csv\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/runLinearRegDS.sh", "new_path": "scripts/perftest/runLinearRegDS.sh", "diff": "@@ -42,7 +42,7 @@ do\n#predict\ntstart=$(date +%s.%N)\n- ${CMD} -f ./algorithms/GLM-predict.dml \\\n+ ${CMD} -f scripts/GLM-predict.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n--nvargs dfam=1 link=1 vpow=0.0 lpow=1.0 fmt=csv X=$1_test B=${BASE}/b Y=$2_test M=${BASE}/m O=${BASE}/out.csv\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/runMultiLogReg.sh", "new_path": "scripts/perftest/runMultiLogReg.sh", "diff": "@@ -42,7 +42,7 @@ for i in 0 1 2; do\n#predict\ntstart=$(date +%s.%N)\n- ${CMD} -f ./algorithms/GLM-predict.dml \\\n+ ${CMD} -f scripts/GLM-predict.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n--nvargs dfam=$DFAM vpow=-1 link=2 lpow=-1 fmt=csv X=$1_test B=${BASE}/b Y=$2_test M=${BASE}/m O=${BASE}/out.csv\n" }, { "change_type": "ADD", "old_path": null, "new_path": "scripts/perftest/scripts/GLM-predict.dml", "diff": "+#-------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+#-------------------------------------------------------------\n+\n+X = read($X);\n+B = read($B);\n+Y = matrix(0,0,0);\n+if($Y != \" \")\n+ Y = read($Y);\n+\n+dfam = ifdef ($dfam, 1); # $dfam = 1;\n+vpow = ifdef ($vpow, 0.0); # $vpow = 0.0;\n+link = ifdef ($link, 0); # $link = 0;\n+lpow = ifdef ($lpow, 1.0); # $lpow = 1.0;\n+disp = ifdef ($disp, 1.0); # $disp = 1.0;\n+\n+\n+[M] = glmPredict(X=X, B=B, ytest=Y,\n+ dfam=dfam, vpow=vpow, link=link, lpow=lpow, disp=disp);\n+\n+if( $M != \" \" )\n+ write(M, $M, format=$fmt);\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/common/Builtins.java", "new_path": "src/main/java/org/apache/sysds/common/Builtins.java", "diff": "@@ -130,6 +130,7 @@ public enum Builtins {\nGAUSSIAN_CLASSIFIER(\"gaussianClassifier\", true),\nGET_ACCURACY(\"getAccuracy\", true),\nGLM(\"glm\", true),\n+ GLM_PREDICT(\"glmPredict\", true),\nGMM(\"gmm\", true),\nGMM_PREDICT(\"gmmPredict\", true),\nGNMF(\"gnmf\", true),\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/builtin/lmpredict.dml", "new_path": "src/test/scripts/functions/builtin/lmpredict.dml", "diff": "@@ -24,4 +24,9 @@ y = read($2) # response values\np = read($3) # random data to predict\nw = lmDS(X = X, y = y, icpt = 1, reg = 1e-12)\np = lmPredict(X = X, B = w, ytest=matrix(0,1,1), icpt = 1)\n+p2 = glmPredict(X = X, B = w, dfam=1, link=1, vpow=0.0, lpow=1.0);\n+\n+if( sum(abs(p2-p) > 1e8) !=0 )\n+ stop(\"Mismatching lmPredict and glmPredict - no output written\");\n+\nwrite(p, $4)\n" } ]

[ { "change_type": "MODIFY", "old_path": "docker/entrypoint.sh", "new_path": "docker/entrypoint.sh", "diff": "cd /github/workspace\n+export MAVEN_OPTS=\"-Xmx512m -XX:MaxPermSize=128m\"\n+\nlog=\"/tmp/sysdstest.log\"\nmvn -ntp test-compile 2>&1 | grep -E \"BUILD|Total time:|---|Building SystemDS\"\nmvn -ntp test -D maven.test.skip=false -D automatedtestbase.outputbuffering=true -D test=$1 2>&1 | grep -v \"already exists in destination.\" | tee $log\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/hops/rewrite/IPAPassRewriteFederatedPlan.java", "new_path": "src/main/java/org/apache/sysds/hops/rewrite/IPAPassRewriteFederatedPlan.java", "diff": "@@ -252,7 +252,6 @@ public class IPAPassRewriteFederatedPlan extends IPAPass {\nif ( hopRels.isEmpty() )\nhopRels.add(new HopRel(currentHop, FEDInstruction.FederatedOutput.NONE, hopRelMemo));\nhopRelMemo.put(currentHop.getHopID(), hopRels);\n- currentHop.setVisited();\n}\n/**\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/lops/Lop.java", "new_path": "src/main/java/org/apache/sysds/lops/Lop.java", "diff": "@@ -117,9 +117,9 @@ public abstract class Lop\nprotected PrivacyConstraint privacyConstraint;\n/**\n- * Boolean defining if the output of the operation should be federated.\n- * If it is true, the output should be kept at federated sites.\n- * If it is false, the output should be retrieved by the coordinator.\n+ * Enum defining if the output of the operation should be forced federated, forced local or neither.\n+ * If it is FOUT, the output should be kept at federated sites.\n+ * If it is LOUT, the output should be retrieved by the coordinator.\n*/\nprotected FederatedOutput _fedOutput = null;\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/instructions/FEDInstructionParser.java", "new_path": "src/main/java/org/apache/sysds/runtime/instructions/FEDInstructionParser.java", "diff": "@@ -63,8 +63,9 @@ public class FEDInstructionParser extends InstructionParser\n// Reorg Instruction Opcodes (repositioning of existing values)\nString2FEDInstructionType.put( \"r'\" , FEDType.Reorg );\nString2FEDInstructionType.put( \"rdiag\" , FEDType.Reorg );\n- String2FEDInstructionType.put( \"rshape\" , FEDType.Reorg );\nString2FEDInstructionType.put( \"rev\" , FEDType.Reorg );\n+ //String2FEDInstructionType.put( \"rshape\" , FEDType.Reorg ); Not supported by ReorgFEDInstruction parser!\n+ //String2FEDInstructionType.put( \"rsort\" , FEDType.Reorg ); Not supported by ReorgFEDInstruction parser!\n// Ternary Instruction Opcodes\nString2FEDInstructionType.put( \"+*\" , FEDType.Ternary);\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/instructions/InstructionUtils.java", "new_path": "src/main/java/org/apache/sysds/runtime/instructions/InstructionUtils.java", "diff": "@@ -85,6 +85,7 @@ import org.apache.sysds.runtime.functionobjects.Builtin.BuiltinCode;\nimport org.apache.sysds.runtime.instructions.cp.CPOperand;\nimport org.apache.sysds.runtime.instructions.cp.CPInstruction.CPType;\nimport org.apache.sysds.runtime.instructions.fed.FEDInstruction.FEDType;\n+import org.apache.sysds.runtime.instructions.fed.FEDInstruction.FederatedOutput;\nimport org.apache.sysds.runtime.instructions.gpu.GPUInstruction.GPUINSTRUCTION_TYPE;\nimport org.apache.sysds.runtime.instructions.spark.SPInstruction.SPType;\nimport org.apache.sysds.runtime.matrix.data.LibCommonsMath;\n@@ -1144,8 +1145,30 @@ public class InstructionUtils\nreturn linst;\n}\n+ /**\n+ * Removes federated output flag from the end of the instruction string if the flag is present.\n+ * @param linst instruction string\n+ * @return instruction string with no federated output flag\n+ */\npublic static String removeFEDOutputFlag(String linst){\n- return linst.substring(0, linst.lastIndexOf(Lop.OPERAND_DELIMITOR));\n+ int lastOperandStartIndex = linst.lastIndexOf(Lop.OPERAND_DELIMITOR);\n+ String lastOperand = linst.substring(lastOperandStartIndex);\n+ if ( containsFEDOutputFlag(lastOperand) )\n+ return linst.substring(0, lastOperandStartIndex);\n+ else return linst;\n+ }\n+\n+ /**\n+ * Checks whether the given operand string contains a federated output flag\n+ * @param operandString which is checked for federated output flag\n+ * @return true if the given operand string contains a federated output flag\n+ */\n+ private static boolean containsFEDOutputFlag(String operandString){\n+ for (FederatedOutput fedOutput : FederatedOutput.values()){\n+ if ( operandString.contains(fedOutput.name()) )\n+ return true;\n+ }\n+ return false;\n}\nprivate static String replaceOperand(String linst, CPOperand oldOperand, String newOperandName){\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/instructions/fed/ReorgFEDInstruction.java", "new_path": "src/main/java/org/apache/sysds/runtime/instructions/fed/ReorgFEDInstruction.java", "diff": "@@ -54,8 +54,6 @@ import org.apache.sysds.runtime.matrix.operators.ReorgOperator;\nimport org.apache.sysds.runtime.meta.MatrixCharacteristics;\npublic class ReorgFEDInstruction extends UnaryFEDInstruction {\n- @SuppressWarnings(\"unused\")\n- private static boolean fedoutFlagInString = false;\npublic ReorgFEDInstruction(Operator op, CPOperand in1, CPOperand out, String opcode, String istr, FederatedOutput fedOut) {\nsuper(FEDType.Reorg, op, in1, out, opcode, istr, fedOut);\n@@ -71,23 +69,25 @@ public class ReorgFEDInstruction extends UnaryFEDInstruction {\nString[] parts = InstructionUtils.getInstructionPartsWithValueType(str);\nString opcode = parts[0];\n+ FederatedOutput fedOut;\nif ( opcode.equalsIgnoreCase(\"r'\") ) {\nInstructionUtils.checkNumFields(str, 2, 3, 4);\nin.split(parts[1]);\nout.split(parts[2]);\nint k = str.startsWith(Types.ExecMode.SPARK.name()) ? 0 : Integer.parseInt(parts[3]);\n- FederatedOutput fedOut = str.startsWith(Types.ExecMode.SPARK.name()) ? FederatedOutput.valueOf(parts[3]) :\n- FederatedOutput.valueOf(parts[4]);\n+ fedOut = str.startsWith(Types.ExecMode.SPARK.name()) ?\n+ FederatedOutput.valueOf(parts[3]) : FederatedOutput.valueOf(parts[4]);\nreturn new ReorgFEDInstruction(new ReorgOperator(SwapIndex.getSwapIndexFnObject(), k), in, out, opcode, str, fedOut);\n}\nelse if ( opcode.equalsIgnoreCase(\"rdiag\") ) {\nparseUnaryInstruction(str, in, out); //max 2 operands\n- return new ReorgFEDInstruction(new ReorgOperator(DiagIndex.getDiagIndexFnObject()), in, out, opcode, str);\n+ fedOut = parseFedOutFlag(str, 3);\n+ return new ReorgFEDInstruction(new ReorgOperator(DiagIndex.getDiagIndexFnObject()), in, out, opcode, str, fedOut);\n}\nelse if ( opcode.equalsIgnoreCase(\"rev\") ) {\n- fedoutFlagInString = parts.length > 3;\nparseUnaryInstruction(str, in, out); //max 2 operands\n- return new ReorgFEDInstruction(new ReorgOperator(RevIndex.getRevIndexFnObject()), in, out, opcode, str);\n+ fedOut = parseFedOutFlag(str, 3);\n+ return new ReorgFEDInstruction(new ReorgOperator(RevIndex.getRevIndexFnObject()), in, out, opcode, str, fedOut);\n}\nelse {\nthrow new DMLRuntimeException(\"ReorgFEDInstruction: unsupported opcode: \"+opcode);\n@@ -117,7 +117,6 @@ public class ReorgFEDInstruction extends UnaryFEDInstruction {\nmo1.getFedMapping().execute(getTID(), true, fr, fr1);\nif (_fedOut != null && !_fedOut.isForcedLocal()){\n- mo1.getFedMapping().execute(getTID(), true, fr1);\n//drive output federated mapping\nMatrixObject out = ec.getMatrixObject(output);\nout.getDataCharacteristics().set(mo1.getNumColumns(), mo1.getNumRows(), (int) mo1.getBlocksize(), mo1.getNnz());\n@@ -146,10 +145,7 @@ public class ReorgFEDInstruction extends UnaryFEDInstruction {\nout.getDataCharacteristics().set(mo1.getNumRows(), mo1.getNumColumns(), (int) mo1.getBlocksize(), mo1.getNnz());\nout.setFedMapping(mo1.getFedMapping().copyWithNewID(fr1.getID()));\n- if ( _fedOut != null && _fedOut.isForcedLocal() ){\n- out.acquireReadAndRelease();\n- out.getFedMapping().cleanup(getTID(), fr1.getID());\n- }\n+ optionalForceLocal(out);\n}\nelse if (instOpcode.equals(\"rdiag\")) {\nRdiagResult result;\n@@ -160,10 +156,27 @@ public class ReorgFEDInstruction extends UnaryFEDInstruction {\nresult = rdiagM2V(mo1, r_op);\n}\n+ FederationMap diagFedMap = updateFedRanges(result);\n+\n+ //update output mapping and data characteristics\n+ MatrixObject rdiag = ec.getMatrixObject(output);\n+ rdiag.getDataCharacteristics()\n+ .set(diagFedMap.getMaxIndexInRange(0), diagFedMap.getMaxIndexInRange(1),\n+ (int) mo1.getBlocksize());\n+ rdiag.setFedMapping(diagFedMap);\n+ optionalForceLocal(rdiag);\n+ }\n+ }\n+\n+ /**\n+ * Update the federated ranges of result and return the updated federation map.\n+ * @param result RdiagResult for which the fedmap is updated\n+ * @return updated federation map\n+ */\n+ private FederationMap updateFedRanges(RdiagResult result){\nFederationMap diagFedMap = result.getFedMap();\nMap<FederatedRange, int[]> dcs = result.getDcs();\n- //update fed ranges\nfor(int i = 0; i < diagFedMap.getFederatedRanges().length; i++) {\nint[] newRange = dcs.get(diagFedMap.getFederatedRanges()[i]);\n@@ -178,17 +191,17 @@ public class ReorgFEDInstruction extends UnaryFEDInstruction {\ndiagFedMap.getFederatedRanges()[i].setEndDim(1,\ndiagFedMap.getFederatedRanges()[i].getBeginDims()[1] + newRange[1]);\n}\n+ return diagFedMap;\n+ }\n- //update output mapping and data characteristics\n- MatrixObject rdiag = ec.getMatrixObject(output);\n- rdiag.getDataCharacteristics()\n- .set(diagFedMap.getMaxIndexInRange(0), diagFedMap.getMaxIndexInRange(1),\n- (int) mo1.getBlocksize());\n- rdiag.setFedMapping(diagFedMap);\n+ /**\n+ * If federated output is forced local, the output will be retrieved and removed from federated workers.\n+ * @param outputMatrixObject which will be retrieved and removed from federated workers\n+ */\n+ private void optionalForceLocal(MatrixObject outputMatrixObject){\nif ( _fedOut != null && _fedOut.isForcedLocal() ){\n- rdiag.acquireReadAndRelease();\n- rdiag.getFedMapping().cleanup(getTID(), rdiag.getFedMapping().getID());\n- }\n+ outputMatrixObject.acquireReadAndRelease();\n+ outputMatrixObject.getFedMapping().cleanup(getTID(), outputMatrixObject.getFedMapping().getID());\n}\n}\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/instructions/fed/UnaryFEDInstruction.java", "new_path": "src/main/java/org/apache/sysds/runtime/instructions/fed/UnaryFEDInstruction.java", "diff": "@@ -110,4 +110,18 @@ public abstract class UnaryFEDInstruction extends ComputationFEDInstruction {\nout.split(parts[parts.length - 2]);\nreturn opcode;\n}\n+\n+ /**\n+ * Parse and return federated output flag from given instr string at given position.\n+ * If the position given is greater than the length of the instruction, FederatedOutput.NONE is returned.\n+ * @param instr instruction string to be parsed\n+ * @param position of federated output flag\n+ * @return parsed federated output flag or FederatedOutput.NONE\n+ */\n+ static FederatedOutput parseFedOutFlag(String instr, int position){\n+ String[] parts = InstructionUtils.getInstructionPartsWithValueType(instr);\n+ if ( parts.length > position )\n+ return FederatedOutput.valueOf(parts[position]);\n+ else return FederatedOutput.NONE;\n+ }\n}\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/federated/primitives/FederatedRdiagTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/federated/primitives/FederatedRdiagTest.java", "diff": "@@ -24,6 +24,7 @@ import java.util.Collection;\nimport org.apache.sysds.api.DMLScript;\nimport org.apache.sysds.common.Types;\n+import org.apache.sysds.hops.OptimizerUtils;\nimport org.apache.sysds.runtime.meta.MatrixCharacteristics;\nimport org.apache.sysds.runtime.util.HDFSTool;\nimport org.apache.sysds.test.AutomatedTestBase;\n@@ -69,7 +70,21 @@ public class FederatedRdiagTest extends AutomatedTestBase {\n@Test\npublic void federatedRdiagSP() { federatedRdiag(Types.ExecMode.SPARK); }\n+ @Test\n+ public void federatedCompilationRDiagCP(){\n+ federatedRdiag(Types.ExecMode.SINGLE_NODE, true);\n+ }\n+\n+ @Test\n+ public void federatedCompilationRdiagSP(){\n+ federatedRdiag(Types.ExecMode.SPARK, true);\n+ }\n+\npublic void federatedRdiag(Types.ExecMode execMode) {\n+ federatedRdiag(execMode, false);\n+ }\n+\n+ public void federatedRdiag(Types.ExecMode execMode, boolean activateFedCompilation) {\nboolean sparkConfigOld = DMLScript.USE_LOCAL_SPARK_CONFIG;\nTypes.ExecMode platformOld = rtplatform;\n@@ -111,6 +126,7 @@ public class FederatedRdiagTest extends AutomatedTestBase {\ninput(\"X1\"), input(\"X2\"), input(\"X3\"), input(\"X4\"), expected(\"S\")};\nrunTest(null);\n+ OptimizerUtils.FEDERATED_COMPILATION = activateFedCompilation;\nTestConfiguration config = availableTestConfigurations.get(TEST_NAME);\nloadTestConfiguration(config);\n@@ -139,5 +155,6 @@ public class FederatedRdiagTest extends AutomatedTestBase {\nTestUtils.shutdownThreads(t1, t2, t3, t4);\nrtplatform = platformOld;\nDMLScript.USE_LOCAL_SPARK_CONFIG = sparkConfigOld;\n+ OptimizerUtils.FEDERATED_COMPILATION = false;\n}\n}\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/federated/primitives/FederatedRevTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/federated/primitives/FederatedRevTest.java", "diff": "@@ -23,7 +23,9 @@ import java.util.Arrays;\nimport java.util.Collection;\nimport org.apache.sysds.api.DMLScript;\n+import org.apache.sysds.common.Types;\nimport org.apache.sysds.common.Types.ExecMode;\n+import org.apache.sysds.hops.OptimizerUtils;\nimport org.apache.sysds.runtime.meta.MatrixCharacteristics;\nimport org.apache.sysds.runtime.util.HDFSTool;\nimport org.apache.sysds.test.AutomatedTestBase;\n@@ -77,7 +79,21 @@ public class FederatedRevTest extends AutomatedTestBase {\nrunRevTest(ExecMode.SPARK);\n}\n+ @Test\n+ public void federatedCompilationRevCP(){\n+ runRevTest(Types.ExecMode.SINGLE_NODE, true);\n+ }\n+\n+ @Test\n+ public void federatedCompilationRevSP(){\n+ runRevTest(Types.ExecMode.SPARK, true);\n+ }\n+\nprivate void runRevTest(ExecMode execMode) {\n+ runRevTest(execMode, false);\n+ }\n+\n+ private void runRevTest(ExecMode execMode, boolean activateFedCompilation) {\nboolean sparkConfigOld = DMLScript.USE_LOCAL_SPARK_CONFIG;\nExecMode platformOld = rtplatform;\n@@ -135,6 +151,7 @@ public class FederatedRevTest extends AutomatedTestBase {\nrunTest(null);\n+ OptimizerUtils.FEDERATED_COMPILATION = activateFedCompilation;\nfullDMLScriptName = HOME + TEST_NAME + \".dml\";\nprogramArgs = new String[] {\"-stats\", \"100\", \"-nvargs\",\n\"in_X1=\" + TestUtils.federatedAddress(port1, input(\"X1\")),\n@@ -160,6 +177,7 @@ public class FederatedRevTest extends AutomatedTestBase {\nrtplatform = platformOld;\nDMLScript.USE_LOCAL_SPARK_CONFIG = sparkConfigOld;\n+ OptimizerUtils.FEDERATED_COMPILATION = false;\n}\n}\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/privacy/CheckedConstraintsLog.java", "new_path": "src/main/java/org/apache/sysds/runtime/privacy/CheckedConstraintsLog.java", "diff": "@@ -53,7 +53,7 @@ public class CheckedConstraintsLog {\n}\n/**\n- * Add an occurence of the given privacy level to the loaded constraints log total.\n+ * Add an occurrence of the given privacy level to the loaded constraints log total.\n* @param level privacy level from loaded privacy constraint\n*/\npublic static void addLoadedConstraint(PrivacyLevel level){\n@@ -84,9 +84,9 @@ public class CheckedConstraintsLog {\npublic static String display(){\nStringBuilder sb = new StringBuilder();\nsb.append(\"Checked Privacy Constraints:\\n\");\n- checkedConstraintsTotal.forEach((k,v)->sb.append(\"\\t\" + k + \": \" + v + \"\\n\"));\n+ checkedConstraintsTotal.forEach((k,v)-> sb.append(String.format(\"\\t%s: %s\\n\", k, v)));\nsb.append(\"Loaded Privacy Constraints:\\n\");\n- loadedConstraintsTotal.forEach((k,v)->sb.append(\"\\t\" + k + \": \" + v + \"\\n\"));\n+ loadedConstraintsTotal.forEach((k,v)->sb.append(String.format(\"\\t%s: %s\\n\", k, v)));\nreturn sb.toString();\n}\n}\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/privacy/PrivacyConstraint.java", "new_path": "src/main/java/org/apache/sysds/runtime/privacy/PrivacyConstraint.java", "diff": "@@ -143,10 +143,9 @@ public class PrivacyConstraint implements Externalizable\n/**\n* Get privacy constraints and put them into JSON object.\n* @param json JSON object in which the privacy constraints are put\n- * @return JSON object including the privacy constraints\n- * @throws JSONException in case of errors in creating JSON object\n+ * @throws JSONException in case of errors in putting into JSON object\n*/\n- public JSONObject toJson(JSONObject json) throws JSONException {\n+ public void toJson(JSONObject json) throws JSONException {\nif ( getPrivacyLevel() != null && getPrivacyLevel() != PrivacyLevel.None )\njson.put(DataExpression.PRIVACY, getPrivacyLevel().name());\nif ( hasFineGrainedConstraints() ) {\n@@ -161,7 +160,6 @@ public class PrivacyConstraint implements Externalizable\nrangesJson.put(PrivacyLevel.PrivateAggregation.name(), aggregateRangesJson);\njson.put(DataExpression.FINE_GRAINED_PRIVACY, rangesJson);\n}\n- return json;\n}\nprivate static JSONArray getJsonArray(DataRange[] ranges) throws JSONException {\n@@ -185,7 +183,7 @@ public class PrivacyConstraint implements Externalizable\nint fineGrainedConstraintLength = is.readInt();\nif ( fineGrainedConstraintLength > 0 ){\nfor (int i = 0; i < fineGrainedConstraintLength; i++){\n- Integer levelIndex = (Integer) is.readInt();\n+ int levelIndex = is.readInt();\nPrivacyLevel rangePrivacy = PrivacyLevel.values()[levelIndex];\nDataRange dataRange = readExternalDataRangeObject(is);\nfineGrainedPrivacy.put(dataRange, rangePrivacy);\n@@ -198,9 +196,9 @@ public class PrivacyConstraint implements Externalizable\nobjectOutput.writeInt(getPrivacyLevel().ordinal());\nif (fineGrainedPrivacy != null && fineGrainedPrivacy.hasConstraints()){\n- List<Entry<DataRange,PrivacyLevel>> finegrainedConstraints = fineGrainedPrivacy.getAllConstraintsList();\n- objectOutput.writeInt(finegrainedConstraints.size());\n- for ( Entry<DataRange,PrivacyLevel> constraint : finegrainedConstraints ) {\n+ List<Entry<DataRange,PrivacyLevel>> fineGrainedConstraints = fineGrainedPrivacy.getAllConstraintsList();\n+ objectOutput.writeInt(fineGrainedConstraints.size());\n+ for ( Entry<DataRange,PrivacyLevel> constraint : fineGrainedConstraints ) {\nobjectOutput.writeInt(constraint.getValue().ordinal());\nDataRange dataRange = constraint.getKey();\nobjectOutput.writeInt(dataRange.getBeginDims().length);\n@@ -217,7 +215,7 @@ public class PrivacyConstraint implements Externalizable\n* Reads a DataRange from ObjectInput.\n* @param is ObjectInput from which the DataRange is read\n* @return DataRange from ObjectInput\n- * @throws IOException\n+ * @throws IOException if an I/O error occurs during read\n*/\nprivate static DataRange readExternalDataRangeObject(ObjectInput is) throws IOException {\nint dimLength = is.readInt();\n@@ -231,7 +229,7 @@ public class PrivacyConstraint implements Externalizable\n* @param is ObjectInput from which the long array is read\n* @param dimLength length of input long array\n* @return the input array as a long array\n- * @throws IOException\n+ * @throws IOException if an I/O error occurs during read\n*/\nprivate static long[] readExternalDataRangeDim(ObjectInput is, int dimLength) throws IOException {\nlong[] dims = new long[dimLength];\n@@ -245,7 +243,7 @@ public class PrivacyConstraint implements Externalizable\n* Write the long array to ObjectOutput.\n* @param objectOutput ObjectOutput in which the long array is written.\n* @param rangeDim long array to write in ObjectOutput.\n- * @throws IOException\n+ * @throws IOException if an I/O error occurs during write\n*/\nprivate static void writeExternalRangeDim(ObjectOutput objectOutput, long[] rangeDim) throws IOException {\nfor ( long beginIndex : rangeDim ){\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/privacy/PrivacyMonitor.java", "new_path": "src/main/java/org/apache/sysds/runtime/privacy/PrivacyMonitor.java", "diff": "@@ -27,7 +27,7 @@ import org.apache.sysds.runtime.privacy.PrivacyConstraint.PrivacyLevel;\npublic class PrivacyMonitor\n{\n- private static EnumMap<PrivacyLevel,LongAdder> checkedConstraints;\n+ private static final EnumMap<PrivacyLevel,LongAdder> checkedConstraints;\nprivate static boolean checkPrivacy = false;\n@@ -71,6 +71,10 @@ public class PrivacyMonitor\n}\n}\n+ /**\n+ * Clears all checked constraints.\n+ * This is used to reset the counter of checked constraints for each PrivacyLevel.\n+ */\npublic static void clearCheckedConstraints(){\ncheckedConstraints.replaceAll((k,v)->new LongAdder());\n}\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/privacy/finegrained/DataRange.java", "new_path": "src/main/java/org/apache/sysds/runtime/privacy/finegrained/DataRange.java", "diff": "@@ -22,7 +22,7 @@ package org.apache.sysds.runtime.privacy.finegrained;\nimport java.util.Arrays;\n/**\n- * A DataRange instance marks a part of a CachableData data object.\n+ * A DataRange instance marks a part of a CacheableData data object.\n* The beginDims marks the beginning for all dimensions and\n* the endDims marks the end for all dimensions.\n* DataRange is very similar to org.apache.sysds.runtime.util.IndexRange,\n@@ -30,8 +30,8 @@ import java.util.Arrays;\n*/\npublic class DataRange {\n- private long[] _beginDims;\n- private long[] _endDims;\n+ private final long[] _beginDims;\n+ private final long[] _endDims;\npublic DataRange(long[] beginDims, long[] endDims){\n_beginDims = beginDims;\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/privacy/finegrained/FineGrainedPrivacy.java", "new_path": "src/main/java/org/apache/sysds/runtime/privacy/finegrained/FineGrainedPrivacy.java", "diff": "@@ -31,60 +31,60 @@ public interface FineGrainedPrivacy {\n* @param dataRange representing the range for which the privacy is set\n* @param privacyLevel the level of privacy for the given data range\n*/\n- public void put(DataRange dataRange, PrivacyLevel privacyLevel);\n+ void put(DataRange dataRange, PrivacyLevel privacyLevel);\n- public void putRow(int rowIndex, int rowLength, PrivacyLevel privacyLevel);\n+ void putRow(int rowIndex, int rowLength, PrivacyLevel privacyLevel);\n- public void putCol(int colIndex, int colLength, PrivacyLevel privacyLevel);\n+ void putCol(int colIndex, int colLength, PrivacyLevel privacyLevel);\n- public void putElement(int rowIndex, int colIndex, PrivacyLevel privacyLevel);\n+ void putElement(int rowIndex, int colIndex, PrivacyLevel privacyLevel);\n/**\n* Get the data ranges and related privacy levels within given data search range.\n* @param searchRange the range from which all privacy levels are retrieved\n* @return all mappings from range to privacy level within the given search range\n*/\n- public Map<DataRange,PrivacyLevel> getPrivacyLevel(DataRange searchRange);\n+ Map<DataRange,PrivacyLevel> getPrivacyLevel(DataRange searchRange);\n/**\n* Get the data ranges and related privacy levels of the element with the given index.\n* @param searchIndex index of element\n* @return all mappings from range to privacy level for the given search element\n*/\n- public Map<DataRange,PrivacyLevel> getPrivacyLevelOfElement(long[] searchIndex);\n+ Map<DataRange,PrivacyLevel> getPrivacyLevelOfElement(long[] searchIndex);\n/**\n* Get all data ranges for the given privacy level.\n* @param privacyLevel for which data ranges are found\n* @return all data ranges with the given privacy level\n*/\n- public DataRange[] getDataRangesOfPrivacyLevel(PrivacyLevel privacyLevel);\n+ DataRange[] getDataRangesOfPrivacyLevel(PrivacyLevel privacyLevel);\n/**\n* Remove all fine-grained privacy constraints.\n*/\n- public void removeAllConstraints();\n+ void removeAllConstraints();\n/**\n* True if any fine-grained constraints has been set.\n* @return true if any fine-grained constraint is set\n*/\n- public boolean hasConstraints();\n+ boolean hasConstraints();\n/**\n* Get all fine-grained constraints as a map from privacy level to\n* an array of data ranges represented as two-dimensional long arrays.\n* @return map from privacy level to array of data ranges\n*/\n- public Map<String, long[][][]> getAllConstraints();\n+ Map<String, long[][][]> getAllConstraints();\n/**\n* Return all fine-grained privacy constraints as an arraylist.\n* @return all constraints\n*/\n- public ArrayList<Map.Entry<DataRange, PrivacyLevel>> getAllConstraintsList();\n+ ArrayList<Map.Entry<DataRange, PrivacyLevel>> getAllConstraintsList();\n- public PrivacyLevel[] getRowPrivacy(int numRows, int numCols);\n+ PrivacyLevel[] getRowPrivacy(int numRows, int numCols);\n- public PrivacyLevel[] getColPrivacy(int numRows, int numCols);\n+ PrivacyLevel[] getColPrivacy(int numRows, int numCols);\n}\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/privacy/finegrained/FineGrainedPrivacyList.java", "new_path": "src/main/java/org/apache/sysds/runtime/privacy/finegrained/FineGrainedPrivacyList.java", "diff": "@@ -33,7 +33,7 @@ import org.apache.sysds.runtime.privacy.PrivacyConstraint.PrivacyLevel;\n*/\npublic class FineGrainedPrivacyList implements FineGrainedPrivacy {\n- private ArrayList<Map.Entry<DataRange, PrivacyLevel>> constraintCollection = new ArrayList<>();\n+ private final ArrayList<Map.Entry<DataRange, PrivacyLevel>> constraintCollection = new ArrayList<>();\n@Override\npublic PrivacyLevel[] getRowPrivacy(int numRows, int numCols) {\n@@ -142,7 +142,7 @@ public class FineGrainedPrivacyList implements FineGrainedPrivacy {\n});\nMap<String, long[][][]> constraintMap = new HashMap<>();\nconstraintMap.put(PrivacyLevel.Private.name(), privateRanges.toArray(new long[0][][]));\n- constraintMap.put(PrivacyLevel.PrivateAggregation.name(), privateRanges.toArray(new long[0][][]));\n+ constraintMap.put(PrivacyLevel.PrivateAggregation.name(), aggregateRanges.toArray(new long[0][][]));\nreturn constraintMap;\n}\n@@ -177,9 +177,8 @@ public class FineGrainedPrivacyList implements FineGrainedPrivacy {\n@Override\npublic String toString(){\nStringBuilder stringBuilder = new StringBuilder();\n- for ( Map.Entry<DataRange,PrivacyLevel> entry : constraintCollection ){\n- stringBuilder.append(entry.getKey().toString() + \" : \" + entry.getValue().name());\n- }\n+ for ( Map.Entry<DataRange,PrivacyLevel> entry : constraintCollection )\n+ stringBuilder.append(String.format(\"%s : %s\",entry.getKey().toString(), entry.getValue().name()));\nreturn stringBuilder.toString();\n}\n}\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/privacy/finegrained/FineGrainedPrivacyMap.java", "new_path": "src/main/java/org/apache/sysds/runtime/privacy/finegrained/FineGrainedPrivacyMap.java", "diff": "@@ -33,7 +33,7 @@ import org.apache.sysds.runtime.privacy.PrivacyConstraint.PrivacyLevel;\n*/\npublic class FineGrainedPrivacyMap implements FineGrainedPrivacy {\n- private Map<DataRange, PrivacyLevel> constraintCollection = new LinkedHashMap<>();\n+ private final Map<DataRange, PrivacyLevel> constraintCollection = new LinkedHashMap<>();\n@Override\npublic void put(DataRange dataRange, PrivacyLevel privacyLevel) {\n@@ -96,15 +96,15 @@ public class FineGrainedPrivacyMap implements FineGrainedPrivacy {\npublic Map<String, long[][][]> getAllConstraints() {\nArrayList<long[][]> privateRanges = new ArrayList<>();\nArrayList<long[][]> aggregateRanges = new ArrayList<>();\n- constraintCollection.forEach((range, privacylevel) -> {\n- if (privacylevel == PrivacyLevel.Private)\n+ constraintCollection.forEach((range, privacyLevel) -> {\n+ if (privacyLevel == PrivacyLevel.Private)\nprivateRanges.add(new long[][] { range.getBeginDims(), range.getEndDims() });\n- else if (privacylevel == PrivacyLevel.PrivateAggregation)\n+ else if (privacyLevel == PrivacyLevel.PrivateAggregation)\naggregateRanges.add(new long[][] { range.getBeginDims(), range.getEndDims() });\n});\nMap<String, long[][][]> constraintMap = new LinkedHashMap<>();\nconstraintMap.put(PrivacyLevel.Private.name(), privateRanges.toArray(new long[0][][]));\n- constraintMap.put(PrivacyLevel.PrivateAggregation.name(), privateRanges.toArray(new long[0][][]));\n+ constraintMap.put(PrivacyLevel.PrivateAggregation.name(), aggregateRanges.toArray(new long[0][][]));\nreturn constraintMap;\n}\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/privacy/propagation/PrivacyPropagator.java", "new_path": "src/main/java/org/apache/sysds/runtime/privacy/propagation/PrivacyPropagator.java", "diff": "@@ -340,7 +340,7 @@ public class PrivacyPropagator\nelse {\nMatrixBlock input1 = ec.getMatrixInput(inst.input1.getName());\nMatrixBlock input2 = ec.getMatrixInput(inst.input2.getName());\n- Propagator propagator = null;\n+ Propagator propagator;\nif ( inst.getAppendType() == AppendCPInstruction.AppendType.RBIND )\npropagator = new RBindPropagator(input1, privacyConstraints[0], input2, privacyConstraints[1]);\nelse if ( inst.getAppendType() == AppendCPInstruction.AppendType.CBIND )\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/privacy/CheckedConstraintsLogTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/privacy/CheckedConstraintsLogTest.java", "diff": "@@ -55,7 +55,7 @@ public class CheckedConstraintsLogTest extends AutomatedTestBase {\npublic void addCheckedConstraintsSingleValue(){\nEnumMap<PrivacyLevel,LongAdder> checked = getMap(PrivacyLevel.Private, 300);\nCheckedConstraintsLog.addCheckedConstraints(checked);\n- assertTrue(CheckedConstraintsLog.getCheckedConstraints().get(PrivacyLevel.Private).longValue() == 300);\n+ assertEquals(300, CheckedConstraintsLog.getCheckedConstraints().get(PrivacyLevel.Private).longValue());\n}\n@Test\n@@ -64,7 +64,7 @@ public class CheckedConstraintsLogTest extends AutomatedTestBase {\nCheckedConstraintsLog.addCheckedConstraints(checked);\nEnumMap<PrivacyLevel,LongAdder> checked2 = getMap(PrivacyLevel.Private, 150);\nCheckedConstraintsLog.addCheckedConstraints(checked2);\n- assertTrue(CheckedConstraintsLog.getCheckedConstraints().get(PrivacyLevel.Private).longValue() == 450);\n+ assertEquals(450, CheckedConstraintsLog.getCheckedConstraints().get(PrivacyLevel.Private).longValue());\n}\n@Test\n@@ -89,9 +89,9 @@ public class CheckedConstraintsLogTest extends AutomatedTestBase {\n@Test\npublic void addLoadedConstraintsSingleValue(){\n- Integer n = 12;\n+ int n = 12;\nfor (int i = 0; i < n; i++)\nCheckedConstraintsLog.addLoadedConstraint(PrivacyLevel.Private);\n- assertEquals(n.longValue(), CheckedConstraintsLog.getLoadedConstraints().get(PrivacyLevel.Private).longValue());\n+ assertEquals(n, CheckedConstraintsLog.getLoadedConstraints().get(PrivacyLevel.Private).longValue());\n}\n}\n" } ]

[ { "change_type": "MODIFY", "old_path": "scripts/builtin/executePipeline.dml", "new_path": "scripts/builtin/executePipeline.dml", "diff": "@@ -35,9 +35,9 @@ s_executePipeline = function(Frame[String] logical = as.frame(\"NULL\"), Frame[Str\nX = rbind(X, Xtest)\nY = rbind(Y, Ytest)\ntestRow = nrow(Xtest)\n+ Xout = X\nt1 = time()\n#print(\"PIPELINE EXECUTION START ... \"+toString(pipeline))\n-\nif(verbose) {\nprint(\"checks rows in X = \"+nrow(X)+\" rows in Y = \"+nrow(Y)+\" cols in X = \"+ncol(X)+\" col in Y = \"+ncol(Y))\nprint(\"pipeline in execution \"+toString(pipeline))\n@@ -51,10 +51,10 @@ s_executePipeline = function(Frame[String] logical = as.frame(\"NULL\"), Frame[Str\nif(test == FALSE | lgOp != \"CI\") {\nXclone = X\n- [hp, dataFlag, yFlag] = matrixToList(X, Y, mask, FD, hyperParameters[i], flagsCount, op)\n+ [hp, dataFlag, yFlag, executeFlag] = matrixToList(X, Y, mask, FD, hyperParameters[i], flagsCount, op)\n+ if(executeFlag == 1) {\nX = eval(op, hp)\nXout = X\n-\nX = confirmData(X, Xclone, mask, dataFlag, yFlag)\n# dataFlag 0 = only on numeric, 1 = on whole data\nif(yFlag)\n@@ -64,6 +64,10 @@ s_executePipeline = function(Frame[String] logical = as.frame(\"NULL\"), Frame[Str\n}\nX = confirmMeta(X, mask)\n}\n+ else {\n+ print(\"not applying \"+op+\" executeFlag = 0\")\n+ }\n+ }\nelse {\nXclone = X\n#print(\"not applying \"+lgOp+\" \"+op+\" on data test flag: \"+test)\n@@ -71,7 +75,9 @@ s_executePipeline = function(Frame[String] logical = as.frame(\"NULL\"), Frame[Str\nYtest = Y[testStIdx:nrow(X), ]\nX = X[1:trainEndIdx, ]\nY = Y[1:trainEndIdx, ]\n- [hp, dataFlag, yFlag] = matrixToList(X, Y, mask, FD, hyperParameters[i], flagsCount, op)\n+ [hp, dataFlag, yFlag, executeFlag] = matrixToList(X, Y, mask, FD, hyperParameters[i], flagsCount, op)\n+ if(executeFlag == 1)\n+ {\nX = eval(op, hp)\nX = confirmData(X, Xclone, mask, dataFlag, yFlag)\n# dataFlag 0 = only on numeric, 1 = on whole data\n@@ -84,6 +90,10 @@ s_executePipeline = function(Frame[String] logical = as.frame(\"NULL\"), Frame[Str\nX = rbind(X, Xtest)\nY = rbind(Y, Ytest)\n}\n+ else {\n+ print(\"not applying \"+op+\" executeFlag = 0\")\n+ }\n+ }\nif(as.scalar(pipeline[1, i]) == \"outlierBySd\" | as.scalar(pipeline[1, i]) == \"outlierByIQR\" | as.scalar(pipeline[1, i]) == \"imputeByFd\") {\nchanges = sum(abs(replace(target=Xout, pattern=NaN, replacement=0) - replace(target=as.matrix(hp[1]), pattern=NaN, replacement=0)) > 0.001 )\n[hpForPruning, changesByOp] = storeDataForPrunning(pipeline, hyperParameters, hpForPruning, changesByOp, changes, i)\n@@ -104,7 +114,7 @@ s_executePipeline = function(Frame[String] logical = as.frame(\"NULL\"), Frame[Str\n# This function will convert the matrix row-vector into list\nmatrixToList = function(Matrix[Double] X, Matrix[Double] Y, Matrix[Double] mask, Matrix[Double] FD,\nMatrix[Double] p, Integer flagsCount, String op)\n- return (List[Unknown] l, Integer dataFlag, Integer yFlag)\n+ return (List[Unknown] l, Integer dataFlag, Integer yFlag, Integer executeFlag)\n{\nNUM_META_FLAGS = flagsCount;\ndataFlag = as.integer(as.scalar(p[1, ncol(p)]))\n@@ -113,18 +123,26 @@ matrixToList = function(Matrix[Double] X, Matrix[Double] Y, Matrix[Double] mask\nfDFlag = as.integer(as.scalar(p[1, ncol(p)-3]))\nmaskFlag = as.integer(as.scalar(p[1, ncol(p)-4]))\n+ executeFlag = 1\n######################################################\n# CHECK FOR DATA FLAG\nif(dataFlag == 0)\n{\n- # take numerics out\n+ if(sum(mask) == ncol(mask))\n+ executeFlag = 0\n+ else {\n+ # take numerics out and remove categorical\nX = removeEmpty(target=X, margin = \"cols\", select = (mask == 0))\n}\n+ }\nelse if(dataFlag == 1)\n{\n- # take categorical out\n+ if(sum(mask) == 0)\n+ executeFlag = 0\n+ else {\n+ # take categorical out and remove numerics\nX = removeEmpty(target=X, margin = \"cols\", select = mask)\n- # print(\"data for execution \\n\"+toString(X, rows=5))\n+ }\n}\nl = list(X)\n@@ -175,7 +193,7 @@ return (Matrix[Double] X)\n# take categorical out\ncat = removeEmpty(target=X, margin=\"cols\", select = mask)\n# round categorical (if there is any floating point)\n- cat = ceil(cat)\n+ cat = round(cat)\n# reconstruct original X\nX = X * (mask == 0)\nq = table(seq(1, ncol(cat)), removeEmpty(target=seq(1, ncol(mask)), margin=\"rows\",\n@@ -199,7 +217,7 @@ return (Matrix[Double] X)\nnX = nX[, 1: ncol(nX) - 1]\n}\n- if(dataFlag == 0 & (sum(mask) > 0) & (sum(mask) != ncol(nX)))\n+ if(dataFlag == 0 & (sum(mask) > 0) & (sum(mask) != ncol(originalX)))\n{\nmaxDummy = max(nX) + 1\nnX = replace(target = nX, pattern = NaN, replacement = maxDummy)\n@@ -219,7 +237,7 @@ return (Matrix[Double] X)\nX = replace(target = X, pattern = maxDummy, replacement = NaN)\nX = replace(target = X, pattern = -1111, replacement = NaN)\n}\n- else if(dataFlag == 1 & (sum(mask) > 0) & (sum(mask) != ncol(nX)))\n+ else if(dataFlag == 1 & (sum(mask) > 0) & (sum(mask) != ncol(originalX)))\n{\nmaxDummy = max(nX) + 1\nnX = replace(target = nX, pattern = NaN, replacement = maxDummy)\n@@ -279,6 +297,7 @@ return (Matrix[Double] X_filled)\n{\nif(sum(fdMask) > 0)\n{\n+ fdMask = removeEmpty(target=fdMask, margin=\"cols\")\nFD = discoverFD(X=replace(target=X, pattern=NaN, replacement=1), Mask=fdMask, threshold=threshold)\nFD = (diag(matrix(1, rows=nrow(FD), cols=1)) ==0) * FD\nFD = FD > 0\n" }, { "change_type": "MODIFY", "old_path": "scripts/builtin/imputeByFD.dml", "new_path": "scripts/builtin/imputeByFD.dml", "diff": "@@ -49,8 +49,15 @@ m_imputeByFD = function(Matrix[Double] X, Integer sourceAttribute, Integer targe\nif(sourceAttribute < 0 | sourceAttribute > ncol(X) | targetAttribute < 0 | targetAttribute > ncol(X))\nstop(\"Stopping due to invalid source and target\")\n+ if(min(X[,sourceAttribute]) < 1 | min(X[,targetAttribute]) < 1)\n+ {\n+ print(\"imputeByFD: source or target contain values less than 1\")\n+\n+ }\n+ else {\n# impute missing values and fix errors\nX[,targetAttribute] = imputeAndCorrect(X[,sourceAttribute], X[,targetAttribute], threshold)\n+ }\nif(verbose)\nprint(\"output \\n\"+toString(X))\n@@ -61,9 +68,10 @@ imputeAndCorrect = function(Matrix[Double] X, Matrix[Double] Y, Double threshold\nXY = cbind(X, Y)\n+ missing_mask = is.na(XY)\n# replace the NaN values with zero\n- XY = replace(target = XY, pattern=NaN, replacement=0)\n- missing_mask = (XY == 0)\n+ XY = replace(target = XY, pattern=NaN, replacement=1)\n+\n# map the missing values to an arbitrary number (i.e., Max values + 1)\nXY = missing_mask * (colMaxs(XY)+1) + XY\n@@ -82,4 +90,5 @@ imputeAndCorrect = function(Matrix[Double] X, Matrix[Double] Y, Double threshold\ntabMax = rowSums(ans) != (ncol(ans) * ((ncol(ans))+1)/2) # vector for controlling max(0)\nfilled = rowMaxs(ans) * tabMax\nimputed_Y = table(seq(1,nrow(X)), XY[,1]) %*% filled;\n+ imputed_Y = replace(target=imputed_Y, pattern=0, replacement=NaN)\n}\n" }, { "change_type": "MODIFY", "old_path": "scripts/pipelines/properties/param.csv", "new_path": "scripts/pipelines/properties/param.csv", "diff": "@@ -14,5 +14,5 @@ fillDefault,0,0,0,0,0,2,,,,,,,,,,,,\ndummycoding,0,1,0,0,0,2,,,,,,,,,,,,\nscale,2,0,0,0,0,0,BOOL,BOOL,0,1,0,1,,,,,,\nforward_fill,1,0,0,0,1,2,BOOL,0,1,,,,,,,,,\n-imputeByFd,1,0,1,0,0,2,FP,0.6,0.9,,,,,,,,,\n+imputeByFd,1,0,1,0,0,1,FP,0.6,0.9,,,,,,,,,\nwtomeklink,0,0,0,1,0,2,,,,,,,,,,,,\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/pipelines/BuiltinTopkCleaningRegressionTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/pipelines/BuiltinTopkCleaningRegressionTest.java", "diff": "@@ -35,7 +35,7 @@ public class BuiltinTopkCleaningRegressionTest extends AutomatedTestBase{\nprivate final static String OUTPUT = RESOURCE+\"intermediates/regression/\";\nprivate static final String PARAM_DIR = \"./scripts/pipelines/properties/\";\nprivate final static String PARAM = PARAM_DIR + \"param.csv\";\n- private final static String PRIMITIVES = PARAM_DIR + \"primitives.csv\";\n+ private final static String PRIMITIVES = PARAM_DIR + \"testPrimitives.csv\";\n@Override\npublic void setUp() {\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/bestAcc.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/bestAcc.csv", "diff": "-65.89595375722543\n-65.3179190751445\n-65.3179190751445\n+94.5945945945946\n+94.5945945945946\n+94.5945945945946\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/dirtyScore.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/dirtyScore.csv", "diff": "-63.72832369942196\n\\ No newline at end of file\n+90.990990990991\n\\ No newline at end of file\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/evalHp.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/evalHp.csv", "diff": "-0,1.0,0.001,10.0\n+2.0,0.001,1.0,1000.0\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/hp.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/hp.csv", "diff": "-24.0,2.0,0.042803849955920424,0.9504400993873047,0,0,0,1.0,0,0,0,0,0,0,1.0,0,2.0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n-24.0,2.0,0.016013893020007757,0.9642527252494045,0,0,0,1.0,0,0,0,0,0,0,1.0,0,2.0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n-24.0,2.0,0.03480400352286382,0.9561745054711843,0,0,0,1.0,0,0,0,0,0,0,1.0,0,2.0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n+32.0,2.0,0.029000277257674192,0.9510406998977287,0,0,0,1.0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,1.0,0,2.0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n+32.0,2.0,0.033677950367757156,0.9519989979315087,0,0,0,1.0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,1.0,0,2.0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n+36.0,3.0,3.0,2.0,1.0,0,0,0,1.0,0,0,0,0,0,1.0,0,0,0,2.0,1.0,0.6200453262235062,0,0,0,0,1.0,1.0,2.0,0,0,0,0,1.0,0,0,0,2.0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/lp.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/lp.csv", "diff": "-OTLR,CI,DUMMY\n+OTLR,EC,CI,DUMMY\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/pipelines/intermediates/classification/pip.csv", "new_path": "src/test/scripts/functions/pipelines/intermediates/classification/pip.csv", "diff": "-winsorize,wtomeklink,dummycoding\n-winsorize,wtomeklink,dummycoding\n-winsorize,wtomeklink,dummycoding\n+winsorize,imputeByMedian,wtomeklink,dummycoding\n+winsorize,imputeByMedian,wtomeklink,dummycoding\n+outlierBySd,imputeByMean,abstain,dummycoding\n" } ]

[ { "change_type": "MODIFY", "old_path": "pom.xml", "new_path": "pom.xml", "diff": "<version>18</version>\n</parent>\n<groupId>org.apache.systemds</groupId>\n- <version>2.2.0-SNAPSHOT</version>\n+ <version>2.2.0</version>\n<artifactId>systemds</artifactId>\n<packaging>jar</packaging>\n<name>SystemDS</name>\n<scm>\n<developerConnection>scm:git:https://github.com/apache/systemds.git</developerConnection>\n- <tag>HEAD</tag>\n+ <tag>2.2.0-rc1</tag>\n</scm>\n<build>\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/python/setup.py", "new_path": "src/main/python/setup.py", "diff": "@@ -48,7 +48,7 @@ java_dir_full_path = python_dir + '/' + java_dir\nsetup(\nname=ARTIFACT_NAME,\n- version=ARTIFACT_VERSION_SHORT,\n+ version=ARTIFACT_VERSION,\ndescription='SystemDS is a distributed and declarative machine learning platform.',\nlong_description=open('README.md', encoding='utf-8').read(),\nlong_description_content_type='text/markdown',\n" } ]

[ { "change_type": "ADD", "old_path": null, "new_path": "dev/release/svn-staging-to-release.sh", "diff": "+#!/usr/bin/env bash\n+#-------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+#-------------------------------------------------------------\n+\n+################################################################################\n+## File: release-to-svn.sh\n+## Desc: Promote release candidate from svn dev/systemds to release/systemds\n+## Note: This file to be run only after the succesful voting\n+################################################################################\n+\n+SELF=$(cd $(dirname $0) && pwd)\n+\n+dry_run_flag=0\n+while getopts \":n\" opt; do\n+ case $opt in\n+ n) dry_run_flag=1 ;;\n+ \\?) error \"Invalid option: $OPTARG\" ;;\n+ esac\n+done\n+\n+\n+RELEASE_STAGING_LOCATION=\"https://dist.apache.org/repos/dist/dev/systemds\"\n+RELEASE_LOCATION=\"https://dist.apache.org/repos/dist/release/systemds\"\n+\n+RELEASE_VERSION=\n+APPROVED_RELEASE_TAG=\n+\n+ASF_USERNAME=\n+\n+read -p \"RELEASE_VERSION : \" RELEASE_VERSION\n+read -p \"APPROVED_RELEASE_TAG : \" APPROVED_RELEASE_TAG\n+\n+read -p \"ASF_USERNAME : \" ASF_USERNAME\n+\n+tmp_repo=$(mktemp -d systemds-repo-tmp-XXXXX)\n+\n+pushd \"${tmp_repo}\"\n+\n+# 1. Checkout only the directory associated with approved release tag\n+svn co --depth=empty $RELEASE_STAGING_LOCATION svn-dev-systemds\n+cd svn-dev-systemds\n+svn update --set-depth files ${APPROVED_RELEASE_TAG}\n+cd ..\n+\n+# 2.1. Checkout the empty repo, and copy the contents from svn dev\n+svn co --depth=empty $RELEASE_LOCATION svn-release-systemds\n+mkdir -p svn-release-systemds/$RELEASE_VERSION\n+\n+cp svn-dev-systemds/${APPROVED_RELEASE_TAG}/systemds-* svn-release-systemds/$RELEASE_VERSION\n+\n+# 2.2. Add the files to svn\n+svn add svn-release-systemds/$RELEASE_VERSION\n+cd svn-release-systemds\n+\n+# 2.3. Commit and upload the files to the svn repository\n+\n+if [[ $dry_run_flag != 1 ]]; then\n+ # This step prompts for the Apache Credentials\n+ svn ci --username $ASF_USERNAME -m'Apache SystemDS $RELEASE_VERSION Released' --no-auth-cache \\n\n+ [[ $? == 0 ]] && printf '\\n Publishing to $RELEASE_LOCATION is complete!\\n'\n+else\n+ printf \"\\n==========\\n\"\n+ printf \"This step would commit to the SVN release repo\\n\"\n+ printf \"At $RELEASE_LOCATION \\n\"\n+ printf \"\\n==========\\n\"\n+ printf \"You might want to manually check the files and run the following:\\n\"\n+ printf \"svn ci --username $ASF_USERNAME -m'Apache SystemDS $RELEASE_VERSION Released' --no-auth-cache \\n\"\n+ printf \"\\n==========\\n\"\n+fi\n+\n+\n" } ]

[ { "change_type": "MODIFY", "old_path": ".github/workflows/applicationTests.yml", "new_path": ".github/workflows/applicationTests.yml", "diff": "@@ -23,12 +23,21 @@ name: Application Test\non:\npush:\n+ paths-ignore:\n+ - 'docs/**'\n+ - '*.md'\n+ - '*.html'\n+ - 'src/main/python/docs/**'\n+ - 'dev/**'\nbranches:\n- master\npull_request:\npaths-ignore:\n- 'docs/**'\n- '*.md'\n+ - '*.html'\n+ - 'src/main/python/docs/**'\n+ - 'dev/**'\nbranches:\n- master\n" }, { "change_type": "MODIFY", "old_path": ".github/workflows/build.yml", "new_path": ".github/workflows/build.yml", "diff": "@@ -23,14 +23,24 @@ name: Build\non:\npush:\n+ paths-ignore:\n+ - 'docs/**'\n+ - '*.md'\n+ - '*.html'\n+ - 'src/main/python/docs/**'\n+ - 'dev/**'\nbranches:\n- master\npull_request:\npaths-ignore:\n- 'docs/**'\n- '*.md'\n+ - '*.html'\n+ - 'src/main/python/docs/**'\n+ - 'dev/**'\nbranches:\n- master\n+\njobs:\nbuild:\nruns-on: ${{ matrix.os }}\n" }, { "change_type": "MODIFY", "old_path": ".github/workflows/componentTests.yml", "new_path": ".github/workflows/componentTests.yml", "diff": "@@ -23,12 +23,21 @@ name: Component Test\non:\npush:\n+ paths-ignore:\n+ - 'docs/**'\n+ - '*.md'\n+ - '*.html'\n+ - 'src/main/python/docs/**'\n+ - 'dev/**'\nbranches:\n- master\npull_request:\npaths-ignore:\n- 'docs/**'\n- '*.md'\n+ - '*.html'\n+ - 'src/main/python/docs/**'\n+ - 'dev/**'\nbranches:\n- master\n" }, { "change_type": "MODIFY", "old_path": ".github/workflows/documentation.yml", "new_path": ".github/workflows/documentation.yml", "diff": "@@ -23,12 +23,15 @@ name: Documentation\non:\npush:\n+ paths-ignore:\n+ - '*.html'\n+ - 'dev/**'\nbranches:\n- master\npull_request:\npaths-ignore:\n- - 'docs/**'\n- - '*.md'\n+ - '*.html'\n+ - 'dev/**'\nbranches:\n- master\n" }, { "change_type": "MODIFY", "old_path": ".github/workflows/functionsTests.yml", "new_path": ".github/workflows/functionsTests.yml", "diff": "@@ -23,12 +23,21 @@ name: Function Test\non:\npush:\n+ paths-ignore:\n+ - 'docs/**'\n+ - '*.md'\n+ - '*.html'\n+ - 'src/main/python/docs/**'\n+ - 'dev/**'\nbranches:\n- master\npull_request:\npaths-ignore:\n- 'docs/**'\n- '*.md'\n+ - '*.html'\n+ - 'src/main/python/docs/**'\n+ - 'dev/**'\nbranches:\n- master\n" }, { "change_type": "MODIFY", "old_path": ".github/workflows/python.yml", "new_path": ".github/workflows/python.yml", "diff": "@@ -23,12 +23,21 @@ name: Python Test\non:\npush:\n+ paths-ignore:\n+ - 'docs/**'\n+ - '*.md'\n+ - '*.html'\n+ - 'src/main/python/docs/**'\n+ - 'dev/**'\nbranches:\n- master\npull_request:\npaths-ignore:\n- 'docs/**'\n- '*.md'\n+ - '*.html'\n+ - 'src/main/python/docs/**'\n+ - 'dev/**'\nbranches:\n- master\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/hops/OptimizerUtils.java", "new_path": "src/main/java/org/apache/sysds/hops/OptimizerUtils.java", "diff": "@@ -190,6 +190,14 @@ public class OptimizerUtils\n*/\npublic static boolean ALLOW_LOOP_UPDATE_IN_PLACE = true;\n+ /**\n+ * Enables the update-in-place for all unary operators with a single\n+ * consumer. In this case we do not allocate the output, but directly\n+ * write the output values back to the input block.\n+ */\n+ //TODO enabling it by default requires modifications in lineage-based reuse\n+ public static boolean ALLOW_UNARY_UPDATE_IN_PLACE = false;\n+\n/**\n* Replace eval second-order function calls with normal function call\n* if the function name is a known string (after constant propagation).\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/hops/UnaryOp.java", "new_path": "src/main/java/org/apache/sysds/hops/UnaryOp.java", "diff": "@@ -43,6 +43,7 @@ import org.apache.sysds.runtime.util.UtilFunctions;\nimport java.util.ArrayList;\n+\n/* Unary (cell operations): e.g, b_ij = round(a_ij)\n* Semantic: given a value, perform the operation (independent of other values)\n*/\n@@ -165,10 +166,20 @@ public class UnaryOp extends MultiThreadedHop\n}\nelse //default unary\n{\n+ boolean inplace = false;\n+\n+ //check in-place\n+ if (OptimizerUtils.ALLOW_UNARY_UPDATE_IN_PLACE\n+ && input.getParent().size() == 1)\n+ {\n+ inplace = !(input instanceof DataOp)\n+ || !((DataOp) input).isRead();\n+ }\n+\nint k = isCumulativeUnaryOperation() || isExpensiveUnaryOperation() ?\nOptimizerUtils.getConstrainedNumThreads( _maxNumThreads ) : 1;\nUnary unary1 = new Unary(input.constructLops(),\n- _op, getDataType(), getValueType(), et, k, false);\n+ _op, getDataType(), getValueType(), et, k, inplace);\nsetOutputDimensions(unary1);\nsetLineNumbers(unary1);\nsetLops(unary1);\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/lops/Unary.java", "new_path": "src/main/java/org/apache/sysds/lops/Unary.java", "diff": "@@ -122,6 +122,7 @@ public class Unary extends Lop\n}\npublic static boolean isMultiThreadedOp(OpOp1 op) {\n+ //TODO extend for all basic unary operations\nreturn op==OpOp1.CUMSUM\n|| op==OpOp1.CUMPROD\n|| op==OpOp1.CUMMIN\n@@ -129,6 +130,10 @@ public class Unary extends Lop\n|| op==OpOp1.CUMSUMPROD\n|| op==OpOp1.EXP\n|| op==OpOp1.LOG\n+ || op==OpOp1.ABS\n+ || op==OpOp1.ROUND\n+ || op==OpOp1.FLOOR\n+ || op==OpOp1.CEIL\n|| op==OpOp1.SIGMOID\n|| op==OpOp1.POW2\n|| op==OpOp1.MULT2;\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/instructions/cp/UnaryCPInstruction.java", "new_path": "src/main/java/org/apache/sysds/runtime/instructions/cp/UnaryCPInstruction.java", "diff": "package org.apache.sysds.runtime.instructions.cp;\n-import java.util.Arrays;\n-\n+import org.apache.sysds.common.Types;\nimport org.apache.sysds.common.Types.DataType;\nimport org.apache.sysds.common.Types.ValueType;\n+import org.apache.sysds.lops.Unary;\nimport org.apache.sysds.runtime.DMLRuntimeException;\nimport org.apache.sysds.runtime.functionobjects.Builtin;\nimport org.apache.sysds.runtime.functionobjects.ValueFunction;\n@@ -61,8 +61,8 @@ public abstract class UnaryCPInstruction extends ComputationCPInstruction {\nin.split(parts[1]);\nout.split(parts[2]);\nfunc = Builtin.getBuiltinFnObject(opcode);\n-\n- if( Arrays.asList(new String[]{\"ucumk+\",\"ucum*\",\"ucumk+*\",\"ucummin\",\"ucummax\",\"exp\",\"log\",\"sigmoid\"}).contains(opcode) ){\n+ Types.OpOp1 op_type = Types.OpOp1.valueOfByOpcode(opcode);\n+ if( Unary.isMultiThreadedOp(op_type)){\nUnaryOperator op = new UnaryOperator(func, Integer.parseInt(parts[3]),Boolean.parseBoolean(parts[4]));\nreturn new UnaryMatrixCPInstruction(op, in, out, opcode, str);\n}\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/matrix/data/LibMatrixAgg.java", "new_path": "src/main/java/org/apache/sysds/runtime/matrix/data/LibMatrixAgg.java", "diff": "@@ -294,6 +294,8 @@ public class LibMatrixAgg\n}\npublic static MatrixBlock cumaggregateUnaryMatrix(MatrixBlock in, MatrixBlock out, UnaryOperator uop, double[] agg) {\n+ //Check this implementation, standard case for cumagg (single threaded)\n+\n//prepare meta data\nAggType aggtype = getAggType(uop);\nfinal int m = in.rlen;\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/matrix/data/MatrixBlock.java", "new_path": "src/main/java/org/apache/sysds/runtime/matrix/data/MatrixBlock.java", "diff": "@@ -2755,6 +2755,7 @@ public class MatrixBlock extends MatrixValue implements CacheBlock, Externalizab\nreturn ret;\n}\n+\n@Override\npublic MatrixBlock unaryOperations(UnaryOperator op, MatrixValue result) {\nMatrixBlock ret = checkType(result);\n@@ -2788,7 +2789,11 @@ public class MatrixBlock extends MatrixValue implements CacheBlock, Externalizab\n//note: we apply multi-threading in a best-effort manner here\n//only for expensive operators such as exp, log, sigmoid, because\n//otherwise allocation, read and write anyway dominates\n+ if (!op.isInplace() || isEmpty())\nret.allocateDenseBlock(false);\n+ else\n+ ret = this;\n+\nDenseBlock a = getDenseBlock();\nDenseBlock c = ret.getDenseBlock();\nfor(int bi=0; bi<a.numBlocks(); bi++) {\n@@ -2797,7 +2802,11 @@ public class MatrixBlock extends MatrixValue implements CacheBlock, Externalizab\n}\nret.recomputeNonZeros();\n}\n- else {\n+ else\n+ {\n+ if (op.isInplace() && !isInSparseFormat() )\n+ ret = this;\n+\n//default execute unary operations\nif(op.sparseSafe)\nsparseUnaryOperations(op, ret);\n@@ -2870,7 +2879,7 @@ public class MatrixBlock extends MatrixValue implements CacheBlock, Externalizab\n}\nelse //DENSE <- DENSE\n{\n- //allocate dense output block\n+ if( this != ret ) //!in-place\nret.allocateDenseBlock(false);\nDenseBlock da = getDenseBlock();\nDenseBlock dc = ret.getDenseBlock();\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/test/java/org/apache/sysds/test/functions/updateinplace/UnaryUpdateInPlaceTest.java", "diff": "+/*\n+ * Licensed to the Apache Software Foundation (ASF) under one\n+ * or more contributor license agreements. See the NOTICE file\n+ * distributed with this work for additional information\n+ * regarding copyright ownership. The ASF licenses this file\n+ * to you under the Apache License, Version 2.0 (the\n+ * \"License\"); you may not use this file except in compliance\n+ * with the License. You may obtain a copy of the License at\n+ *\n+ * http://www.apache.org/licenses/LICENSE-2.0\n+ *\n+ * Unless required by applicable law or agreed to in writing,\n+ * software distributed under the License is distributed on an\n+ * \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+ * KIND, either express or implied. See the License for the\n+ * specific language governing permissions and limitations\n+ * under the License.\n+ */\n+\n+package org.apache.sysds.test.functions.updateinplace;\n+\n+import org.apache.sysds.common.Types;\n+import org.apache.sysds.hops.OptimizerUtils;\n+import org.apache.sysds.runtime.matrix.data.MatrixValue;\n+import org.apache.sysds.test.AutomatedTestBase;\n+import org.apache.sysds.test.TestConfiguration;\n+import org.apache.sysds.test.TestUtils;\n+import org.apache.sysds.test.functions.builtin.BuiltinSplitTest;\n+import org.junit.Test;\n+\n+import java.util.HashMap;\n+\n+\n+public class UnaryUpdateInPlaceTest extends AutomatedTestBase{\n+ private final static String TEST_NAME = \"UnaryUpdateInplace\";\n+ private final static String TEST_DIR = \"functions/updateinplace/\";\n+ private final static String TEST_CLASS_DIR = TEST_DIR + BuiltinSplitTest.class.getSimpleName() + \"/\";\n+ private final static double eps = 1e-3;\n+\n+ @Override\n+ public void setUp() {\n+ TestUtils.clearAssertionInformation();\n+ addTestConfiguration(TEST_NAME, new TestConfiguration(TEST_CLASS_DIR, TEST_NAME, new String[]{\"B\",}));\n+ }\n+\n+ @Test\n+ public void testInPlace() {\n+ runInPlaceTest(Types.ExecType.CP);\n+ }\n+\n+\n+ private void runInPlaceTest(Types.ExecType instType) {\n+ Types.ExecMode platformOld = setExecMode(instType);\n+ boolean oldFlag = OptimizerUtils.ALLOW_UNARY_UPDATE_IN_PLACE;\n+\n+ try {\n+ loadTestConfiguration(getTestConfiguration(TEST_NAME));\n+ String HOME = SCRIPT_DIR + TEST_DIR;\n+ fullDMLScriptName = HOME + TEST_NAME + \".dml\";\n+ programArgs = new String[]{\"-explain\",\"-nvargs\",\"Out=\" + output(\"Out\") };\n+\n+ OptimizerUtils.ALLOW_UNARY_UPDATE_IN_PLACE = true;\n+ runTest(true, false, null, -1);\n+ HashMap<MatrixValue.CellIndex, Double> dmlfileOut1 = readDMLMatrixFromOutputDir(\"Out\");\n+ OptimizerUtils.ALLOW_UNARY_UPDATE_IN_PLACE = false;\n+ runTest(true, false, null, -1);\n+ HashMap<MatrixValue.CellIndex, Double> dmlfileOut2 = readDMLMatrixFromOutputDir(\"Out\");\n+\n+ //compare matrices\n+ TestUtils.compareMatrices(dmlfileOut1,dmlfileOut2,eps,\"Stat-DML1\",\"Stat-DML2\");\n+ }\n+ catch(Exception e) {\n+ e.printStackTrace();\n+ }\n+ finally {\n+ rtplatform = platformOld;\n+ OptimizerUtils.ALLOW_UNARY_UPDATE_IN_PLACE = oldFlag;\n+ }\n+ }\n+}\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/test/scripts/functions/updateinplace/UnaryUpdateInplace.dml", "diff": "+/*\n+ * Licensed to the Apache Software Foundation (ASF) under one\n+ * or more contributor license agreements. See the NOTICE file\n+ * distributed with this work for additional information\n+ * regarding copyright ownership. The ASF licenses this file\n+ * to you under the Apache License, Version 2.0 (the\n+ * \"License\"); you may not use this file except in compliance\n+ * with the License. You may obtain a copy of the License at\n+ *\n+ * http://www.apache.org/licenses/LICENSE-2.0\n+ *\n+ * Unless required by applicable law or agreed to in writing,\n+ * software distributed under the License is distributed on an\n+ * \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+ * KIND, either express or implied. See the License for the\n+ * specific language governing permissions and limitations\n+ * under the License.\n+ */\n+\n+\n+#A = rand(rows = 100, cols = 100)\n+#C = rand(rows = 100, cols = 100)\n+\n+A = matrix(1, 10, 10);\n+C = matrix(1, 10, 10);\n+while(FALSE){}\n+A = A * seq(1.1,10.1);\n+while(FALSE){}\n+B = round(A) # does not apply\n+C = C * seq(1.1,10.1);\n+D = log(C) # applies\n+while(FALSE){}\n+C = A + B + D*3\n+Out = C\n+write(Out, $Out);\n+print(as.scalar(C[2, 1]))\n" } ]

[ { "change_type": "MODIFY", "old_path": "README.md", "new_path": "README.md", "diff": "@@ -45,4 +45,3 @@ To build from source visit [SystemDS Install from source](https://apache.github.\n[](https://github.com/apache/systemds/actions?query=workflow%3A%22Application+Test%22+branch%3Amain+event%3Apush)\n[](https://github.com/apache/systemds/actions?query=workflow%3A%22Function+Test%22+branch%3Amain+event%3Apush)\n[](https://github.com/apache/systemds/actions?query=workflow%3A%22Python+Test%22+branch%3Amain+event%3Apush)\n-[](https://github.com/apache/systemds/actions?query=workflow%3A%22Federated+Python+Test%22+branch%3Amain+event%3Apush)\n" } ]

[ { "change_type": "MODIFY", "old_path": ".github/workflows/build.yml", "new_path": ".github/workflows/build.yml", "diff": "@@ -57,9 +57,10 @@ jobs:\nuses: actions/checkout@v2\n- name: Setup Java 11\n- uses: actions/setup-java@v1\n+ uses: actions/setup-java@v2\nwith:\n- java-version: 11\n+ distribution: 'adopt'\n+ java-version: '11'\n- name: Cache Maven Dependencies\nuses: actions/cache@v1\n" }, { "change_type": "MODIFY", "old_path": ".github/workflows/componentTests.yml", "new_path": ".github/workflows/componentTests.yml", "diff": "@@ -53,10 +53,11 @@ jobs:\n- name: Checkout Repository\nuses: actions/checkout@v2\n- - name: Setup Java 1.8\n- uses: actions/setup-java@v1\n+ - name: Setup Java 11\n+ uses: actions/setup-java@v2\nwith:\n- java-version: 1.8\n+ distribution: 'adopt'\n+ java-version: '11'\n- name: Cache Maven Dependencies\nuses: actions/cache@v1\n" }, { "change_type": "MODIFY", "old_path": ".github/workflows/documentation.yml", "new_path": ".github/workflows/documentation.yml", "diff": "@@ -43,10 +43,11 @@ jobs:\n- name: Checkout Repository\nuses: actions/checkout@v2\n- - name: Setup Java 1.8\n- uses: actions/setup-java@v1\n+ - name: Setup Java 11\n+ uses: actions/setup-java@v2\nwith:\n- java-version: 1.8\n+ distribution: 'adopt'\n+ java-version: '11'\n- name: Cache Maven Dependencies\nuses: actions/cache@v1\n" } ]

[ { "change_type": "MODIFY", "old_path": ".github/workflows/build.yml", "new_path": ".github/workflows/build.yml", "diff": "@@ -49,18 +49,20 @@ jobs:\nmatrix:\nos: [\nubuntu-latest,\n- # macOS-latest,\n+ macOS-latest,\nwindows-latest\n]\n+ java: ['1.8', '11', '16']\n+ javadist: ['adopt', 'adopt-openj9', 'zulu', 'temurin']\nsteps:\n- name: Checkout Repository\nuses: actions/checkout@v2\n- - name: Setup Java 11\n+ - name: Setup Java ${{ matrix.java }} ${{ matrix.javadist }}\nuses: actions/setup-java@v2\nwith:\n- distribution: 'adopt'\n- java-version: '11'\n+ distribution: ${{ matrix.javadist }}\n+ java-version: ${{ matrix.java }}\n- name: Cache Maven Dependencies\nuses: actions/cache@v1\n" } ]

[ { "change_type": "MODIFY", "old_path": ".github/workflows/applicationTests.yml", "new_path": ".github/workflows/applicationTests.yml", "diff": "@@ -42,19 +42,19 @@ on:\n- main\njobs:\n- applicationsTests:\n+ test:\n+ name: ${{ matrix.os }}\nruns-on: ${{ matrix.os }}\nstrategy:\nfail-fast: false\nmatrix:\nos: [ubuntu-latest]\n- name: Ap Test ${{ matrix.tests }}\nsteps:\n- name: Checkout Repository\nuses: actions/checkout@v2\n- name: Cache Maven Dependencies\n- uses: actions/cache@v1\n+ uses: actions/cache@v2\nwith:\npath: ~/.m2/repository\nkey: ${{ runner.os }}-maven-test-${{ hashFiles('**/pom.xml') }}\n" }, { "change_type": "MODIFY", "old_path": ".github/workflows/build.yml", "new_path": ".github/workflows/build.yml", "diff": "@@ -43,6 +43,7 @@ on:\njobs:\nbuild:\n+ name: ${{ matrix.os }} Java ${{ matrix.java }} ${{ matrix.javadist }}\nruns-on: ${{ matrix.os }}\nstrategy:\nfail-fast: false\n@@ -52,8 +53,17 @@ jobs:\nmacOS-latest,\nwindows-latest\n]\n- java: ['1.8', '11', '16']\n- javadist: ['adopt', 'adopt-openj9', 'zulu', 'temurin']\n+ java: [\n+ # '8',\n+ '11',\n+ # '16'\n+ ]\n+ javadist: [\n+ # 'adopt',\n+ 'adopt-openj9',\n+ # 'zulu',\n+ # 'temurin'\n+ ]\nsteps:\n- name: Checkout Repository\nuses: actions/checkout@v2\n@@ -63,14 +73,7 @@ jobs:\nwith:\ndistribution: ${{ matrix.javadist }}\njava-version: ${{ matrix.java }}\n-\n- - name: Cache Maven Dependencies\n- uses: actions/cache@v1\n- with:\n- path: ~/.m2/repository\n- key: ${{ runner.os }}-maven-${{ hashFiles('**/pom.xml') }}\n- restore-keys: |\n- ${{ runner.os }}-maven-\n+ cache: 'maven'\n- name: Build\n- run: mvn package -P rat\n+ run: mvn package\n" }, { "change_type": "MODIFY", "old_path": ".github/workflows/componentTests.yml", "new_path": ".github/workflows/componentTests.yml", "diff": "@@ -42,30 +42,25 @@ on:\n- main\njobs:\n- componentTests:\n+ test:\nruns-on: ${{ matrix.os }}\nstrategy:\nfail-fast: false\nmatrix:\nos: [ubuntu-latest]\n- name: Component Tests ${{ matrix.os }}\n+ java: ['11']\n+ javadist: ['adopt-openj9']\n+ name: ${{ matrix.os }}\nsteps:\n- name: Checkout Repository\nuses: actions/checkout@v2\n- - name: Setup Java 11\n+ - name: Setup Java ${{ matrix.java }} ${{ matrix.javadist }}\nuses: actions/setup-java@v2\nwith:\n- distribution: 'adopt'\n- java-version: '11'\n-\n- - name: Cache Maven Dependencies\n- uses: actions/cache@v1\n- with:\n- path: ~/.m2/repository\n- key: ${{ runner.os }}-maven-test-${{ hashFiles('**/pom.xml') }}\n- restore-keys: |\n- ${{ runner.os }}-maven-test-\n+ distribution: ${{ matrix.javadist }}\n+ java-version: ${{ matrix.java }}\n+ cache: 'maven'\n- name: Component Tests\nrun: ./docker/entrypoint.sh org.apache.sysds.test.component.**\n" }, { "change_type": "MODIFY", "old_path": ".github/workflows/documentation.yml", "new_path": ".github/workflows/documentation.yml", "diff": "@@ -36,33 +36,32 @@ on:\n- main\njobs:\n- documentation1:\n- runs-on: ubuntu-latest\n- name: Documentation Java\n+ doc1:\n+ runs-on: ${{ matrix.os }}\n+ strategy:\n+ fail-fast: false\n+ matrix:\n+ os: [ubuntu-latest]\n+ java: ['11']\n+ javadist: ['adopt-openj9']\n+ name: Java\nsteps:\n- name: Checkout Repository\nuses: actions/checkout@v2\n- - name: Setup Java 11\n+ - name: Setup Java ${{ matrix.java }} ${{ matrix.javadist }}\nuses: actions/setup-java@v2\nwith:\n- distribution: 'adopt'\n- java-version: '11'\n-\n- - name: Cache Maven Dependencies\n- uses: actions/cache@v1\n- with:\n- path: ~/.m2/repository\n- key: ${{ runner.os }}-maven-${{ hashFiles('**/pom.xml') }}\n- restore-keys: |\n- ${{ runner.os }}-maven-\n+ distribution: ${{ matrix.javadist }}\n+ java-version: ${{ matrix.java }}\n+ cache: 'maven'\n- name: Make Documentation SystemDS Java\nrun: mvn -ntp -P distribution package\n- documentation2:\n+ doc2:\nruns-on: ubuntu-latest\n- name: Documentation Python\n+ name: Python\nsteps:\n- name: Checkout Repository\nuses: actions/checkout@v2\n@@ -74,7 +73,7 @@ jobs:\narchitecture: 'x64'\n- name: Cache Pip Dependencies\n- uses: actions/cache@v1\n+ uses: actions/cache@v2\nwith:\npath: ~/.cache/pip\nkey: ${{ runner.os }}-pip-docs-${{ hashFiles('src/main/python/docs/requires-docs.txt') }}\n" }, { "change_type": "MODIFY", "old_path": ".github/workflows/functionsTests.yml", "new_path": ".github/workflows/functionsTests.yml", "diff": "@@ -42,7 +42,7 @@ on:\n- main\njobs:\n- applicationsTests:\n+ test:\nruns-on: ${{ matrix.os }}\ntimeout-minutes: 90\nstrategy:\n@@ -65,13 +65,13 @@ jobs:\n\"**.functions.unary.matrix.**\"\n]\nos: [ubuntu-latest]\n- name: Function Test ${{ matrix.tests }}\n+ name: ${{ matrix.tests }}\nsteps:\n- name: Checkout Repository\nuses: actions/checkout@v2\n- name: Cache Maven Dependencies\n- uses: actions/cache@v1\n+ uses: actions/cache@v2\nwith:\npath: ~/.m2/repository\nkey: ${{ runner.os }}-maven-test-${{ hashFiles('**/pom.xml') }}\n" }, { "change_type": "ADD", "old_path": null, "new_path": ".github/workflows/license.yml", "diff": "+#-------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+#-------------------------------------------------------------\n+\n+name: LicenseCheck\n+\n+on:\n+ push:\n+ paths-ignore:\n+ - 'docs/**'\n+ - '*.md'\n+ - '*.html'\n+ - 'src/main/python/docs/**'\n+ - 'dev/**'\n+ branches:\n+ - main\n+ pull_request:\n+ paths-ignore:\n+ - 'docs/**'\n+ - '*.md'\n+ - '*.html'\n+ - 'src/main/python/docs/**'\n+ - 'dev/**'\n+ branches:\n+ - main\n+\n+jobs:\n+ build:\n+ name: ${{ matrix.os }}\n+ runs-on: ${{ matrix.os }}\n+ strategy:\n+ fail-fast: false\n+ matrix:\n+ os: [ubuntu-latest]\n+ java: ['11']\n+ javadist: ['adopt-openj9']\n+\n+ steps:\n+ - name: Checkout Repository\n+ uses: actions/checkout@v2\n+\n+ - name: Setup Java ${{ matrix.java }} ${{ matrix.javadist }}\n+ uses: actions/setup-java@v2\n+ with:\n+ distribution: ${{ matrix.javadist }}\n+ java-version: ${{ matrix.java }}\n+ cache: 'maven'\n+\n+ - name: Build\n+ run: mvn package -P rat\n" }, { "change_type": "MODIFY", "old_path": ".github/workflows/python.yml", "new_path": ".github/workflows/python.yml", "diff": "@@ -42,50 +42,44 @@ on:\n- main\njobs:\n- applicationsTests:\n+ test:\nruns-on: ${{ matrix.os }}\nstrategy:\nfail-fast: false\nmatrix:\npython-version: [3.8]\nos: [ubuntu-latest]\n- java: [ 11 ]\n- name: Python Test\n+ java: ['11']\n+ javadist: ['adopt-openj9']\n+\n+ name: ${{ matrix.os }} Java ${{ matrix.java }} ${{ matrix.javadist }} Python ${{ matrix.python-version }}\nsteps:\n- name: Checkout Repository\nuses: actions/checkout@v2\n- - name: Setup Java\n- uses: actions/setup-java@v1\n+ - name: Setup Java ${{ matrix.java }} ${{ matrix.javadist }}\n+ uses: actions/setup-java@v2\nwith:\n+ distribution: ${{ matrix.javadist }}\njava-version: ${{ matrix.java }}\n-\n- - name: Cache Maven Dependencies\n- uses: actions/cache@v1\n- with:\n- path: ~/.m2/repository\n- key: ${{ runner.os }}-maven-${{ hashFiles('**/pom.xml') }}\n+ cache: 'maven'\n- name: Cache Pip Dependencies\n- uses: actions/cache@v1\n+ uses: actions/cache@v2\nwith:\npath: ~/.cache/pip\nkey: ${{ runner.os }}-pip-${{ matrix.python-version }}-${{ hashFiles('src/main/python/setup.py') }}\n- - name: Cache MNIST\n- uses: actions/cache@v1\n- with:\n- path: src/main/python/systemds/examples/tutorials/mnist\n- key: ${{ runner.os }}-mnist-${{ hashFiles('src/main/python/systemds/examples/tutorials/mnist.py') }}\n-\n- - name: Cache Adult/Census\n- uses: actions/cache@v1\n+ - name: Cache Datasets\n+ uses: actions/cache@v2\nwith:\n- path: src/main/python/systemds/examples/tutorials/adult/data.zip\n- key: ${{ runner.os }}-adult-${{ hashFiles('src/main/python/systemds/examples/tutorials/adult.py') }}\n+ path: |\n+ src/main/python/systemds/examples/tutorials/mnist\n+ src/main/python/systemds/examples/tutorials/adult/data.zip\n+ key: ${{ runner.os }}-mnist-${{ hashFiles('src/main/python/systemds/examples/tutorials/mnist.py') }}-${{ hashFiles('src/main/python/systemds/examples/tutorials/adult.py') }}\n- name: Cache Deb Dependencies\n- uses: actions/cache@v1\n+ uses: actions/cache@v2\nwith:\npath: /var/cache/apt/archives\nkey: ${{ runner.os }}-${{ hashFiles('.github/workflows/python.yml') }}\n@@ -93,8 +87,8 @@ jobs:\n- name: Maven clean & package\nrun: mvn -ntp clean package -P distribution\n- - name: Setup Python\n- uses: actions/setup-python@v1\n+ - name: Setup Python ${{ matrix.python-version }}\n+ uses: actions/setup-python@v2\nwith:\npython-version: ${{ matrix.python-version }}\narchitecture: 'x64'\n" } ]

[ { "change_type": "MODIFY", "old_path": "docs/_config.yml", "new_path": "docs/_config.yml", "diff": "@@ -39,7 +39,7 @@ exclude:\n- updateAPI.sh\n# These allow the documentation to be updated with newer releases\n-SYSTEMDS_VERSION: 2.2.0-SNAPSHOT\n+SYSTEMDS_VERSION: 2.3.0-SNAPSHOT\n# if 'analytics_on' is true, analytics section will be rendered on the HTML pages\nanalytics_on: true\n" } ]

[ { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/docs/source/code/federatedTutorial_part1.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+# Python\n+import numpy as np\n+import os\n+if not os.path.isdir(\"temp\"):\n+ os.mkdir(\"temp\")\n+a = np.asarray([[1, 2, 3], [4, 5, 6], [7, 8, 9]])\n+np.savetxt(\"temp/test.csv\", a, delimiter=\",\")\n+with open(\"temp/test.csv.mtd\", \"w\") as mtd:\n+ mtd.write('{ \"format\":\"csv\", \"header\":false, \"rows\":3, \"cols\":3 }')\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/docs/source/code/federatedTutorial_part2.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+# Python\n+import numpy as np\n+from systemds.context import SystemDSContext\n+\n+# Create a federated matrix\n+# Indicate the dimensions of the data:\n+# Here the first list in the tuple is the top left Coordinate,\n+# and the second the bottom left coordinate.\n+# It is ordered as [col,row].\n+dims = ([0, 0], [3, 3])\n+\n+# Specify the address + file path from worker:\n+address = \"localhost:8001/temp/test.csv\"\n+\n+with SystemDSContext() as sds:\n+ fed_a = sds.federated([address], [dims])\n+ # Sum the federated matrix and call compute to execute\n+ print(fed_a.sum().compute())\n+ # Result should be 45.\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/docs/source/code/federatedTutorial_part3.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+# Python\n+import numpy as np\n+from systemds.context import SystemDSContext\n+\n+addr1 = \"localhost:8001/temp/test.csv\"\n+addr2 = \"localhost:8002/temp/test.csv\"\n+addr3 = \"localhost:8003/temp/test.csv\"\n+\n+# Create a federated matrix using two federated environments\n+# Note that the two federated matrices are stacked on top of each other\n+\n+with SystemDSContext() as sds:\n+ # federated data on three locations\n+ fed = sds.federated([addr1, addr2, addr3], [\n+ ([0, 0], [3, 3]),\n+ ([3, 0], [6, 3]),\n+ ([6, 0], [9, 3])])\n+ # local matrix to multiply with\n+ loc = sds.from_numpy(np.array([\n+ [1,2,3,4,5,6,7,8,9],\n+ [1,2,3,4,5,6,7,8,9],\n+ [1,2,3,4,5,6,7,8,9]\n+ ]))\n+ # Multiply local and federated\n+ ret = loc @ fed\n+ # execute the lazy script and print\n+ print(ret.compute())\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/docs/source/code/federatedTutorial_part3_old.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+# Python\n+import numpy as np\n+from systemds.context import SystemDSContext\n+\n+addr1 = \"localhost:8001/temp/test.csv\"\n+addr2 = \"localhost:8002/temp/test.csv\"\n+addr3 = \"localhost:8003/temp/test.csv\"\n+\n+# Create a federated matrix using two federated environments\n+# Note that the two federated matrices are stacked on top of each other\n+\n+with SystemDSContext() as sds:\n+ fed_a = sds.federated(\n+ [addr1, addr2],\n+ [([0, 0], [3, 3]), ([0, 3], [3, 6])])\n+\n+ fed_b = sds.federated(\n+ [addr1, addr3],\n+ [([0, 0], [3, 3]), ([0, 3], [3, 6])])\n+\n+ # Multiply, compute and print.\n+ res = (fed_a * fed_b).compute()\n+\n+print(res)\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/docs/source/code/federatedTutorial_part3_old2.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+# Python\n+import numpy as np\n+from systemds.context import SystemDSContext\n+\n+addr1 = \"localhost:8001/temp/test.csv\"\n+addr2 = \"localhost:8002/temp/test.csv\"\n+addr3 = \"localhost:8003/temp/test.csv\"\n+\n+# Create a federated matrix using two federated environments\n+# Note that the two federated matrices are stacked on top of each other\n+\n+with SystemDSContext() as sds:\n+\n+ fed_a = sds.federated([addr1],[([0, 0], [3, 3])])\n+ fed_b = sds.federated([addr2],[([0, 0], [3, 3])])\n+ # fed_c = sds.federated([addr3],[([0, 0], [3, 3])])\n+\n+ np_array = np.array([[1,2,3],[4,5,6],[7,8,9]])\n+\n+ loc_a = sds.from_numpy(np_array)\n+ loc_b = sds.from_numpy(np_array)\n+\n+ fed_res = fed_a @ fed_b\n+ loc_res = loc_a @ loc_b\n+\n+ hybrid_res_1 = fed_a @ loc_b\n+ hybrid_res_2 = loc_a @ fed_b\n+\n+ # compute and print\n+ print(fed_a.compute())\n+ print(fed_b.compute())\n+ print(fed_res.compute(verbose=True))\n+ print(loc_res.compute(verbose=True))\n+ print(hybrid_res_1.compute())\n+ print(hybrid_res_1.compute())\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/docs/source/guide/federated.rst", "new_path": "src/main/python/docs/source/guide/federated.rst", "diff": "@@ -37,7 +37,7 @@ A simple guide to do this is in the SystemDS Repository_.\nIf that is setup correctly simply start a worker using the following command.\nHere the ``8001`` refer to the port used by the worker.\n-.. code-block:: python\n+.. code-block::\nsystemds WORKER 8001\n@@ -47,45 +47,22 @@ Simple Aggregation Example\nIn this example we use a single federated worker, and aggregate the sum of its data.\nFirst we need to create some data for our federated worker to use.\n-In this example we simply use Numpy to create a ``test.csv`` file\n-\n-.. code-block:: python\n-\n- # Import numpy\n- import numpy as np\n- a = np.asarray([[1,2,3], [4,5,6], [7,8,9]])\n- np.savetxt(\"temp/test.csv\", a, delimiter=\",\")\n+In this example we simply use Numpy to create a ``test.csv`` file.\nCurrently we also require a metadata file for the federated worker.\nThis should be located next to the ``test.csv`` file called ``test.csv.mtd``.\n-To make this simply execute the following::\n+To make both the data and metadata simply execute the following\n- echo '{ \"format\":\"csv\", \"header\":false, \"rows\":3, \"cols\":3 }' > temp/test.csv.mtd\n+.. include:: ../code/federatedTutorial_part1.py\n+ :start-line: 20\n+ :code: python\n-After creating our data we the federated worker becomes able to execute federated instructions.\n+After creating our data the federated worker becomes able to execute federated instructions.\nThe aggregated sum using federated instructions in python SystemDS is done as follows\n-.. code-block:: python\n-\n- # Import numpy and SystemDS\n- import numpy as np\n- from systemds.context import SystemDSContext\n-\n- # Create a federated matrix\n- ## Indicate the dimensions of the data:\n- ### Here the first list in the tuple is the top left Coordinate,\n- ### and the second the bottom left coordinate.\n- ### It is ordered as [col,row].\n- dims = ([0,0], [3,3])\n-\n- ## Specify the address + file path from worker:\n- address = \"localhost:8001/temp/test.csv\"\n-\n- with SystemDSContext() as sds:\n- fed_a = sds.federated([address], [dims])\n- # Sum the federated matrix and call compute to execute\n- print(fed_a.sum().compute())\n- # Result should be 45.\n+.. include:: ../code/federatedTutorial_part2.py\n+ :start-line: 20\n+ :code: python\nMultiple Federated Environments\n-------------------------------\n@@ -96,7 +73,7 @@ Using the data created from the last example we can simulate\nmultiple federated workers by starting multiple ones on different ports.\nStart with 3 different terminals, and run one federated environment in each.\n-.. code-block:: python\n+.. code-block::\nsystemds WORKER 8001\nsystemds WORKER 8002\n@@ -104,35 +81,13 @@ Start with 3 different terminals, and run one federated environment in each.\nOnce all three workers are up and running we can leverage all three in the following example\n-.. code-block:: python\n-\n- import numpy as np\n- from systemds.context import SystemDSContext\n-\n- addr1 = \"localhost:8001/temp/test.csv\"\n- addr2 = \"localhost:8002/temp/test.csv\"\n- addr3 = \"localhost:8003/temp/test.csv\"\n-\n- # Create a federated matrix using two federated environments\n- # Note that the two federated matrices are stacked on top of each other\n-\n- with SystemDSContext() as sds:\n- fed_a = sds.federated(\n- [addr1, addr2],\n- [([0,0], [3,3]), ([0,3], [3,6])])\n-\n- fed_b = sds.federated(\n- [addr1, addr3],\n- [([0,0], [3,3]), ([0,3], [3,6])])\n-\n- # Multiply, compute and print.\n- res = (fed_a * fed_b).compute()\n-\n- print(res)\n+.. include:: ../code/federatedTutorial_part3.py\n+ :start-line: 20\n+ :code: python\nThe print should look like\n-.. code-block:: python\n+.. code-block::\n[[ 1. 4. 9. 1. 4. 9.]\n[16. 25. 36. 16. 25. 36.]\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/tests/federated/runFedTest.sh", "new_path": "src/main/python/tests/federated/runFedTest.sh", "diff": "@@ -34,6 +34,7 @@ mkdir -p $workerdir\nmkdir -p $outputdir\nw1_Output=\"$workerdir/w1\"\nw2_Output=\"$workerdir/w2\"\n+w3_Output=\"$workerdir/w3\"\nlog=\"$outputdir/out.log\"\n# Make the workers start quietly and pipe their output to a file to print later\n@@ -42,12 +43,15 @@ systemds WORKER 8001 >$w1_Output 2>&1 &\nFed1=$!\nsystemds WORKER 8002 >$w2_Output 2>&1 &\nFed2=$!\n+systemds WORKER 8003 >$w3_Output 2>&1 &\n+Fed3=$!\necho \"Starting workers\" && sleep 3 && echo \"Starting tests\"\n# Run test\npython -m unittest discover -s tests/federated -p 'test_*.py' $1 >$log 2>&1\npkill -P $Fed1\npkill -P $Fed2\n+pkill -P $Fed3\n# Print output\necho -e \"\\n---------------\\nWorkers Output:\\n---------------\"\n@@ -55,6 +59,8 @@ echo -e \"\\nWorker 1:\"\ncat $w1_Output\necho -e \"\\nWorker 2:\"\ncat $w2_Output\n+echo -e \"\\nWorker 3:\"\n+cat $w3_Output\necho -e \"\\n------------\\nTest output:\\n------------\"\ncat $log\ngrepvals=\"$(tail -n 10 $log | grep OK)\"\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/tests/federated/test_federated_tutorial.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+\n+import shutil\n+import unittest\n+\n+\n+class TestFederatedAggFn(unittest.TestCase):\n+\n+ @classmethod\n+ def tearDownClass(cls):\n+ shutil.rmtree(\"temp\")\n+\n+ def test_part1(self):\n+ import docs.source.code.federatedTutorial_part1\n+\n+ def test_part2(self):\n+ import docs.source.code.federatedTutorial_part2\n+\n+ def test_part3(self):\n+ import docs.source.code.federatedTutorial_part3\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/meta/MetaDataAll.java", "new_path": "src/main/java/org/apache/sysds/runtime/meta/MetaDataAll.java", "diff": "@@ -100,7 +100,7 @@ public class MetaDataAll extends DataIdentifier {\n{\nJSONObject retVal = new JSONObject();\nboolean exists = HDFSTool.existsFileOnHDFS(filename);\n- boolean isDir = HDFSTool.isDirectory(filename);\n+ boolean isDir = exists ? HDFSTool.isDirectory(filename) : false;\n// CASE: filename is a directory -- process as a directory\nif( exists && isDir )\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/util/HDFSTool.java", "new_path": "src/main/java/org/apache/sysds/runtime/util/HDFSTool.java", "diff": "@@ -121,9 +121,8 @@ public class HDFSTool\n.getFileSystem(path).getFileStatus(path).isDirectory();\n}\ncatch(Exception ex) {\n- LOG.error(\"Failed check isDirectory.\", ex);\n+ throw new DMLRuntimeException(\"Failed to check if file is directory\", ex);\n}\n- return false;\n}\npublic static FileStatus[] getDirectoryListing(String fname) {\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/frame/FrameValueSwapTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/frame/FrameValueSwapTest.java", "diff": "@@ -53,14 +53,16 @@ public class FrameValueSwapTest extends AutomatedTestBase\naddTestConfiguration(TEST_NAME, new TestConfiguration(TEST_CLASS_DIR, TEST_NAME, new String[] {\"B\"}));\n}\n//\n+ @Ignore\n@Test\n- public void tesSwapValueTestCP() {\n+ public void testSwapValueTestCP() {\nrunValueSwapTest(ExecType.CP);\n}\n// TODO fix frame comparisons in spark context\n@Ignore\n- public void tesSwapValueTestSP() {\n+ @Test\n+ public void testSwapValueTestSP() {\nrunValueSwapTest(ExecType.SPARK);\n}\n" } ]

[ { "change_type": "MODIFY", "old_path": "pom.xml", "new_path": "pom.xml", "diff": "<parent>\n<groupId>org.apache</groupId>\n<artifactId>apache</artifactId>\n- <version>18</version>\n+ <version>24</version>\n</parent>\n<groupId>org.apache.systemds</groupId>\n<version>2.3.0-SNAPSHOT</version>\n<scala.version>2.12.0</scala.version>\n<scala.binary.version>2.12</scala.binary.version>\n<maven.build.timestamp.format>yyyy-MM-dd HH:mm:ss z</maven.build.timestamp.format>\n+ <project.build.outputTimestamp>1</project.build.outputTimestamp>\n<enableGPU>false</enableGPU>\n<jcuda.scope>provided</jcuda.scope>\n<jcuda.version>10.2.0</jcuda.version>\n" }, { "change_type": "MODIFY", "old_path": "src/assembly/bin.xml", "new_path": "src/assembly/bin.xml", "diff": "<include>*:commons-cli*</include>\n<include>*:commons-collections*</include>\n<include>*:commons-configuration*</include>\n- <include>*:commons-httpclient*</include>\n+ <!-- <include>*:commons-httpclient*</include> -->\n<include>*:commons-io*</include>\n<include>*:commons-lang</include>\n<include>*:commons-lang3</include>\n" } ]

[ { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/tests/federated/test_federated_matrix_mult.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+\n+import io\n+import json\n+import os\n+import time\n+import unittest\n+\n+import numpy as np\n+from systemds.context import SystemDSContext\n+\n+os.environ['SYSDS_QUIET'] = \"1\"\n+\n+dim = 3\n+\n+m = np.reshape(np.arange(1, dim * dim + 1, 1), (dim, dim))\n+m_c2 = np.column_stack((m, m))\n+m_c3 = np.column_stack((m, m_c2))\n+m_r2 = np.row_stack((m, m))\n+m_r3 = np.row_stack((m, m_r2))\n+\n+tempdir = \"./tests/federated/tmp/test_federated_matrixmult/\"\n+mtd = {\"format\": \"csv\", \"header\": False, \"rows\": dim,\n+ \"cols\": dim, \"data_type\": \"matrix\", \"value_type\": \"double\"}\n+\n+# Create the testing directory if it does not exist.\n+if not os.path.exists(tempdir):\n+ os.makedirs(tempdir)\n+\n+# Save data files for the Federated workers.\n+np.savetxt(tempdir + \"m.csv\", m, delimiter=\",\")\n+with io.open(tempdir + \"m.csv.mtd\", \"w\", encoding=\"utf-8\") as f:\n+ f.write(json.dumps(mtd, ensure_ascii=False))\n+\n+# Federated workers + file locations\n+fed1 = \"localhost:8001/\" + tempdir + \"m.csv\"\n+fed2 = \"localhost:8002/\" + tempdir + \"m.csv\"\n+fed3 = \"localhost:8003/\" + tempdir + \"m.csv\"\n+\n+fed1_file = tempdir+\"m1.fed\"\n+fed_c2_file = tempdir+\"m_c2.fed\"\n+fed_c3_file = tempdir+\"m_c3.fed\"\n+fed_r2_file = tempdir+\"m_r2.fed\"\n+fed_r3_file = tempdir+\"m_r3.fed\"\n+\n+\n+class TestFederatedAggFn(unittest.TestCase):\n+\n+ sds: SystemDSContext = None\n+\n+ @classmethod\n+ def setUpClass(cls):\n+ cls.sds = SystemDSContext()\n+ cls.sds.federated([fed1], [([0, 0], [dim, dim])]\n+ ).write(fed1_file, format=\"federated\").compute()\n+ cls.sds.federated([fed1, fed2], [\n+ ([0, 0], [dim, dim]),\n+ ([0, dim], [dim, dim*2])]).write(fed_c2_file, format=\"federated\").compute()\n+ cls.sds.federated([fed1, fed2, fed3], [\n+ ([0, 0], [dim, dim]),\n+ ([0, dim], [dim, dim*2]),\n+ ([0, dim*2], [dim, dim*3])]).write(fed_c3_file, format=\"federated\").compute()\n+ cls.sds.federated([fed1, fed2], [\n+ ([0, 0], [dim, dim]),\n+ ([dim, 0], [dim*2, dim])]).write(fed_r2_file, format=\"federated\").compute()\n+ cls.sds.federated([fed1, fed2, fed3], [\n+ ([0, 0], [dim, dim]),\n+ ([dim, 0], [dim*2, dim]),\n+ ([dim*2, 0], [dim*3, dim])]).write(fed_r3_file, format=\"federated\").compute()\n+\n+ @classmethod\n+ def tearDownClass(cls):\n+ cls.sds.close()\n+\n+ #####################\n+ # Single site tests #\n+ #####################\n+\n+ def test_single_fed_site_same_matrix(self):\n+ f_m = self.sds.read(fed1_file)\n+ self.exec_test(m, m, f_m, f_m)\n+\n+ def test_single_fed_left_same_size(self):\n+ f_m = self.sds.read(fed1_file)\n+ m_s = self.sds.from_numpy(m)\n+ self.exec_test(m, m, m_s, f_m)\n+\n+ def test_single_fed_left_plus_one_row(self):\n+ f_m = self.sds.read(fed1_file)\n+ m_row_plus1 = np.reshape(\n+ np.arange(1, dim*(dim+1) + 1, 1), (dim+1, dim))\n+ m_s = self.sds.from_numpy(m_row_plus1)\n+ self.exec_test(m_row_plus1, m, m_s, f_m)\n+\n+ def test_single_fed_left_minus_one_row(self):\n+ f_m = self.sds.read(fed1_file)\n+ m_row_minus1 = np.reshape(\n+ np.arange(1, dim*(dim-1) + 1, 1), (dim-1, dim))\n+ m_s = self.sds.from_numpy(m_row_minus1)\n+ self.exec_test(m_row_minus1, m, m_s, f_m)\n+\n+ def test_single_fed_left_vector_row(self):\n+ f_m = self.sds.read(fed1_file)\n+ v_row = np.arange(1, dim + 1, 1)\n+ v_s = self.sds.from_numpy(v_row).t()\n+ self.exec_test(v_row, m, v_s, f_m)\n+\n+ def test_single_fed_right_same_size(self):\n+ f_m = self.sds.read(fed1_file)\n+ m_s = self.sds.from_numpy(m)\n+ self.exec_test(m, m, f_m, m_s)\n+\n+ def test_single_fed_right_plus_one_row(self):\n+ f_m = self.sds.read(fed1_file)\n+ m_col_plus1 = np.reshape(\n+ np.arange(1, dim*(dim+1) + 1, 1), (dim, dim+1))\n+ m_s = self.sds.from_numpy(m_col_plus1)\n+ self.exec_test(m, m_col_plus1, f_m, m_s)\n+\n+ def test_single_fed_right_minus_one_row(self):\n+ f_m = self.sds.read(fed1_file)\n+ m_col_minus1 = np.reshape(\n+ np.arange(1, dim*(dim-1) + 1, 1), (dim, dim-1))\n+ m_s = self.sds.from_numpy(m_col_minus1)\n+ self.exec_test(m, m_col_minus1, f_m, m_s)\n+\n+ def test_single_fed_right_vector(self):\n+ f_m = self.sds.read(fed1_file)\n+ v_col = np.reshape(np.arange(1, dim + 1, 1), (1, dim))\n+ v_col_sds = self.sds.from_numpy(v_col).t()\n+ self.exec_test(m, np.transpose(v_col), f_m, v_col_sds)\n+\n+ ##################################\n+ # start two federated site tests #\n+ ##################################\n+\n+ def test_two_fed_standard(self):\n+ f_m2 = self.sds.read(fed_c2_file)\n+ m = np.reshape(np.arange(1, dim*(dim + dim) + 1, 1), (dim*2, dim))\n+ m_s = self.sds.from_numpy(m)\n+ self.exec_test(m, m_c2, m_s, f_m2)\n+\n+ def test_two_fed_left_minus_one_row(self):\n+ f_m2 = self.sds.read(fed_c2_file)\n+ m = np.reshape(np.arange(1, dim*(dim + dim-1)+1, 1), (dim*2 - 1, dim))\n+ m_s = self.sds.from_numpy(m)\n+ self.exec_test(m, m_c2, m_s, f_m2)\n+\n+ def test_two_fed_left_plus_one_row(self):\n+ f_m2 = self.sds.read(fed_c2_file)\n+ m = np.reshape(np.arange(1, dim*(dim + dim+1)+1, 1), (dim*2 + 1, dim))\n+ m_s = self.sds.from_numpy(m)\n+ self.exec_test(m, m_c2, m_s, f_m2)\n+\n+ def test_two_fed_left_vector_row(self):\n+ f_m2 = self.sds.read(fed_c2_file)\n+ m = np.arange(1, dim+1, 1)\n+ m_s = self.sds.from_numpy(m).t()\n+ self.exec_test(m, m_c2, m_s, f_m2)\n+\n+ def test_two_fed_right_standard(self):\n+ f_m2 = self.sds.read(fed_c2_file)\n+ m_s = self.sds.from_numpy(m_r2)\n+ self.exec_test(m_c2, m_r2, f_m2, m_s)\n+\n+ def test_two_fed_right_col_minus_1(self):\n+ f_m2 = self.sds.read(fed_c2_file)\n+ m = np.reshape(np.arange(1, (dim-1)*(dim + dim)+1, 1),\n+ (dim * 2, dim-1))\n+ m_s = self.sds.from_numpy(m)\n+ self.exec_test(m_c2, m, f_m2, m_s)\n+\n+ def test_two_fed_right_col_plus_1(self):\n+ f_m2 = self.sds.read(fed_c2_file)\n+ m = np.reshape(np.arange(1, (dim+1)*(dim + dim)+1, 1),\n+ (dim * 2, dim+1))\n+ m_s = self.sds.from_numpy(m)\n+ self.exec_test(m_c2, m, f_m2, m_s)\n+\n+ def test_two_fed_right_vector(self):\n+ f_m2 = self.sds.read(fed_c2_file)\n+ m = np.reshape(np.arange(1, (dim + dim)+1, 1), (dim * 2, 1))\n+ m_s = self.sds.from_numpy(m)\n+ self.exec_test(m_c2, m, f_m2, m_s)\n+\n+ ####################################\n+ # Start three federated site tests #\n+ ####################################\n+\n+ def test_three_fed_standard(self):\n+ f_m3 = self.sds.read(fed_c3_file)\n+ m = np.reshape(np.arange(1, dim*(dim * 3) + 1, 1), (dim*3, dim))\n+ m_s = self.sds.from_numpy(m)\n+ self.exec_test(m, m_c3, m_s, f_m3)\n+\n+ def test_three_fed_left_minus_one_row(self):\n+ f_m3 = self.sds.read(fed_c3_file)\n+ m = np.reshape(np.arange(1, dim*(dim * 3-1)+1, 1), (dim*3 - 1, dim))\n+ m_s = self.sds.from_numpy(m)\n+ self.exec_test(m, m_c3, m_s, f_m3)\n+\n+ def test_three_fed_left_plus_one_row(self):\n+ f_m3 = self.sds.read(fed_c3_file)\n+ m = np.reshape(np.arange(1, dim*(dim *3+1)+1, 1), (dim*3 + 1, dim))\n+ m_s = self.sds.from_numpy(m)\n+ self.exec_test(m, m_c3, m_s, f_m3)\n+\n+ def test_three_fed_left_vector_row(self):\n+ f_m3 = self.sds.read(fed_c3_file)\n+ m = np.arange(1, dim+1, 1)\n+ m_s = self.sds.from_numpy(m).t()\n+ self.exec_test(m, m_c3, m_s, f_m3)\n+\n+ def test_three_fed_right_standard(self):\n+ f_m3 = self.sds.read(fed_c3_file)\n+ m_s = self.sds.from_numpy(m_r3)\n+ self.exec_test(m_c3, m_r3, f_m3, m_s)\n+\n+ def test_three_fed_right_col_minus_1(self):\n+ f_m3 = self.sds.read(fed_c3_file)\n+ m = np.reshape(np.arange(1, (dim-1)*(dim*3)+1, 1), (dim * 3, dim-1))\n+ m_s = self.sds.from_numpy(m)\n+ self.exec_test(m_c3, m, f_m3, m_s)\n+\n+ def test_three_fed_right_col_plus_1(self):\n+ f_m3 = self.sds.read(fed_c3_file)\n+ m = np.reshape(np.arange(1, (dim+1)*(dim *3)+1, 1), (dim * 3, dim+1))\n+ m_s = self.sds.from_numpy(m)\n+ self.exec_test(m_c3, m, f_m3, m_s)\n+\n+ def test_three_fed_right_vector(self):\n+ f_m3 = self.sds.read(fed_c3_file)\n+ m = np.reshape(np.arange(1, (dim *3)+1, 1), (dim * 3, 1))\n+ m_s = self.sds.from_numpy(m)\n+ self.exec_test(m_c3, m, f_m3, m_s)\n+\n+ ###################\n+ # row bind matrix #\n+ ###################\n+\n+ def test_federated_row2_binded(self):\n+ fed = self.sds.read(fed_r2_file)\n+ s_m = self.sds.from_numpy(m_c2)\n+ self.exec_test(m_c2, m_r2, s_m, fed)\n+\n+ def test_federated_row3_binded(self):\n+ fed = self.sds.read(fed_r3_file)\n+ s_m = self.sds.from_numpy(m_c3)\n+ self.exec_test(m_c3, m_r3, s_m, fed)\n+\n+\n+\n+\n+ def test_previously_failing(self):\n+ # local matrix to multiply with\n+ loc = np.array([\n+ [1, 2, 3, 4, 5, 6, 7, 8, 9],\n+ [1, 2, 3, 4, 5, 6, 7, 8, 9],\n+ [1, 2, 3, 4, 5, 6, 7, 8, 9]])\n+ # Multiply local and federated\n+ ret_loc = loc @ m_r3\n+\n+ for i in range(1, 100):\n+ loc_systemds = self.sds.from_numpy(loc)\n+ fed = self.sds.read(fed_r3_file)\n+ ret_fed = (loc_systemds @ fed).compute()\n+ if not np.allclose(ret_fed, ret_loc):\n+ self.fail(\n+ \"not equal outputs of federated matrix multiplications\")\n+\n+ def exec_test(self, left, right, f_left, f_right):\n+ fed = f_left @ f_right\n+ loc = left @ right\n+ fed_res = fed.compute()\n+ self.assertTrue(np.allclose(fed_res, loc))\n+\n+\n+if __name__ == \"__main__\":\n+ unittest.main(exit=False)\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/tests/federated/test_federated_read.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+\n+import io\n+import json\n+import os\n+import shutil\n+import sys\n+import unittest\n+\n+import numpy as np\n+from systemds.context import SystemDSContext\n+\n+os.environ['SYSDS_QUIET'] = \"1\"\n+\n+dim = 3\n+\n+m = np.reshape(np.arange(1, dim * dim + 1, 1), (dim, dim))\n+\n+tempdir = \"./tests/federated/tmp/test_federated_matrixmult/\"\n+mtd = {\"format\": \"csv\", \"header\": False, \"rows\": dim,\n+ \"cols\": dim, \"data_type\": \"matrix\", \"value_type\": \"double\"}\n+\n+# Create the testing directory if it does not exist.\n+if not os.path.exists(tempdir):\n+ os.makedirs(tempdir)\n+\n+# Save data files for the Federated workers.\n+np.savetxt(tempdir + \"m.csv\", m, delimiter=\",\")\n+with io.open(tempdir + \"m.csv.mtd\", \"w\", encoding=\"utf-8\") as f:\n+ f.write(json.dumps(mtd, ensure_ascii=False))\n+\n+# Federated workers + file locations\n+fed1 = \"localhost:8001/\" + tempdir + \"m.csv\"\n+fed2 = \"localhost:8002/\" + tempdir + \"m.csv\"\n+fed3 = \"localhost:8003/\" + tempdir + \"m.csv\"\n+\n+fed1_file = tempdir+\"m1.fed\"\n+fed2_file = tempdir+\"m2.fed\"\n+fed3_file = tempdir+\"m3.fed\"\n+\n+\n+class TestFederatedAggFn(unittest.TestCase):\n+\n+ sds: SystemDSContext = None\n+\n+ @classmethod\n+ def setUpClass(cls):\n+ cls.sds = SystemDSContext()\n+ cls.sds.federated([fed1], [\n+ ([0, 0], [dim, dim])]).write(fed1_file, format=\"federated\").compute()\n+ cls.sds.federated([fed1, fed2], [\n+ ([0, 0], [dim, dim]),\n+ ([0, dim], [dim, dim*2])]).write(fed2_file, format=\"federated\").compute()\n+ cls.sds.federated([fed1, fed2, fed3], [\n+ ([0, 0], [dim, dim]),\n+ ([0, dim], [dim, dim*2]),\n+ ([0, dim*2], [dim, dim*3])]).write(fed3_file, format=\"federated\").compute()\n+\n+ @classmethod\n+ def tearDownClass(cls):\n+ cls.sds.close()\n+\n+ def test_verify_same_input(self):\n+ f_m = self.sds.federated([fed1], [([0, 0], [dim, dim])]).compute()\n+ self.assertTrue(np.allclose(f_m, m))\n+\n+ def test_verify_same_input_if_reading_fed(self):\n+ f_m = self.sds.read(fed1_file).compute()\n+ self.assertTrue(np.allclose(f_m, m))\n+\n+ def test_verify_same_input_if_reading_fed2(self):\n+ f_m = self.sds.read(fed2_file).compute()\n+ m2 = np.column_stack((m,m))\n+ self.assertTrue(np.allclose(f_m, m2))\n+\n+ def test_verify_same_input_if_reading_fed3(self):\n+ f_m = self.sds.read(fed3_file).compute()\n+ m2 = np.column_stack((m,m))\n+ m3 = np.column_stack((m,m2))\n+ self.assertTrue(np.allclose(f_m, m3))\n+\n+\n+if __name__ == \"__main__\":\n+ unittest.main(exit=False)\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/lineage/FedFullReuseTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/lineage/FedFullReuseTest.java", "diff": "@@ -31,6 +31,7 @@ import org.apache.sysds.test.TestConfiguration;\nimport org.apache.sysds.test.TestUtils;\nimport org.apache.sysds.utils.Statistics;\nimport org.junit.Assert;\n+import org.junit.Ignore;\nimport org.junit.Test;\nimport org.junit.runner.RunWith;\nimport org.junit.runners.Parameterized;\n@@ -69,6 +70,7 @@ public class FedFullReuseTest extends AutomatedTestBase {\n}\n@Test\n+ @Ignore\npublic void federatedOutputReuse() {\n//don't cache federated outputs in the coordinator\n//reuse inside federated workers\n@@ -76,6 +78,7 @@ public class FedFullReuseTest extends AutomatedTestBase {\n}\n@Test\n+ @Ignore\npublic void nonfederatedOutputReuse() {\n//cache non-federated outputs in the coordinator\nfederatedReuse(TEST_NAME2);\n" } ]

[ { "change_type": "MODIFY", "old_path": "dev/release/pypi-upload.sh", "new_path": "dev/release/pypi-upload.sh", "diff": "@@ -86,7 +86,7 @@ python3 -m twine check dist/*\n# password: pypi-DU5y...\nif [[ $dry_run_flag != 1 ]]; then\n- python twine upload dist/*\n+ python -m twine upload dist/*\nelse\npython -m twine upload --repository testpypi dist/*\nfi\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/pre_setup.py", "new_path": "src/main/python/pre_setup.py", "diff": "@@ -39,7 +39,7 @@ if os.path.exists(TMP_DIR):\nshutil.rmtree(TMP_DIR, True)\nos.mkdir(TMP_DIR)\n-SYSTEMDS_BIN = 'systemds-*-SNAPSHOT-bin.zip'\n+SYSTEMDS_BIN = 'systemds-*-bin.zip'\nfor file in os.listdir(os.path.join(root_dir, 'target')):\nif fnmatch.fnmatch(file, SYSTEMDS_BIN):\nnew_path = os.path.join(TMP_DIR, file)\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/hops/rewrite/IPAPassRewriteFederatedPlan.java", "new_path": "src/main/java/org/apache/sysds/hops/rewrite/IPAPassRewriteFederatedPlan.java", "diff": "@@ -24,6 +24,7 @@ import org.apache.sysds.hops.AggBinaryOp;\nimport org.apache.sysds.hops.AggUnaryOp;\nimport org.apache.sysds.hops.BinaryOp;\nimport org.apache.sysds.hops.DataOp;\n+import org.apache.sysds.hops.FunctionOp;\nimport org.apache.sysds.hops.Hop;\nimport org.apache.sysds.hops.OptimizerUtils;\nimport org.apache.sysds.hops.ReorgOp;\n@@ -84,7 +85,7 @@ public class IPAPassRewriteFederatedPlan extends IPAPass {\n*/\n@Override\npublic boolean rewriteProgram(DMLProgram prog, FunctionCallGraph fgraph, FunctionCallSizeInfo fcallSizes) {\n- rewriteStatementBlocks(prog.getStatementBlocks());\n+ rewriteStatementBlocks(prog, prog.getStatementBlocks());\nreturn false;\n}\n@@ -93,13 +94,14 @@ public class IPAPassRewriteFederatedPlan extends IPAPass {\n* by setting the federated output value of each hop in the statement blocks.\n* The method calls the contained statement blocks recursively.\n*\n+ * @param prog dml program\n* @param sbs list of statement blocks\n* @return list of statement blocks with the federated output value updated for each hop\n*/\n- public ArrayList<StatementBlock> rewriteStatementBlocks(List<StatementBlock> sbs) {\n+ public ArrayList<StatementBlock> rewriteStatementBlocks(DMLProgram prog, List<StatementBlock> sbs) {\nArrayList<StatementBlock> rewrittenStmBlocks = new ArrayList<>();\nfor ( StatementBlock stmBlock : sbs )\n- rewrittenStmBlocks.addAll(rewriteStatementBlock(stmBlock));\n+ rewrittenStmBlocks.addAll(rewriteStatementBlock(prog, stmBlock));\nreturn rewrittenStmBlocks;\n}\n@@ -108,66 +110,80 @@ public class IPAPassRewriteFederatedPlan extends IPAPass {\n* by setting the federated output value of each hop in the statement blocks.\n* The method calls the contained statement blocks recursively.\n*\n+ * @param prog dml program\n* @param sb statement block\n* @return list of statement blocks with the federated output value updated for each hop\n*/\n- public ArrayList<StatementBlock> rewriteStatementBlock(StatementBlock sb) {\n+ public ArrayList<StatementBlock> rewriteStatementBlock(DMLProgram prog, StatementBlock sb) {\nif ( sb instanceof WhileStatementBlock)\n- return rewriteWhileStatementBlock((WhileStatementBlock) sb);\n+ return rewriteWhileStatementBlock(prog, (WhileStatementBlock) sb);\nelse if ( sb instanceof IfStatementBlock)\n- return rewriteIfStatementBlock((IfStatementBlock) sb);\n+ return rewriteIfStatementBlock(prog, (IfStatementBlock) sb);\nelse if ( sb instanceof ForStatementBlock){\n// This also includes ParForStatementBlocks\n- return rewriteForStatementBlock((ForStatementBlock) sb);\n+ return rewriteForStatementBlock(prog, (ForStatementBlock) sb);\n}\nelse if ( sb instanceof FunctionStatementBlock)\n- return rewriteFunctionStatementBlock((FunctionStatementBlock) sb);\n+ return rewriteFunctionStatementBlock(prog, (FunctionStatementBlock) sb);\nelse {\n// StatementBlock type (no subclass)\n- selectFederatedExecutionPlan(sb.getHops());\n+ return rewriteDefaultStatementBlock(prog, sb);\n}\n- return new ArrayList<>(Collections.singletonList(sb));\n}\n- private ArrayList<StatementBlock> rewriteWhileStatementBlock(WhileStatementBlock whileSB){\n+ private ArrayList<StatementBlock> rewriteWhileStatementBlock(DMLProgram prog, WhileStatementBlock whileSB){\nHop whilePredicateHop = whileSB.getPredicateHops();\nselectFederatedExecutionPlan(whilePredicateHop);\nfor ( Statement stm : whileSB.getStatements() ){\nWhileStatement whileStm = (WhileStatement) stm;\n- whileStm.setBody(rewriteStatementBlocks(whileStm.getBody()));\n+ whileStm.setBody(rewriteStatementBlocks(prog, whileStm.getBody()));\n}\nreturn new ArrayList<>(Collections.singletonList(whileSB));\n}\n- private ArrayList<StatementBlock> rewriteIfStatementBlock(IfStatementBlock ifSB){\n+ private ArrayList<StatementBlock> rewriteIfStatementBlock(DMLProgram prog, IfStatementBlock ifSB){\nselectFederatedExecutionPlan(ifSB.getPredicateHops());\nfor ( Statement statement : ifSB.getStatements() ){\nIfStatement ifStatement = (IfStatement) statement;\n- ifStatement.setIfBody(rewriteStatementBlocks(ifStatement.getIfBody()));\n- ifStatement.setElseBody(rewriteStatementBlocks(ifStatement.getElseBody()));\n+ ifStatement.setIfBody(rewriteStatementBlocks(prog, ifStatement.getIfBody()));\n+ ifStatement.setElseBody(rewriteStatementBlocks(prog, ifStatement.getElseBody()));\n}\nreturn new ArrayList<>(Collections.singletonList(ifSB));\n}\n- private ArrayList<StatementBlock> rewriteForStatementBlock(ForStatementBlock forSB){\n+ private ArrayList<StatementBlock> rewriteForStatementBlock(DMLProgram prog, ForStatementBlock forSB){\nselectFederatedExecutionPlan(forSB.getFromHops());\nselectFederatedExecutionPlan(forSB.getToHops());\nselectFederatedExecutionPlan(forSB.getIncrementHops());\nfor ( Statement statement : forSB.getStatements() ){\nForStatement forStatement = ((ForStatement)statement);\n- forStatement.setBody(rewriteStatementBlocks(forStatement.getBody()));\n+ forStatement.setBody(rewriteStatementBlocks(prog, forStatement.getBody()));\n}\nreturn new ArrayList<>(Collections.singletonList(forSB));\n}\n- private ArrayList<StatementBlock> rewriteFunctionStatementBlock(FunctionStatementBlock funcSB){\n+ private ArrayList<StatementBlock> rewriteFunctionStatementBlock(DMLProgram prog, FunctionStatementBlock funcSB){\nfor ( Statement statement : funcSB.getStatements() ){\nFunctionStatement funcStm = (FunctionStatement) statement;\n- funcStm.setBody(rewriteStatementBlocks(funcStm.getBody()));\n+ funcStm.setBody(rewriteStatementBlocks(prog, funcStm.getBody()));\n}\nreturn new ArrayList<>(Collections.singletonList(funcSB));\n}\n+ private ArrayList<StatementBlock> rewriteDefaultStatementBlock(DMLProgram prog, StatementBlock sb){\n+ if ( sb.getHops() != null && !sb.getHops().isEmpty() ){\n+ for ( Hop sbHop : sb.getHops() ){\n+ if ( sbHop instanceof FunctionOp ){\n+ String funcName = ((FunctionOp) sbHop).getFunctionName();\n+ FunctionStatementBlock sbFuncBlock = prog.getBuiltinFunctionDictionary().getFunction(funcName);\n+ rewriteStatementBlock(prog, sbFuncBlock);\n+ }\n+ else selectFederatedExecutionPlan(sbHop);\n+ }\n+ }\n+ return new ArrayList<>(Collections.singletonList(sb));\n+ }\n+\n/**\n* Sets FederatedOutput field of all hops in DAG starting from given root.\n* The FederatedOutput chosen for root is the minimum cost HopRel found in memo table for the given root.\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/hops/rewrite/RewriteFederatedExecution.java", "new_path": "src/main/java/org/apache/sysds/hops/rewrite/RewriteFederatedExecution.java", "diff": "@@ -56,15 +56,16 @@ public class RewriteFederatedExecution extends HopRewriteRule {\n@Override\npublic ArrayList<Hop> rewriteHopDAGs(ArrayList<Hop> roots, ProgramRewriteStatus state) {\n- if ( roots == null )\n- return null;\n+ if ( roots != null )\nfor ( Hop root : roots )\n- visitHop(root);\n+ rewriteHopDAG(root, state);\nreturn roots;\n}\n@Override public Hop rewriteHopDAG(Hop root, ProgramRewriteStatus state) {\n- return null;\n+ if ( root != null )\n+ visitHop(root);\n+ return root;\n}\nprivate void visitHop(Hop hop){\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/privacy/algorithms/FederatedL2SVMTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/privacy/algorithms/FederatedL2SVMTest.java", "diff": "package org.apache.sysds.test.functions.privacy.algorithms;\n+import edu.emory.mathcs.backport.java.util.Arrays;\n+import org.apache.sysds.hops.OptimizerUtils;\nimport org.apache.sysds.runtime.DMLRuntimeException;\nimport org.junit.Assert;\n-import org.junit.Ignore;\nimport org.junit.Test;\nimport org.apache.sysds.api.DMLScript;\nimport org.apache.sysds.common.Types;\n@@ -31,12 +32,15 @@ import org.apache.sysds.runtime.privacy.PrivacyConstraint.PrivacyLevel;\nimport org.apache.sysds.test.AutomatedTestBase;\nimport org.apache.sysds.test.TestConfiguration;\nimport org.apache.sysds.test.TestUtils;\n-import org.apache.wink.json4j.JSONException;\n+import org.junit.runner.RunWith;\n+import org.junit.runners.Parameterized;\n+import java.util.Collection;\nimport java.util.HashMap;\nimport java.util.Map;\[email protected]\n+@RunWith(value = Parameterized.class)\npublic class FederatedL2SVMTest extends AutomatedTestBase {\nprivate final static String TEST_DIR = \"functions/federated/\";\n@@ -47,57 +51,62 @@ public class FederatedL2SVMTest extends AutomatedTestBase {\nprivate int rows = 100;\nprivate int cols = 10;\n- @Override\n- public void setUp() {\n+ @Parameterized.Parameter()\n+ public boolean fedOutCompilation;\n+\n+ @Parameterized.Parameters\n+ public static Collection<Object[]> data() {\n+ return Arrays.asList(new Object[][]{\n+ {false},\n+ {true}\n+ });\n+ }\n+\n+ @Override public void setUp() {\nTestUtils.clearAssertionInformation();\naddTestConfiguration(TEST_NAME, new TestConfiguration(TEST_CLASS_DIR, TEST_NAME, new String[] {\"Z\"}));\n}\n// PrivateAggregation Single Input\n- @Test\n- @Ignore\n- public void federatedL2SVMCPPrivateAggregationX1() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivateAggregationX1() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"X1\", new PrivacyConstraint(PrivacyLevel.PrivateAggregation));\n- federatedL2SVMNoException(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.PrivateAggregation);\n+ federatedL2SVMNoException(Types.ExecMode.SINGLE_NODE, privacyConstraints, null,\n+ PrivacyLevel.PrivateAggregation);\n}\n- @Test\n- @Ignore\n- public void federatedL2SVMCPPrivateAggregationX2() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivateAggregationX2() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"X2\", new PrivacyConstraint(PrivacyLevel.PrivateAggregation));\n- federatedL2SVMNoException(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.PrivateAggregation);\n+ federatedL2SVMNoException(Types.ExecMode.SINGLE_NODE, privacyConstraints, null,\n+ PrivacyLevel.PrivateAggregation);\n}\n- @Test\n- public void federatedL2SVMCPPrivateAggregationY() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivateAggregationY() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"Y\", new PrivacyConstraint(PrivacyLevel.PrivateAggregation));\n- federatedL2SVMNoException(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.PrivateAggregation);\n+ federatedL2SVMNoException(Types.ExecMode.SINGLE_NODE, privacyConstraints, null,\n+ PrivacyLevel.PrivateAggregation);\n}\n// Private Single Input\n- @Test\n- public void federatedL2SVMCPPrivateFederatedX1() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivateFederatedX1() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"X1\", new PrivacyConstraint(PrivacyLevel.Private));\n- federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private,\n- false, null, true, DMLRuntimeException.class);\n+ federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private, false, null, true,\n+ DMLRuntimeException.class);\n}\n- @Test\n- public void federatedL2SVMCPPrivateFederatedX2() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivateFederatedX2() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"X2\", new PrivacyConstraint(PrivacyLevel.Private));\n- federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private,\n- false, null, true, DMLRuntimeException.class);\n+ federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private, false, null, true,\n+ DMLRuntimeException.class);\n}\n- @Test\n- public void federatedL2SVMCPPrivateFederatedY() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivateFederatedY() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"Y\", new PrivacyConstraint(PrivacyLevel.Private));\nfederatedL2SVMNoException(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private);\n@@ -105,230 +114,212 @@ public class FederatedL2SVMTest extends AutomatedTestBase {\n// Setting Privacy of Matrix (Throws Exception)\n- @Test\n- public void federatedL2SVMCPPrivateMatrixX1() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivateMatrixX1() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"X1\", new PrivacyConstraint(PrivacyLevel.Private));\n- federatedL2SVM(Types.ExecMode.SINGLE_NODE, null, privacyConstraints, PrivacyLevel.Private,\n- false, null, false, null);\n+ federatedL2SVM(Types.ExecMode.SINGLE_NODE, null, privacyConstraints, PrivacyLevel.Private, false, null, false,\n+ null);\n}\n- @Test\n- public void federatedL2SVMCPPrivateMatrixX2() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivateMatrixX2() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"X2\", new PrivacyConstraint(PrivacyLevel.Private));\n- federatedL2SVM(Types.ExecMode.SINGLE_NODE, null, privacyConstraints, PrivacyLevel.Private,\n- false, null, false, null);\n+ federatedL2SVM(Types.ExecMode.SINGLE_NODE, null, privacyConstraints, PrivacyLevel.Private, false, null, false,\n+ null);\n}\n- @Test\n- public void federatedL2SVMCPPrivateMatrixY() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivateMatrixY() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"Y\", new PrivacyConstraint(PrivacyLevel.Private));\n- federatedL2SVM(Types.ExecMode.SINGLE_NODE, null, privacyConstraints, PrivacyLevel.Private,\n- false, null, false, null);\n+ federatedL2SVM(Types.ExecMode.SINGLE_NODE, null, privacyConstraints, PrivacyLevel.Private, false, null, false,\n+ null);\n}\n- @Test\n- public void federatedL2SVMCPPrivateFederatedAndMatrixX1() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivateFederatedAndMatrixX1() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"X1\", new PrivacyConstraint(PrivacyLevel.Private));\n- federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, privacyConstraints, PrivacyLevel.Private,\n- false, null, true, DMLRuntimeException.class);\n+ federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, privacyConstraints, PrivacyLevel.Private, false,\n+ null, true, DMLRuntimeException.class);\n}\n- @Test\n- public void federatedL2SVMCPPrivateFederatedAndMatrixX2() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivateFederatedAndMatrixX2() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"X2\", new PrivacyConstraint(PrivacyLevel.Private));\n- federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, privacyConstraints, PrivacyLevel.Private,\n- false, null, true, DMLRuntimeException.class);\n+ federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, privacyConstraints, PrivacyLevel.Private, false,\n+ null, true, DMLRuntimeException.class);\n}\n- @Test\n- public void federatedL2SVMCPPrivateFederatedAndMatrixY() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivateFederatedAndMatrixY() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"Y\", new PrivacyConstraint(PrivacyLevel.Private));\n- federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, privacyConstraints, PrivacyLevel.Private,\n- false, null, false, null);\n+ federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, privacyConstraints, PrivacyLevel.Private, false,\n+ null, false, null);\n}\n// Privacy Level Private Combinations\n- @Test\n- public void federatedL2SVMCPPrivateFederatedX1X2() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivateFederatedX1X2() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"X1\", new PrivacyConstraint(PrivacyLevel.Private));\nprivacyConstraints.put(\"X2\", new PrivacyConstraint(PrivacyLevel.Private));\n- federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private,\n- false, null, true, DMLRuntimeException.class);\n+ federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private, false, null, true,\n+ DMLRuntimeException.class);\n}\n- @Test\n- public void federatedL2SVMCPPrivateFederatedX1Y() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivateFederatedX1Y() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"X1\", new PrivacyConstraint(PrivacyLevel.Private));\nprivacyConstraints.put(\"Y\", new PrivacyConstraint(PrivacyLevel.Private));\n- federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private,\n- false, null, true, DMLRuntimeException.class);\n+ federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private, false, null, true,\n+ DMLRuntimeException.class);\n}\n- @Test\n- public void federatedL2SVMCPPrivateFederatedX2Y() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivateFederatedX2Y() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"X2\", new PrivacyConstraint(PrivacyLevel.Private));\nprivacyConstraints.put(\"Y\", new PrivacyConstraint(PrivacyLevel.Private));\n- federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private,\n- false, null, true, DMLRuntimeException.class);\n+ federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private, false, null, true,\n+ DMLRuntimeException.class);\n}\n- @Test\n- public void federatedL2SVMCPPrivateFederatedX1X2Y() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivateFederatedX1X2Y() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"X1\", new PrivacyConstraint(PrivacyLevel.Private));\nprivacyConstraints.put(\"X2\", new PrivacyConstraint(PrivacyLevel.Private));\nprivacyConstraints.put(\"Y\", new PrivacyConstraint(PrivacyLevel.Private));\n- federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private,\n- false, null, true, DMLRuntimeException.class);\n+ federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private, false, null, true,\n+ DMLRuntimeException.class);\n}\n// Privacy Level PrivateAggregation Combinations\n- @Test\n- public void federatedL2SVMCPPrivateAggregationFederatedX1X2() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivateAggregationFederatedX1X2() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"X1\", new PrivacyConstraint(PrivacyLevel.PrivateAggregation));\nprivacyConstraints.put(\"X2\", new PrivacyConstraint(PrivacyLevel.PrivateAggregation));\n- federatedL2SVMNoException(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.PrivateAggregation);\n+ federatedL2SVMNoException(Types.ExecMode.SINGLE_NODE, privacyConstraints, null,\n+ PrivacyLevel.PrivateAggregation);\n}\n- @Test\n- public void federatedL2SVMCPPrivateAggregationFederatedX1Y() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivateAggregationFederatedX1Y() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"X1\", new PrivacyConstraint(PrivacyLevel.PrivateAggregation));\nprivacyConstraints.put(\"Y\", new PrivacyConstraint(PrivacyLevel.PrivateAggregation));\n- federatedL2SVMNoException(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.PrivateAggregation);\n+ federatedL2SVMNoException(Types.ExecMode.SINGLE_NODE, privacyConstraints, null,\n+ PrivacyLevel.PrivateAggregation);\n}\n- @Test\n- public void federatedL2SVMCPPrivateAggregationFederatedX2Y() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivateAggregationFederatedX2Y() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"X2\", new PrivacyConstraint(PrivacyLevel.PrivateAggregation));\nprivacyConstraints.put(\"Y\", new PrivacyConstraint(PrivacyLevel.PrivateAggregation));\n- federatedL2SVMNoException(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.PrivateAggregation);\n+ federatedL2SVMNoException(Types.ExecMode.SINGLE_NODE, privacyConstraints, null,\n+ PrivacyLevel.PrivateAggregation);\n}\n- @Test\n- public void federatedL2SVMCPPrivateAggregationFederatedX1X2Y() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivateAggregationFederatedX1X2Y() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"X1\", new PrivacyConstraint(PrivacyLevel.PrivateAggregation));\nprivacyConstraints.put(\"X2\", new PrivacyConstraint(PrivacyLevel.PrivateAggregation));\nprivacyConstraints.put(\"Y\", new PrivacyConstraint(PrivacyLevel.PrivateAggregation));\n- federatedL2SVMNoException(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.PrivateAggregation);\n+ federatedL2SVMNoException(Types.ExecMode.SINGLE_NODE, privacyConstraints, null,\n+ PrivacyLevel.PrivateAggregation);\n}\n// Privacy Level Combinations\n- @Test\n- public void federatedL2SVMCPPrivatePrivateAggregationFederatedX1X2() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivatePrivateAggregationFederatedX1X2() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"X1\", new PrivacyConstraint(PrivacyLevel.Private));\nprivacyConstraints.put(\"X2\", new PrivacyConstraint(PrivacyLevel.PrivateAggregation));\n- federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private,\n- false, null, true, DMLRuntimeException.class);\n+ federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private, false, null, true,\n+ DMLRuntimeException.class);\n}\n- @Test\n- public void federatedL2SVMCPPrivatePrivateAggregationFederatedX1Y() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivatePrivateAggregationFederatedX1Y() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"X1\", new PrivacyConstraint(PrivacyLevel.Private));\nprivacyConstraints.put(\"Y\", new PrivacyConstraint(PrivacyLevel.PrivateAggregation));\n- federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private,\n- false, null, true, DMLRuntimeException.class);\n+ federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private, false, null, true,\n+ DMLRuntimeException.class);\n}\n- @Test\n- public void federatedL2SVMCPPrivatePrivateAggregationFederatedX2Y() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivatePrivateAggregationFederatedX2Y() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"X2\", new PrivacyConstraint(PrivacyLevel.Private));\nprivacyConstraints.put(\"Y\", new PrivacyConstraint(PrivacyLevel.PrivateAggregation));\n- federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private,\n- false, null, true, DMLRuntimeException.class);\n+ federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private, false, null, true,\n+ DMLRuntimeException.class);\n}\n- @Test\n- public void federatedL2SVMCPPrivatePrivateAggregationFederatedYX1() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivatePrivateAggregationFederatedYX1() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"Y\", new PrivacyConstraint(PrivacyLevel.Private));\nprivacyConstraints.put(\"X1\", new PrivacyConstraint(PrivacyLevel.PrivateAggregation));\nfederatedL2SVMNoException(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private);\n}\n- @Test\n- public void federatedL2SVMCPPrivatePrivateAggregationFederatedYX2() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivatePrivateAggregationFederatedYX2() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"Y\", new PrivacyConstraint(PrivacyLevel.Private));\nprivacyConstraints.put(\"X2\", new PrivacyConstraint(PrivacyLevel.PrivateAggregation));\nfederatedL2SVMNoException(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private);\n}\n- @Test\n- public void federatedL2SVMCPPrivatePrivateAggregationFederatedX2X1() throws JSONException {\n+ @Test public void federatedL2SVMCPPrivatePrivateAggregationFederatedX2X1() {\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"X2\", new PrivacyConstraint(PrivacyLevel.Private));\nprivacyConstraints.put(\"X1\", new PrivacyConstraint(PrivacyLevel.PrivateAggregation));\n- federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private,\n- false, null, true, DMLRuntimeException.class);\n+ federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private, false, null, true,\n+ DMLRuntimeException.class);\n}\n// Require Federated Workers to return matrix\n- @Test\n- public void federatedL2SVMCPPrivateAggregationX1Exception() throws JSONException {\n- rows = 1000; cols = 1;\n+ @Test public void federatedL2SVMCPPrivateAggregationX1Exception() {\n+ rows = 1000;\n+ cols = 1;\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"X1\", new PrivacyConstraint(PrivacyLevel.PrivateAggregation));\n- federatedL2SVMNoException(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.PrivateAggregation);\n+ federatedL2SVMNoException(Types.ExecMode.SINGLE_NODE, privacyConstraints, null,\n+ PrivacyLevel.PrivateAggregation);\n}\n- @Test\n- public void federatedL2SVMCPPrivateAggregationX2Exception() throws JSONException {\n- rows = 1000; cols = 1;\n+ @Test public void federatedL2SVMCPPrivateAggregationX2Exception() {\n+ rows = 1000;\n+ cols = 1;\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"X2\", new PrivacyConstraint(PrivacyLevel.PrivateAggregation));\n- federatedL2SVMNoException(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.PrivateAggregation);\n+ federatedL2SVMNoException(Types.ExecMode.SINGLE_NODE, privacyConstraints, null,\n+ PrivacyLevel.PrivateAggregation);\n}\n- @Test\n- public void federatedL2SVMCPPrivateX1Exception() throws JSONException {\n- rows = 1000; cols = 1;\n+ @Test public void federatedL2SVMCPPrivateX1Exception() {\n+ rows = 1000;\n+ cols = 1;\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"X1\", new PrivacyConstraint(PrivacyLevel.Private));\n- federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private,\n- false, null, true, DMLRuntimeException.class);\n+ federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private, false, null, true,\n+ DMLRuntimeException.class);\n}\n- @Test\n- public void federatedL2SVMCPPrivateX2Exception() throws JSONException {\n- rows = 1000; cols = 1;\n+ @Test public void federatedL2SVMCPPrivateX2Exception() {\n+ rows = 1000;\n+ cols = 1;\nMap<String, PrivacyConstraint> privacyConstraints = new HashMap<>();\nprivacyConstraints.put(\"X2\", new PrivacyConstraint(PrivacyLevel.Private));\n- federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private,\n- false, null, true, DMLRuntimeException.class);\n+ federatedL2SVM(Types.ExecMode.SINGLE_NODE, privacyConstraints, null, PrivacyLevel.Private, false, null, true,\n+ DMLRuntimeException.class);\n}\n- private void federatedL2SVMNoException(Types.ExecMode execMode, Map<String,\n- PrivacyConstraint> privacyConstraintsFederated, Map<String, PrivacyConstraint> privacyConstraintsMatrix,\n- PrivacyLevel expectedPrivacyLevel)\n- throws JSONException\n- {\n- federatedL2SVM(execMode, privacyConstraintsFederated, privacyConstraintsMatrix, expectedPrivacyLevel,\n- false, null, false, null);\n+ private void federatedL2SVMNoException(Types.ExecMode execMode,\n+ Map<String, PrivacyConstraint> privacyConstraintsFederated,\n+ Map<String, PrivacyConstraint> privacyConstraintsMatrix, PrivacyLevel expectedPrivacyLevel) {\n+ federatedL2SVM(execMode, privacyConstraintsFederated, privacyConstraintsMatrix, expectedPrivacyLevel, false,\n+ null, false, null);\n}\nprivate void federatedL2SVM(Types.ExecMode execMode, Map<String, PrivacyConstraint> privacyConstraintsFederated,\n- Map<String, PrivacyConstraint> privacyConstraintsMatrix, PrivacyLevel expectedPrivacyLevel,\n- boolean exception1, Class<?> expectedException1, boolean exception2, Class<?> expectedException2 )\n- throws JSONException\n- {\n+ Map<String, PrivacyConstraint> privacyConstraintsMatrix, PrivacyLevel expectedPrivacyLevel, boolean exception1,\n+ Class<?> expectedException1, boolean exception2, Class<?> expectedException2) {\nboolean sparkConfigOld = DMLScript.USE_LOCAL_SPARK_CONFIG;\nTypes.ExecMode platformOld = rtplatform;\nrtplatform = execMode;\n@@ -352,23 +343,39 @@ public class FederatedL2SVMTest extends AutomatedTestBase {\n// Write privacy constraints of normal matrix\nif(privacyConstraintsMatrix != null) {\n- writeInputMatrixWithMTD(\"MX1\", X1, false, new MatrixCharacteristics(halfRows, cols, blocksize, halfRows * cols), privacyConstraintsMatrix.get(\"X1\"));\n- writeInputMatrixWithMTD(\"MX2\", X2, false, new MatrixCharacteristics(halfRows, cols, blocksize, halfRows * cols), privacyConstraintsMatrix.get(\"X2\"));\n- writeInputMatrixWithMTD(\"MY\", Y, false, new MatrixCharacteristics(rows, 1, blocksize, rows), privacyConstraintsMatrix.get(\"Y\"));\n- } else {\n- writeInputMatrixWithMTD(\"MX1\", X1, false, new MatrixCharacteristics(halfRows, cols, blocksize, halfRows * cols));\n- writeInputMatrixWithMTD(\"MX2\", X2, false, new MatrixCharacteristics(halfRows, cols, blocksize, halfRows * cols));\n+ writeInputMatrixWithMTD(\"MX1\", X1, false,\n+ new MatrixCharacteristics(halfRows, cols, blocksize, halfRows * cols),\n+ privacyConstraintsMatrix.get(\"X1\"));\n+ writeInputMatrixWithMTD(\"MX2\", X2, false,\n+ new MatrixCharacteristics(halfRows, cols, blocksize, halfRows * cols),\n+ privacyConstraintsMatrix.get(\"X2\"));\n+ writeInputMatrixWithMTD(\"MY\", Y, false, new MatrixCharacteristics(rows, 1, blocksize, rows),\n+ privacyConstraintsMatrix.get(\"Y\"));\n+ }\n+ else {\n+ writeInputMatrixWithMTD(\"MX1\", X1, false,\n+ new MatrixCharacteristics(halfRows, cols, blocksize, halfRows * cols));\n+ writeInputMatrixWithMTD(\"MX2\", X2, false,\n+ new MatrixCharacteristics(halfRows, cols, blocksize, halfRows * cols));\nwriteInputMatrixWithMTD(\"MY\", Y, false, new MatrixCharacteristics(rows, 1, blocksize, rows));\n}\n// Write privacy constraints of federated matrix\nif(privacyConstraintsFederated != null) {\n- writeInputMatrixWithMTD(\"X1\", X1, false, new MatrixCharacteristics(halfRows, cols, blocksize, halfRows * cols), privacyConstraintsFederated.get(\"X1\"));\n- writeInputMatrixWithMTD(\"X2\", X2, false, new MatrixCharacteristics(halfRows, cols, blocksize, halfRows * cols), privacyConstraintsFederated.get(\"X2\"));\n- writeInputMatrixWithMTD(\"Y\", Y, false, new MatrixCharacteristics(rows, 1, blocksize, rows), privacyConstraintsFederated.get(\"Y\"));\n- } else {\n- writeInputMatrixWithMTD(\"X1\", X1, false, new MatrixCharacteristics(halfRows, cols, blocksize, halfRows * cols));\n- writeInputMatrixWithMTD(\"X2\", X2, false, new MatrixCharacteristics(halfRows, cols, blocksize, halfRows * cols));\n+ writeInputMatrixWithMTD(\"X1\", X1, false,\n+ new MatrixCharacteristics(halfRows, cols, blocksize, halfRows * cols),\n+ privacyConstraintsFederated.get(\"X1\"));\n+ writeInputMatrixWithMTD(\"X2\", X2, false,\n+ new MatrixCharacteristics(halfRows, cols, blocksize, halfRows * cols),\n+ privacyConstraintsFederated.get(\"X2\"));\n+ writeInputMatrixWithMTD(\"Y\", Y, false, new MatrixCharacteristics(rows, 1, blocksize, rows),\n+ privacyConstraintsFederated.get(\"Y\"));\n+ }\n+ else {\n+ writeInputMatrixWithMTD(\"X1\", X1, false,\n+ new MatrixCharacteristics(halfRows, cols, blocksize, halfRows * cols));\n+ writeInputMatrixWithMTD(\"X2\", X2, false,\n+ new MatrixCharacteristics(halfRows, cols, blocksize, halfRows * cols));\nwriteInputMatrixWithMTD(\"Y\", Y, false, new MatrixCharacteristics(rows, 1, blocksize, rows));\n}\n@@ -388,9 +395,10 @@ public class FederatedL2SVMTest extends AutomatedTestBase {\nrunTest(true, exception1, expectedException1, -1);\n// Run actual dml script with federated matrix\n+ OptimizerUtils.FEDERATED_COMPILATION = fedOutCompilation;\nfullDMLScriptName = HOME + TEST_NAME + \".dml\";\n- programArgs = new String[] {\"-checkPrivacy\",\n- \"-nvargs\", \"in_X1=\" + TestUtils.federatedAddress(port1, input(\"X1\")),\n+ programArgs = new String[] {\"-stats\", \"-checkPrivacy\", \"-nvargs\",\n+ \"in_X1=\" + TestUtils.federatedAddress(port1, input(\"X1\")),\n\"in_X2=\" + TestUtils.federatedAddress(port2, input(\"X2\")), \"rows=\" + rows, \"cols=\" + cols,\n\"in_Y=\" + input(\"Y\"), \"single=FALSE\", \"out=\" + output(\"Z\")};\n@@ -407,6 +415,7 @@ public class FederatedL2SVMTest extends AutomatedTestBase {\nTestUtils.shutdownThreads(t1, t2);\nrtplatform = platformOld;\nDMLScript.USE_LOCAL_SPARK_CONFIG = sparkConfigOld;\n+ OptimizerUtils.FEDERATED_COMPILATION = false;\n}\n}\n}\n" } ]

[ { "change_type": "ADD", "old_path": null, "new_path": "scripts/perftest/fed/runL2SVMFed.sh", "diff": "+#!/bin/bash\n+#-------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+#-------------------------------------------------------------\n+\n+# Read Parameters\n+FILENAME=$0\n+CMD=${1:-\"systemds\"}\n+DATADIR=${2:-\"temp\"}/L2SVM\n+NUMFED=${3:-2}\n+MAXITR=${4:-100}\n+\n+# Error Prints\n+err_report() {\n+ echo \"Error in $FILENAME on line $1\"\n+}\n+trap 'err_report $LINENO' ERR\n+\n+# Set Properties\n+export SYSDS_QUIET=1\n+BASEPATH=$(dirname \"$0\")\n+\n+# Generate Data\n+${BASEPATH}/../genL2SVMData.sh systemds $DATADIR;\n+\n+# Start the Federated Workers on Localhost\n+${BASEPATH}/utils/startFedWorkers.sh systemds $DATADIR $NUMFED \"localhost\";\n+\n+for d in \"10k_1k_dense\" \"10k_1k_sparse\"\n+do\n+ # Split the generated data into partitions and create a federated object\n+ ${CMD} -f ${BASEPATH}/data/splitAndMakeFederated.dml \\\n+ --config ${BASEPATH}/../conf/SystemDS-config.xml \\\n+ --nvargs data=${DATADIR}/X${d} nSplit=$NUMFED transposed=FALSE \\\n+ target=${DATADIR}/X${d}_fed.json hosts=${DATADIR}/workers/hosts fmt=\"csv\"\n+\n+ ${CMD} -f ${BASEPATH}/data/splitAndMakeFederated.dml \\\n+ --config ${BASEPATH}/../conf/SystemDS-config.xml \\\n+ --nvargs data=${DATADIR}/Y${d} nSplit=$NUMFED transposed=FALSE \\\n+ target=${DATADIR}/Y${d}_fed.json hosts=${DATADIR}/workers/hosts fmt=\"csv\"\n+\n+\n+\n+ for fedCompile in \"\" \"--federatedCompilation\"\n+ do\n+ runningMessage=\"-- Running L2SVM \"$fedCompile\" with federated data (\"$d\") on \"$NUMFED\" federated workers\";\n+ echo \"$runningMessage\" >> results/times.txt\n+ echo \"$runningMessage\" >> results/compiletimes.txt\n+ # Run the L2SVM algorithm on the federated object\n+ # $1 X, $2 Y, $3 unknown, $4 BASE, $5 maxiter, $6 CMD, $7 RunPrediction, $8 FEDERATEDCOMPILATION\n+ ${BASEPATH}/../runL2SVM.sh ${DATADIR}/X${d}_fed.json ${DATADIR}/Y${d}_fed.json 2 $DATADIR ${MAXITR} systemds false $fedCompile | egrep -w 'compilation|L2SVM' | tee -a results/compiletimes.txt;\n+ done\n+done\n+\n+# Kill the Federated Workers\n+${BASEPATH}/utils/killFedWorkers.sh $DATADIR;\n" }, { "change_type": "ADD", "old_path": null, "new_path": "scripts/perftest/genL2SVMData.sh", "diff": "+#!/bin/bash\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+\n+CMD=$1\n+DATADIR=$2\n+\n+FORMAT=\"binary\" # can be csv, mm, text, binary\n+DENSE_SP=0.9\n+SPARSE_SP=0.01\n+\n+BASEPATH=$(dirname $0)\n+\n+#generate XS scenarios (80MB)\n+${CMD} -f ${BASEPATH}/../datagen/genRandData4LogisticRegression.dml --args 10000 1000 5 5 ${DATADIR}/w10k_1k_dense ${DATADIR}/X10k_1k_dense ${DATADIR}/Y10k_1k_dense 1 0 $DENSE_SP $FORMAT 1\n+${CMD} -f ${BASEPATH}/../datagen/genRandData4LogisticRegression.dml --args 10000 1000 5 5 ${DATADIR}/w10k_1k_sparse ${DATADIR}/X10k_1k_sparse ${DATADIR}/Y10k_1k_sparse 1 0 $SPARSE_SP $FORMAT 1\n" }, { "change_type": "MODIFY", "old_path": "scripts/perftest/runL2SVM.sh", "new_path": "scripts/perftest/runL2SVM.sh", "diff": "@@ -23,6 +23,8 @@ set -e\nCMD=$6\nBASE=$4\n+RUNPrediction=${7:-true}\n+FEDERATEDCOMPILATION=${8:-\"\"}\n#for all intercept values\nfor i in 0 1; do\n@@ -30,7 +32,8 @@ for i in 0 1; do\ntstart=$(date +%s.%N)\n# /algorithms/l2-svm.dml already calls a built-in function for the l2 svm.\n- ${CMD} -f ./algorithms/l2-svm.dml \\\n+ ${CMD} -f ./../algorithms/l2-svm.dml \\\n+ \"$FEDERATEDCOMPILATION\" \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n--nvargs X=$1 Y=$2 icpt=$i tol=0.0001 reg=0.01 maxiter=$5 model=${BASE}/b fmt=\"csv\"\n@@ -38,9 +41,10 @@ for i in 0 1; do\nttrain=$(echo \"$(date +%s.%N) - $tstart - .4\" | bc)\necho \"L2SVM train ict=\"$i\" on \"$1\": \"$ttrain >> results/times.txt\n+ if [ $RUNPrediction = true ]\n+ then\n#predict\ntstart=$(date +%s.%N)\n- #${CMD} -f ./algorithms/l2-svm-predict.dml \\\n${CMD} -f scripts/l2-svm-predict.dml \\\n--config conf/SystemDS-config.xml \\\n--stats \\\n@@ -48,4 +52,5 @@ for i in 0 1; do\ntpredict=$(echo \"$(date +%s.%N) - $tstart - .4\" | bc)\necho \"L2SVM predict ict=\"$i\" on \"$1\": \"$tpredict >> results/times.txt\n+ fi\ndone\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/api/DMLOptions.java", "new_path": "src/main/java/org/apache/sysds/api/DMLOptions.java", "diff": "@@ -72,6 +72,7 @@ public class DMLOptions {\npublic boolean fedWorker = false;\npublic int fedWorkerPort = -1;\npublic boolean checkPrivacy = false; // Check which privacy constraints are loaded and checked during federated execution\n+ public boolean federatedCompilation = false; // Compile federated instructions based on input federation state and privacy constraints.\npublic final static DMLOptions defaultOptions = new DMLOptions(null);\n@@ -100,6 +101,7 @@ public class DMLOptions {\n\", help=\" + help +\n\", lineage=\" + lineage +\n\", w=\" + fedWorker +\n+ \", federatedCompilation=\" + federatedCompilation +\n'}';\n}\n@@ -259,6 +261,10 @@ public class DMLOptions {\n}\ndmlOptions.checkPrivacy = line.hasOption(\"checkPrivacy\");\n+ if (line.hasOption(\"federatedCompilation\")){\n+ OptimizerUtils.FEDERATED_COMPILATION = true;\n+ dmlOptions.federatedCompilation = true;\n+ }\nreturn dmlOptions;\n}\n@@ -313,6 +319,9 @@ public class DMLOptions {\nOption checkPrivacy = OptionBuilder\n.withDescription(\"Check which privacy constraints are loaded and checked during federated execution\")\n.create(\"checkPrivacy\");\n+ Option federatedCompilation = OptionBuilder\n+ .withDescription(\"Compile federated instructions based on input federation state and privacy constraints.\")\n+ .create(\"federatedCompilation\");\noptions.addOption(configOpt);\noptions.addOption(cleanOpt);\n@@ -327,6 +336,7 @@ public class DMLOptions {\noptions.addOption(lineageOpt);\noptions.addOption(fedOpt);\noptions.addOption(checkPrivacy);\n+ options.addOption(federatedCompilation);\n// Either a clean(-clean), a file(-f), a script(-s) or help(-help) needs to be specified\nOptionGroup fileOrScriptOpt = new OptionGroup()\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/api/DMLScript.java", "new_path": "src/main/java/org/apache/sysds/api/DMLScript.java", "diff": "@@ -595,7 +595,7 @@ public class DMLScript\nfinal String ANSI_RESET = \"\\u001B[0m\";\nStringBuilder sb = new StringBuilder();\nsb.append(ANSI_RED + \"\\n\");\n- sb.append(\"An Error Occured : \");\n+ sb.append(\"An Error Occurred : \");\nsb.append(\"\\n\" );\nsb.append(StringUtils.leftPad(e.getClass().getSimpleName(),25));\nsb.append(\" -- \");\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/usertest/UserInterfaceTest.java", "new_path": "src/test/java/org/apache/sysds/test/usertest/UserInterfaceTest.java", "diff": "@@ -38,7 +38,7 @@ public class UserInterfaceTest extends Base {\npublic void testStop(){\nPair<String,String> res = runThread(\"Stop.dml\");\nassertEquals(\"\",res.getRight());\n- assertTrue(res.getLeft().contains(\"An Error Occured :\"));\n+ assertTrue(res.getLeft().contains(\"An Error Occurred :\"));\nassertTrue(res.getLeft().contains(\"DMLScriptException -- Stop Message!\"));\n}\n@@ -46,7 +46,7 @@ public class UserInterfaceTest extends Base {\npublic void SyntaxError(){\nPair<String,String> res = runThread(\"SyntaxError.dml\");\nassertEquals(\"\",res.getRight());\n- assertTrue(res.getLeft().contains(\"An Error Occured :\"));\n+ assertTrue(res.getLeft().contains(\"An Error Occurred :\"));\nassertTrue(res.getLeft().contains(\"[Syntax error]\"));\nassertTrue(res.getLeft().contains(\"ParseException --\"));\n}\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/parser/ParameterizedBuiltinFunctionExpression.java", "new_path": "src/main/java/org/apache/sysds/parser/ParameterizedBuiltinFunctionExpression.java", "diff": "@@ -584,7 +584,8 @@ public class ParameterizedBuiltinFunctionExpression extends DataIdentifier\n+ Arrays.toString(invalid.toArray(new String[0])), false);\n//check existence and correctness of arguments\n- checkTargetParam(getVarParam(\"target\"), conditional);\n+ Expression target = getVarParam(\"target\");\n+ checkEmptyTargetParam(target, conditional);\nExpression margin = getVarParam(\"margin\");\nif( margin==null ){\n@@ -608,7 +609,10 @@ public class ParameterizedBuiltinFunctionExpression extends DataIdentifier\n_varParams.put(\"empty.return\", new BooleanIdentifier(true));\n// Output is a matrix with unknown dims\n- output.setDataType(DataType.MATRIX);\n+ output.setDataType(target.getOutput().getDataType());\n+ if(target.getOutput().getDataType() == DataType.FRAME)\n+ output.setValueType(ValueType.STRING);\n+ else\noutput.setValueType(ValueType.FP64);\noutput.setDimensions(-1, -1);\n}\n@@ -727,6 +731,12 @@ public class ParameterizedBuiltinFunctionExpression extends DataIdentifier\n+\"'. Please specify the input matrix.\", conditional, LanguageErrorCodes.INVALID_PARAMETERS);\n}\n+ private void checkEmptyTargetParam(Expression target, boolean conditional) {\n+ if( target==null )\n+ raiseValidateError(\"Named parameter 'target' missing. Please specify the input matrix.\",\n+ conditional, LanguageErrorCodes.INVALID_PARAMETERS);\n+ }\n+\nprivate void checkOptionalBooleanParam(Expression param, String name, boolean conditional) {\nif( param!=null && (!param.getOutput().getDataType().isScalar() || param.getOutput().getValueType() != ValueType.BOOLEAN) ){\nraiseValidateError(\"Boolean parameter '\"+name+\"' is of type \"+param.getOutput().getDataType()\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/instructions/cp/ParameterizedBuiltinCPInstruction.java", "new_path": "src/main/java/org/apache/sysds/runtime/instructions/cp/ParameterizedBuiltinCPInstruction.java", "diff": "@@ -208,15 +208,24 @@ public class ParameterizedBuiltinCPInstruction extends ComputationCPInstruction\nString margin = params.get(\"margin\");\nif(!(margin.equals(\"rows\") || margin.equals(\"cols\")))\nthrow new DMLRuntimeException(\"Unspupported margin identifier '\" + margin + \"'.\");\n+ if(ec.isFrameObject(params.get(\"target\"))) {\n+ FrameBlock target = ec.getFrameInput(params.get(\"target\"));\n+ MatrixBlock select = params.containsKey(\"select\") ? ec.getMatrixInput(params.get(\"select\")) : null;\n+ boolean emptyReturn = Boolean.parseBoolean(params.get(\"empty.return\").toLowerCase());\n+ FrameBlock soresBlock = target.removeEmptyOperations(margin.equals(\"rows\"), emptyReturn, select);\n+ ec.setFrameOutput(output.getName(), soresBlock);\n+ ec.releaseFrameInput(params.get(\"target\"));\n+ if(params.containsKey(\"select\"))\n+ ec.releaseMatrixInput(params.get(\"select\"));\n+ } else {\n// acquire locks\nMatrixBlock target = ec.getMatrixInput(params.get(\"target\"));\nMatrixBlock select = params.containsKey(\"select\") ? ec.getMatrixInput(params.get(\"select\")) : null;\n// compute the result\nboolean emptyReturn = Boolean.parseBoolean(params.get(\"empty.return\").toLowerCase());\n- MatrixBlock soresBlock = target\n- .removeEmptyOperations(new MatrixBlock(), margin.equals(\"rows\"), emptyReturn, select);\n+ MatrixBlock soresBlock = target.removeEmptyOperations(new MatrixBlock(), margin.equals(\"rows\"), emptyReturn, select);\n// release locks\nec.setMatrixOutput(output.getName(), soresBlock);\n@@ -224,6 +233,7 @@ public class ParameterizedBuiltinCPInstruction extends ComputationCPInstruction\nif(params.containsKey(\"select\"))\nec.releaseMatrixInput(params.get(\"select\"));\n}\n+ }\nelse if(opcode.equalsIgnoreCase(\"replace\")) {\nif(ec.isFrameObject(params.get(\"target\"))){\nFrameBlock target = ec.getFrameInput(params.get(\"target\"));\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/instructions/fed/ParameterizedBuiltinFEDInstruction.java", "new_path": "src/main/java/org/apache/sysds/runtime/instructions/fed/ParameterizedBuiltinFEDInstruction.java", "diff": "@@ -28,10 +28,12 @@ import java.util.LinkedHashMap;\nimport java.util.List;\nimport java.util.Map;\nimport java.util.concurrent.Future;\n+import java.util.stream.IntStream;\nimport java.util.stream.Stream;\nimport java.util.zip.Adler32;\nimport java.util.zip.Checksum;\n+import org.apache.commons.lang.ArrayUtils;\nimport org.apache.commons.lang3.SerializationUtils;\nimport org.apache.commons.lang3.tuple.Pair;\nimport org.apache.sysds.common.Types;\n@@ -73,6 +75,7 @@ import org.apache.sysds.runtime.transform.decode.DecoderFactory;\nimport org.apache.sysds.runtime.transform.encode.EncoderFactory;\nimport org.apache.sysds.runtime.transform.encode.EncoderOmit;\nimport org.apache.sysds.runtime.transform.encode.MultiColumnEncoder;\n+import org.apache.sysds.runtime.util.UtilFunctions;\npublic class ParameterizedBuiltinFEDInstruction extends ComputationFEDInstruction {\nprotected final LinkedHashMap<String, String> params;\n@@ -151,7 +154,10 @@ public class ParameterizedBuiltinFEDInstruction extends ComputationFEDInstructio\nout.setFedMapping(mo.getFedMapping().copyWithNewID(fr1.getID()));\n}\nelse if(opcode.equals(\"rmempty\"))\n- rmempty(ec);\n+ if (getTarget(ec) instanceof FrameObject)\n+ rmemptyFrame(ec);\n+ else\n+ rmemptyMatrix(ec);\nelse if(opcode.equals(\"lowertri\") || opcode.equals(\"uppertri\"))\ntriangle(ec, opcode);\nelse if(opcode.equalsIgnoreCase(\"transformdecode\"))\n@@ -329,7 +335,170 @@ public class ParameterizedBuiltinFEDInstruction extends ComputationFEDInstructio\n}\n}\n- private void rmempty(ExecutionContext ec) {\n+ private void rmemptyFrame(ExecutionContext ec) {\n+ String margin = params.get(\"margin\");\n+ if(!(margin.equals(\"rows\") || margin.equals(\"cols\")))\n+ throw new DMLRuntimeException(\"Unsupported margin identifier '\" + margin + \"'.\");\n+\n+ FrameObject mo = (FrameObject) getTarget(ec);\n+ MatrixObject select = params.containsKey(\"select\") ? ec.getMatrixObject(params.get(\"select\")) : null;\n+ FrameObject out = ec.getFrameObject(output);\n+\n+ boolean marginRow = params.get(\"margin\").equals(\"rows\");\n+ boolean isNotAligned = ((marginRow && mo.getFedMapping().getType().isColPartitioned()) ||\n+ (!marginRow && mo.getFedMapping().getType().isRowPartitioned()));\n+\n+ MatrixBlock s = new MatrixBlock();\n+ if(select == null && isNotAligned) {\n+ List<MatrixBlock> colSums = new ArrayList<>();\n+ mo.getFedMapping().forEachParallel((range, data) -> {\n+ try {\n+ FederatedResponse response = data\n+ .executeFederatedOperation(new FederatedRequest(FederatedRequest.RequestType.EXEC_UDF, -1,\n+ new GetFrameVector(data.getVarID(), margin.equals(\"rows\"))))\n+ .get();\n+\n+ if(!response.isSuccessful())\n+ response.throwExceptionFromResponse();\n+ MatrixBlock vector = (MatrixBlock) response.getData()[0];\n+ synchronized(colSums) {\n+ colSums.add(vector);\n+ }\n+ }\n+ catch(Exception e) {\n+ throw new DMLRuntimeException(e);\n+ }\n+ return null;\n+ });\n+ // find empty in matrix\n+ BinaryOperator plus = InstructionUtils.parseBinaryOperator(\"+\");\n+ BinaryOperator greater = InstructionUtils.parseBinaryOperator(\">\");\n+ s = colSums.get(0);\n+ for(int i = 1; i < colSums.size(); i++)\n+ s = s.binaryOperationsInPlace(plus, colSums.get(i));\n+ s = s.binaryOperationsInPlace(greater, new MatrixBlock(s.getNumRows(), s.getNumColumns(), 0.0));\n+ select = ExecutionContext.createMatrixObject(s);\n+\n+ long varID = FederationUtils.getNextFedDataID();\n+ ec.setVariable(String.valueOf(varID), select);\n+ params.put(\"select\", String.valueOf(varID));\n+ // construct new string\n+ String[] oldString = InstructionUtils.getInstructionParts(instString);\n+ String[] newString = new String[oldString.length + 1];\n+ newString[2] = \"select=\" + varID;\n+ System.arraycopy(oldString, 0, newString, 0, 2);\n+ System.arraycopy(oldString, 2, newString, 3, newString.length - 3);\n+ instString = instString.replace(InstructionUtils.concatOperands(oldString),\n+ InstructionUtils.concatOperands(newString));\n+ }\n+\n+ if(select == null) {\n+ FederatedRequest fr1 = FederationUtils.callInstruction(instString,\n+ output,\n+ new CPOperand[] {getTargetOperand()},\n+ new long[] {mo.getFedMapping().getID()});\n+ mo.getFedMapping().execute(getTID(), true, fr1);\n+ out.setFedMapping(mo.getFedMapping().copyWithNewID(fr1.getID()));\n+ }\n+ else if(!isNotAligned) {\n+ // construct commands: broadcast , fed rmempty, clean broadcast\n+ FederatedRequest[] fr1 = mo.getFedMapping().broadcastSliced(select, !marginRow);\n+ FederatedRequest fr2 = FederationUtils.callInstruction(instString,\n+ output,\n+ new CPOperand[] {getTargetOperand(),\n+ new CPOperand(params.get(\"select\"), ValueType.FP64, DataType.MATRIX)},\n+ new long[] {mo.getFedMapping().getID(), fr1[0].getID()});\n+\n+ // execute federated operations and set output\n+ mo.getFedMapping().execute(getTID(), true, fr1, fr2);\n+ out.setFedMapping(mo.getFedMapping().copyWithNewID(fr2.getID()));\n+ }\n+ else {\n+ // construct commands: broadcast , fed rmempty, clean broadcast\n+ FederatedRequest fr1 = mo.getFedMapping().broadcast(select);\n+ FederatedRequest fr2 = FederationUtils.callInstruction(instString,\n+ output,\n+ new CPOperand[] {getTargetOperand(),\n+ new CPOperand(params.get(\"select\"), ValueType.FP64, DataType.MATRIX)},\n+ new long[] {mo.getFedMapping().getID(), fr1.getID()});\n+\n+ // execute federated operations and set output\n+ mo.getFedMapping().execute(getTID(), true, fr1, fr2);\n+ out.setFedMapping(mo.getFedMapping().copyWithNewID(fr2.getID()));\n+ }\n+\n+ // new ranges\n+ Map<FederatedRange, int[]> dcs = new HashMap<>();\n+ Map<FederatedRange, int[]> finalDcs1 = dcs;\n+ Map<FederatedRange, ValueType[]> finalSchema = new HashMap<>();\n+ out.getFedMapping().forEachParallel((range, data) -> {\n+ try {\n+ FederatedResponse response = data\n+ .executeFederatedOperation(new FederatedRequest(FederatedRequest.RequestType.EXEC_UDF, -1,\n+ new GetFrameCharacteristics(data.getVarID())))\n+ .get();\n+\n+ if(!response.isSuccessful())\n+ response.throwExceptionFromResponse();\n+ Object[] ret = response.getData();\n+ int[] subRangeCharacteristics = new int[]{(int) ret[0], (int) ret[1]};\n+ ValueType[] schema = (ValueType[]) ret[2];\n+ synchronized(finalDcs1) {\n+ finalDcs1.put(range, subRangeCharacteristics);\n+ }\n+ synchronized(finalSchema) {\n+ finalSchema.put(range, schema);\n+ }\n+ }\n+ catch(Exception e) {\n+ throw new DMLRuntimeException(e);\n+ }\n+ return null;\n+ });\n+\n+ dcs = finalDcs1;\n+ out.getDataCharacteristics().set(mo.getDataCharacteristics());\n+ int len = marginRow ? mo.getSchema().length : (int) (mo.isFederated(FederationMap.FType.ROW) ? s\n+ .getNonZeros() : finalSchema.values().stream().mapToInt(e -> e.length).sum());\n+ ValueType[] schema = new ValueType[len];\n+ int pos = 0;\n+ for(int i = 0; i < mo.getFedMapping().getFederatedRanges().length; i++) {\n+ FederatedRange federatedRange = new FederatedRange(out.getFedMapping().getFederatedRanges()[i]);\n+\n+ if(marginRow) {\n+ schema = mo.getSchema();\n+ } else if(mo.isFederated(FederationMap.FType.ROW)) {\n+ schema = finalSchema.get(federatedRange);\n+ } else {\n+ ValueType[] tmp = finalSchema.get(federatedRange);\n+ System.arraycopy(tmp, 0, schema, pos, tmp.length);\n+ pos += tmp.length;\n+ }\n+\n+ int[] newRange = dcs.get(federatedRange);\n+ out.getFedMapping().getFederatedRanges()[i].setBeginDim(0,\n+ (out.getFedMapping().getFederatedRanges()[i].getBeginDims()[0] == 0 ||\n+ i == 0) ? 0 : out.getFedMapping().getFederatedRanges()[i - 1].getEndDims()[0]);\n+\n+ out.getFedMapping().getFederatedRanges()[i].setEndDim(0,\n+ out.getFedMapping().getFederatedRanges()[i].getBeginDims()[0] + newRange[0]);\n+\n+ out.getFedMapping().getFederatedRanges()[i].setBeginDim(1,\n+ (out.getFedMapping().getFederatedRanges()[i].getBeginDims()[1] == 0 ||\n+ i == 0) ? 0 : out.getFedMapping().getFederatedRanges()[i - 1].getEndDims()[1]);\n+\n+ out.getFedMapping().getFederatedRanges()[i].setEndDim(1,\n+ out.getFedMapping().getFederatedRanges()[i].getBeginDims()[1] + newRange[1]);\n+ }\n+\n+ out.setSchema(schema);\n+ out.getDataCharacteristics().set(out.getFedMapping().getMaxIndexInRange(0),\n+ out.getFedMapping().getMaxIndexInRange(1),\n+ (int) mo.getBlocksize());\n+ }\n+\n+\n+ private void rmemptyMatrix(ExecutionContext ec) {\nString margin = params.get(\"margin\");\nif(!(margin.equals(\"rows\") || margin.equals(\"cols\")))\nthrow new DMLRuntimeException(\"Unsupported margin identifier '\" + margin + \"'.\");\n@@ -428,7 +597,7 @@ public class ParameterizedBuiltinFEDInstruction extends ComputationFEDInstructio\ntry {\nFederatedResponse response = data\n.executeFederatedOperation(new FederatedRequest(FederatedRequest.RequestType.EXEC_UDF, -1,\n- new GetDataCharacteristics(data.getVarID())))\n+ new GetMatrixCharacteristics(data.getVarID())))\n.get();\nif(!response.isSuccessful())\n@@ -724,11 +893,11 @@ public class ParameterizedBuiltinFEDInstruction extends ComputationFEDInstructio\n}\n}\n- private static class GetDataCharacteristics extends FederatedUDF {\n+ private static class GetMatrixCharacteristics extends FederatedUDF {\nprivate static final long serialVersionUID = 578461386177730925L;\n- public GetDataCharacteristics(long varID) {\n+ public GetMatrixCharacteristics(long varID) {\nsuper(new long[] {varID});\n}\n@@ -746,6 +915,28 @@ public class ParameterizedBuiltinFEDInstruction extends ComputationFEDInstructio\n}\n}\n+ private static class GetFrameCharacteristics extends FederatedUDF {\n+\n+ private static final long serialVersionUID = 578461386177730925L;\n+\n+ public GetFrameCharacteristics(long varID) {\n+ super(new long[] {varID});\n+ }\n+\n+ @Override\n+ public FederatedResponse execute(ExecutionContext ec, Data... data) {\n+ FrameBlock fb = ((FrameObject) data[0]).acquireReadAndRelease();\n+ int r = fb.getNumRows() != 0 || fb.getNumRows() != -1 ? fb.getNumRows() : 0;\n+ int c = fb.getNumColumns() != 0 || fb.getNumColumns() != -1 ? fb.getNumColumns() : 0;\n+ return new FederatedResponse(ResponseType.SUCCESS, new Object[] {r, c, fb.getSchema()});\n+ }\n+\n+ @Override\n+ public Pair<String, LineageItem> getLineageItem(ExecutionContext ec) {\n+ return null;\n+ }\n+ }\n+\nprivate static class GetVector extends FederatedUDF {\nprivate static final long serialVersionUID = -1003061862215703768L;\n@@ -779,4 +970,54 @@ public class ParameterizedBuiltinFEDInstruction extends ComputationFEDInstructio\nreturn null;\n}\n}\n+\n+ private static class GetFrameVector extends FederatedUDF {\n+\n+ private static final long serialVersionUID = -1003061862215703768L;\n+ private final boolean _marginRow;\n+\n+ public GetFrameVector(long varID, boolean marginRow) {\n+ super(new long[] {varID});\n+ _marginRow = marginRow;\n+ }\n+\n+ @Override\n+ public FederatedResponse execute(ExecutionContext ec, Data... data) {\n+ FrameBlock fb = ((FrameObject) data[0]).acquireReadAndRelease();\n+\n+ MatrixBlock ret = _marginRow ? new MatrixBlock(fb.getNumRows(), 1, 0.0) : new MatrixBlock(1,fb.getNumColumns(), 0.0);\n+\n+ if(_marginRow) {\n+ for(int i = 0; i < fb.getNumRows(); i++) {\n+ boolean isEmpty = true;\n+\n+ for(int j = 0; j < fb.getNumColumns(); j++) {\n+ ValueType type = fb.getSchema()[j];\n+ isEmpty = isEmpty && (ArrayUtils.contains(new double[]{0.0, Double.NaN}, UtilFunctions.objectToDoubleSafe(type, fb.get(i, j))));\n+\n+ }\n+\n+ if(!isEmpty)\n+ ret.setValue(i, 0, 1.0);\n+ }\n+ } else {\n+ for(int i = 0; i < fb.getNumColumns(); i++) {\n+ int finalI = i;\n+ ValueType type = fb.getSchema()[i];\n+ boolean isEmpty = IntStream.range(0, fb.getNumRows()).mapToObj(j -> fb.get(j, finalI))\n+ .allMatch(e -> ArrayUtils.contains(new double[]{0.0, Double.NaN}, UtilFunctions.objectToDoubleSafe(type, e)));\n+\n+ if(!isEmpty)\n+ ret.setValue(0, i,1.0);\n+ }\n+ }\n+\n+ return new FederatedResponse(ResponseType.SUCCESS, ret);\n+ }\n+\n+ @Override\n+ public Pair<String, LineageItem> getLineageItem(ExecutionContext ec) {\n+ return null;\n+ }\n+ }\n}\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/matrix/data/FrameBlock.java", "new_path": "src/main/java/org/apache/sysds/runtime/matrix/data/FrameBlock.java", "diff": "@@ -37,6 +37,8 @@ import java.util.concurrent.ExecutorService;\nimport java.util.concurrent.Future;\nimport java.util.concurrent.ThreadLocalRandom;\nimport java.util.function.Function;\n+import java.util.function.IntFunction;\n+import java.util.stream.IntStream;\nimport org.apache.commons.lang.ArrayUtils;\nimport org.apache.commons.lang.NotImplementedException;\n@@ -46,6 +48,7 @@ import org.apache.commons.logging.Log;\nimport org.apache.commons.logging.LogFactory;\nimport org.apache.hadoop.io.Writable;\nimport org.apache.sysds.api.DMLException;\n+import org.apache.sysds.common.Types;\nimport org.apache.sysds.common.Types.ValueType;\nimport org.apache.sysds.runtime.DMLRuntimeException;\nimport org.apache.sysds.runtime.codegen.CodegenUtils;\n@@ -598,6 +601,31 @@ public class FrameBlock implements CacheBlock, Externalizable {\n_msize = -1;\n}\n+ public void appendColumn(ValueType vt, Array col) {\n+ switch (vt) {\n+ case STRING:\n+ appendColumn(((StringArray) col).get());\n+ break;\n+ case BOOLEAN:\n+ appendColumn(((BooleanArray) col).get());\n+ break;\n+ case INT32:\n+ appendColumn(((IntegerArray) col).get());\n+ break;\n+ case INT64:\n+ appendColumn(((LongArray) col).get());\n+ break;\n+ case FP32:\n+ appendColumn(((FloatArray) col).get());\n+ break;\n+ case FP64:\n+ appendColumn(((DoubleArray) col).get());\n+ break;\n+ default:\n+ throw new RuntimeException(\"Unsupported value type: \" + vt);\n+ }\n+ }\n+\npublic Object getColumnData(int c) {\nswitch(_schema[c]) {\ncase STRING: return ((StringArray)_coldata[c])._data;\n@@ -1640,10 +1668,13 @@ public class FrameBlock implements CacheBlock, Externalizable {\n_data = data;\n_size = _data.length;\n}\n+ public String[] get() { return _data; }\n+\n@Override\npublic String get(int index) {\nreturn _data[index];\n}\n+\n@Override\npublic void set(int index, String value) {\n_data[index] = value;\n@@ -1705,10 +1736,13 @@ public class FrameBlock implements CacheBlock, Externalizable {\n_data = data;\n_size = _data.length;\n}\n+ public boolean[] get() { return _data; }\n+\n@Override\npublic Boolean get(int index) {\nreturn _data[index];\n}\n+\n@Override\npublic void set(int index, Boolean value) {\n_data[index] = (value!=null) ? value : false;\n@@ -1772,6 +1806,7 @@ public class FrameBlock implements CacheBlock, Externalizable {\n_data = data;\n_size = _data.length;\n}\n+ public long[] get() { return _data; }\n@Override\npublic Long get(int index) {\nreturn _data[index];\n@@ -1839,6 +1874,7 @@ public class FrameBlock implements CacheBlock, Externalizable {\n_data = data;\n_size = _data.length;\n}\n+ public int[] get() { return _data; }\n@Override\npublic Integer get(int index) {\n@@ -1906,6 +1942,8 @@ public class FrameBlock implements CacheBlock, Externalizable {\n_data = data;\n_size = _data.length;\n}\n+ public float[] get() { return _data; }\n+\n@Override\npublic Float get(int index) {\nreturn _data[index];\n@@ -1972,6 +2010,7 @@ public class FrameBlock implements CacheBlock, Externalizable {\n_data = data;\n_size = _data.length;\n}\n+ public double[] get() { return _data; }\n@Override\npublic Double get(int index) {\nreturn _data[index];\n@@ -2473,6 +2512,77 @@ public class FrameBlock implements CacheBlock, Externalizable {\nreturn ret;\n}\n+ public FrameBlock removeEmptyOperations(boolean rows, boolean emptyReturn, MatrixBlock select) {\n+ if( rows )\n+ return removeEmptyRows(select, emptyReturn);\n+ else //cols\n+ return removeEmptyColumns(select, emptyReturn);\n+ }\n+\n+ private FrameBlock removeEmptyRows(MatrixBlock select, boolean emptyReturn) {\n+ FrameBlock ret = new FrameBlock(_schema, _colnames);\n+\n+ for(int i = 0; i < _numRows; i++) {\n+ boolean isEmpty = true;\n+ Object[] row = new Object[getNumColumns()];\n+\n+ for(int j = 0; j < getNumColumns(); j++) {\n+ Array colData = _coldata[j].clone();\n+ row[j] = colData.get(i);\n+ ValueType type = _schema[j];\n+ isEmpty = isEmpty && (ArrayUtils.contains(new double[]{0.0, Double.NaN}, UtilFunctions.objectToDoubleSafe(type, colData.get(i))));\n+ }\n+\n+ if((!isEmpty && select == null) || (select != null && select.getValue(i, 0) == 1)) {\n+ ret.appendRow(row);\n+ }\n+ }\n+\n+ if(ret.getNumRows() == 0 && emptyReturn) {\n+ String[][] arr = new String[1][getNumColumns()];\n+ Arrays.fill(arr, new String[]{null});\n+ ValueType[] schema = new ValueType[getNumColumns()];\n+ Arrays.fill(schema, ValueType.STRING);\n+ return new FrameBlock(schema, arr);\n+ }\n+\n+ return ret;\n+ }\n+\n+ private FrameBlock removeEmptyColumns(MatrixBlock select, boolean emptyReturn) {\n+ FrameBlock ret = new FrameBlock();\n+ List<ColumnMetadata> columnMetadata = new ArrayList<>();\n+\n+ for(int i = 0; i < getNumColumns(); i++) {\n+ Array colData = _coldata[i];\n+\n+ boolean isEmpty = false;\n+ if(select == null) {\n+ ValueType type = _schema[i];\n+ isEmpty = IntStream.range(0, colData._size).mapToObj((IntFunction<Object>) colData::get)\n+ .allMatch(e -> ArrayUtils.contains(new double[]{0.0, Double.NaN}, UtilFunctions.objectToDoubleSafe(type, e)));\n+ }\n+\n+ if((select != null && select.getValue(0, i) == 1) || (!isEmpty && select == null)) {\n+ Types.ValueType vt = _schema[i];\n+ ret.appendColumn(vt, _coldata[i].clone());\n+ columnMetadata.add(new ColumnMetadata(_colmeta[i]));\n+ }\n+ }\n+\n+ if(ret.getNumColumns() == 0 && emptyReturn) {\n+ String[][] arr = new String[_numRows][];\n+ Arrays.fill(arr, new String[]{null});\n+ return new FrameBlock(new ValueType[]{ValueType.STRING}, arr);\n+ }\n+\n+ ret._colmeta = new ColumnMetadata[columnMetadata.size()];\n+ columnMetadata.toArray(ret._colmeta);\n+ ret.setColumnMetadata(ret._colmeta);\n+\n+ return ret;\n+ }\n+\n@Override\npublic String toString(){\nStringBuilder sb = new StringBuilder();\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/util/UtilFunctions.java", "new_path": "src/main/java/org/apache/sysds/runtime/util/UtilFunctions.java", "diff": "package org.apache.sysds.runtime.util;\nimport org.apache.commons.lang.ArrayUtils;\n+import org.apache.commons.lang3.math.NumberUtils;\nimport org.apache.commons.math3.random.RandomDataGenerator;\nimport org.apache.sysds.common.Types.ValueType;\nimport org.apache.sysds.runtime.DMLRuntimeException;\n@@ -487,6 +488,12 @@ public class UtilFunctions {\n}\n}\n+ public static double objectToDoubleSafe(ValueType vt, Object in) {\n+ if(vt == ValueType.STRING && !NumberUtils.isCreatable((String) in)) {\n+ return 1.0;\n+ } else return objectToDouble(vt, in);\n+ }\n+\npublic static double objectToDouble(ValueType vt, Object in) {\nif( in == null ) return Double.NaN;\nswitch( vt ) {\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/test/java/org/apache/sysds/test/component/frame/FrameRemoveEmptyTest.java", "diff": "+/*\n+ * Licensed to the Apache Software Foundation (ASF) under one\n+ * or more contributor license agreements. See the NOTICE file\n+ * distributed with this work for additional information\n+ * regarding copyright ownership. The ASF licenses this file\n+ * to you under the Apache License, Version 2.0 (the\n+ * \"License\"); you may not use this file except in compliance\n+ * with the License. You may obtain a copy of the License at\n+ *\n+ * http://www.apache.org/licenses/LICENSE-2.0\n+ *\n+ * Unless required by applicable law or agreed to in writing,\n+ * software distributed under the License is distributed on an\n+ * \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+ * KIND, either express or implied. See the License for the\n+ * specific language governing permissions and limitations\n+ * under the License.\n+ */\n+\n+package org.apache.sysds.test.component.frame;\n+\n+import org.apache.sysds.api.DMLScript;\n+import org.apache.sysds.common.Types;\n+import org.apache.sysds.runtime.matrix.data.MatrixBlock;\n+import org.apache.sysds.runtime.meta.MatrixCharacteristics;\n+import org.apache.sysds.test.AutomatedTestBase;\n+import org.apache.sysds.test.TestConfiguration;\n+import org.apache.sysds.test.TestUtils;\n+import org.apache.sysds.test.functions.unary.matrix.RemoveEmptyTest;\n+import org.junit.Ignore;\n+import org.junit.Test;\n+\n+public class FrameRemoveEmptyTest extends AutomatedTestBase {\n+ private final static String TEST_NAME1 = \"removeEmpty1\";\n+ private final static String TEST_DIR = \"functions/frame/\";\n+ private static final String TEST_CLASS_DIR = TEST_DIR + RemoveEmptyTest.class.getSimpleName() + \"/\";\n+\n+ private final static int _rows = 10;\n+ private final static int _cols = 6;\n+\n+ private final static double _sparsityDense = 0.7;\n+\n+ @Override\n+ public void setUp() {\n+ addTestConfiguration(TEST_NAME1, new TestConfiguration(TEST_CLASS_DIR, TEST_NAME1, new String[] {\"V\"}));\n+ }\n+\n+ @Test\n+ public void testRemoveEmptyRowsDenseCP() {\n+ runTestRemoveEmpty(TEST_NAME1, \"rows\", Types.ExecType.CP, false);\n+ }\n+\n+ @Test\n+ public void testRemoveEmptyRowsSparseCP() {\n+ runTestRemoveEmpty(TEST_NAME1, \"cols\", Types.ExecType.CP, true);\n+ }\n+\n+ @Test\n+ @Ignore\n+ public void testRemoveEmptyRowsDenseSP() {\n+ runTestRemoveEmpty(TEST_NAME1, \"rows\", Types.ExecType.SPARK, false);\n+ }\n+\n+ @Test\n+ @Ignore\n+ public void testRemoveEmptyRowsSparseSP() {\n+ runTestRemoveEmpty(TEST_NAME1, \"rows\", Types.ExecType.SPARK, true);\n+ }\n+\n+ private void runTestRemoveEmpty(String testname, String margin, Types.ExecType et, boolean bSelectIndex) {\n+ // rtplatform for MR\n+ Types.ExecMode platformOld = rtplatform;\n+ switch(et) {\n+ case SPARK:\n+ rtplatform = Types.ExecMode.SPARK;\n+ break;\n+ default:\n+ rtplatform = Types.ExecMode.HYBRID;\n+ break;\n+ }\n+\n+ boolean sparkConfigOld = DMLScript.USE_LOCAL_SPARK_CONFIG;\n+ if(rtplatform == Types.ExecMode.SPARK)\n+ DMLScript.USE_LOCAL_SPARK_CONFIG = true;\n+\n+ try {\n+ // register test configuration\n+ TestConfiguration config = getTestConfiguration(testname);\n+ config.addVariable(\"rows\", _rows);\n+ config.addVariable(\"cols\", _cols);\n+ loadTestConfiguration(config);\n+\n+ /* This is for running the junit test the new way, i.e., construct the arguments directly */\n+ String HOME = SCRIPT_DIR + TEST_DIR;\n+ fullDMLScriptName = HOME + testname + \".dml\";\n+ programArgs = new String[] {\"-explain\", \"-args\", input(\"V\"), margin, output(\"V\")};\n+\n+ MatrixBlock in = createInputMatrix(margin, _rows, _cols, _sparsityDense, bSelectIndex);\n+\n+ runTest(true, false, null, -1);\n+ double[][] outArray = TestUtils.convertHashMapToDoubleArray(readDMLMatrixFromOutputDir(\"V\"));\n+ MatrixBlock out = new MatrixBlock(outArray.length, outArray[0].length, false);\n+ out.init(outArray, outArray.length, outArray[0].length);\n+\n+ MatrixBlock in2 = new MatrixBlock(_rows, _cols + 2, 0.0);\n+ in2.copy(0, _rows - 1, 0, _cols - 1, in, true);\n+ in2.copy(0, (_rows / 2) - 1, _cols, _cols + 1, new MatrixBlock(_rows / 2, 2, 1.0), true);\n+ MatrixBlock expected = in2.removeEmptyOperations(new MatrixBlock(), margin.equals(\"rows\"), false, null);\n+ expected = expected.slice(0, expected.getNumRows() - 1, 0, expected.getNumColumns() - 3);\n+\n+ TestUtils.compareMatrices(expected, out, 0);\n+ }\n+ finally {\n+ // reset platform for additional tests\n+ rtplatform = platformOld;\n+ DMLScript.USE_LOCAL_SPARK_CONFIG = sparkConfigOld;\n+ }\n+ }\n+\n+ private MatrixBlock createInputMatrix(String margin, int rows, int cols, double sparsity, boolean bSelectIndex) {\n+ int rowsp = -1, colsp = -1;\n+ if(margin.equals(\"rows\")) {\n+ rowsp = rows / 2;\n+ colsp = cols;\n+ }\n+ else {\n+ rowsp = rows;\n+ colsp = cols / 2;\n+ }\n+\n+ // long seed = System.nanoTime();\n+ double[][] V = getRandomMatrix(rows, cols, 0, 1, sparsity, 7);\n+ double[][] Vp = new double[rowsp][colsp];\n+ double[][] Ix = null;\n+ int innz = 0, vnnz = 0;\n+\n+ // clear out every other row/column\n+ if(margin.equals(\"rows\")) {\n+ Ix = new double[rows][1];\n+ for(int i = 0; i < rows; i++) {\n+ boolean clear = i % 2 != 0;\n+ if(clear) {\n+ for(int j = 0; j < cols; j++)\n+ V[i][j] = 0;\n+ Ix[i][0] = 0;\n+ }\n+ else {\n+ boolean bNonEmpty = false;\n+ for(int j = 0; j < cols; j++) {\n+ Vp[i / 2][j] = V[i][j];\n+ bNonEmpty |= (V[i][j] != 0.0) ? true : false;\n+ vnnz += (V[i][j] == 0.0) ? 0 : 1;\n+ }\n+ Ix[i][0] = (bNonEmpty) ? 1 : 0;\n+ innz += Ix[i][0];\n+ }\n+ }\n+ }\n+ else {\n+ Ix = new double[1][cols];\n+ for(int j = 0; j < cols; j++) {\n+ boolean clear = j % 2 != 0;\n+ if(clear) {\n+ for(int i = 0; i < rows; i++)\n+ V[i][j] = 0;\n+ Ix[0][j] = 0;\n+ }\n+ else {\n+ boolean bNonEmpty = false;\n+ for(int i = 0; i < rows; i++) {\n+ Vp[i][j / 2] = V[i][j];\n+ bNonEmpty |= (V[i][j] != 0.0) ? true : false;\n+ vnnz += (V[i][j] == 0.0) ? 0 : 1;\n+ }\n+ Ix[0][j] = (bNonEmpty) ? 1 : 0;\n+ innz += Ix[0][j];\n+ }\n+ }\n+ }\n+\n+ MatrixCharacteristics imc = new MatrixCharacteristics(margin.equals(\"rows\") ? rows : 1,\n+ margin.equals(\"rows\") ? 1 : cols, 1000, innz);\n+ MatrixCharacteristics vmc = new MatrixCharacteristics(rows, cols, 1000, vnnz);\n+\n+ MatrixBlock in = new MatrixBlock(rows, cols, false);\n+ in.init(V, _rows, _cols);\n+\n+ writeInputMatrixWithMTD(\"V\", V, false, vmc); // always text\n+ writeExpectedMatrix(\"V\", Vp);\n+ if(bSelectIndex)\n+ writeInputMatrixWithMTD(\"I\", Ix, false, imc);\n+\n+ return in;\n+ }\n+}\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/test/scripts/functions/frame/removeEmpty1.dml", "diff": "+#-------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+#-------------------------------------------------------------\n+\n+\n+A = read($1, naStrings= [\"NA\", \"null\",\" \",\"NaN\", \"nan\", \"\", \"?\", \"99999\"])\n+B = frame(data=[\"TRUE\", \"abc\"], rows=nrow(A) / 2, cols=2, schema=[\"BOOLEAN\", \"STRING\"])\n+C = frame(data=[\"FALSE\", \"0.0\"], rows=nrow(A) / 2, cols=2, schema=[\"BOOLEAN\", \"STRING\"])\n+D = rbind(B, C)\n+V = cbind(as.frame(A), D)\n+Vp = removeEmpty(target=V, margin=$2)\n+X = as.matrix(Vp[, 1:(ncol(Vp)-2)])\n+write(X, $3);\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/python/tests/federated/runFedTest.sh", "new_path": "src/main/python/tests/federated/runFedTest.sh", "diff": "@@ -45,7 +45,7 @@ systemds WORKER 8002 >$w2_Output 2>&1 &\nFed2=$!\nsystemds WORKER 8003 >$w3_Output 2>&1 &\nFed3=$!\n-echo \"Starting workers\" && sleep 3 && echo \"Starting tests\"\n+echo \"Starting workers\" && sleep 6 && echo \"Starting tests\"\n# Run test\npython -m unittest discover -s tests/federated -p 'test_*.py' $1 >$log 2>&1\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/nodes/frame.py", "new_path": "src/main/python/systemds/operator/nodes/frame.py", "diff": "@@ -28,12 +28,12 @@ from typing import (TYPE_CHECKING, Dict, Iterable, Optional, Sequence, Tuple,\nimport numpy as np\nimport pandas as pd\nfrom py4j.java_gateway import JavaObject, JVMView\n-from systemds.operator import Matrix, MultiReturn, OperationNode\n+from systemds.operator import Matrix, MultiReturn, OperationNode, Scalar\nfrom systemds.script_building.dag import DAGNode, OutputType\nfrom systemds.utils.consts import VALID_INPUT_TYPES\nfrom systemds.utils.converters import (frame_block_to_pandas,\npandas_to_frame_block)\n-from systemds.utils.helpers import get_slice_string\n+from systemds.utils.helpers import check_is_empty_slice, check_no_less_than_zero, get_slice_string\nif TYPE_CHECKING:\n# to avoid cyclic dependencies during runtime\n@@ -73,7 +73,7 @@ class Frame(OperationNode):\ncode_line = code_line.format(file_name=var_name)\nreturn code_line\n- def compute(self, verbose: bool = False, lineage: bool = False) -> Union[pd.DataFrame]:\n+ def compute(self, verbose: bool = False, lineage: bool = False) -> pd.DataFrame:\nif self._is_pandas():\nif verbose:\nprint(\"[Pandas Frame - No Compilation necessary]\")\n@@ -139,6 +139,33 @@ class Frame(OperationNode):\ndef __str__(self):\nreturn \"FrameNode\"\n+ def nRow(self) -> 'Scalar':\n+ return Scalar(self.sds_context, 'nrow', [self])\n+\n+ def nCol(self) -> 'Scalar':\n+ return Scalar(self.sds_context, 'ncol', [self])\n+\ndef __getitem__(self, i) -> 'Frame':\n+ if isinstance(i, tuple) and len(i) > 2:\n+ raise ValueError(\"Maximum of two dimensions are allowed\")\n+ elif isinstance(i, list):\n+ check_no_less_than_zero(i)\n+ slice = self.sds_context.from_numpy(np.array(i)) + 1\n+ select = Matrix(self.sds_context, \"table\",\n+ [slice, 1, self.nRow(), 1])\n+ ret = Frame(self.sds_context, \"removeEmpty\", [], {\n+ 'target': self, 'margin': '\"rows\"', 'select': select})\n+ return ret\n+ elif isinstance(i, tuple) and isinstance(i[0], list) and isinstance(i[1], list):\n+ raise NotImplementedError(\"double slicing is not supported yet\")\n+ elif isinstance(i, tuple) and check_is_empty_slice(i[0]) and isinstance(i[1], list):\n+ check_no_less_than_zero(i[1])\n+ slice = self.sds_context.from_numpy(np.array(i[1])) + 1\n+ select = Matrix(self.sds_context, \"table\",\n+ [slice, 1, self.nCol(), 1])\n+ ret = Frame(self.sds_context, \"removeEmpty\", [], {\n+ 'target': self, 'margin': '\"cols\"', 'select': select})\n+ return ret\n+ else:\nsliceIns = get_slice_string(i)\nreturn Frame(self.sds_context, '', [self, sliceIns], brackets=True)\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/nodes/matrix.py", "new_path": "src/main/python/systemds/operator/nodes/matrix.py", "diff": "__all__ = [\"Matrix\"]\n-import os\n-from typing import (TYPE_CHECKING, Dict, Iterable, Optional, Sequence, Tuple,\n- Union)\n+from typing import TYPE_CHECKING, Dict, Iterable, Sequence, Union\nimport numpy as np\n-from py4j.java_gateway import JavaObject, JVMView\n+from py4j.java_gateway import JavaObject\nfrom systemds.operator import OperationNode, Scalar\nfrom systemds.script_building.dag import OutputType\nfrom systemds.utils.consts import (BINARY_OPERATIONS, VALID_ARITHMETIC_TYPES,\nVALID_INPUT_TYPES)\nfrom systemds.utils.converters import (matrix_block_to_numpy,\nnumpy_to_matrix_block)\n-from systemds.utils.helpers import get_slice_string\n+from systemds.utils.helpers import check_is_empty_slice, check_no_less_than_zero, get_slice_string\nclass Matrix(OperationNode):\n_np_array: np.array\n- def __init__(self, sds_context: 'SystemDSContext', operation: str,\n+ def __init__(self, sds_context, operation: str,\nunnamed_input_nodes: Union[str,\nIterable[VALID_INPUT_TYPES]] = None,\nnamed_input_nodes: Dict[str, VALID_INPUT_TYPES] = None,\n@@ -68,7 +66,7 @@ class Matrix(OperationNode):\ncode_line = code_line.format(file_name=var_name)\nreturn code_line\n- def compute(self, verbose: bool = False, lineage: bool = False) -> Union[np.array]:\n+ def compute(self, verbose: bool = False, lineage: bool = False) -> np.array:\nif self._is_numpy():\nif verbose:\nprint('[Numpy Array - No Compilation necessary]')\n@@ -154,7 +152,34 @@ class Matrix(OperationNode):\ndef __matmul__(self, other: 'Matrix') -> 'Matrix':\nreturn Matrix(self.sds_context, '%*%', [self, other])\n+ def nRow(self) -> 'Scalar':\n+ return Scalar(self.sds_context, 'nrow', [self])\n+\n+ def nCol(self) -> 'Scalar':\n+ return Scalar(self.sds_context, 'ncol', [self])\n+\ndef __getitem__(self, i):\n+ if isinstance(i, tuple) and len(i) > 2:\n+ raise ValueError(\"Maximum of two dimensions are allowed\")\n+ elif isinstance(i, list):\n+ check_no_less_than_zero(i)\n+ slice = self.sds_context.from_numpy(np.array(i)) + 1\n+ select = Matrix(self.sds_context, \"table\",\n+ [slice, 1, self.nRow(), 1])\n+ ret = Matrix(self.sds_context, \"removeEmpty\", [], {\n+ 'target': self, 'margin': '\"rows\"', 'select': select})\n+ return ret\n+ elif isinstance(i, tuple) and isinstance(i[0], list) and isinstance(i[1], list):\n+ raise NotImplementedError(\"double slicing is not supported yet\")\n+ elif isinstance(i, tuple) and check_is_empty_slice(i[0]) and isinstance(i[1], list):\n+ check_no_less_than_zero(i[1])\n+ slice = self.sds_context.from_numpy(np.array(i[1])) + 1\n+ select = Matrix(self.sds_context, \"table\",\n+ [slice, 1, self.nCol(), 1])\n+ ret = Matrix(self.sds_context, \"removeEmpty\", [], {\n+ 'target': self, 'margin': '\"cols\"', 'select': select})\n+ return ret\n+ else:\nsliceIns = get_slice_string(i)\nreturn Matrix(self.sds_context, '', [self, sliceIns], brackets=True)\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/script_building/script.py", "new_path": "src/main/python/systemds/script_building/script.py", "diff": "@@ -81,7 +81,7 @@ class DMLScript:\nret = self.prepared_script.executeScript()\nreturn ret\nexcept Py4JNetworkError:\n- exception_str = \"Py4JNetworkError: no connection to JVM, most likely due to previous crash\"\n+ exception_str = \"Py4JNetworkError: no connection to JVM, most likely due to previous crash or closed JVM from calls to close()\"\ntrace_back_limit = 0\nexcept Exception as e:\nexception_str = str(e)\n@@ -111,7 +111,7 @@ class DMLScript:\nreturn ret, traces\nexcept Py4JNetworkError:\n- exception_str = \"Py4JNetworkError: no connection to JVM, most likely due to previous crash\"\n+ exception_str = \"Py4JNetworkError: no connection to JVM, most likely due to previous crash or closed JVM from calls to close()\"\ntrace_back_limit = 0\nexcept Exception as e:\nexception_str = str(e)\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/utils/helpers.py", "new_path": "src/main/python/systemds/utils/helpers.py", "diff": "@@ -51,11 +51,9 @@ def get_module_dir() -> os.PathLike:\ndef get_slice_string(i):\n+ if isinstance(i, list):\n+ raise ValueError(\"Not Supported list query\")\nif isinstance(i, tuple):\n- if len(i) > 2:\n- raise ValueError(\n- f'Invalid number of dimensions to slice {len(i)}, Only 2 dimensions allowed')\n- else:\nreturn f'{get_slice_string(i[0])},{get_slice_string(i[1])}'\nelif isinstance(i, slice):\nif i.step:\n@@ -71,3 +69,13 @@ def get_slice_string(i):\n# + 1 since R and systemDS is 1 indexed.\nsliceIns = i+1\nreturn sliceIns\n+\n+\n+def check_is_empty_slice(i):\n+ return isinstance(i, slice) and i.start == None and i.stop == None and i.step == None\n+\n+\n+def check_no_less_than_zero(i: list):\n+ for x in i:\n+ if(x < 0):\n+ raise ValueError(\"Negative index not supported in systemds\")\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/tests/frame/test_slice.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+\n+import unittest\n+\n+import numpy as np\n+import pandas as pd\n+from systemds.context import SystemDSContext\n+\n+df = pd.DataFrame(\n+ {\n+ \"col1\": [\"col1_hello_3\", \"col1_world_3\", \"col1_hello_3\"],\n+ \"col2\": [6, 7, 8],\n+ \"col3\": [0.6, 0.7, 0.8],\n+ }\n+)\n+\n+\n+class TestFederatedAggFn(unittest.TestCase):\n+\n+ sds: SystemDSContext = None\n+\n+ @classmethod\n+ def setUpClass(cls):\n+ cls.sds = SystemDSContext()\n+\n+ @classmethod\n+ def tearDownClass(cls):\n+ cls.sds.close()\n+\n+ def test_setup(self):\n+ sm = self.sds.from_pandas(df)\n+ sr = sm.compute()\n+ self.assertTrue(isinstance(sr, pd.DataFrame))\n+ e = df\n+ self.assertTrue((e.values == sr.values).all())\n+\n+ def test_slice_first_third_row(self):\n+ sm = self.sds.from_pandas(df)[[0, 2]]\n+ sr = sm.compute()\n+ e = df.loc[[0, 2]]\n+ self.assertTrue((e.values == sr.values).all())\n+\n+ def test_slice_single_row(self):\n+ sm = self.sds.from_pandas(df)[[1]]\n+ sr = sm.compute()\n+ e = df.loc[[1]]\n+ self.assertTrue((e.values == sr.values).all())\n+\n+ def test_slice_last_row(self):\n+ with self.assertRaises(ValueError):\n+ self.sds.from_pandas(df)[[-1]]\n+\n+ # https://issues.apache.org/jira/browse/SYSTEMDS-3203\n+ # def test_slice_first_third_col(self):\n+ # sm = self.sds.from_pandas(df)[:, [0, 2]]\n+ # sr = sm.compute()\n+ # e = df.loc[:, [0, 2]]\n+ # self.assertTrue((e.values == sr.values).all())\n+\n+ # def test_slice_single_col(self):\n+ # sm = self.sds.from_pandas(df)[:, [1]]\n+ # sr = sm.compute()\n+ # e = df.loc[:, [1]]\n+ # self.assertTrue((e.values == sr.values).all())\n+\n+ def test_slice_row_col_both(self):\n+ with self.assertRaises(NotImplementedError):\n+ self.sds.from_pandas(df)[[1, 2], [0, 2]]\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/tests/matrix/test_slice.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+\n+import unittest\n+\n+import numpy as np\n+from systemds.context import SystemDSContext\n+\n+# Seed the randomness.\n+np.random.seed(7)\n+\n+m = np.random.rand(3, 4)\n+\n+\n+class TestFederatedAggFn(unittest.TestCase):\n+\n+ sds: SystemDSContext = None\n+\n+ @classmethod\n+ def setUpClass(cls):\n+ cls.sds = SystemDSContext()\n+\n+ @classmethod\n+ def tearDownClass(cls):\n+ cls.sds.close()\n+\n+ def test_setup(self):\n+ sm = self.sds.from_numpy(m)\n+ sr = sm.compute()\n+ e = m\n+ self.assertTrue(np.allclose(e, sr))\n+\n+ def test_slice_first_third_row(self):\n+ sm = self.sds.from_numpy(m)[[0, 2]]\n+ sr = sm.compute()\n+ e = m[[0, 2]]\n+ self.assertTrue(np.allclose(e, sr))\n+\n+ def test_slice_single_row(self):\n+ sm = self.sds.from_numpy(m)[[1]]\n+ sr = sm.compute()\n+ e = m[[1]]\n+ self.assertTrue(np.allclose(e, sr))\n+\n+ def test_slice_last_row(self):\n+ with self.assertRaises(ValueError):\n+ self.sds.from_numpy(m)[[-1]]\n+\n+ def test_slice_first_third_col(self):\n+ sm = self.sds.from_numpy(m)[:, [0, 2]]\n+ sr = sm.compute()\n+ e = m[:, [0, 2]]\n+ self.assertTrue(np.allclose(e, sr))\n+\n+ def test_slice_single_col(self):\n+ sm = self.sds.from_numpy(m)[:, [1]]\n+ sr = sm.compute()\n+ e = m[:, [1]]\n+ self.assertTrue(np.allclose(e, sr))\n+\n+ def test_slice_row_col_both(self):\n+ with self.assertRaises(NotImplementedError):\n+ self.sds.from_numpy(m)[[1, 2], [0, 3]]\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/conf/ConfigurationManager.java", "new_path": "src/main/java/org/apache/sysds/conf/ConfigurationManager.java", "diff": "@@ -175,6 +175,10 @@ public class ConfigurationManager\nreturn getDMLConfig().getBooleanValue(DMLConfig.PARALLEL_ENCODE);\n}\n+ public static boolean isStagedParallelTransform() {\n+ return getDMLConfig().getBooleanValue(DMLConfig.PARALLEL_ENCODE_STAGED);\n+ }\n+\npublic static int getParallelApplyBlocks(){\nreturn getDMLConfig().getIntValue(DMLConfig.PARALLEL_ENCODE_APPLY_BLOCKS);\n}\n@@ -183,6 +187,10 @@ public class ConfigurationManager\nreturn getDMLConfig().getIntValue(DMLConfig.PARALLEL_ENCODE_BUILD_BLOCKS);\n}\n+ public static int getNumThreads() {\n+ return getDMLConfig().getIntValue(DMLConfig.PARALLEL_ENCODE_NUM_THREADS);\n+ }\n+\npublic static boolean isParallelParFor() {\nreturn getCompilerConfigFlag(ConfigType.PARALLEL_LOCAL_OR_REMOTE_PARFOR);\n}\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/conf/DMLConfig.java", "new_path": "src/main/java/org/apache/sysds/conf/DMLConfig.java", "diff": "@@ -68,8 +68,10 @@ public class DMLConfig\npublic static final String CP_PARALLEL_OPS = \"sysds.cp.parallel.ops\";\npublic static final String CP_PARALLEL_IO = \"sysds.cp.parallel.io\";\npublic static final String PARALLEL_ENCODE = \"sysds.parallel.encode\"; // boolean: enable multi-threaded transformencode and apply\n+ public static final String PARALLEL_ENCODE_STAGED = \"sysds.parallel.encode.staged\";\npublic static final String PARALLEL_ENCODE_APPLY_BLOCKS = \"sysds.parallel.encode.applyBlocks\";\npublic static final String PARALLEL_ENCODE_BUILD_BLOCKS = \"sysds.parallel.encode.buildBlocks\";\n+ public static final String PARALLEL_ENCODE_NUM_THREADS = \"sysds.parallel.encode.numThreads\";\npublic static final String COMPRESSED_LINALG = \"sysds.compressed.linalg\";\npublic static final String COMPRESSED_LOSSY = \"sysds.compressed.lossy\";\npublic static final String COMPRESSED_VALID_COMPRESSIONS = \"sysds.compressed.valid.compressions\";\n@@ -130,8 +132,10 @@ public class DMLConfig\n_defaultVals.put(CP_PARALLEL_OPS, \"true\" );\n_defaultVals.put(CP_PARALLEL_IO, \"true\" );\n_defaultVals.put(PARALLEL_ENCODE, \"false\" );\n+ _defaultVals.put(PARALLEL_ENCODE_STAGED, \"false\" );\n_defaultVals.put(PARALLEL_ENCODE_APPLY_BLOCKS, \"1\");\n_defaultVals.put(PARALLEL_ENCODE_BUILD_BLOCKS, \"1\");\n+ _defaultVals.put(PARALLEL_ENCODE_NUM_THREADS, \"-1\");\n_defaultVals.put(COMPRESSED_LINALG, Compression.CompressConfig.FALSE.name() );\n_defaultVals.put(COMPRESSED_LOSSY, \"false\" );\n_defaultVals.put(COMPRESSED_VALID_COMPRESSIONS, \"SDC,DDC\");\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/hops/OptimizerUtils.java", "new_path": "src/main/java/org/apache/sysds/hops/OptimizerUtils.java", "diff": "@@ -1017,10 +1017,11 @@ public class OptimizerUtils\nreturn ret;\n}\n- public static int getTransformNumThreads(int maxNumThreads)\n+ public static int getTransformNumThreads()\n{\n//by default max local parallelism (vcores)\nint ret = InfrastructureAnalyzer.getLocalParallelism();\n+ int maxNumThreads = ConfigurationManager.getNumThreads();\n//apply external max constraint (e.g., set by parfor or other rewrites)\nif( maxNumThreads > 0 ) {\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/instructions/cp/MultiReturnParameterizedBuiltinCPInstruction.java", "new_path": "src/main/java/org/apache/sysds/runtime/instructions/cp/MultiReturnParameterizedBuiltinCPInstruction.java", "diff": "@@ -87,7 +87,7 @@ public class MultiReturnParameterizedBuiltinCPInstruction extends ComputationCPI\n// execute block transform encode\nMultiColumnEncoder encoder = EncoderFactory.createEncoder(spec, colnames, fin.getNumColumns(), null);\n// TODO: Assign #threads in compiler and pass via the instruction string\n- MatrixBlock data = encoder.encode(fin, OptimizerUtils.getTransformNumThreads(-1)); // build and apply\n+ MatrixBlock data = encoder.encode(fin, OptimizerUtils.getTransformNumThreads()); // build and apply\nFrameBlock meta = encoder.getMetaData(new FrameBlock(fin.getNumColumns(), ValueType.STRING));\nmeta.setColumnNames(colnames);\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/transform/encode/MultiColumnEncoder.java", "new_path": "src/main/java/org/apache/sysds/runtime/transform/encode/MultiColumnEncoder.java", "diff": "@@ -40,6 +40,7 @@ import org.apache.commons.logging.Log;\nimport org.apache.commons.logging.LogFactory;\nimport org.apache.sysds.api.DMLScript;\nimport org.apache.sysds.common.Types;\n+import org.apache.sysds.conf.ConfigurationManager;\nimport org.apache.sysds.runtime.DMLRuntimeException;\nimport org.apache.sysds.runtime.controlprogram.caching.CacheBlock;\nimport org.apache.sysds.runtime.data.SparseBlock;\n@@ -56,9 +57,8 @@ import org.apache.sysds.utils.Statistics;\npublic class MultiColumnEncoder implements Encoder {\nprotected static final Log LOG = LogFactory.getLog(MultiColumnEncoder.class.getName());\n- private static final boolean MULTI_THREADED = true;\n// If true build and apply separately by placing a synchronization barrier\n- public static boolean MULTI_THREADED_STAGES = false;\n+ public static boolean MULTI_THREADED_STAGES = ConfigurationManager.isStagedParallelTransform();\n// Only affects if MULTI_THREADED_STAGES is true\n// if true apply tasks for each column will complete\n@@ -87,7 +87,7 @@ public class MultiColumnEncoder implements Encoder {\npublic MatrixBlock encode(CacheBlock in, int k) {\nMatrixBlock out;\ntry {\n- if(MULTI_THREADED && k > 1 && !MULTI_THREADED_STAGES && !hasLegacyEncoder()) {\n+ if(k > 1 && !MULTI_THREADED_STAGES && !hasLegacyEncoder()) {\nout = new MatrixBlock();\nDependencyThreadPool pool = new DependencyThreadPool(k);\nLOG.debug(\"Encoding with full DAG on \" + k + \" Threads\");\n@@ -187,7 +187,7 @@ public class MultiColumnEncoder implements Encoder {\npublic void build(CacheBlock in, int k) {\nif(hasLegacyEncoder() && !(in instanceof FrameBlock))\nthrow new DMLRuntimeException(\"LegacyEncoders do not support non FrameBlock Inputs\");\n- if(MULTI_THREADED && k > 1) {\n+ if(k > 1) {\nbuildMT(in, k);\n}\nelse {\n@@ -255,7 +255,7 @@ public class MultiColumnEncoder implements Encoder {\n// TODO smart checks\n// Block allocation for MT access\noutputMatrixPreProcessing(out, in);\n- if(MULTI_THREADED && k > 1) {\n+ if(k > 1) {\napplyMT(in, out, outputCol, k);\n}\nelse {\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/transform/TransformFrameEncodeMultithreadedTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/transform/TransformFrameEncodeMultithreadedTest.java", "diff": "@@ -208,7 +208,7 @@ public class TransformFrameEncodeMultithreadedTest extends AutomatedTestBase {\nFiles.readAllLines(Paths.get(SPEC)).forEach(s -> specSb.append(s).append(\"\\n\"));\nMultiColumnEncoder encoder = EncoderFactory.createEncoder(specSb.toString(), input.getColumnNames(),\ninput.getNumColumns(), null);\n- MultiColumnEncoder.MULTI_THREADED_STAGES = staged;\n+ //MultiColumnEncoder.MULTI_THREADED_STAGES = staged;\nMatrixBlock outputS = encoder.encode(input, 1);\nMatrixBlock outputM = encoder.encode(input, 12);\n" } ]

[ { "change_type": "MODIFY", "old_path": "docs/api/java/org/apache/sysds/hops/rewrite/IPAPassRewriteFederatedPlan.html", "new_path": "docs/api/java/org/apache/sysds/hops/rewrite/IPAPassRewriteFederatedPlan.html", "diff": "@@ -103,7 +103,7 @@ var activeTableTab = \"activeTableTab\";\n<li><a href=\"../../../../../org/apache/sysds/hops/ipa/IPAPass.html\" title=\"class in org.apache.sysds.hops.ipa\">org.apache.sysds.hops.ipa.IPAPass</a></li>\n<li>\n<ul class=\"inheritance\">\n-<li>org.apache.sysds.hops.rewrite.IPAPassRewriteFederatedPlan</li>\n+<li>org.apache.sysds.hops.ipa.IPAPassRewriteFederatedPlan</li>\n</ul>\n</li>\n</ul>\n" }, { "change_type": "MODIFY", "old_path": "docs/api/java/org/apache/sysds/hops/rewrite/class-use/IPAPassRewriteFederatedPlan.html", "new_path": "docs/api/java/org/apache/sysds/hops/rewrite/class-use/IPAPassRewriteFederatedPlan.html", "diff": "<head>\n<!-- Generated by javadoc -->\n<meta http-equiv=\"Content-Type\" content=\"text/html; charset=UTF-8\">\n-<title>Uses of Class org.apache.sysds.hops.rewrite.IPAPassRewriteFederatedPlan (SystemDS 2.3.0-SNAPSHOT API)</title>\n+<title>Uses of Class org.apache.sysds.hops.ipa.IPAPassRewriteFederatedPlan (SystemDS 2.3.0-SNAPSHOT API)</title>\n<link rel=\"stylesheet\" type=\"text/css\" href=\"../../../../../../stylesheet.css\" title=\"Style\">\n<script type=\"text/javascript\" src=\"../../../../../../script.js\"></script>\n</head>\n<script type=\"text/javascript\"><!--\ntry {\nif (location.href.indexOf('is-external=true') == -1) {\n- parent.document.title=\"Uses of Class org.apache.sysds.hops.rewrite.IPAPassRewriteFederatedPlan (SystemDS 2.3.0-SNAPSHOT API)\";\n+ parent.document.title=\"Uses of Class org.apache.sysds.hops.ipa.IPAPassRewriteFederatedPlan (SystemDS 2.3.0-SNAPSHOT API)\";\n}\n}\ncatch(err) {\n</a></div>\n<!-- ========= END OF TOP NAVBAR ========= -->\n<div class=\"header\">\n-<h2 title=\"Uses of Class org.apache.sysds.hops.rewrite.IPAPassRewriteFederatedPlan\" class=\"title\">Uses of Class<br>org.apache.sysds.hops.rewrite.IPAPassRewriteFederatedPlan</h2>\n+<h2 title=\"Uses of Class org.apache.sysds.hops.ipa.IPAPassRewriteFederatedPlan\" class=\"title\">Uses of Class<br>org.apache.sysds.hops.ipa.IPAPassRewriteFederatedPlan</h2>\n</div>\n-<div class=\"classUseContainer\">No usage of org.apache.sysds.hops.rewrite.IPAPassRewriteFederatedPlan</div>\n+<div class=\"classUseContainer\">No usage of org.apache.sysds.hops.ipa.IPAPassRewriteFederatedPlan</div>\n<!-- ======= START OF BOTTOM NAVBAR ====== -->\n<div class=\"bottomNav\"><a name=\"navbar.bottom\">\n<!-- -->\n" }, { "change_type": "RENAME", "old_path": "src/main/java/org/apache/sysds/hops/rewrite/IPAPassRewriteFederatedPlan.java", "new_path": "src/main/java/org/apache/sysds/hops/ipa/IPAPassRewriteFederatedPlan.java", "diff": "* under the License.\n*/\n-package org.apache.sysds.hops.rewrite;\n+package org.apache.sysds.hops.ipa;\nimport org.apache.sysds.api.DMLException;\nimport org.apache.sysds.hops.AggBinaryOp;\n@@ -30,9 +30,6 @@ import org.apache.sysds.hops.OptimizerUtils;\nimport org.apache.sysds.hops.ReorgOp;\nimport org.apache.sysds.hops.TernaryOp;\nimport org.apache.sysds.hops.cost.HopRel;\n-import org.apache.sysds.hops.ipa.FunctionCallGraph;\n-import org.apache.sysds.hops.ipa.FunctionCallSizeInfo;\n-import org.apache.sysds.hops.ipa.IPAPass;\nimport org.apache.sysds.parser.DMLProgram;\nimport org.apache.sysds.parser.ForStatement;\nimport org.apache.sysds.parser.ForStatementBlock;\n@@ -69,8 +66,7 @@ public class IPAPassRewriteFederatedPlan extends IPAPass {\n* @param fgraph function call graph\n* @return true if federated compilation is activated.\n*/\n- @Override\n- public boolean isApplicable(FunctionCallGraph fgraph) {\n+ @Override public boolean isApplicable(FunctionCallGraph fgraph) {\nreturn OptimizerUtils.FEDERATED_COMPILATION;\n}\n@@ -83,8 +79,8 @@ public class IPAPassRewriteFederatedPlan extends IPAPass {\n* @param fcallSizes function call size infos\n* @return false since the function call graph never has to be rebuilt\n*/\n- @Override\n- public boolean rewriteProgram(DMLProgram prog, FunctionCallGraph fgraph, FunctionCallSizeInfo fcallSizes) {\n+ @Override public boolean rewriteProgram(DMLProgram prog, FunctionCallGraph fgraph,\n+ FunctionCallSizeInfo fcallSizes) {\nrewriteStatementBlocks(prog, prog.getStatementBlocks());\nreturn false;\n}\n@@ -178,7 +174,8 @@ public class IPAPassRewriteFederatedPlan extends IPAPass {\nFunctionStatementBlock sbFuncBlock = prog.getBuiltinFunctionDictionary().getFunction(funcName);\nrewriteStatementBlock(prog, sbFuncBlock);\n}\n- else selectFederatedExecutionPlan(sbHop);\n+ else\n+ selectFederatedExecutionPlan(sbHop);\n}\n}\nreturn new ArrayList<>(Collections.singletonList(sb));\n@@ -188,16 +185,19 @@ public class IPAPassRewriteFederatedPlan extends IPAPass {\n* Sets FederatedOutput field of all hops in DAG starting from given root.\n* The FederatedOutput chosen for root is the minimum cost HopRel found in memo table for the given root.\n* The FederatedOutput values chosen for the inputs to the root are chosen based on the input dependencies.\n+ *\n* @param root hop for which FederatedOutput needs to be set\n*/\nprivate void setFinalFedout(Hop root) {\n- HopRel optimalRootHopRel = hopRelMemo.get(root.getHopID()).stream().min(Comparator.comparingDouble(HopRel::getCost))\n+ HopRel optimalRootHopRel = hopRelMemo.get(root.getHopID()).stream()\n+ .min(Comparator.comparingDouble(HopRel::getCost))\n.orElseThrow(() -> new DMLException(\"Hop root \" + root + \" has no feasible federated output alternatives\"));\nsetFinalFedout(root, optimalRootHopRel);\n}\n/**\n* Update the FederatedOutput value and cost based on information stored in given rootHopRel.\n+ *\n* @param root hop for which FederatedOutput is set\n* @param rootHopRel from which FederatedOutput value and cost is retrieved\n*/\n@@ -208,6 +208,7 @@ public class IPAPassRewriteFederatedPlan extends IPAPass {\n/**\n* Sets FederatedOutput value for each of the inputs of rootHopRel\n+ *\n* @param rootHopRel which has its input values updated\n*/\nprivate void visitInputDependency(HopRel rootHopRel) {\n@@ -218,6 +219,7 @@ public class IPAPassRewriteFederatedPlan extends IPAPass {\n/**\n* Updates FederatedOutput value and cost estimate based on updateHopRel values.\n+ *\n* @param root which has its values updated\n* @param updateHopRel from which the values are retrieved\n*/\n@@ -229,6 +231,7 @@ public class IPAPassRewriteFederatedPlan extends IPAPass {\n/**\n* Select federated execution plan for every Hop in the DAG starting from given roots.\n* The cost estimates of the hops are also updated when FederatedOutput is updated in the hops.\n+ *\n* @param roots starting point for going through the Hop DAG to update the FederatedOutput fields.\n*/\n@SuppressWarnings(\"unused\")\n@@ -239,6 +242,7 @@ public class IPAPassRewriteFederatedPlan extends IPAPass {\n/**\n* Select federated execution plan for every Hop in the DAG starting from given root.\n+ *\n* @param root starting point for going through the Hop DAG to update the federatedOutput fields\n*/\nprivate void selectFederatedExecutionPlan(Hop root) {\n@@ -248,6 +252,7 @@ public class IPAPassRewriteFederatedPlan extends IPAPass {\n/**\n* Go through the Hop DAG and set the FederatedOutput field and cost estimate for each Hop from leaf to given currentHop.\n+ *\n* @param currentHop the Hop from which the DAG is visited\n*/\nprivate void visitFedPlanHop(Hop currentHop) {\n@@ -273,6 +278,7 @@ public class IPAPassRewriteFederatedPlan extends IPAPass {\n/**\n* Checks if the instructions related to the given hop supports FOUT/LOUT processing.\n+ *\n* @param hop to check for federated support\n* @return true if federated instructions related to hop supports FOUT/LOUT processing\n*/\n@@ -284,6 +290,7 @@ public class IPAPassRewriteFederatedPlan extends IPAPass {\n/**\n* Checks if the associatedHop supports the given federated output value.\n+ *\n* @param associatedHop to check support of\n* @param fedOut federated output value\n* @return true if associatedHop supports fedOut\n@@ -303,6 +310,7 @@ public class IPAPassRewriteFederatedPlan extends IPAPass {\n/**\n* Checks to see if the associatedHop supports FOUT.\n+ *\n* @param associatedHop for which FOUT support is checked\n* @return true if FOUT is supported by the associatedHop\n*/\n@@ -311,9 +319,8 @@ public class IPAPassRewriteFederatedPlan extends IPAPass {\nif(associatedHop instanceof AggUnaryOp && associatedHop.isScalar())\nreturn false;\n// It can only be FOUT if at least one of the inputs are FOUT, except if it is a federated DataOp\n- if ( associatedHop.getInput().stream().noneMatch(\n- input -> hopRelMemo.get(input.getHopID()).stream().anyMatch(HopRel::hasFederatedOutput) )\n- && !associatedHop.isFederatedDataOp() )\n+ if(associatedHop.getInput().stream().noneMatch(input -> hopRelMemo.get(input.getHopID()).stream()\n+ .anyMatch(HopRel::hasFederatedOutput)) && !associatedHop.isFederatedDataOp())\nreturn false;\nreturn true;\n}\n@@ -321,6 +328,7 @@ public class IPAPassRewriteFederatedPlan extends IPAPass {\n/**\n* Checks to see if the associatedHop supports LOUT.\n* It supports LOUT if the output has no privacy constraints.\n+ *\n* @param associatedHop for which LOUT support is checked.\n* @return true if LOUT is supported by the associatedHop\n*/\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/hops/ipa/InterProceduralAnalysis.java", "new_path": "src/main/java/org/apache/sysds/hops/ipa/InterProceduralAnalysis.java", "diff": "@@ -34,7 +34,6 @@ import org.apache.sysds.hops.HopsException;\nimport org.apache.sysds.hops.LiteralOp;\nimport org.apache.sysds.hops.OptimizerUtils;\nimport org.apache.sysds.hops.recompile.Recompiler;\n-import org.apache.sysds.hops.rewrite.IPAPassRewriteFederatedPlan;\nimport org.apache.sysds.parser.DMLProgram;\nimport org.apache.sysds.parser.DMLTranslator;\nimport org.apache.sysds.parser.DataIdentifier;\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/controlprogram/ForProgramBlock.java", "new_path": "src/main/java/org/apache/sysds/runtime/controlprogram/ForProgramBlock.java", "diff": "@@ -165,12 +165,12 @@ public class ForProgramBlock extends ProgramBlock\n}\n//execute exit instructions\n- executeExitInstructions(_exitInstruction, \"for\", ec);\n+ executeExitInstructions(\"for\", ec);\n}\nprotected ScalarObject executePredicateInstructions( int pos, ArrayList<Instruction> instructions, ExecutionContext ec, boolean downCast )\n{\n- ScalarObject ret = null;\n+ ScalarObject ret;\nValueType vt = downCast ? ValueType.INT64 : null;\ntry\n@@ -221,7 +221,7 @@ public class ForProgramBlock extends ProgramBlock\n/**\n* Utility class for iterating over positive or negative predicate sequences.\n*/\n- protected class SequenceIterator implements Iterator<IntObject>, Iterable<IntObject>\n+ protected static class SequenceIterator implements Iterator<IntObject>, Iterable<IntObject>\n{\nprivate long _cur = -1;\nprivate long _to = -1;\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/controlprogram/IfProgramBlock.java", "new_path": "src/main/java/org/apache/sysds/runtime/controlprogram/IfProgramBlock.java", "diff": "@@ -128,12 +128,12 @@ public class IfProgramBlock extends ProgramBlock\n}\n//execute exit instructions\n- executeExitInstructions(_exitInstruction, \"if\", ec);\n+ executeExitInstructions(\"if\", ec);\n}\nprivate BooleanObject executePredicate(ExecutionContext ec)\n{\n- BooleanObject result = null;\n+ BooleanObject result;\ntry {\nif( _sb != null ) {\nIfStatementBlock isb = (IfStatementBlock)_sb;\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/controlprogram/ParForProgramBlock.java", "new_path": "src/main/java/org/apache/sysds/runtime/controlprogram/ParForProgramBlock.java", "diff": "@@ -292,7 +292,6 @@ public class ParForProgramBlock extends ForProgramBlock\npublic static final boolean ALLOW_NESTED_PARALLELISM = true; // if not, transparently change parfor to for on program conversions (local,remote)\npublic static final boolean USE_PARALLEL_RESULT_MERGE = false; // if result merge is run in parallel or serial\npublic static final boolean USE_PARALLEL_RESULT_MERGE_REMOTE = true; // if remote result merge should be run in parallel for multiple result vars\n- public static final boolean ALLOW_DATA_COLOCATION = true;\npublic static final boolean CREATE_UNSCOPED_RESULTVARS = true;\npublic static boolean ALLOW_REUSE_PARTITION_VARS = true; //reuse partition input matrices, applied only if read-only in surrounding loops\npublic static final int WRITE_REPLICATION_FACTOR = 1;\n@@ -322,18 +321,18 @@ public class ParForProgramBlock extends ForProgramBlock\nprotected int _numThreads = -1;\nprotected boolean _fixedDOP = false; //guard for numThreads\nprotected long _taskSize = -1;\n- protected PTaskPartitioner _taskPartitioner = null;\n- protected PDataPartitioner _dataPartitioner = null;\n- protected PResultMerge _resultMerge = null;\n- protected PExecMode _execMode = null;\n- protected POptMode _optMode = null;\n+ protected PTaskPartitioner _taskPartitioner;\n+ protected PDataPartitioner _dataPartitioner;\n+ protected PResultMerge _resultMerge;\n+ protected PExecMode _execMode;\n+ protected POptMode _optMode;\n//specifics used for optimization\nprotected long _numIterations = -1;\n//specifics used for data partitioning\n- protected LocalVariableMap _variablesDPOriginal = null;\n- protected LocalVariableMap _variablesDPReuse = null;\n+ protected LocalVariableMap _variablesDPOriginal;\n+ protected LocalVariableMap _variablesDPReuse;\nprotected String _colocatedDPMatrix = null;\nprotected boolean _tSparseCol = false;\nprotected int _replicationDP = WRITE_REPLICATION_FACTOR;\n@@ -647,15 +646,15 @@ public class ParForProgramBlock extends ForProgramBlock\nswitch( _execMode )\n{\ncase LOCAL: //create parworkers as local threads\n- executeLocalParFor(ec, iterVar, from, to, incr);\n+ executeLocalParFor(ec, from, to, incr);\nbreak;\ncase REMOTE_SPARK: // create parworkers as Spark tasks (one job per parfor)\n- executeRemoteSparkParFor(ec, iterVar, from, to, incr);\n+ executeRemoteSparkParFor(ec, from, to, incr);\nbreak;\ncase REMOTE_SPARK_DP: // create parworkers as Spark tasks (one job per parfor)\n- executeRemoteSparkParForDP(ec, iterVar, from, to, incr);\n+ executeRemoteSparkParForDP(ec, from, to, incr);\nbreak;\ndefault:\n@@ -688,7 +687,7 @@ public class ParForProgramBlock extends ForProgramBlock\n}\n//execute exit instructions\n- executeExitInstructions(_exitInstruction, \"parfor\", ec);\n+ executeExitInstructions(\"parfor\", ec);\n///////\n//end PARALLEL EXECUTION of (PAR)FOR body\n@@ -715,13 +714,12 @@ public class ParForProgramBlock extends ForProgramBlock\n* below for details of the realization.\n*\n* @param ec execution context\n- * @param itervar ?\n* @param from ?\n* @param to ?\n* @param incr ?\n* @throws InterruptedException if InterruptedException occurs\n*/\n- private void executeLocalParFor( ExecutionContext ec, IntObject itervar, IntObject from, IntObject to, IntObject incr )\n+ private void executeLocalParFor( ExecutionContext ec, IntObject from, IntObject to, IntObject incr )\nthrows InterruptedException\n{\nLOG.trace(\"Local Par For (multi-threaded) with degree of parallelism : \" + _numThreads);\n@@ -850,7 +848,7 @@ public class ParForProgramBlock extends ForProgramBlock\n}\n}\n- private void executeRemoteSparkParFor(ExecutionContext ec, IntObject itervar, IntObject from, IntObject to, IntObject incr)\n+ private void executeRemoteSparkParFor(ExecutionContext ec, IntObject from, IntObject to, IntObject incr)\n{\nTiming time = ( _monitor ? new Timing(true) : null );\n@@ -916,7 +914,7 @@ public class ParForProgramBlock extends ForProgramBlock\n}\n}\n- private void executeRemoteSparkParForDP( ExecutionContext ec, IntObject itervar, IntObject from, IntObject to, IntObject incr ) {\n+ private void executeRemoteSparkParForDP( ExecutionContext ec, IntObject from, IntObject to, IntObject incr ) {\nTiming time = ( _monitor ? new Timing(true) : null );\n// Step 0) check and compile to CP (if forced remote parfor)\n@@ -991,7 +989,7 @@ public class ParForProgramBlock extends ForProgramBlock\nData dat = ec.getVariable(var);\n//skip non-existing input matrices (which are due to unknown sizes marked for\n//partitioning but typically related branches are never executed)\n- if( dat != null && dat instanceof MatrixObject )\n+ if( dat instanceof MatrixObject )\n{\nMatrixObject moVar = (MatrixObject) dat; //unpartitioned input\n@@ -1076,7 +1074,7 @@ public class ParForProgramBlock extends ForProgramBlock\nStream<CacheableData<?>> results = Arrays.stream(in).filter(m -> m!=null && m!=out);\n//perform cleanup (parallel to mitigate file deletion bottlenecks)\n(parallel ? results.parallel() : results)\n- .forEach(m -> ec.cleanupCacheableData(m));\n+ .forEach(ec::cleanupCacheableData);\n}\n/**\n@@ -1254,7 +1252,7 @@ public class ParForProgramBlock extends ForProgramBlock\n*/\nprivate TaskPartitioner createTaskPartitioner( IntObject from, IntObject to, IntObject incr )\n{\n- TaskPartitioner tp = null;\n+ TaskPartitioner tp;\nswitch( _taskPartitioner ) {\ncase FIXED:\n@@ -1301,7 +1299,7 @@ public class ParForProgramBlock extends ForProgramBlock\n*/\nprivate DataPartitioner createDataPartitioner(PartitionFormat dpf, PDataPartitioner dataPartitioner, ExecutionContext ec)\n{\n- DataPartitioner dp = null;\n+ DataPartitioner dp;\n//determine max degree of parallelism\nint numRed = OptimizerUtils.isSparkExecutionMode() ?\n@@ -1325,7 +1323,7 @@ public class ParForProgramBlock extends ForProgramBlock\nprivate ResultMerge<?> createResultMerge( PResultMerge prm,\nCacheableData<?> out, CacheableData<?>[] in, String fname, boolean accum, ExecutionContext ec )\n{\n- ResultMerge<?> rm = null;\n+ ResultMerge<?> rm;\nif( out instanceof FrameObject ) {\nrm = new ResultMergeFrameLocalMemory((FrameObject)out, (FrameObject[])in, fname, accum);\n@@ -1352,7 +1350,7 @@ public class ParForProgramBlock extends ForProgramBlock\n(MatrixObject[])in, fname, accum, ec, numMap, numRed );\nbreak;\ndefault:\n- throw new DMLRuntimeException(\"Undefined result merge: '\" +prm.toString()+\"'.\");\n+ throw new DMLRuntimeException(\"Undefined result merge: '\" + prm +\"'.\");\n}\n}\nelse {\n@@ -1650,9 +1648,9 @@ public class ParForProgramBlock extends ForProgramBlock\n*/\nprivate class ResultMergeWorker extends Thread\n{\n- private LocalTaskQueue<ResultVar> _q = null;\n- private LocalVariableMap[] _refVars = null;\n- private ExecutionContext _ec = null;\n+ private final LocalTaskQueue<ResultVar> _q;\n+ private final LocalVariableMap[] _refVars;\n+ private final ExecutionContext _ec;\nprivate boolean _success = false;\npublic ResultMergeWorker( LocalTaskQueue<ResultVar> q, LocalVariableMap[] results, ExecutionContext ec )\n@@ -1689,7 +1687,7 @@ public class ParForProgramBlock extends ForProgramBlock\nString fname = constructResultMergeFileName();\nResultMerge<?> rm = createResultMerge(_resultMerge, out, in, fname, var._isAccum, _ec);\n- CacheableData<?> outNew = null;\n+ CacheableData<?> outNew;\nif( USE_PARALLEL_RESULT_MERGE )\noutNew = rm.executeParallelMerge( _numThreads );\nelse\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/controlprogram/Program.java", "new_path": "src/main/java/org/apache/sysds/runtime/controlprogram/Program.java", "diff": "@@ -35,8 +35,8 @@ public class Program\npublic static final String KEY_DELIM = \"::\";\nprivate DMLProgram _prog;\n- private ArrayList<ProgramBlock> _programBlocks;\n- private HashMap<String, FunctionDictionary<FunctionProgramBlock>> _namespaces;\n+ private final ArrayList<ProgramBlock> _programBlocks;\n+ private final HashMap<String, FunctionDictionary<FunctionProgramBlock>> _namespaces;\npublic Program() {\n_namespaces = new HashMap<>();\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/controlprogram/ProgramBlock.java", "new_path": "src/main/java/org/apache/sysds/runtime/controlprogram/ProgramBlock.java", "diff": "@@ -181,7 +181,7 @@ public abstract class ProgramBlock implements ParseInfo {\nreturn executePredicateInstructions(tmp, retType, ec);\n}\n- protected void executeExitInstructions(Instruction inst, String ctx, ExecutionContext ec) {\n+ protected void executeExitInstructions(String ctx, ExecutionContext ec) {\ntry {\nif(_exitInstruction != null)\nexecuteSingleInstruction(_exitInstruction, ec);\n@@ -279,7 +279,7 @@ public abstract class ProgramBlock implements ParseInfo {\n// variables in symbol table (for tracking source of wrong representation)\nif(CHECK_MATRIX_PROPERTIES) {\ncheckSparsity(tmp, ec.getVariables(), ec);\n- checkFederated(tmp, ec.getVariables());\n+ checkFederated(ec.getVariables());\n}\n}\ncatch(DMLScriptException e) {\n@@ -377,7 +377,7 @@ public abstract class ProgramBlock implements ParseInfo {\n}\n}\n- private static void checkFederated(Instruction lastInst, LocalVariableMap vars) {\n+ private static void checkFederated(LocalVariableMap vars) {\nfor(String varname : vars.keySet()) {\nData dat = vars.get(varname);\nif(!(dat instanceof CacheableData))\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/controlprogram/WhileProgramBlock.java", "new_path": "src/main/java/org/apache/sysds/runtime/controlprogram/WhileProgramBlock.java", "diff": "@@ -69,7 +69,7 @@ public class WhileProgramBlock extends ProgramBlock\nprivate BooleanObject executePredicate(ExecutionContext ec)\n{\n- BooleanObject result = null;\n+ BooleanObject result;\ntry\n{\nif( _sb!=null )\n@@ -140,7 +140,7 @@ public class WhileProgramBlock extends ProgramBlock\n}\n//execute exit instructions\n- executeExitInstructions(_exitInstruction, \"while\", ec);\n+ executeExitInstructions(\"while\", ec);\n}\npublic void setChildBlocks(ArrayList<ProgramBlock> childs) {\n" } ]

[ { "change_type": "ADD", "old_path": null, "new_path": "scripts/builtin/underSampling.dml", "diff": "+#-------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+#-------------------------------------------------------------\n+# # # following built-in performs random under sampling on data\n+\n+underSampling = function(Matrix[Double] data, Double ratio)\n+return(Matrix[Double] data)\n+{\n+ if(ratio < 0 | ratio > 0.5) {\n+ ratio = 0.1\n+ print(\"ratio should be greater than 0 and less than 0.5 setting ratio = 0.1\")\n+ }\n+ # # separate Y\n+ Y = data[, ncol(data)]\n+ # # get the minority class\n+ classes = table(Y, 1)\n+ # # # get the minority class\n+ minority = as.scalar(rowIndexMin(t(classes)))\n+ # # # separate the minority class\n+ notMin = (Y != matrix(minority, rows=nrow(Y), cols=1))\n+ dX = cbind(seq(1, nrow(data)), data)\n+ majority = removeEmpty(target=dX, margin=\"rows\", select=notMin)\n+ # # # formulate the undersampling ratio\n+ u_ratio = floor(nrow(majority) * ratio)\n+ # take the samples for oversampling\n+ u_sample = sample(nrow(majority), u_ratio)\n+ u_select = table(u_sample, 1, 1, nrow(majority), 1)\n+ u_select = u_select * majority[, 1]\n+ u_select = removeEmpty(target = u_select, margin = \"rows\")\n+ u_select1 = table(u_select, 1, 1, nrow(data), 1)\n+ data = removeEmpty(target=data, margin=\"rows\", select = (u_select1 == 0))\n+}\n\\ No newline at end of file\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/common/Builtins.java", "new_path": "src/main/java/org/apache/sysds/common/Builtins.java", "diff": "@@ -309,6 +309,7 @@ public enum Builtins {\nTRANSFORMDECODE(\"transformdecode\", false, true),\nTRANSFORMENCODE(\"transformencode\", false, true),\nTRANSFORMMETA(\"transformmeta\", false, true),\n+ UNDER_SAMPLING(\"underSampling\", true),\nUPPER_TRI(\"upper.tri\", false, true),\nXDUMMY1(\"xdummy1\", true), //error handling test\nXDUMMY2(\"xdummy2\", true); //error handling test\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/test/java/org/apache/sysds/test/functions/builtin/BuiltinUnderSamplingTest.java", "diff": "+/*\n+ * Licensed to the Apache Software Foundation (ASF) under one\n+ * or more contributor license agreements. See the NOTICE file\n+ * distributed with this work for additional information\n+ * regarding copyright ownership. The ASF licenses this file\n+ * to you under the Apache License, Version 2.0 (the\n+ * \"License\"); you may not use this file except in compliance\n+ * with the License. You may obtain a copy of the License at\n+ *\n+ * http://www.apache.org/licenses/LICENSE-2.0\n+ *\n+ * Unless required by applicable law or agreed to in writing,\n+ * software distributed under the License is distributed on an\n+ * \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+ * KIND, either express or implied. See the License for the\n+ * specific language governing permissions and limitations\n+ * under the License.\n+ */\n+\n+package org.apache.sysds.test.functions.builtin;\n+\n+import org.apache.sysds.common.Types;\n+import org.apache.sysds.test.AutomatedTestBase;\n+import org.apache.sysds.test.TestConfiguration;\n+import org.apache.sysds.test.TestUtils;\n+import org.junit.Assert;\n+import org.junit.Test;\n+\n+public class BuiltinUnderSamplingTest extends AutomatedTestBase {\n+ private final static String TEST_NAME = \"underSamplingTest\";\n+ private final static String TEST_DIR = \"functions/builtin/\";\n+ private final static String TEST_CLASS_DIR = TEST_DIR + BuiltinUnderSamplingTest.class.getSimpleName() + \"/\";\n+\n+ @Override\n+ public void setUp() {\n+ TestUtils.clearAssertionInformation();\n+ addTestConfiguration(TEST_NAME, new TestConfiguration(TEST_CLASS_DIR, TEST_NAME, new String[] {\"B\",}));\n+ }\n+\n+ @Test\n+ public void test_CP1() {\n+\n+ runUnderSamplingTest(0.3, Types.ExecType.CP);\n+\n+ }\n+\n+ @Test\n+ public void test_CP2() {\n+\n+ runUnderSamplingTest(0.5, Types.ExecType.CP);\n+\n+ }\n+\n+ @Test\n+ public void test_Spark() {\n+ runUnderSamplingTest(0.4,Types.ExecType.SPARK);\n+ }\n+\n+ private void runUnderSamplingTest(double ratio, Types.ExecType instType) {\n+ Types.ExecMode platformOld = setExecMode(instType);\n+\n+ try {\n+ setOutputBuffering(true);\n+\n+ loadTestConfiguration(getTestConfiguration(TEST_NAME));\n+\n+ String HOME = SCRIPT_DIR + TEST_DIR;\n+\n+ fullDMLScriptName = HOME + TEST_NAME + \".dml\";\n+ programArgs = new String[] {\"-args\", String.valueOf(ratio)};\n+\n+ String out = runTest(null).toString();\n+ Assert.assertTrue(out.contains(\"TRUE\"));\n+ }\n+ finally {\n+ rtplatform = platformOld;\n+ }\n+ }\n+}\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/test/scripts/functions/builtin/underSamplingTest.dml", "diff": "+#-------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+#-------------------------------------------------------------\n+\n+ratio = as.double($1)\n+X = rand(rows=20, cols=4, min=1, max =100)\n+Y = rbind(matrix(1, rows=15, cols=1), matrix(2, rows=5, cols=1))\n+data = cbind(X, Y)\n+classesUnBalanced = table(Y[, ncol(Y)], 1)\n+# # # randomize the data\n+IX = sample(nrow(data), nrow(data))\n+P = table(seq(1,nrow(IX)), IX, nrow(IX), nrow(data));\n+data = P %*% data\n+balanced = underSampling(data, ratio)\n+classesBalanced = table(balanced[, ncol(balanced)], 1)\n+out = as.scalar(classesUnBalanced[1] - classesBalanced[1]) == floor(15.0*ratio)\n+print(out)\n+\n+\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/api/DMLScript.java", "new_path": "src/main/java/org/apache/sysds/api/DMLScript.java", "diff": "@@ -67,13 +67,13 @@ import org.apache.sysds.runtime.controlprogram.federated.FederatedWorker;\nimport org.apache.sysds.runtime.controlprogram.parfor.stat.InfrastructureAnalyzer;\nimport org.apache.sysds.runtime.controlprogram.parfor.util.IDHandler;\nimport org.apache.sysds.runtime.instructions.gpu.context.GPUContextPool;\n-import org.apache.sysds.runtime.instructions.spark.utils.SparkUtils;\nimport org.apache.sysds.runtime.io.IOUtilFunctions;\nimport org.apache.sysds.runtime.lineage.LineageCacheConfig;\nimport org.apache.sysds.runtime.lineage.LineageCacheConfig.LineageCachePolicy;\nimport org.apache.sysds.runtime.lineage.LineageCacheConfig.ReuseCacheType;\nimport org.apache.sysds.runtime.privacy.CheckedConstraintsLog;\nimport org.apache.sysds.runtime.util.LocalFileUtils;\n+import org.apache.sysds.runtime.util.CommonThreadPool;\nimport org.apache.sysds.runtime.util.HDFSTool;\nimport org.apache.sysds.utils.Explain;\nimport org.apache.sysds.utils.NativeHelper;\n@@ -519,7 +519,7 @@ public class DMLScript\nFederatedData.clearFederatedWorkers();\n//0) shutdown prefetch/broadcast thread pool if necessary\n- SparkUtils.shutdownPool();\n+ CommonThreadPool.shutdownAsyncRDDPool();\n//1) cleanup scratch space (everything for current uuid)\n//(required otherwise export to hdfs would skip assumed unnecessary writes if same name)\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/instructions/cp/BroadcastCPInstruction.java", "new_path": "src/main/java/org/apache/sysds/runtime/instructions/cp/BroadcastCPInstruction.java", "diff": "@@ -23,8 +23,8 @@ import java.util.concurrent.Executors;\nimport org.apache.sysds.runtime.controlprogram.context.ExecutionContext;\nimport org.apache.sysds.runtime.instructions.InstructionUtils;\n-import org.apache.sysds.runtime.instructions.spark.utils.SparkUtils;\nimport org.apache.sysds.runtime.matrix.operators.Operator;\n+import org.apache.sysds.runtime.util.CommonThreadPool;\npublic class BroadcastCPInstruction extends UnaryCPInstruction {\nprivate BroadcastCPInstruction(Operator op, CPOperand in, CPOperand out, String opcode, String istr) {\n@@ -44,8 +44,8 @@ public class BroadcastCPInstruction extends UnaryCPInstruction {\npublic void processInstruction(ExecutionContext ec) {\nec.setVariable(output.getName(), ec.getMatrixObject(input1));\n- if (SparkUtils.triggerRDDPool == null)\n- SparkUtils.triggerRDDPool = Executors.newCachedThreadPool();\n- SparkUtils.triggerRDDPool.submit(new TriggerBroadcastTask(ec, ec.getMatrixObject(output)));\n+ if (CommonThreadPool.triggerRDDPool == null)\n+ CommonThreadPool.triggerRDDPool = Executors.newCachedThreadPool();\n+ CommonThreadPool.triggerRDDPool.submit(new TriggerBroadcastTask(ec, ec.getMatrixObject(output)));\n}\n}\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/instructions/cp/PrefetchCPInstruction.java", "new_path": "src/main/java/org/apache/sysds/runtime/instructions/cp/PrefetchCPInstruction.java", "diff": "@@ -23,8 +23,8 @@ import java.util.concurrent.Executors;\nimport org.apache.sysds.runtime.controlprogram.context.ExecutionContext;\nimport org.apache.sysds.runtime.instructions.InstructionUtils;\n-import org.apache.sysds.runtime.instructions.spark.utils.SparkUtils;\nimport org.apache.sysds.runtime.matrix.operators.Operator;\n+import org.apache.sysds.runtime.util.CommonThreadPool;\npublic class PrefetchCPInstruction extends UnaryCPInstruction {\nprivate PrefetchCPInstruction(Operator op, CPOperand in, CPOperand out, String opcode, String istr) {\n@@ -49,8 +49,8 @@ public class PrefetchCPInstruction extends UnaryCPInstruction {\n// If the next instruction which takes this output as an input comes before\n// the prefetch thread triggers, that instruction will start the operations.\n// In that case this Prefetch instruction will act like a NOOP.\n- if (SparkUtils.triggerRDDPool == null)\n- SparkUtils.triggerRDDPool = Executors.newCachedThreadPool();\n- SparkUtils.triggerRDDPool.submit(new TriggerRDDOperationsTask(ec.getMatrixObject(output)));\n+ if (CommonThreadPool.triggerRDDPool == null)\n+ CommonThreadPool.triggerRDDPool = Executors.newCachedThreadPool();\n+ CommonThreadPool.triggerRDDPool.submit(new TriggerRDDOperationsTask(ec.getMatrixObject(output)));\n}\n}\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/instructions/spark/utils/SparkUtils.java", "new_path": "src/main/java/org/apache/sysds/runtime/instructions/spark/utils/SparkUtils.java", "diff": "@@ -57,14 +57,11 @@ import scala.Tuple2;\nimport java.util.Iterator;\nimport java.util.List;\n-import java.util.concurrent.ExecutorService;\nimport java.util.stream.Collectors;\nimport java.util.stream.LongStream;\npublic class SparkUtils\n{\n- public static ExecutorService triggerRDDPool = null;\n-\n//internal configuration\npublic static final StorageLevel DEFAULT_TMP = Checkpoint.DEFAULT_STORAGE_LEVEL;\n@@ -296,14 +293,6 @@ public class SparkUtils\nmo.acquireReadAndRelease();\n}\n- public static void shutdownPool() {\n- if (triggerRDDPool != null) {\n- //shutdown prefetch/broadcast thread pool\n- triggerRDDPool.shutdown();\n- triggerRDDPool = null;\n- }\n- }\n-\nprivate static class CheckSparsityFunction implements VoidFunction<Tuple2<MatrixIndexes,MatrixBlock>>\n{\nprivate static final long serialVersionUID = 4150132775681848807L;\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/util/CommonThreadPool.java", "new_path": "src/main/java/org/apache/sysds/runtime/util/CommonThreadPool.java", "diff": "@@ -49,6 +49,7 @@ public class CommonThreadPool implements ExecutorService\nprivate static final int size = InfrastructureAnalyzer.getLocalParallelism();\nprivate static final ExecutorService shared = ForkJoinPool.commonPool();\nprivate final ExecutorService _pool;\n+ public static ExecutorService triggerRDDPool = null;\npublic CommonThreadPool(ExecutorService pool) {\n_pool = pool;\n@@ -78,6 +79,14 @@ public class CommonThreadPool implements ExecutorService\nshared.shutdownNow();\n}\n+ public static void shutdownAsyncRDDPool() {\n+ if (triggerRDDPool != null) {\n+ //shutdown prefetch/broadcast thread pool\n+ triggerRDDPool.shutdown();\n+ triggerRDDPool = null;\n+ }\n+ }\n+\n@Override\npublic void shutdown() {\nif( _pool != shared )\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/async/PrefetchRDDTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/async/PrefetchRDDTest.java", "diff": "@@ -101,7 +101,9 @@ public class PrefetchRDDTest extends AutomatedTestBase {\nHashMap<MatrixValue.CellIndex, Double> R_pf = readDMLScalarFromOutputDir(\"R\");\n//compare matrices\n- TestUtils.compareMatrices(R, R_pf, 1e-6, \"Origin\", \"withPrefetch\");\n+ Boolean matchVal = TestUtils.compareMatrices(R, R_pf, 1e-6, \"Origin\", \"withPrefetch\");\n+ if (!matchVal)\n+ System.out.println(\"Value w/o Prefetch \"+R+\" w/ Prefetch \"+R_pf);\n//assert Prefetch instructions and number of success.\nlong expected_numPF = !testname.equalsIgnoreCase(TEST_NAME+\"3\") ? 1 : 0;\nlong expected_successPF = !testname.equalsIgnoreCase(TEST_NAME+\"3\") ? 1 : 0;\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/frame/FrameValueSwapTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/frame/FrameValueSwapTest.java", "diff": "@@ -26,7 +26,6 @@ import org.apache.sysds.test.AutomatedTestBase;\nimport org.apache.sysds.test.TestConfiguration;\nimport org.apache.sysds.test.TestUtils;\nimport org.junit.Assert;\n-import org.junit.Ignore;\nimport org.junit.Test;\npublic class FrameValueSwapTest extends AutomatedTestBase\n@@ -53,14 +52,12 @@ public class FrameValueSwapTest extends AutomatedTestBase\naddTestConfiguration(TEST_NAME, new TestConfiguration(TEST_CLASS_DIR, TEST_NAME, new String[] {\"B\"}));\n}\n//\n- @Ignore\n+\n@Test\npublic void testSwapValueTestCP() {\nrunValueSwapTest(ExecType.CP);\n}\n- // TODO fix frame comparisons in spark context\n- @Ignore\n@Test\npublic void testSwapValueTestSP() {\nrunValueSwapTest(ExecType.SPARK);\n@@ -68,16 +65,17 @@ public class FrameValueSwapTest extends AutomatedTestBase\nprivate void runValueSwapTest(ExecType et)\n{\n+ setOutputBuffering(true);\nTypes.ExecMode platformOld = setExecMode(et);\ntry {\ngetAndLoadTestConfiguration(TEST_NAME);\nString HOME = SCRIPT_DIR + TEST_DIR;\nfullDMLScriptName = HOME + TEST_NAME + \".dml\";\n- programArgs = new String[] {\"-args\", INPUT, output(\"B\")};\n+ programArgs = new String[] {\"-args\", INPUT};\nrunTest(true, false, null, -1);\n- boolean retCondition = HDFSTool.readBooleanFromHDFSFile(output(\"B\"));\n- Assert.assertEquals(true, retCondition);\n+ String out = runTest(null).toString();\n+ Assert.assertTrue(out.contains(\"TRUE\"));\n}\ncatch (Exception ex) {\n" }, { "change_type": "MODIFY", "old_path": "src/test/scripts/functions/frame/valueSwaps.dml", "new_path": "src/test/scripts/functions/frame/valueSwaps.dml", "diff": "# read the inputs\nF = read($1, data_type=\"frame\", format=\"csv\", header=TRUE, naStrings= [\"NA\", \"null\",\" \",\"NaN\", \"nan\", \"\", \"?\", \"99999\"]);\n-\n+F = F[1:50]\nd = detectSchema(F)\n-idx = sample(20, 10)\n+idx = sample(nrow(F), 10)\nF1 = F\n# # swap values\nfor(i in 1:10)\n@@ -36,6 +36,8 @@ for(i in 1:10)\nR = valueSwap(F1, d)\nf1 = as.matrix(F == R)\n+print(\"f1: \"+toString(f1))\n+print(\"sum of f1: \"+sum(f1))\n+print(\"dims: \"+(ncol(F) * nrow(F)))\nresult = ((ncol(F) * nrow(F)) == sum(f1))\n-\n-write(result, $2)\n\\ No newline at end of file\n+print(result)\n\\ No newline at end of file\n" } ]

[ { "change_type": "MODIFY", "old_path": ".gitignore", "new_path": ".gitignore", "diff": "@@ -41,19 +41,6 @@ buildNumber.properties\n.jekyll-cache/\n_site/\n-# Python artifacts\n-src/main/python/systemds/systemds-java/\n-src/main/python/systemds.egg-info/\n-src/main/python/build/\n-src/main/python/LICENSE\n-src/main/python/NOTICE\n-src/main/python/dist\n-src/main/python/docs/build\n-src/main/python/docs/source/_build\n-src/main/python/tests/onnx_systemds/output_test\n-src/main/python/tests/onnx_systemds/dml_output\n-src/main/python/tests/onnx_systemds/test_models/*.onnx\n-\n# Tutorial data mnist\nsrc/main/python/systemds/examples/tutorials/*/\n" }, { "change_type": "MODIFY", "old_path": "pom.xml", "new_path": "pom.xml", "diff": "<excludes>\n<exclude>scripts/perftest/results/**</exclude>\n<exclude>.gitignore</exclude>\n+ <exclude>src/main/python/.gitignore</exclude>\n<exclude>.gitmodules</exclude>\n<exclude>.repository/</exclude>\n<exclude>.idea/</exclude>\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/.gitignore", "diff": "+\n+\n+# Git ignore for python files.\n+systemds/lib/\n+systemds.egg-info/\n+systemds/conf/\n+build/\n+LICENSE\n+NOTICE\n+generator.log\n+dist\n+docs/build\n+docs/source/_build\n+tests/onnx_systemds/output_test\n+tests/onnx_systemds/dml_output\n+tests/onnx_systemds/test_models/*.onnx\n+\n+# git ignore tmp federated files\n+tests/federated/output\n+tests/federated/worker\n+tests/federated/tmp\n+\n+tests/list/tmp\n\\ No newline at end of file\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/assembly/bin.xml", "new_path": "src/assembly/bin.xml", "diff": "<includes>\n<include>*:${artifactId}*</include>\n<include>*:avro*</include>\n+ <include>*:commons-beanutils*</include>\n<include>*:commons-cli*</include>\n<include>*:commons-collections*</include>\n<include>*:commons-configuration*</include>\n<include>*:commons-lang3</include>\n<include>*:commons-logging*</include>\n<include>*:commons-math3*</include>\n+ <include>*:commons-text*</include>\n<include>*:guava*</include>\n<include>*:hadoop-auth*</include>\n<include>*:hadoop-client*</include>\n<include>*:hadoop-hdfs*</include>\n<include>*:hadoop-mapreduce-client*</include>\n<include>*:hadoop-yarn*</include>\n+ <include>*:htrace-core*</include>\n<include>*:jackson-core-asl*</include>\n<include>*:jackson-mapper-asl*</include>\n<include>*:janino*</include>\n<include>*:netty*</include>\n<include>*:protobuf-java*</include>\n<include>*:py4j*</include>\n+ <include>*:re2j*</include>\n<include>*:slf4j-api*</include>\n<include>*:slf4j-log4j*</include>\n<include>*:spark-core*</include>\n+ <include>*:stax2-api*</include>\n+ <include>*:woodstox*</include>\n</includes>\n<outputDirectory>./lib</outputDirectory>\n<scope>compile</scope>\n" } ]

[ { "change_type": "MODIFY", "old_path": ".github/workflows/applicationTests.yml", "new_path": ".github/workflows/applicationTests.yml", "diff": "@@ -27,7 +27,7 @@ on:\n- 'docs/**'\n- '*.md'\n- '*.html'\n- - 'src/main/python/docs/**'\n+ - 'src/main/python/**'\n- 'dev/**'\nbranches:\n- main\n@@ -36,7 +36,7 @@ on:\n- 'docs/**'\n- '*.md'\n- '*.html'\n- - 'src/main/python/docs/**'\n+ - 'src/main/python/**'\n- 'dev/**'\nbranches:\n- main\n" }, { "change_type": "MODIFY", "old_path": ".github/workflows/build.yml", "new_path": ".github/workflows/build.yml", "diff": "@@ -27,7 +27,7 @@ on:\n- 'docs/**'\n- '*.md'\n- '*.html'\n- - 'src/main/python/docs/**'\n+ - 'src/main/python/**'\n- 'dev/**'\nbranches:\n- main\n@@ -36,7 +36,7 @@ on:\n- 'docs/**'\n- '*.md'\n- '*.html'\n- - 'src/main/python/docs/**'\n+ - 'src/main/python/**'\n- 'dev/**'\nbranches:\n- main\n" }, { "change_type": "MODIFY", "old_path": ".github/workflows/componentTests.yml", "new_path": ".github/workflows/componentTests.yml", "diff": "@@ -27,7 +27,7 @@ on:\n- 'docs/**'\n- '*.md'\n- '*.html'\n- - 'src/main/python/docs/**'\n+ - 'src/main/python/**'\n- 'dev/**'\nbranches:\n- main\n@@ -36,7 +36,7 @@ on:\n- 'docs/**'\n- '*.md'\n- '*.html'\n- - 'src/main/python/docs/**'\n+ - 'src/main/python/**'\n- 'dev/**'\nbranches:\n- main\n" }, { "change_type": "MODIFY", "old_path": ".github/workflows/functionsTests.yml", "new_path": ".github/workflows/functionsTests.yml", "diff": "@@ -27,7 +27,7 @@ on:\n- 'docs/**'\n- '*.md'\n- '*.html'\n- - 'src/main/python/docs/**'\n+ - 'src/main/python/**'\n- 'dev/**'\nbranches:\n- main\n@@ -36,7 +36,7 @@ on:\n- 'docs/**'\n- '*.md'\n- '*.html'\n- - 'src/main/python/docs/**'\n+ - 'src/main/python/**'\n- 'dev/**'\nbranches:\n- main\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/nodes/list.py", "new_path": "src/main/python/systemds/operator/nodes/list.py", "diff": "@@ -80,7 +80,7 @@ class List(OperationNode):\nunnamed_input_vars, named_input_vars)\nreturn f'{var_name}={self.operation}({inputs_comma_sep});'\n- def compute(self, verbose: bool = False, lineage: bool = False) -> Union[np.array]:\n+ def compute(self, verbose: bool = False, lineage: bool = False) -> np.array:\nreturn super().compute(verbose, lineage)\ndef __str__(self):\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/nodes/list_access.py", "new_path": "src/main/python/systemds/operator/nodes/list_access.py", "diff": "__all__ = [\"ListAccess\"]\n-from typing import Dict, Iterable, Sequence, Tuple, Union\n-\n-import numpy as np\n-from py4j.java_gateway import JavaObject\n+from typing import Dict, Sequence\nfrom systemds.operator import Frame, Matrix, OperationNode, Scalar\nfrom systemds.script_building.dag import OutputType\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/tests/algorithms/test_kmeans.py", "new_path": "src/main/python/tests/algorithms/test_kmeans.py", "diff": "@@ -82,7 +82,7 @@ class TestKMeans(unittest.TestCase):\nself.assertTrue(len(corners) == 4)\n- def generate_matrices_for_k_means(self, dims: (int, int), seed: int = 1234):\n+ def generate_matrices_for_k_means(self, dims, seed: int = 1234):\nnp.random.seed(seed)\nmu, sigma = 0, 0.1\ns = np.random.normal(mu, sigma, dims[0] * dims[1])\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/tests/basics/test_context_creation.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+\n+import unittest\n+\n+from systemds.context import SystemDSContext\n+\n+\n+class TestContextCreation(unittest.TestCase):\n+\n+ def test_same_port(self):\n+ # Same port should graciously change port\n+ sds1 = SystemDSContext(port=9415)\n+ sds2 = SystemDSContext(port=9415)\n+ sds1.close()\n+ sds2.close()\n+\n+ def test_create_10_contexts(self):\n+ # Creating multiple contexts and closing them should be no problem.\n+ for _ in range(0, 10):\n+ SystemDSContext().close()\n+\n+ def test_create_multiple_context(self):\n+ # Creating multiple contexts in sequence but open at the same time is okay.\n+ a = SystemDSContext()\n+ b = SystemDSContext()\n+ c = SystemDSContext()\n+ d = SystemDSContext()\n+\n+ a.close()\n+ b.close()\n+ c.close()\n+ d.close()\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/tests/frame/test_hyperband.py", "new_path": "src/main/python/tests/frame/test_hyperband.py", "diff": "@@ -54,7 +54,6 @@ class TestHyperband(unittest.TestCase):\npass\ndef test_hyperband(self):\n- if \"SYSTEMDS_ROOT\" in os.environ:\nx_train = self.sds.from_numpy(self.X_train)\ny_train = self.sds.from_numpy(self.y_train)\nx_val = self.sds.from_numpy(self.X_val)\n@@ -78,8 +77,6 @@ class TestHyperband(unittest.TestCase):\nfor i, hyper_param in enumerate(opt_hyper_params_df.values.flatten().tolist()):\nself.assertTrue(\nself.min_max_params[i][0] <= hyper_param <= self.min_max_params[i][1])\n- else:\n- print(\"to enable hyperband tests, set SYSTEMDS_ROOT\")\nif __name__ == \"__main__\":\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/tests/frame/test_write_read.py", "new_path": "src/main/python/tests/frame/test_write_read.py", "diff": "#\n# -------------------------------------------------------------\n-import os\nimport shutil\n-import sys\nimport unittest\nimport pandas as pd\n@@ -63,7 +61,8 @@ class TestWriteRead(unittest.TestCase):\ndef test_write_read_csv(self):\nframe = self.sds.from_pandas(self.df)\nframe.write(self.temp_dir + \"02\", header=True, format=\"csv\").compute()\n- NX = self.sds.read(self.temp_dir + \"02\", data_type=\"frame\", format=\"csv\")\n+ NX = self.sds.read(self.temp_dir + \"02\",\n+ data_type=\"frame\", format=\"csv\")\nresult_df = NX.compute()\nself.assertTrue(isinstance(result_df, pd.DataFrame))\nself.assertTrue(self.df.equals(result_df))\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/tests/lineage/test_lineagetrace.py", "new_path": "src/main/python/tests/lineage/test_lineagetrace.py", "diff": "import os\nimport shutil\n-import sys\nimport unittest\nfrom systemds.context import SystemDSContext\n-from systemds.utils.helpers import get_module_dir\nos.environ['SYSDS_QUIET'] = \"1\"\n@@ -48,8 +46,8 @@ class TestLineageTrace(unittest.TestCase):\ndef tearDown(self):\nshutil.rmtree(temp_dir, ignore_errors=True)\n+ @unittest.skipIf(\"SYSTEMDS_ROOT\" not in os.environ, \"The test is skipped if SYSTEMDS_ROOT is not set, this is required for this tests since it use the bin/systemds file to execute a reference\")\ndef test_compare_trace1(self): # test getLineageTrace() on an intermediate\n- if \"SYSTEMDS_ROOT\" in os.environ:\nm = self.sds.full((10, 10), 1)\nm_res = m + m\n@@ -69,8 +67,6 @@ class TestLineageTrace(unittest.TestCase):\npython_trace_commands = [x[:1] for x in python_trace]\ndml_script_commands = [x[:1] for x in sysds_trace]\nself.assertEqual(python_trace_commands[0], dml_script_commands[0])\n- else:\n- print(\"to enable lineage tests, set SYSTEMDS_ROOT\")\n# TODO add more tests cases.\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/tests/list/test_list_readwrite.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+\n+import shutil\n+import unittest\n+\n+import numpy as np\n+from systemds.context import SystemDSContext\n+\n+\n+class TestListOperations(unittest.TestCase):\n+\n+ sds: SystemDSContext = None\n+ temp_dir: str = \"tests/list/tmp/readwrite/\"\n+\n+ @classmethod\n+ def setUpClass(cls):\n+ cls.sds = SystemDSContext()\n+\n+ @classmethod\n+ def tearDownClass(cls):\n+ cls.sds.close()\n+ shutil.rmtree(cls.temp_dir)\n+\n+ def test_write_followed_by_read(self):\n+ ''' Test write and read of lists variables in python.\n+ Since we do not support serializing a list (from java to python) yet we\n+ read and compute each list element when reading again\n+ '''\n+ m1 = np.array([[1., 2., 3.]])\n+ m1p = self.sds.from_numpy(m1)\n+ m2 = np.array([[4., 5., 6.]])\n+ m2p = self.sds.from_numpy(m2)\n+ list_obj = self.sds.array(m1p, m2p)\n+\n+ path = self.temp_dir + \"01\"\n+ list_obj.write(path).compute()\n+ ret_m1 = self.sds.read(path)[1].as_matrix().compute()\n+ ret_m2 = self.sds.read(path)[2].as_matrix().compute()\n+ self.assertTrue(np.allclose(m1, ret_m1))\n+ self.assertTrue(np.allclose(m2, ret_m2))\n+\n+\n+if __name__ == \"__main__\":\n+ unittest.main(exit=False)\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/tests/matrix/test_cholesky.py", "new_path": "src/main/python/tests/matrix/test_cholesky.py", "diff": "@@ -30,6 +30,7 @@ A = np.random.rand(shape, shape)\n# set A = MM^T and A is a positive definite matrix\nA = np.matmul(A, A.transpose())\n+\nclass TestCholesky(unittest.TestCase):\nsds: SystemDSContext = None\n@@ -43,7 +44,7 @@ class TestCholesky(unittest.TestCase):\ncls.sds.close()\n-class TestCholesky_0(TestCholesky):\n+class TestCholeskyValid(TestCholesky):\ndef test_basic1(self):\nL = self.sds.from_numpy(A).cholesky().compute()\n@@ -54,24 +55,27 @@ class TestCholesky_0(TestCholesky):\n# L * L.H = A\nself.assertTrue(np.allclose(A, np.dot(L, L.T.conj())))\n-class TestCholesky_1(TestCholesky):\n+\n+class TestCholeskyInvalid_1(TestCholesky):\ndef test_pos_def(self):\nm1 = -np.random.rand(shape, shape)\n- with self.assertRaises(RuntimeError) as context:\n+ with self.assertRaises(Exception):\nself.sds.from_numpy(m1).cholesky().compute()\n-class TestCholesky_2(TestCholesky):\n+\n+class TestCholeskyInvalid_2(TestCholesky):\ndef test_symmetric_matrix(self):\nm2 = np.asarray([[4, 9], [1, 4]])\nnp.linalg.cholesky(m2)\n- with self.assertRaises(RuntimeError) as context:\n+ with self.assertRaises(Exception):\nself.sds.from_numpy(m2).cholesky().compute()\n-class TestCholesky_3(TestCholesky):\n+\n+class TestCholeskyInvalid_3(TestCholesky):\ndef test_asymetric_dim(self):\nm3 = np.random.rand(shape, shape + 1)\n- with self.assertRaises(RuntimeError) as context:\n+ with self.assertRaises(Exception):\nself.sds.from_numpy(m3).cholesky().compute()\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/tests/matrix/test_order.py", "new_path": "src/main/python/tests/matrix/test_order.py", "diff": "#\n# -------------------------------------------------------------\n-import unittest\nimport random\n+import unittest\nimport numpy as np\nfrom systemds.context import SystemDSContext\n@@ -33,7 +33,8 @@ mx = np.random.rand(1, shape[1])\nmy = np.random.rand(shape[0], 1)\nby = random.randrange(1, np.size(m, 1)+1)\n-class TestOrder(unittest.TestCase):\n+\n+class TestOrderBase(unittest.TestCase):\nsds: SystemDSContext = None\n@@ -45,26 +46,35 @@ class TestOrder(unittest.TestCase):\ndef tearDownClass(cls):\ncls.sds.close()\n+\n+class TestOrderValid(TestOrderBase):\n+\ndef test_basic(self):\n- o = self.sds.from_numpy(m).order(by=by, decreasing=False, index_return=False).compute()\n+ o = self.sds.from_numpy(m).order(\n+ by=by, decreasing=False, index_return=False).compute()\ns = m[np.argsort(m[:, by-1])]\nself.assertTrue(np.allclose(o, s))\ndef test_index(self):\n- o = self.sds.from_numpy(m).order(by=by, decreasing=False, index_return=True).compute()\n+ o = self.sds.from_numpy(m).order(\n+ by=by, decreasing=False, index_return=True).compute()\ns = np.argsort(m[:, by - 1]) + 1\nself.assertTrue(np.allclose(np.transpose(o), s))\ndef test_decreasing(self):\n- o = self.sds.from_numpy(m).order(by=by, decreasing=True, index_return=True).compute()\n+ o = self.sds.from_numpy(m).order(\n+ by=by, decreasing=True, index_return=True).compute()\ns = np.argsort(-m[:, by - 1]) + 1\nself.assertTrue(np.allclose(np.transpose(o), s))\n-class TestOrder_1(TestOrder):\n+\n+class TestOrderInvalid(TestOrderBase):\n+\ndef test_out_of_bounds(self):\nby_max = np.size(m, 1) + 2\n- with self.assertRaises(RuntimeError) as context:\n+ with self.assertRaises(Exception):\nself.sds.from_numpy(m).order(by=by_max).compute()\n+\nif __name__ == \"__main__\":\nunittest.main(exit=False)\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/tests/matrix/test_print.py", "new_path": "src/main/python/tests/matrix/test_print.py", "diff": "import unittest\nimport numpy as np\n+from time import sleep\nfrom systemds.context import SystemDSContext\n@@ -32,6 +33,10 @@ class TestPrint(unittest.TestCase):\n@classmethod\ndef setUpClass(cls):\ncls.sds = SystemDSContext()\n+ sleep(1.0)\n+ # Clear stdout ...\n+ cls.sds.get_stdout()\n+ cls.sds.get_stdout()\n@classmethod\ndef tearDownClass(cls):\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/tests/script/test_dml_script.py", "new_path": "src/main/python/tests/script/test_dml_script.py", "diff": "# -------------------------------------------------------------\nimport unittest\n-import time\n+from time import sleep\nfrom systemds.context import SystemDSContext\nfrom systemds.script_building import DMLScript\n@@ -35,6 +35,9 @@ class Test_DMLScript(unittest.TestCase):\n@classmethod\ndef setUpClass(cls):\ncls.sds = SystemDSContext()\n+ sleep(1)\n+ cls.sds.get_stdout()\n+ cls.sds.get_stdout()\n@classmethod\ndef tearDownClass(cls):\n@@ -44,7 +47,7 @@ class Test_DMLScript(unittest.TestCase):\nscript = DMLScript(self.sds)\nscript.add_code('print(\"Hello\")')\nscript.execute()\n- time.sleep(0.5)\n+ sleep(0.5)\nstdout = self.sds.get_stdout(100)\nself.assertListEqual([\"Hello\"], stdout)\n@@ -54,7 +57,7 @@ class Test_DMLScript(unittest.TestCase):\nscript.add_code('print(\"World\")')\nscript.add_code('print(\"!\")')\nscript.execute()\n- time.sleep(0.5)\n+ sleep(0.5)\nstdout = self.sds.get_stdout(100)\nself.assertListEqual(['Hello', 'World', '!'], stdout)\n@@ -65,7 +68,7 @@ class Test_DMLScript(unittest.TestCase):\nscr_a.add_code('y = x + 1')\nscr_a.add_code('print(y)')\nscr_a.execute()\n- time.sleep(0.5)\n+ sleep(0.5)\nstdout = self.sds.get_stdout(100)\nself.assertEqual(\"4\", stdout[0])\nself.assertEqual(\"5\", stdout[1])\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/io/IOUtilFunctions.java", "new_path": "src/main/java/org/apache/sysds/runtime/io/IOUtilFunctions.java", "diff": "@@ -93,7 +93,7 @@ public class IOUtilFunctions\ntry{\nreturn FileSystem.get(conf);\n} catch(NoClassDefFoundError err) {\n- throw new IOException(err.getMessage());\n+ throw new IOException(err.getMessage(), err);\n}\n}\n@@ -101,7 +101,7 @@ public class IOUtilFunctions\ntry {\nreturn FileSystem.get(fname.toUri(), conf);\n} catch(NoClassDefFoundError err) {\n- throw new IOException(err.getMessage());\n+ throw new IOException(err.getMessage(), err);\n}\n}\n" } ]

[ { "change_type": "MODIFY", "old_path": "bin/systemds", "new_path": "bin/systemds", "diff": "@@ -55,6 +55,14 @@ print_out()\nfi\n}\n+if [[ -z $SYSTEMDS_ROOT ]] ; then\n+ SYSTEMDS_ROOT=.\n+ print_out \"SYSTEMDS_ROOT not set defaulting to current dir $(pwd)\"\n+else\n+ # construct a relative path\n+ SYSTEMDS_ROOT=$(realpath --relative-to=. ${SYSTEMDS_ROOT})\n+fi;\n+\n# when using find, look in the directories in this order\nDIR_SEARCH_ORDER=\". $SYSTEMDS_ROOT $SYSTEMDS_ROOT/conf $SYSTEMDS_ROOT/lib $SYSTEMDS_ROOT/src $SYSTEMDS_ROOT/target\"\nordered_find() {\n@@ -264,13 +272,6 @@ if [ -z \"$WORKER\" ] ; then\nWORKER=0\nfi\n-if [[ -z $SYSTEMDS_ROOT ]] ; then\n- SYSTEMDS_ROOT=.\n- print_out \"SYSTEMDS_ROOT not set defaulting to current dir $(pwd)\"\n-else\n- # construct a relative path\n- SYSTEMDS_ROOT=$(realpath --relative-to=. ${SYSTEMDS_ROOT})\n-fi;\n# find me a SystemDS jar file to run\nif [ -z \"$SYSTEMDS_JAR_FILE\" ];then\n@@ -320,6 +321,9 @@ if [[ \"$*\" == *-config* ]]; then\nelif [ -z \"$CONFIG_FILE\" ] ; then\n# same as above: set config file param if the file exists\nCONFIG_FILE=$(ordered_find \"SystemDS-config-defaults.xml\")\n+ if [ -z \"$CONFIG_FILE\" ]; then\n+ CONFIG_FILE=$(ordered_find \"SystemDS-config.xml\")\n+ fi\nif [ -z \"$CONFIG_FILE\" ]; then\nCONFIG_FILE=\"\"\nelse\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/component/matrix/CountDistinctTest.java", "new_path": "src/test/java/org/apache/sysds/test/component/matrix/CountDistinctTest.java", "diff": "package org.apache.sysds.test.component.matrix;\n+import static org.junit.Assert.assertThrows;\n+import static org.junit.Assert.fail;\n+\nimport java.util.ArrayList;\nimport java.util.Collection;\n@@ -31,10 +34,7 @@ import org.apache.sysds.runtime.matrix.operators.CountDistinctOperator.CountDist\nimport org.apache.sysds.runtime.util.DataConverter;\nimport org.apache.sysds.test.TestUtils;\nimport org.apache.sysds.utils.Hash.HashType;\n-import org.junit.Assert;\n-import org.junit.Rule;\nimport org.junit.Test;\n-import org.junit.rules.ExpectedException;\nimport org.junit.runner.RunWith;\nimport org.junit.runners.Parameterized;\nimport org.junit.runners.Parameterized.Parameters;\n@@ -91,11 +91,11 @@ public class CountDistinctTest {\nif((ht == HashType.ExpHash && et == CountDistinctTypes.KMV) ||\n(ht == HashType.StandardJava && et == CountDistinctTypes.KMV)) {\nString errorMessage = \"Invalid hashing configuration using \" + ht + \" and \" + et;\n- tests.add(new Object[] {et, inputs.get(0), actualUnique.get(0), ht, DMLException.class,\n+ tests.add(new Object[] {et, inputs.get(0), actualUnique.get(0), ht, new DMLException(),\nerrorMessage, 0.0});\n}\nelse if(et == CountDistinctTypes.HLL) {\n- tests.add(new Object[] {et, inputs.get(0), actualUnique.get(0), ht, NotImplementedException.class,\n+ tests.add(new Object[] {et, inputs.get(0), actualUnique.get(0), ht, new NotImplementedException(),\n\"HyperLogLog not implemented\", 0.0});\n}\nelse if(et != CountDistinctTypes.COUNT) {\n@@ -125,13 +125,10 @@ public class CountDistinctTest {\n// Exception handling\[email protected](4)\n- public Class<? extends Exception> expectedException;\n+ public Exception expectedException;\[email protected](5)\npublic String expectedExceptionMsg;\n- @Rule\n- public ExpectedException thrown = ExpectedException.none();\n-\n// allowing the estimate to be within 20% of target.\[email protected](6)\npublic double epsilon;\n@@ -139,15 +136,14 @@ public class CountDistinctTest {\n@Test\npublic void testEstimation() {\n- // setup expected exception\n- if(expectedException != null) {\n- thrown.expect(expectedException);\n- thrown.expectMessage(expectedExceptionMsg);\n- }\n-\nInteger out = 0;\nCountDistinctOperator op = new CountDistinctOperator(et, ht);\ntry {\n+ if(expectedException != null){\n+ assertThrows(expectedException.getClass(), () -> {LibMatrixCountDistinct.estimateDistinctValues(in, op);});\n+ return;\n+ }\n+ else\nout = LibMatrixCountDistinct.estimateDistinctValues(in, op);\n}\ncatch(DMLException e) {\n@@ -158,15 +154,17 @@ public class CountDistinctTest {\n}\ncatch(Exception e) {\ne.printStackTrace();\n- Assert.assertTrue(this.toString(), false);\n+ fail(this.toString());\n}\nint count = out;\nboolean success = Math.abs(nrUnique - count) <= nrUnique * epsilon;\n+ if(!success){\nStringBuilder sb = new StringBuilder();\nsb.append(this.toString());\nsb.append(\"\\n\" + count + \" unique values, actual:\" + nrUnique + \" with eps of \" + epsilon);\n- Assert.assertTrue(sb.toString(), success);\n+ fail(sb.toString());\n+ }\n}\n@Override\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/frame/FrameValueSwapTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/frame/FrameValueSwapTest.java", "diff": "@@ -21,7 +21,6 @@ package org.apache.sysds.test.functions.frame;\nimport org.apache.sysds.common.Types;\nimport org.apache.sysds.common.Types.ExecType;\n-import org.apache.sysds.runtime.util.HDFSTool;\nimport org.apache.sysds.test.AutomatedTestBase;\nimport org.apache.sysds.test.TestConfiguration;\nimport org.apache.sysds.test.TestUtils;\n@@ -51,7 +50,6 @@ public class FrameValueSwapTest extends AutomatedTestBase\nTestUtils.clearAssertionInformation();\naddTestConfiguration(TEST_NAME, new TestConfiguration(TEST_CLASS_DIR, TEST_NAME, new String[] {\"B\"}));\n}\n- //\n@Test\npublic void testSwapValueTestCP() {\n" } ]

[ { "change_type": "MODIFY", "old_path": "dev/release/release-utils.sh", "new_path": "dev/release/release-utils.sh", "diff": "@@ -215,7 +215,7 @@ get_release_info() {\n# Git configuration info\n# The ASF ID is obtained from\n- # https://people.apache.org/phonebook.html?unix=systemds\n+ # https://people.apache.org/committers-by-project.html#systemds\nif [[ -z \"$ASF_USERNAME\" ]]; then\nexport ASF_USERNAME=$(read_config \"ASF ID\" \"$LOGNAME\")\nfi\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/compress/CompressedMatrixBlock.java", "new_path": "src/main/java/org/apache/sysds/runtime/compress/CompressedMatrixBlock.java", "diff": "@@ -1076,9 +1076,9 @@ public class CompressedMatrixBlock extends MatrixBlock {\n}\nif(m2 instanceof CompressedMatrixBlock)\n- m2 = ((CompressedMatrixBlock) m2).getUncompressed(\"Ternay Operator arg2 \" + op.fn.getClass().getSimpleName());\n+ m2 = ((CompressedMatrixBlock) m2).getUncompressed(\"Ternary Operator arg2 \" + op.fn.getClass().getSimpleName());\nif(m3 instanceof CompressedMatrixBlock)\n- m3 = ((CompressedMatrixBlock) m3).getUncompressed(\"Ternay Operator arg3 \" + op.fn.getClass().getSimpleName());\n+ m3 = ((CompressedMatrixBlock) m3).getUncompressed(\"Ternary Operator arg3 \" + op.fn.getClass().getSimpleName());\nif(s2 != s3 && (op.fn instanceof PlusMultiply || op.fn instanceof MinusMultiply)) {\n// SPECIAL CASE for sparse-dense combinations of common +* and -*\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/instructions/cp/MultiReturnParameterizedBuiltinCPInstruction.java", "new_path": "src/main/java/org/apache/sysds/runtime/instructions/cp/MultiReturnParameterizedBuiltinCPInstruction.java", "diff": "@@ -88,7 +88,8 @@ public class MultiReturnParameterizedBuiltinCPInstruction extends ComputationCPI\nMultiColumnEncoder encoder = EncoderFactory.createEncoder(spec, colnames, fin.getNumColumns(), null);\n// TODO: Assign #threads in compiler and pass via the instruction string\nMatrixBlock data = encoder.encode(fin, OptimizerUtils.getTransformNumThreads()); // build and apply\n- FrameBlock meta = encoder.getMetaData(new FrameBlock(fin.getNumColumns(), ValueType.STRING));\n+ FrameBlock meta = encoder.getMetaData(new FrameBlock(fin.getNumColumns(), ValueType.STRING),\n+ OptimizerUtils.getTransformNumThreads());\nmeta.setColumnNames(colnames);\n// release input and outputs\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/matrix/data/MatrixBlock.java", "new_path": "src/main/java/org/apache/sysds/runtime/matrix/data/MatrixBlock.java", "diff": "@@ -3032,10 +3032,10 @@ public class MatrixBlock extends MatrixValue implements CacheBlock, Externalizab\nif(m2 instanceof CompressedMatrixBlock)\nm2 = ((CompressedMatrixBlock) m2)\n- .getUncompressed(\"Ternay Operator arg2 \" + op.fn.getClass().getSimpleName());\n+ .getUncompressed(\"Ternary Operator arg2 \" + op.fn.getClass().getSimpleName());\nif(m3 instanceof CompressedMatrixBlock)\nm3 = ((CompressedMatrixBlock) m3)\n- .getUncompressed(\"Ternay Operator arg3 \" + op.fn.getClass().getSimpleName());\n+ .getUncompressed(\"Ternary Operator arg3 \" + op.fn.getClass().getSimpleName());\nret.reset(m, n, sparseOutput);\n@@ -5102,11 +5102,11 @@ public class MatrixBlock extends MatrixValue implements CacheBlock, Externalizab\npublic MatrixBlock aggregateTernaryOperations(MatrixBlock m1, MatrixBlock m2, MatrixBlock m3, MatrixBlock ret,\nAggregateTernaryOperator op, boolean inCP) {\nif(m1 instanceof CompressedMatrixBlock)\n- m1 = ((CompressedMatrixBlock) m1).getUncompressed(\"Aggregate Ternay Operator arg1 \" + op.getClass().getSimpleName());\n+ m1 = ((CompressedMatrixBlock) m1).getUncompressed(\"Aggregate Ternary Operator arg1 \" + op.getClass().getSimpleName());\nif(m2 instanceof CompressedMatrixBlock)\n- m2 = ((CompressedMatrixBlock) m2).getUncompressed(\"Aggregate Ternay Operator arg2 \" + op.getClass().getSimpleName());\n+ m2 = ((CompressedMatrixBlock) m2).getUncompressed(\"Aggregate Ternary Operator arg2 \" + op.getClass().getSimpleName());\nif(m3 instanceof CompressedMatrixBlock)\n- m3 = ((CompressedMatrixBlock) m3).getUncompressed(\"Aggregate Ternay Operator arg3 \" + op.getClass().getSimpleName());\n+ m3 = ((CompressedMatrixBlock) m3).getUncompressed(\"Aggregate Ternary Operator arg3 \" + op.getClass().getSimpleName());\n//create output matrix block w/ corrections\nint rl = (op.indexFn instanceof ReduceRow) ? 2 : 1;\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderBin.java", "new_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderBin.java", "diff": "@@ -204,6 +204,11 @@ public class ColumnEncoderBin extends ColumnEncoder {\nsuper.mergeAt(other);\n}\n+ @Override\n+ public void allocateMetaData(FrameBlock meta) {\n+ meta.ensureAllocatedColumns(_binMaxs.length);\n+ }\n+\n@Override\npublic FrameBlock getMetaData(FrameBlock meta) {\n// allocate frame if necessary\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderComposite.java", "new_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderComposite.java", "diff": "@@ -268,6 +268,14 @@ public class ColumnEncoderComposite extends ColumnEncoder {\n}\n}\n+ @Override\n+ public void allocateMetaData(FrameBlock meta) {\n+ if(_meta != null)\n+ return;\n+ for(ColumnEncoder columnEncoder : _columnEncoders)\n+ columnEncoder.allocateMetaData(meta);\n+ }\n+\n@Override\npublic FrameBlock getMetaData(FrameBlock out) {\nif(_meta != null)\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderDummycode.java", "new_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderDummycode.java", "diff": "@@ -188,6 +188,11 @@ public class ColumnEncoderDummycode extends ColumnEncoder {\n}\n}\n+ @Override\n+ public void allocateMetaData(FrameBlock meta) {\n+ return;\n+ }\n+\n@Override\npublic FrameBlock getMetaData(FrameBlock meta) {\nreturn meta;\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderFeatureHash.java", "new_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderFeatureHash.java", "diff": "@@ -98,6 +98,12 @@ public class ColumnEncoderFeatureHash extends ColumnEncoder {\nsuper.mergeAt(other);\n}\n+ @Override\n+ public void allocateMetaData(FrameBlock meta) {\n+ if (isApplicable())\n+ meta.ensureAllocatedColumns(1);\n+ }\n+\n@Override\npublic FrameBlock getMetaData(FrameBlock meta) {\nif(!isApplicable())\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderPassThrough.java", "new_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderPassThrough.java", "diff": "@@ -99,6 +99,12 @@ public class ColumnEncoderPassThrough extends ColumnEncoder {\nsuper.mergeAt(other);\n}\n+ @Override\n+ public void allocateMetaData(FrameBlock meta) {\n+ // do nothing\n+ return;\n+ }\n+\n@Override\npublic FrameBlock getMetaData(FrameBlock meta) {\n// do nothing\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderRecode.java", "new_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderRecode.java", "diff": "@@ -179,6 +179,20 @@ public class ColumnEncoderRecode extends ColumnEncoder {\nreturn (code < 0) ? Double.NaN : code;\n}\n+ protected double[] getCodeCol(CacheBlock in) {\n+ Object[] coldata = (Object[]) ((FrameBlock)in).getColumnData(_colID-1);\n+ double codes[] = new double[in.getNumRows()];\n+ for (int i=0; i<coldata.length; i++) {\n+ Object okey = coldata[i];\n+ String key = (okey != null) ? okey.toString() : null;\n+ if(key == null || key.isEmpty())\n+ codes[i] = Double.NaN;\n+ long code = lookupRCDMap(key);\n+ codes[i] = code;\n+ }\n+ return codes;\n+ }\n+\n@Override\npublic void prepareBuildPartial() {\n// ensure allocated partial recode map\n@@ -232,6 +246,12 @@ public class ColumnEncoderRecode extends ColumnEncoder {\nreturn _rcdMap.size();\n}\n+ @Override\n+ public void allocateMetaData(FrameBlock meta) {\n+ // allocate output rows\n+ meta.ensureAllocatedColumns(getNumDistinctValues());\n+ }\n+\n@Override\npublic FrameBlock getMetaData(FrameBlock meta) {\nif(!isApplicable())\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/transform/encode/Encoder.java", "new_path": "src/main/java/org/apache/sysds/runtime/transform/encode/Encoder.java", "diff": "@@ -48,6 +48,12 @@ public interface Encoder extends Externalizable {\n*/\nMatrixBlock apply(CacheBlock in, MatrixBlock out, int outputCol);\n+ /**\n+ * Pre-allocate a FrameBlock for metadata collection.\n+ * @param meta frame block\n+ */\n+ void allocateMetaData(FrameBlock meta);\n+\n/**\n* Construct a frame block out of the transform meta data.\n*\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/transform/encode/MultiColumnEncoder.java", "new_path": "src/main/java/org/apache/sysds/runtime/transform/encode/MultiColumnEncoder.java", "diff": "@@ -32,6 +32,8 @@ import java.util.Objects;\nimport java.util.Set;\nimport java.util.concurrent.Callable;\nimport java.util.concurrent.ExecutionException;\n+import java.util.concurrent.ExecutorService;\n+import java.util.concurrent.Future;\nimport java.util.function.Consumer;\nimport java.util.function.Function;\nimport java.util.stream.Collectors;\n@@ -48,6 +50,7 @@ import org.apache.sysds.runtime.data.SparseBlockMCSR;\nimport org.apache.sysds.runtime.data.SparseRowVector;\nimport org.apache.sysds.runtime.matrix.data.FrameBlock;\nimport org.apache.sysds.runtime.matrix.data.MatrixBlock;\n+import org.apache.sysds.runtime.util.CommonThreadPool;\nimport org.apache.sysds.runtime.util.DependencyTask;\nimport org.apache.sysds.runtime.util.DependencyThreadPool;\nimport org.apache.sysds.runtime.util.DependencyWrapperTask;\n@@ -103,7 +106,10 @@ public class MultiColumnEncoder implements Encoder {\n}\nelse {\nLOG.debug(\"Encoding with staged approach on: \" + k + \" Threads\");\n+ long t0 = System.nanoTime();\nbuild(in, k);\n+ long t1 = System.nanoTime();\n+ LOG.debug(\"Elapsed time for build phase: \"+ ((double) t1 - t0) / 1000000 + \" ms\");\nif(_legacyMVImpute != null) {\n// These operations are redundant for every encoder excluding the legacyMVImpute, the workaround to\n// fix it for this encoder would be very dirty. This will only have a performance impact if there\n@@ -113,7 +119,10 @@ public class MultiColumnEncoder implements Encoder {\ninitMetaData(_meta);\n}\n// apply meta data\n+ t0 = System.nanoTime();\nout = apply(in, k);\n+ t1 = System.nanoTime();\n+ LOG.debug(\"Elapsed time for apply phase: \"+ ((double) t1 - t0) / 1000000 + \" ms\");\n}\n}\ncatch(Exception ex) {\n@@ -350,16 +359,51 @@ public class MultiColumnEncoder implements Encoder {\nStatistics.incTransformOutMatrixPostProcessingTime(System.nanoTime()-t0);\n}\n+ @Override\n+ public void allocateMetaData(FrameBlock meta) {\n+ for(ColumnEncoder columnEncoder : _columnEncoders) {\n+ columnEncoder.allocateMetaData(meta);\n+ }\n+ }\n+\n@Override\npublic FrameBlock getMetaData(FrameBlock meta) {\n+ getMetaData(meta, 1);\n+ return meta;\n+ }\n+\n+ public FrameBlock getMetaData(FrameBlock meta, int k) {\n+ long t0 = System.nanoTime();\nif(_meta != null)\nreturn _meta;\n+ this.allocateMetaData(meta);\n+ if (k > 1) {\n+ try {\n+ ExecutorService pool = CommonThreadPool.get(k);\n+ ArrayList<ColumnMetaDataTask<? extends ColumnEncoder>> tasks = new ArrayList<>();\n+ for(ColumnEncoder columnEncoder : _columnEncoders)\n+ tasks.add(new ColumnMetaDataTask<>(columnEncoder, meta));\n+ List<Future<Object>> taskret = pool.invokeAll(tasks);\n+ pool.shutdown();\n+ for (Future<Object> task : taskret)\n+ task.get();\n+ }\n+ catch(Exception ex) {\n+ throw new DMLRuntimeException(ex);\n+ }\n+ }\n+ else {\nfor(ColumnEncoder columnEncoder : _columnEncoders)\ncolumnEncoder.getMetaData(meta);\n+ }\n+\n+ //_columnEncoders.stream().parallel().forEach(columnEncoder ->\n+ // columnEncoder.getMetaData(meta));\nif(_legacyOmit != null)\n_legacyOmit.getMetaData(meta);\nif(_legacyMVImpute != null)\n_legacyMVImpute.getMetaData(meta);\n+ LOG.debug(\"Time spent getting metadata \"+((double) System.nanoTime() - t0) / 1000000 + \" ms\");\nreturn meta;\n}\n@@ -853,4 +897,20 @@ public class MultiColumnEncoder implements Encoder {\n}\n}\n+ private static class ColumnMetaDataTask<T extends ColumnEncoder> implements Callable<Object> {\n+ private final T _colEncoder;\n+ private final FrameBlock _out;\n+\n+ protected ColumnMetaDataTask(T encoder, FrameBlock out) {\n+ _colEncoder = encoder;\n+ _out = out;\n+ }\n+\n+ @Override\n+ public Object call() throws Exception {\n+ _colEncoder.getMetaData(_out);\n+ return null;\n+ }\n+ }\n+\n}\n" }, { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/functions/transform/TransformFrameEncodeMultithreadedTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/transform/TransformFrameEncodeMultithreadedTest.java", "diff": "@@ -208,7 +208,7 @@ public class TransformFrameEncodeMultithreadedTest extends AutomatedTestBase {\nFiles.readAllLines(Paths.get(SPEC)).forEach(s -> specSb.append(s).append(\"\\n\"));\nMultiColumnEncoder encoder = EncoderFactory.createEncoder(specSb.toString(), input.getColumnNames(),\ninput.getNumColumns(), null);\n- //MultiColumnEncoder.MULTI_THREADED_STAGES = staged;\n+ MultiColumnEncoder.MULTI_THREADED_STAGES = staged;\nMatrixBlock outputS = encoder.encode(input, 1);\nMatrixBlock outputM = encoder.encode(input, 12);\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/python/systemds/context/systemds_context.py", "new_path": "src/main/python/systemds/context/systemds_context.py", "diff": "@@ -64,8 +64,7 @@ class SystemDSContext(object):\nif process.poll() is None:\nself.__start_gateway(actual_port)\nelse:\n- self.exception_and_close(\n- \"Java process stopped before gateway could connect\")\n+ self.exception_and_close(\"Java process stopped before gateway could connect\")\ndef get_stdout(self, lines: int = -1):\n\"\"\"Getter for the stdout of the java subprocess\n@@ -89,7 +88,7 @@ class SystemDSContext(object):\nelse:\nreturn [self.__stderr.get() for x in range(lines)]\n- def exception_and_close(self, exception_str: str, trace_back_limit: int = None):\n+ def exception_and_close(self, exception, trace_back_limit: int = None):\n\"\"\"\nMethod for printing exception, printing stdout and error, while also closing the context correctly.\n@@ -104,7 +103,7 @@ class SystemDSContext(object):\nif stdErr:\nmessage += \"standard error :\\n\" + \"\\n\".join(stdErr)\nmessage += \"\\n\\n\"\n- message += exception_str\n+ message += str(exception)\nsys.tracebacklimit = trace_back_limit\nself.close()\nraise RuntimeError(message)\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/nodes/list.py", "new_path": "src/main/python/systemds/operator/nodes/list.py", "diff": "@@ -76,9 +76,8 @@ class List(OperationNode):\ndef code_line(self, var_name: str, unnamed_input_vars: Sequence[str],\nnamed_input_vars: Dict[str, str]) -> str:\n- inputs_comma_sep = create_params_string(\n- unnamed_input_vars, named_input_vars)\n- return f'{var_name}={self.operation}({inputs_comma_sep});'\n+ code_line = super().code_line(var_name, unnamed_input_vars, named_input_vars)\n+ return code_line\ndef compute(self, verbose: bool = False, lineage: bool = False) -> np.array:\nreturn super().compute(verbose, lineage)\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/nodes/scalar.py", "new_path": "src/main/python/systemds/operator/nodes/scalar.py", "diff": "@@ -57,7 +57,7 @@ class Scalar(OperationNode):\nelse:\nreturn super().code_line(var_name, unnamed_input_vars, named_input_vars)\n- def compute(self, verbose: bool = False, lineage: bool = False) -> Union[np.array]:\n+ def compute(self, verbose: bool = False, lineage: bool = False):\nreturn super().compute(verbose, lineage)\ndef _parse_output_result_variables(self, result_variables):\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/script_building/script.py", "new_path": "src/main/python/systemds/script_building/script.py", "diff": "@@ -200,15 +200,13 @@ class DMLScript:\n# for each node do the dfs operation and save the variable names in `input_var_names`\n# get variable names of unnamed parameters\n- unnamed_input_vars = [self._dfs_dag_nodes(\n- input_node) for input_node in dag_node.unnamed_input_nodes]\n+ unnamed_input_vars = []\n+ for un_node in dag_node.unnamed_input_nodes:\n+ unnamed_input_vars.append(self._dfs_dag_nodes(un_node))\nnamed_input_vars = {}\nfor name, input_node in dag_node.named_input_nodes.items():\nnamed_input_vars[name] = self._dfs_dag_nodes(input_node)\n- if isinstance(input_node, DAGNode) and input_node._output_type == OutputType.LIST:\n- dag_node.dml_name = named_input_vars[name] + name\n- return dag_node.dml_name\n# check if the node gets a name after multireturns\n# If it has, great, return that name\n@@ -222,6 +220,7 @@ class DMLScript:\ncode_line = dag_node.code_line(\ndag_node.dml_name, unnamed_input_vars, named_input_vars)\n+\nself.add_code(code_line)\nreturn dag_node.dml_name\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/tests/source/source_with_list_input.dml", "diff": "+#-------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+#-------------------------------------------------------------\n+\n+func = function(list[unknown] a) return (matrix[double] b){\n+ b = as.matrix(a[1])\n+}\n+\n+func2 = function(list[unknown] a) return (matrix[double] b, matrix[double] c){\n+ b = as.matrix(a[1])\n+ c = as.matrix(a[2])\n+}\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/tests/source/test_source_list.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+\n+import unittest\n+\n+import numpy as np\n+from systemds.context import SystemDSContext\n+from systemds.operator.algorithm.builtin.scale import scale\n+\n+\n+class TestSource_01(unittest.TestCase):\n+\n+ sds: SystemDSContext = None\n+ source_path: str = \"./tests/source/source_with_list_input.dml\"\n+\n+ @classmethod\n+ def setUpClass(cls):\n+ cls.sds = SystemDSContext()\n+\n+ @classmethod\n+ def tearDownClass(cls):\n+ cls.sds.close()\n+\n+ def test_single_return(self):\n+ arr = self.sds.array(self.sds.full((10, 10), 4))\n+ c = self.sds.source(self.source_path, \"test\").func(arr)\n+ res = c.sum().compute()\n+ self.assertTrue(res == 10*10*4)\n+\n+ def test_input_multireturn(self):\n+ m = self.sds.full((10, 10), 2)\n+ [a, b, c] = scale(m, True, True)\n+ arr = self.sds.array(a, b, c)\n+ c = self.sds.source(self.source_path, \"test\").func(arr)\n+ res = c.sum().compute(verbose=True)\n+ self.assertTrue(res == 0)\n+\n+ # [SYSTEMDS-3224] https://issues.apache.org/jira/browse/SYSTEMDS-3224\n+ # def test_multi_return(self):\n+ # arr = self.sds.array(\n+ # self.sds.full((10, 10), 4),\n+ # self.sds.full((3, 3), 5))\n+ # [b, c] = self.sds.source(self.source_path, \"test\", True).func2(arr)\n+ # res = c.sum().compute()\n+ # self.assertTrue(res == 10*10*4)\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/data/SparseBlockCSR.java", "new_path": "src/main/java/org/apache/sysds/runtime/data/SparseBlockCSR.java", "diff": "@@ -865,12 +865,15 @@ public class SparseBlockCSR extends SparseBlock\nsb.append(size());\nsb.append(\"\\n\");\nfor(int i = 0; i < numRows(); i++) {\n+ // append row\n+ final int pos = pos(i);\n+ final int len = size(i);\n+ if(pos < pos + len) {\n+\nsb.append(\"row +\");\nsb.append(i);\nsb.append(\": \");\n- //append row\n- int pos = pos(i);\n- int len = size(i);\n+\nfor(int j = pos; j < pos + len; j++) {\nsb.append(_indexes[j]);\nsb.append(\": \");\n@@ -879,6 +882,7 @@ public class SparseBlockCSR extends SparseBlock\n}\nsb.append(\"\\n\");\n}\n+ }\nreturn sb.toString();\n}\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/data/SparseBlockMCSR.java", "new_path": "src/main/java/org/apache/sysds/runtime/data/SparseBlockMCSR.java", "diff": "@@ -420,6 +420,8 @@ public class SparseBlockMCSR extends SparseBlock\nsb.append(size());\nsb.append(\"\\n\");\nfor( int i=0; i<numRows(); i++ ) {\n+ if(isEmpty(i))\n+ continue;\nsb.append(\"row +\");\nsb.append(i);\nsb.append(\": \");\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/test/java/org/apache/sysds/test/component/matrix/UnaryOpTest.java", "diff": "+/*\n+ * Licensed to the Apache Software Foundation (ASF) under one\n+ * or more contributor license agreements. See the NOTICE file\n+ * distributed with this work for additional information\n+ * regarding copyright ownership. The ASF licenses this file\n+ * to you under the Apache License, Version 2.0 (the\n+ * \"License\"); you may not use this file except in compliance\n+ * with the License. You may obtain a copy of the License at\n+ *\n+ * http://www.apache.org/licenses/LICENSE-2.0\n+ *\n+ * Unless required by applicable law or agreed to in writing,\n+ * software distributed under the License is distributed on an\n+ * \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+ * KIND, either express or implied. See the License for the\n+ * specific language governing permissions and limitations\n+ * under the License.\n+ */\n+package org.apache.sysds.test.component.matrix;\n+\n+import static org.junit.Assert.assertTrue;\n+\n+import org.apache.commons.logging.Log;\n+import org.apache.commons.logging.LogFactory;\n+import org.apache.sysds.runtime.data.SparseBlockCSR;\n+import org.apache.sysds.runtime.functionobjects.Builtin;\n+import org.apache.sysds.runtime.functionobjects.Builtin.BuiltinCode;\n+import org.apache.sysds.runtime.matrix.data.MatrixBlock;\n+import org.apache.sysds.runtime.matrix.operators.UnaryOperator;\n+import org.junit.Test;\n+\n+public class UnaryOpTest {\n+ protected static final Log LOG = LogFactory.getLog(UnaryOpTest.class.getName());\n+ static final MatrixBlock m = new MatrixBlock(100, 100, false);\n+ static final UnaryOperator op = new UnaryOperator(Builtin.getBuiltinFnObject(BuiltinCode.ROUND));\n+\n+ static {\n+ m.setValue(3, 3, 4.2);\n+ }\n+\n+ @Test\n+ public void testBasic() {\n+ assertTrue(m.getValue(3, 3) == 4.2);\n+ }\n+\n+ @Test\n+ public void testDirectUnaryOp() {\n+ MatrixBlock mr = m.unaryOperations(op, null);\n+ assertTrue(mr.getValue(3, 3) == 4);\n+ }\n+\n+ @Test\n+ public void testFromSparseCSRUnaryOp() {\n+ MatrixBlock sb = new MatrixBlock(1, 1, false);\n+ sb.copy(m);\n+ sb.setSparseBlock(new SparseBlockCSR(sb.getSparseBlock()));\n+ MatrixBlock mr = sb.unaryOperations(op, null);\n+ assertTrue(mr.getValue(3, 3) == 4);\n+ }\n+\n+ @Test\n+ public void testFromSparseMCSRUnaryOp() {\n+ MatrixBlock sb = new MatrixBlock(1, 1, false);\n+ sb.copy(m);\n+ MatrixBlock mr = sb.unaryOperations(op, null);\n+ assertTrue(mr.getValue(3, 3) == 4);\n+ }\n+}\n" } ]

[ { "change_type": "MODIFY", "old_path": ".github/workflows/functionsTests.yml", "new_path": ".github/workflows/functionsTests.yml", "diff": "@@ -50,7 +50,8 @@ jobs:\nmatrix:\ntests: [\n\"**.functions.a**.**,**.functions.binary.frame.**,**.functions.binary.matrix.**,**.functions.binary.scalar.**,**.functions.binary.tensor.**\",\n- \"**.functions.blocks.**,**.functions.compress.**,**.functions.countDistinct.**,**.functions.data.misc.**,**.functions.data.rand.**,**.functions.data.tensor.**,**.functions.codegenalg.parttwo.**,**.functions.codegen.**,**.functions.caching.**\",\n+ \"**.functions.blocks.**,**.functions.data.rand.**,**.functions.countDistinct.**,**.functions.data.misc.**\",\n+ \"**.functions.compress.**,,**.functions.data.tensor.**,**.functions.codegenalg.parttwo.**,**.functions.codegen.**,**.functions.caching.**\",\n\"**.functions.binary.matrix_full_cellwise.**,**.functions.binary.matrix_full_other.**\",\n\"**.functions.federated.algorithms.**\",\n\"**.functions.federated.io.**,**.functions.federated.paramserv.**,**.functions.federated.primitives.**,**.functions.federated.transform.**\",\n@@ -61,7 +62,8 @@ jobs:\n\"**.functions.dnn.**,**.functions.paramserv.**\",\n\"**.functions.misc.**,**.functions.mlcontext.**\",\n\"**.functions.nary.**,**.functions.quaternary.**\",\n- \"**.functions.parfor.**,**.functions.pipelines.**,**.functions.privacy.**,**.functions.unary.scalar.**,**.functions.updateinplace.**,**.functions.vect.**\",\n+ \"**.functions.parfor.**,**.functions.pipelines.**,**.functions.privacy.**\",\n+ \"**.functions.unary.scalar.**,**.functions.updateinplace.**,**.functions.vect.**\",\n\"**.functions.reorg.**,**.functions.rewrite.**,**.functions.ternary.**,**.functions.transform.**\",\n\"**.functions.unary.matrix.**\"\n]\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/hops/rewrite/RewriteCompressedReblock.java", "new_path": "src/main/java/org/apache/sysds/hops/rewrite/RewriteCompressedReblock.java", "diff": "@@ -128,11 +128,13 @@ public class RewriteCompressedReblock extends StatementBlockRewriteRule {\npublic static boolean satisfiesSizeConstraintsForCompression(Hop hop) {\nif(hop.getDim2() >= 1) {\n+ final long x = hop.getDim1();\n+ final long y = hop.getDim2();\nreturn\n- // If number of rows is above 1000 and either very sparse or number of columns is less than 100.\n- (hop.getDim1() >= 1000 && (hop.getDim2() < 100) || hop.getSparsity() < 0.0001)\n- // If relative ratio between number of rows and columns is better than 75, aka 75 rows per one column.\n- || hop.getDim1() / hop.getDim2() >= 75;\n+ // If the Cube of the number of rows is greater than multiplying the number of columns by 1024.\n+ y << 10 <= x * x\n+ // is very sparse and at least 100 rows.\n+ || (hop.getSparsity() < 0.0001 && y > 100);\n}\nreturn false;\n}\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/codegen/SpoofCellwise.java", "new_path": "src/main/java/org/apache/sysds/runtime/codegen/SpoofCellwise.java", "diff": "@@ -28,6 +28,7 @@ import java.util.concurrent.ExecutorService;\nimport java.util.concurrent.Future;\nimport org.apache.sysds.runtime.DMLRuntimeException;\n+import org.apache.sysds.runtime.compress.CompressedMatrixBlock;\nimport org.apache.sysds.runtime.data.DenseBlock;\nimport org.apache.sysds.runtime.data.SparseBlock;\nimport org.apache.sysds.runtime.functionobjects.Builtin;\n@@ -44,8 +45,8 @@ import org.apache.sysds.runtime.matrix.data.MatrixBlock;\nimport org.apache.sysds.runtime.util.CommonThreadPool;\nimport org.apache.sysds.runtime.util.UtilFunctions;\n-public abstract class SpoofCellwise extends SpoofOperator\n-{\n+public abstract class SpoofCellwise extends SpoofOperator {\n+\nprivate static final long serialVersionUID = 3442528770573293590L;\n// these values need to match with their native counterparts (spoof cuda ops)\n@@ -146,6 +147,9 @@ public abstract class SpoofCellwise extends SpoofOperator\n//input preparation\nMatrixBlock a = inputs.get(0);\n+ if(a instanceof CompressedMatrixBlock)\n+ a = CompressedMatrixBlock.getUncompressed(a);\n+\nSideInput[] b = prepInputMatrices(inputs);\ndouble[] scalars = prepInputScalars(scalarObjects);\nfinal int m = a.getNumRows();\n@@ -164,7 +168,7 @@ public abstract class SpoofCellwise extends SpoofOperator\ndouble ret = 0;\nif( k <= 1 ) //SINGLE-THREADED\n{\n- if( !inputs.get(0).isInSparseFormat() )\n+ if( !a.isInSparseFormat() )\nret = executeDenseAndAgg(a.getDenseBlock(), b, scalars, m, n, sparseSafe, 0, m, rix);\nelse\nret = executeSparseAndAgg(a.getSparseBlock(), b, scalars, m, n, sparseSafe, 0, m, rix);\n@@ -226,6 +230,8 @@ public abstract class SpoofCellwise extends SpoofOperator\n//input preparation\nMatrixBlock a = inputs.get(0);\n+ if(a instanceof CompressedMatrixBlock)\n+ a = CompressedMatrixBlock.getUncompressed(a);\nSideInput[] b = prepInputMatrices(inputs);\ndouble[] scalars = prepInputScalars(scalarObjects);\nfinal int m = a.getNumRows();\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/codegen/SpoofMultiAggregate.java", "new_path": "src/main/java/org/apache/sysds/runtime/codegen/SpoofMultiAggregate.java", "diff": "@@ -27,6 +27,7 @@ import java.util.concurrent.Future;\nimport org.apache.sysds.runtime.DMLRuntimeException;\nimport org.apache.sysds.runtime.codegen.SpoofCellwise.AggOp;\n+import org.apache.sysds.runtime.compress.CompressedMatrixBlock;\nimport org.apache.sysds.runtime.data.DenseBlock;\nimport org.apache.sysds.runtime.data.SparseBlock;\nimport org.apache.sysds.runtime.functionobjects.Builtin;\n@@ -101,16 +102,20 @@ public abstract class SpoofMultiAggregate extends SpoofOperator\n//input preparation\nSideInput[] b = prepInputMatrices(inputs);\ndouble[] scalars = prepInputScalars(scalarObjects);\n- final int m = inputs.get(0).getNumRows();\n- final int n = inputs.get(0).getNumColumns();\n+ MatrixBlock a = inputs.get(0);\n+ final int m = a.getNumRows();\n+ final int n = a.getNumColumns();\nboolean sparseSafe = isSparseSafe();\n+ if(a instanceof CompressedMatrixBlock)\n+ a = CompressedMatrixBlock.getUncompressed(a);\n+\nif( k <= 1 ) //SINGLE-THREADED\n{\n- if( !inputs.get(0).isInSparseFormat() )\n- executeDense(inputs.get(0).getDenseBlock(), b, scalars, c, m, n, sparseSafe, 0, m, rix);\n+ if( !a.isInSparseFormat() )\n+ executeDense(a.getDenseBlock(), b, scalars, c, m, n, sparseSafe, 0, m, rix);\nelse\n- executeSparse(inputs.get(0).getSparseBlock(), b, scalars, c, m, n, sparseSafe, 0, m, rix);\n+ executeSparse(a.getSparseBlock(), b, scalars, c, m, n, sparseSafe, 0, m, rix);\n}\nelse //MULTI-THREADED\n{\n@@ -120,7 +125,7 @@ public abstract class SpoofMultiAggregate extends SpoofOperator\nint nk = UtilFunctions.roundToNext(Math.min(8*k,m/32), k);\nint blklen = (int)(Math.ceil((double)m/nk));\nfor( int i=0; i<nk & i*blklen<m; i++ )\n- tasks.add(new ParAggTask(inputs.get(0), b, scalars,\n+ tasks.add(new ParAggTask(a, b, scalars,\nm, n, sparseSafe, i*blklen, Math.min((i+1)*blklen, m)));\n//execute tasks\nList<Future<double[]>> taskret = pool.invokeAll(tasks);\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/codegen/SpoofOperator.java", "new_path": "src/main/java/org/apache/sysds/runtime/codegen/SpoofOperator.java", "diff": "@@ -26,6 +26,7 @@ import java.util.Arrays;\nimport org.apache.commons.logging.Log;\nimport org.apache.commons.logging.LogFactory;\nimport org.apache.sysds.runtime.DMLRuntimeException;\n+import org.apache.sysds.runtime.compress.CompressedMatrixBlock;\nimport org.apache.sysds.runtime.data.DenseBlock;\nimport org.apache.sysds.runtime.data.SparseBlock;\nimport org.apache.sysds.runtime.instructions.cp.ScalarObject;\n@@ -37,7 +38,7 @@ import org.apache.sysds.runtime.util.UtilFunctions;\npublic abstract class SpoofOperator implements Serializable\n{\nprivate static final long serialVersionUID = 3834006998853573319L;\n- private static final Log LOG = LogFactory.getLog(SpoofOperator.class.getName());\n+ protected static final Log LOG = LogFactory.getLog(SpoofOperator.class.getName());\nprotected static final long PAR_NUMCELL_THRESHOLD = 1024*1024; //Min 1M elements\nprotected static final long PAR_MINFLOP_THRESHOLD = 2L*1024*1024; //MIN 2 MFLOP\n@@ -83,9 +84,11 @@ public abstract class SpoofOperator implements Serializable\nfor(int i=offset; i<offset+len; i++) {\n//transpose if necessary\nint clen = inputs.get(i).getNumColumns();\n- MatrixBlock in = (tB1 && i==1 ) ? LibMatrixReorg.transpose(inputs.get(i),\n- new MatrixBlock(clen, inputs.get(i).getNumRows(), false)) : inputs.get(i);\n-\n+ MatrixBlock inn = inputs.get(i);\n+ if(inn instanceof CompressedMatrixBlock)\n+ inn = CompressedMatrixBlock.getUncompressed(inn);\n+ MatrixBlock in = (tB1 && i==1 ) ? LibMatrixReorg.transpose(inn,\n+ new MatrixBlock(clen, inn.getNumRows(), false)) : inn;\n//create side input\nif( denseOnly && (in.isInSparseFormat() || !in.isAllocated()) ) {\n//convert empty or sparse to dense temporary block (note: we don't do\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/codegen/SpoofOuterProduct.java", "new_path": "src/main/java/org/apache/sysds/runtime/codegen/SpoofOuterProduct.java", "diff": "@@ -28,6 +28,7 @@ import java.util.concurrent.Future;\nimport org.apache.sysds.hops.OptimizerUtils;\nimport org.apache.sysds.runtime.DMLRuntimeException;\n+import org.apache.sysds.runtime.compress.CompressedMatrixBlock;\nimport org.apache.sysds.runtime.data.DenseBlock;\nimport org.apache.sysds.runtime.data.SparseBlock;\nimport org.apache.sysds.runtime.instructions.cp.DoubleObject;\n@@ -96,6 +97,9 @@ public abstract class SpoofOuterProduct extends SpoofOperator\nMatrixBlock out = new MatrixBlock(1, 1, false);\nout.allocateDenseBlock();\n+ if(a instanceof CompressedMatrixBlock)\n+ a = CompressedMatrixBlock.getUncompressed(a);\n+\nif( !a.isInSparseFormat() )\nexecuteCellwiseDense(a.getDenseBlock(), ab[0], ab[1], b, scalars, out.getDenseBlock(), m, n, k, _outerProductType, 0, m, 0, n);\nelse\n@@ -474,7 +478,7 @@ public abstract class SpoofOuterProduct extends SpoofOperator\n//NOTE: we don't create sparse side inputs w/ row-major cursors because\n//cache blocking would lead to non-sequential access\n- final int blocksizeIJ = (int) (8L*m*n/nnz);\n+ final int blocksizeIJ = (int) (8L*m*n/Math.max(nnz,1));\nint[] curk = new int[Math.min(blocksizeIJ, ru-rl)];\nif( !out.isInSparseFormat() ) //DENSE\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/codegen/SpoofRowwise.java", "new_path": "src/main/java/org/apache/sysds/runtime/codegen/SpoofRowwise.java", "diff": "@@ -28,6 +28,7 @@ import java.util.concurrent.Future;\nimport java.util.stream.IntStream;\nimport org.apache.sysds.runtime.DMLRuntimeException;\n+import org.apache.sysds.runtime.compress.CompressedMatrixBlock;\nimport org.apache.sysds.runtime.controlprogram.caching.MatrixObject;\nimport org.apache.sysds.runtime.data.DenseBlock;\nimport org.apache.sysds.runtime.data.DenseBlockFactory;\n@@ -175,6 +176,9 @@ public abstract class SpoofRowwise extends SpoofOperator\n//core sequential execute\nMatrixBlock a = inputs.get(0);\n+ if(a instanceof CompressedMatrixBlock)\n+ a = CompressedMatrixBlock.getUncompressed(a);\n+\nif( !a.isInSparseFormat() )\nexecuteDense(a.getDenseBlock(), b, scalars, c, n, 0, m, rix);\nelse\n" }, { "change_type": "RENAME", "old_path": "src/test/java/org/apache/sysds/test/functions/codegen/SparseSideInputTest.java", "new_path": "src/test/java/org/apache/sysds/test/functions/compress/codegen/SparseSideInputTest.java", "diff": "* under the License.\n*/\n-package org.apache.sysds.test.functions.codegen;\n+package org.apache.sysds.test.functions.compress.codegen;\n+\n+import static org.junit.Assert.assertTrue;\nimport java.io.File;\nimport java.util.HashMap;\n@@ -25,17 +27,15 @@ import java.util.HashMap;\nimport org.apache.commons.logging.Log;\nimport org.apache.commons.logging.LogFactory;\nimport org.apache.sysds.common.Types.ExecMode;\n-import org.apache.sysds.hops.OptimizerUtils;\nimport org.apache.sysds.common.Types.ExecType;\n+import org.apache.sysds.hops.OptimizerUtils;\nimport org.apache.sysds.runtime.matrix.data.MatrixValue.CellIndex;\nimport org.apache.sysds.test.AutomatedTestBase;\nimport org.apache.sysds.test.TestConfiguration;\nimport org.apache.sysds.test.TestUtils;\n-import org.junit.Assert;\nimport org.junit.Test;\n-public class SparseSideInputTest extends AutomatedTestBase\n-{\n+public class SparseSideInputTest extends AutomatedTestBase {\nprivate static final Log LOG = LogFactory.getLog(SparseSideInputTest.class.getName());\nprivate static final String TEST_NAME = \"SparseSideInput\";\n@@ -59,7 +59,8 @@ public class SparseSideInputTest extends AutomatedTestBase\npublic void setUp() {\nTestUtils.clearAssertionInformation();\nfor(int i = 1; i <= 4; i++)\n- addTestConfiguration( TEST_NAME+i, new TestConfiguration(TEST_CLASS_DIR, TEST_NAME+i, new String[] { String.valueOf(i) }) );\n+ addTestConfiguration(TEST_NAME + i,\n+ new TestConfiguration(TEST_CLASS_DIR, TEST_NAME + i, new String[] {String.valueOf(i)}));\n}\n@Test\n@@ -142,13 +143,11 @@ public class SparseSideInputTest extends AutomatedTestBase\ntestCodegenIntegration(TEST_NAME4, true, ExecType.SPARK);\n}\n- private void testCodegenIntegration( String testname, boolean compress, ExecType instType )\n- {\n+ private void testCodegenIntegration(String testname, boolean compress, ExecType instType) {\nboolean oldFlag = OptimizerUtils.ALLOW_ALGEBRAIC_SIMPLIFICATION;\nExecMode platformOld = setExecMode(instType);\n- try\n- {\n+ try {\nTEST_CONF = compress ? TEST_CONF2 : TEST_CONF1;\nOptimizerUtils.ALLOW_ALGEBRAIC_SIMPLIFICATION = false;\n@@ -157,8 +156,7 @@ public class SparseSideInputTest extends AutomatedTestBase\nString HOME = SCRIPT_DIR + TEST_DIR;\nfullDMLScriptName = HOME + testname + \".dml\";\n- programArgs = new String[]{\"-stats\",\"-explain\", \"-args\",\n- input(\"X\"), input(\"Y\"), output(\"R\") };\n+ programArgs = new String[] {\"-stats\", \"-explain\", \"-args\", input(\"X\"), input(\"Y\"), output(\"R\")};\nfullRScriptName = HOME + testname + \".R\";\nrCmd = getRCmd(inputDir(), expectedDir());\n@@ -170,16 +168,18 @@ public class SparseSideInputTest extends AutomatedTestBase\nwriteInputMatrixWithMTD(\"Y\", Y, true);\n// run dml and r scripts\n- LOG.debug(fullDMLScriptName);\n- LOG.debug(runTest(true, false, null, -1));\n+ String ret = runTest(null).toString();\nrunRScript(true);\n+ LOG.debug(ret);\n+\n// compare matrices\n- HashMap<CellIndex, Double> dmlfile = readDMLMatrixFromOutputDir(\"R\");\n- HashMap<CellIndex, Double> rfile = readRMatrixFromExpectedDir(\"R\");\n- TestUtils.compareMatrices(dmlfile, rfile, eps, \"Stat-DML\", \"Stat-R\");\n- Assert.assertTrue(heavyHittersContainsSubString(\"spoof\")\n- || heavyHittersContainsSubString(\"sp_spoof\"));\n+ HashMap<CellIndex, Double> dmlResult = readDMLMatrixFromOutputDir(\"R\");\n+ HashMap<CellIndex, Double> rResult = readRMatrixFromExpectedDir(\"R\");\n+ TestUtils.compareMatrices(dmlResult, rResult, eps, \"Stat-DML\", \"Stat-R\");\n+ assertTrue(heavyHittersContainsSubString(\"spoof\") || heavyHittersContainsSubString(\"sp_spoof\"));\n+ if(compress)\n+ assertTrue(heavyHittersContainsSubString(\"compress\") || heavyHittersContainsSubString(\"sp_compress\"));\n}\nfinally {\nresetExecMode(platformOld);\n@@ -193,7 +193,6 @@ public class SparseSideInputTest extends AutomatedTestBase\nprotected File getConfigTemplateFile() {\n// Instrumentation in this test's output log to show custom configuration file used for template.\nFile f = new File(SCRIPT_DIR + TEST_DIR, TEST_CONF);\n- LOG.info(\"This test case overrides default configuration with \" + f.getPath());\nreturn f;\n}\n}\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/python/docs/source/guide/python_end_to_end_tut.rst", "new_path": "src/main/python/docs/source/guide/python_end_to_end_tut.rst", "diff": "@@ -177,7 +177,7 @@ adding column names at the top of the files such that the first line looks like:\nage,workclass,fnlwgt,education,education-num,marital-status,occupation,relationship,race,sex,capital-gain,capital-loss,hours-per-week,native-country,income\n-We also delete the line holding the string value |1x3 Cross validator inside the test dataset.\n+We also delete the line holding the string value Cross validator inside the test dataset.\nAfter these modifications, we have to define a mtd file for each file we want to read. This mtd file has to be in the same directory as the dataset.\nIn this particular example, the dataset is split into two files \"train_data.csv\" and \"test_data.csv\". We want to read both, which means that we will define a mtd file for\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/docs/source/index.rst", "new_path": "src/main/python/docs/source/index.rst", "diff": "@@ -53,6 +53,7 @@ tensors (multi-dimensional arrays) whose first dimension may have a heterogeneou\nguide/federated.rst\nguide/algorithms_basics.rst\n+ guide/python_end_to_end_tut.rst\n.. toctree::\n:maxdepth: 1\n" } ]

[ { "change_type": "MODIFY", "old_path": "docker/pythonsysds.Dockerfile", "new_path": "docker/pythonsysds.Dockerfile", "diff": "@@ -50,7 +50,7 @@ RUN apt-get update -qq \\\n&& mv apache-maven-$MAVEN_VERSION /usr/lib/mvn \\\n&& git clone --depth 1 https://github.com/apache/systemds.git systemds && \\\ncd /usr/src/systemds/ && \\\n- mvn clean package -P distribution && \\\n+ mvn -ntp clean package -P distribution && \\\ncd /usr/src/systemds/src/main/python && \\\napt-get install -y --no-install-recommends \\\npython3 python3-pip && \\\n" }, { "change_type": "MODIFY", "old_path": "docker/sysds.Dockerfile", "new_path": "docker/sysds.Dockerfile", "diff": "@@ -50,7 +50,7 @@ RUN apt-get update -qq \\\n&& mv apache-maven-$MAVEN_VERSION /usr/lib/mvn \\\n&& git clone --depth 1 https://github.com/apache/systemds.git systemds && \\\ncd /usr/src/systemds/ && \\\n- mvn clean package -P distribution && \\\n+ mvn -ntp clean package -P distribution && \\\nrm -r .git && \\\nrm -r .github && \\\nrm -r target/javadoc** && \\\n" } ]

[ { "change_type": "MODIFY", "old_path": "docker/pythonsysds.Dockerfile", "new_path": "docker/pythonsysds.Dockerfile", "diff": "@@ -50,7 +50,7 @@ RUN apt-get update -qq \\\n&& mv apache-maven-$MAVEN_VERSION /usr/lib/mvn \\\n&& git clone --depth 1 https://github.com/apache/systemds.git systemds && \\\ncd /usr/src/systemds/ && \\\n- mvn -ntp clean package -P distribution && \\\n+ mvn clean package -P distribution && \\\ncd /usr/src/systemds/src/main/python && \\\napt-get install -y --no-install-recommends \\\npython3 python3-pip && \\\n" }, { "change_type": "MODIFY", "old_path": "docker/sysds.Dockerfile", "new_path": "docker/sysds.Dockerfile", "diff": "@@ -50,7 +50,7 @@ RUN apt-get update -qq \\\n&& mv apache-maven-$MAVEN_VERSION /usr/lib/mvn \\\n&& git clone --depth 1 https://github.com/apache/systemds.git systemds && \\\ncd /usr/src/systemds/ && \\\n- mvn -ntp clean package -P distribution && \\\n+ mvn clean package -P distribution && \\\nrm -r .git && \\\nrm -r .github && \\\nrm -r target/javadoc** && \\\n" } ]

[ { "change_type": "ADD", "old_path": null, "new_path": "src/test/java/org/apache/sysds/test/gpu/BuiltinUnaryGPUInstructionTest.java", "diff": "+/*\n+ * Licensed to the Apache Software Foundation (ASF) under one\n+ * or more contributor license agreements. See the NOTICE file\n+ * distributed with this work for additional information\n+ * regarding copyright ownership. The ASF licenses this file\n+ * to you under the Apache License, Version 2.0 (the\n+ * \"License\"); you may not use this file except in compliance\n+ * with the License. You may obtain a copy of the License at\n+ *\n+ * http://www.apache.org/licenses/LICENSE-2.0\n+ *\n+ * Unless required by applicable law or agreed to in writing,\n+ * software distributed under the License is distributed on an\n+ * \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+ * KIND, either express or implied. See the License for the\n+ * specific language governing permissions and limitations\n+ * under the License.\n+ */\n+\n+package org.apache.sysds.test.gpu;\n+\n+import org.apache.sysds.test.AutomatedTestBase;\n+import org.apache.sysds.test.functions.builtin.BuiltinSigmoidTest;\n+import org.apache.sysds.test.functions.unary.matrix.ACosTest;\n+import org.apache.sysds.test.functions.unary.matrix.ASinTest;\n+import org.apache.sysds.test.functions.unary.matrix.ATanTest;\n+import org.apache.sysds.test.functions.unary.matrix.AbsTest;\n+import org.apache.sysds.test.functions.unary.matrix.CosTest;\n+import org.apache.sysds.test.functions.unary.matrix.FullCummaxTest;\n+import org.apache.sysds.test.functions.unary.matrix.FullCumminTest;\n+import org.apache.sysds.test.functions.unary.matrix.FullCumprodTest;\n+import org.apache.sysds.test.functions.unary.matrix.FullCumsumTest;\n+import org.apache.sysds.test.functions.unary.matrix.FullCumsumprodTest;\n+import org.apache.sysds.test.functions.unary.matrix.FullSignTest;\n+import org.apache.sysds.test.functions.unary.matrix.RoundTest;\n+import org.apache.sysds.test.functions.unary.matrix.SinTest;\n+import org.apache.sysds.test.functions.unary.matrix.SqrtTest;\n+import org.apache.sysds.test.functions.unary.matrix.TanTest;\n+import org.junit.Assert;\n+import org.junit.Test;\n+\n+public class BuiltinUnaryGPUInstructionTest extends AutomatedTestBase {\n+ @Override public void setUp() {\n+ TEST_GPU = true;\n+ VERBOSE_STATS = true;\n+ }\n+\n+ // ToDo:\n+ // @Test public void ExponentTest() {}\n+ // @Test public void LogarithmTest() {}\n+ // @Test public void SoftmaxTest() {}\n+ // @Test public void CoshTest() {}\n+ // @Test public void SinhTest() {}\n+ // @Test public void TanhTest() {}\n+\n+ @Test public void AbsTest() {\n+ AbsTest dmlTestCase = new org.apache.sysds.test.functions.unary.matrix.AbsTest();\n+ dmlTestCase.setUpBase();\n+ dmlTestCase.setUp();\n+ dmlTestCase.testPositive();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_abs\"));\n+ dmlTestCase.testNegative();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_abs\"));\n+ dmlTestCase.testRandom();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_abs\"));\n+ }\n+\n+ @Test public void ACosTest() {\n+ ACosTest dmlTestCase = new org.apache.sysds.test.functions.unary.matrix.ACosTest();\n+ dmlTestCase.setUpBase();\n+ dmlTestCase.setUp();\n+ dmlTestCase.testPositive();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_acos\"));\n+ dmlTestCase.testNegative();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_acos\"));\n+ dmlTestCase.testRandom();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_acos\"));\n+ }\n+\n+ @Test public void ASinTest() {\n+ ASinTest dmlTestCase = new org.apache.sysds.test.functions.unary.matrix.ASinTest();\n+ dmlTestCase.setUpBase();\n+ dmlTestCase.setUp();\n+ dmlTestCase.testPositive();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_asin\"));\n+ dmlTestCase.testNegative();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_asin\"));\n+ dmlTestCase.testRandom();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_asin\"));\n+ }\n+\n+ @Test public void ATanTest() {\n+ ATanTest dmlTestCase = new org.apache.sysds.test.functions.unary.matrix.ATanTest();\n+ dmlTestCase.setUpBase();\n+ dmlTestCase.setUp();\n+ dmlTestCase.testPositive();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_atan\"));\n+ dmlTestCase.testNegative();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_atan\"));\n+ dmlTestCase.testRandom();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_atan\"));\n+ }\n+\n+ @Test public void CeilTest() {\n+ RoundTest dmlTestCase = new org.apache.sysds.test.functions.unary.matrix.RoundTest();\n+ dmlTestCase.setUpBase();\n+ dmlTestCase.setUp();\n+ dmlTestCase.testCeil1();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ceil\"));\n+ dmlTestCase.testCeil2();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ceil\"));\n+ dmlTestCase.testCeil3();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ceil\"));\n+ dmlTestCase.testCeil4();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ceil\"));\n+ dmlTestCase.testCeil5();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ceil\"));\n+ dmlTestCase.testCeil6();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ceil\"));\n+ }\n+\n+ @Test public void CosTest() {\n+ CosTest dmlTestCase = new org.apache.sysds.test.functions.unary.matrix.CosTest();\n+ dmlTestCase.setUpBase();\n+ dmlTestCase.setUp();\n+ dmlTestCase.testPositive();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_cos\"));\n+ dmlTestCase.testNegative();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_cos\"));\n+ dmlTestCase.testRandom();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_cos\"));\n+ }\n+\n+ @Test public void CummaxTest() {\n+ FullCummaxTest dmlTestCase = new org.apache.sysds.test.functions.unary.matrix.FullCummaxTest();\n+ dmlTestCase.setUpBase();\n+ dmlTestCase.setUp();\n+ dmlTestCase.testCummaxColVectorDenseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucummax\"));\n+ dmlTestCase.testCummaxColVectorSparseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucummax\"));\n+ dmlTestCase.testCummaxMatrixDenseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucummax\"));\n+ dmlTestCase.testCummaxMatrixSparseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucummax\"));\n+ dmlTestCase.testCummaxRowVectorDenseCP();\n+ Assert.assertFalse(heavyHittersContainsSubString(\"gpu_ucummax\"));\n+ dmlTestCase.testCummaxRowVectorDenseNoRewritesCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucummax\"));\n+ dmlTestCase.testCummaxRowVectorSparseCP();\n+ Assert.assertFalse(heavyHittersContainsSubString(\"gpu_ucummax\"));\n+ dmlTestCase.testCummaxRowVectorSparseNoRewritesCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucummax\"));\n+ }\n+\n+ @Test public void CumminTest() {\n+ FullCumminTest dmlTestCase = new org.apache.sysds.test.functions.unary.matrix.FullCumminTest();\n+ dmlTestCase.setUpBase();\n+ dmlTestCase.setUp();\n+ dmlTestCase.testCumminColVectorDenseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucummin\"));\n+ dmlTestCase.testCumminColVectorSparseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucummin\"));\n+ dmlTestCase.testCumminMatrixDenseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucummin\"));\n+ dmlTestCase.testCumminMatrixSparseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucummin\"));\n+ dmlTestCase.testCumminRowVectorDenseCP();\n+ Assert.assertFalse(heavyHittersContainsSubString(\"gpu_ucummin\"));\n+ dmlTestCase.testCumminRowVectorDenseNoRewritesCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucummin\"));\n+ dmlTestCase.testCumminRowVectorSparseCP();\n+ Assert.assertFalse(heavyHittersContainsSubString(\"gpu_ucummin\"));\n+ dmlTestCase.testCumminRowVectorSparseNoRewritesCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucummin\"));\n+ }\n+\n+ @Test public void CumprodTest() {\n+ FullCumprodTest dmlTestCase = new org.apache.sysds.test.functions.unary.matrix.FullCumprodTest();\n+ dmlTestCase.setUpBase();\n+ dmlTestCase.setUp();\n+ dmlTestCase.testCumprodColVectorDenseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucum*\"));\n+ dmlTestCase.testCumprodColVectorSparseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucum*\"));\n+ dmlTestCase.testCumprodMatrixDenseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucum*\"));\n+ dmlTestCase.testCumprodMatrixSparseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucum*\"));\n+ dmlTestCase.testCumprodRowVectorDenseCP();\n+ Assert.assertFalse(heavyHittersContainsSubString(\"gpu_ucum*\"));\n+ dmlTestCase.testCumprodRowVectorDenseNoRewritesCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucum*\"));\n+ dmlTestCase.testCumprodRowVectorSparseCP();\n+ Assert.assertFalse(heavyHittersContainsSubString(\"gpu_ucum*\"));\n+ dmlTestCase.testCumprodRowVectorSparseNoRewritesCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucum*\"));\n+ }\n+\n+ @Test public void CumsumTest() {\n+ FullCumsumTest dmlTestCase = new org.apache.sysds.test.functions.unary.matrix.FullCumsumTest();\n+ dmlTestCase.setUpBase();\n+ dmlTestCase.setUp();\n+ dmlTestCase.testCumsumColVectorDenseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucumk+\"));\n+ dmlTestCase.testCumsumColVectorSparseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucumk+\"));\n+ dmlTestCase.testCumsumMatrixDenseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucumk+\"));\n+ dmlTestCase.testCumsumMatrixSparseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucumk+\"));\n+ dmlTestCase.testCumsumRowVectorDenseCP();\n+ Assert.assertFalse(heavyHittersContainsSubString(\"gpu_ucumk+\"));\n+ dmlTestCase.testCumsumRowVectorDenseNoRewritesCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucumk+\"));\n+ dmlTestCase.testCumsumRowVectorSparseCP();\n+ Assert.assertFalse(heavyHittersContainsSubString(\"gpu_ucumk+\"));\n+ dmlTestCase.testCumsumRowVectorSparseNoRewritesCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucumk+\"));\n+ }\n+\n+ @Test public void CumsumprodTest() {\n+ FullCumsumprodTest dmlTestCase = new org.apache.sysds.test.functions.unary.matrix.FullCumsumprodTest();\n+ dmlTestCase.setUpBase();\n+ dmlTestCase.setUp();\n+ dmlTestCase.testCumsumprodBackwardDenseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucumk+*\"));\n+ dmlTestCase.testCumsumprodBackwardSparseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucumk+*\"));\n+ dmlTestCase.testCumsumprodForwardDenseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucumk+*\"));\n+ dmlTestCase.testCumsumprodForwardSparseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_ucumk+*\"));\n+ }\n+\n+ @Test public void FloorTest() {\n+ RoundTest dmlTestCase = new org.apache.sysds.test.functions.unary.matrix.RoundTest();\n+ dmlTestCase.setUpBase();\n+ dmlTestCase.setUp();\n+ dmlTestCase.testFloor1();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_floor\"));\n+ dmlTestCase.testFloor2();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_floor\"));\n+ dmlTestCase.testFloor3();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_floor\"));\n+ dmlTestCase.testFloor4();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_floor\"));\n+ dmlTestCase.testFloor5();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_floor\"));\n+ dmlTestCase.testFloor6();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_floor\"));\n+ }\n+\n+ @Test public void RoundTest() {\n+ RoundTest dmlTestCase = new org.apache.sysds.test.functions.unary.matrix.RoundTest();\n+ dmlTestCase.setUpBase();\n+ dmlTestCase.setUp();\n+ dmlTestCase.testRound1();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_round\"));\n+ dmlTestCase.testRound2();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_round\"));\n+ dmlTestCase.testRound3();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_round\"));\n+ dmlTestCase.testRound4();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_round\"));\n+ dmlTestCase.testRound5();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_round\"));\n+ dmlTestCase.testRound6();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_round\"));\n+ }\n+\n+ @Test public void SinTest() {\n+ SinTest dmlTestCase = new org.apache.sysds.test.functions.unary.matrix.SinTest();\n+ dmlTestCase.setUpBase();\n+ dmlTestCase.setUp();\n+ dmlTestCase.testPositive();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_sin\"));\n+ dmlTestCase.testNegative();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_sin\"));\n+ dmlTestCase.testRandom();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_sin\"));\n+ }\n+\n+ @Test public void SqrtTest() {\n+ SqrtTest dmlTestCase = new org.apache.sysds.test.functions.unary.matrix.SqrtTest();\n+ dmlTestCase.setUpBase();\n+ dmlTestCase.setUp();\n+ dmlTestCase.testPositive();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_sqrt\"));\n+ dmlTestCase.testNegativeMatrix();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_sqrt\"));\n+ dmlTestCase.testNegativeVector();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_sqrt\"));\n+ }\n+\n+ @Test public void SignTest() {\n+ FullSignTest dmlTestCase = new org.apache.sysds.test.functions.unary.matrix.FullSignTest();\n+ dmlTestCase.setUpBase();\n+ dmlTestCase.setUp();\n+ dmlTestCase.testRewriteSignDenseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_sign\"));\n+ dmlTestCase.testRewriteSignSparseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_sign\"));\n+ dmlTestCase.testSignDenseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_sign\"));\n+ dmlTestCase.testSignSparseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_sign\"));\n+ }\n+\n+ @Test public void SigmoidTest() {\n+ BuiltinSigmoidTest dmlTestCase = new org.apache.sysds.test.functions.builtin.BuiltinSigmoidTest();\n+ dmlTestCase.setUpBase();\n+ dmlTestCase.setUp();\n+ dmlTestCase.testSigmoidMatrixDenseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_sigmoid\"));\n+ dmlTestCase.testSigmoidMatrixSparseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_sigmoid\"));\n+ dmlTestCase.testSigmoidScalarDenseCP();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_sigmoid\"));\n+ }\n+\n+ @Test public void TanTest() {\n+ TanTest dmlTestCase = new org.apache.sysds.test.functions.unary.matrix.TanTest();\n+ dmlTestCase.setUpBase();\n+ dmlTestCase.setUp();\n+ dmlTestCase.testPositive();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_tan\"));\n+ dmlTestCase.testNegative();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_tan\"));\n+ dmlTestCase.testRandom();\n+ Assert.assertTrue(heavyHittersContainsSubString(\"gpu_tan\"));\n+ }\n+}\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/test/java/org/apache/sysds/test/gpu/codegen/CellwiseTmplTest.java", "diff": "+/*\n+ * Licensed to the Apache Software Foundation (ASF) under one\n+ * or more contributor license agreements. See the NOTICE file\n+ * distributed with this work for additional information\n+ * regarding copyright ownership. The ASF licenses this file\n+ * to you under the Apache License, Version 2.0 (the\n+ * \"License\"); you may not use this file except in compliance\n+ * with the License. You may obtain a copy of the License at\n+ *\n+ * http://www.apache.org/licenses/LICENSE-2.0\n+ *\n+ * Unless required by applicable law or agreed to in writing,\n+ * software distributed under the License is distributed on an\n+ * \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+ * KIND, either express or implied. See the License for the\n+ * specific language governing permissions and limitations\n+ * under the License.\n+ */\n+\n+package org.apache.sysds.test.gpu.codegen;\n+\n+import org.apache.sysds.test.AutomatedTestBase;\n+import org.junit.Test;\n+\n+public class CellwiseTmplTest extends AutomatedTestBase {\n+ org.apache.sysds.test.functions.codegen.CellwiseTmplTest dmlTestCase;\n+\n+ @Override public void setUp() {\n+ TEST_GPU = true;\n+ dmlTestCase = new org.apache.sysds.test.functions.codegen.CellwiseTmplTest();\n+ dmlTestCase.setUpBase();\n+ dmlTestCase.setUp();\n+ }\n+\n+ @Test public void testCodegenCellwise1() { dmlTestCase.testCodegenCellwise1(); }\n+ @Test public void testCodegenCellwise2() { dmlTestCase.testCodegenCellwise2(); }\n+ @Test public void testCodegenCellwise3() { dmlTestCase.testCodegenCellwise3(); }\n+ @Test public void testCodegenCellwise4() { dmlTestCase.testCodegenCellwise4(); }\n+ @Test public void testCodegenCellwise5() { dmlTestCase.testCodegenCellwise5(); }\n+ @Test public void testCodegenCellwise6() { dmlTestCase.testCodegenCellwise6(); }\n+ @Test public void testCodegenCellwise7() { dmlTestCase.testCodegenCellwise7(); }\n+ @Test public void testCodegenCellwise8() { dmlTestCase.testCodegenCellwise8(); }\n+ @Test public void testCodegenCellwise9() { dmlTestCase.testCodegenCellwise9(); }\n+ @Test public void testCodegenCellwise10() { dmlTestCase.testCodegenCellwise10(); }\n+ @Test public void testCodegenCellwise11() { dmlTestCase.testCodegenCellwise11(); }\n+ @Test public void testCodegenCellwise12() { dmlTestCase.testCodegenCellwise12(); }\n+ @Test public void testCodegenCellwise13() { dmlTestCase.testCodegenCellwise13(); }\n+ @Test public void testCodegenCellwise14() { dmlTestCase.testCodegenCellwise14(); }\n+ @Test public void testCodegenCellwise15() { dmlTestCase.testCodegenCellwise15(); }\n+ @Test public void testCodegenCellwise16() { dmlTestCase.testCodegenCellwise16(); }\n+ @Test public void testCodegenCellwise17() { dmlTestCase.testCodegenCellwise17(); }\n+ @Test public void testCodegenCellwise18() { dmlTestCase.testCodegenCellwise18(); }\n+ @Test public void testCodegenCellwise19() { dmlTestCase.testCodegenCellwise19(); }\n+ @Test public void testCodegenCellwise20() { dmlTestCase.testCodegenCellwise20(); }\n+ @Test public void testCodegenCellwise21() { dmlTestCase.testCodegenCellwise21(); }\n+ @Test public void testCodegenCellwise22() { dmlTestCase.testCodegenCellwise22(); }\n+ @Test public void testCodegenCellwise23() { dmlTestCase.testCodegenCellwise23(); }\n+ @Test public void testCodegenCellwise24() { dmlTestCase.testCodegenCellwise24(); }\n+ @Test public void testCodegenCellwise25() { dmlTestCase.testCodegenCellwise25(); }\n+ @Test public void testCodegenCellwise26() { dmlTestCase.testCodegenCellwise26(); }\n+ @Test public void testCodegenCellwise27() { dmlTestCase.testCodegenCellwise27(); }\n+\n+ @Test public void testCodegenCellwiseRewrite1() { dmlTestCase.testCodegenCellwiseRewrite1(); }\n+ @Test public void testCodegenCellwiseRewrite2() { dmlTestCase.testCodegenCellwiseRewrite2(); }\n+ @Test public void testCodegenCellwiseRewrite3() { dmlTestCase.testCodegenCellwiseRewrite3(); }\n+ @Test public void testCodegenCellwiseRewrite4() { dmlTestCase.testCodegenCellwiseRewrite4(); }\n+ @Test public void testCodegenCellwiseRewrite5() { dmlTestCase.testCodegenCellwiseRewrite5(); }\n+ @Test public void testCodegenCellwiseRewrite6() { dmlTestCase.testCodegenCellwiseRewrite6(); }\n+ @Test public void testCodegenCellwiseRewrite7() { dmlTestCase.testCodegenCellwiseRewrite7(); }\n+ @Test public void testCodegenCellwiseRewrite8() { dmlTestCase.testCodegenCellwiseRewrite8(); }\n+ @Test public void testCodegenCellwiseRewrite9() { dmlTestCase.testCodegenCellwiseRewrite9(); }\n+ @Test public void testCodegenCellwiseRewrite10() { dmlTestCase.testCodegenCellwiseRewrite10(); }\n+ @Test public void testCodegenCellwiseRewrite11() { dmlTestCase.testCodegenCellwiseRewrite11(); }\n+ @Test public void testCodegenCellwiseRewrite12() { dmlTestCase.testCodegenCellwiseRewrite12(); }\n+ @Test public void testCodegenCellwiseRewrite13() { dmlTestCase.testCodegenCellwiseRewrite13(); }\n+ @Test public void testCodegenCellwiseRewrite14() { dmlTestCase.testCodegenCellwiseRewrite14(); }\n+ @Test public void testCodegenCellwiseRewrite15() { dmlTestCase.testCodegenCellwiseRewrite15(); }\n+ @Test public void testCodegenCellwiseRewrite16() { dmlTestCase.testCodegenCellwiseRewrite16(); }\n+ @Test public void testCodegenCellwiseRewrite17() { dmlTestCase.testCodegenCellwiseRewrite17(); }\n+ @Test public void testCodegenCellwiseRewrite18() { dmlTestCase.testCodegenCellwiseRewrite18(); }\n+ @Test public void testCodegenCellwiseRewrite19() { dmlTestCase.testCodegenCellwiseRewrite19(); }\n+ @Test public void testCodegenCellwiseRewrite20() { dmlTestCase.testCodegenCellwiseRewrite20(); }\n+ @Test public void testCodegenCellwiseRewrite21() { dmlTestCase.testCodegenCellwiseRewrite21(); }\n+ @Test public void testCodegenCellwiseRewrite22() { dmlTestCase.testCodegenCellwiseRewrite22(); }\n+ @Test public void testCodegenCellwiseRewrite23() { dmlTestCase.testCodegenCellwiseRewrite23(); }\n+ @Test public void testCodegenCellwiseRewrite24() { dmlTestCase.testCodegenCellwiseRewrite24(); }\n+ @Test public void testCodegenCellwiseRewrite25() { dmlTestCase.testCodegenCellwiseRewrite25(); }\n+ @Test public void testCodegenCellwiseRewrite26() { dmlTestCase.testCodegenCellwiseRewrite26(); }\n+ @Test public void testCodegenCellwiseRewrite27() { dmlTestCase.testCodegenCellwiseRewrite27(); }\n+}\n" }, { "change_type": "ADD", "old_path": null, "new_path": "src/test/java/org/apache/sysds/test/gpu/codegen/RowAggTmplTest.java", "diff": "+/*\n+ * Licensed to the Apache Software Foundation (ASF) under one\n+ * or more contributor license agreements. See the NOTICE file\n+ * distributed with this work for additional information\n+ * regarding copyright ownership. The ASF licenses this file\n+ * to you under the Apache License, Version 2.0 (the\n+ * \"License\"); you may not use this file except in compliance\n+ * with the License. You may obtain a copy of the License at\n+ *\n+ * http://www.apache.org/licenses/LICENSE-2.0\n+ *\n+ * Unless required by applicable law or agreed to in writing,\n+ * software distributed under the License is distributed on an\n+ * \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+ * KIND, either express or implied. See the License for the\n+ * specific language governing permissions and limitations\n+ * under the License.\n+ */\n+\n+package org.apache.sysds.test.gpu.codegen;\n+\n+import org.apache.sysds.test.AutomatedTestBase;\n+import org.junit.Test;\n+\n+public class RowAggTmplTest extends AutomatedTestBase {\n+ org.apache.sysds.test.functions.codegen.RowAggTmplTest dmlTestCase;\n+\n+ @Override public void setUp() {\n+ TEST_GPU = true;\n+ dmlTestCase = new org.apache.sysds.test.functions.codegen.RowAggTmplTest();\n+ dmlTestCase.setUpBase();\n+ dmlTestCase.setUp();\n+ }\n+\n+ @Test public void testCodegenRowAgg1CP() { dmlTestCase.testCodegenRowAgg1CP(); }\n+ @Test public void testCodegenRowAgg2CP() { dmlTestCase.testCodegenRowAgg2CP(); }\n+ @Test public void testCodegenRowAgg3CP() { dmlTestCase.testCodegenRowAgg3CP(); }\n+ @Test public void testCodegenRowAgg4CP() { dmlTestCase.testCodegenRowAgg4CP(); }\n+ @Test public void testCodegenRowAgg5CP() { dmlTestCase.testCodegenRowAgg5CP(); }\n+ @Test public void testCodegenRowAgg6CP() { dmlTestCase.testCodegenRowAgg6CP(); }\n+ @Test public void testCodegenRowAgg7CP() { dmlTestCase.testCodegenRowAgg7CP(); }\n+ @Test public void testCodegenRowAgg8CP() { dmlTestCase.testCodegenRowAgg8CP(); }\n+ @Test public void testCodegenRowAgg9CP() { dmlTestCase.testCodegenRowAgg9CP(); }\n+ @Test public void testCodegenRowAgg10CP() { dmlTestCase.testCodegenRowAgg10CP(); }\n+ @Test public void testCodegenRowAgg11CP() { dmlTestCase.testCodegenRowAgg11CP(); }\n+ @Test public void testCodegenRowAgg12CP() { dmlTestCase.testCodegenRowAgg12CP(); }\n+ @Test public void testCodegenRowAgg13CP() { dmlTestCase.testCodegenRowAgg13CP(); }\n+ @Test public void testCodegenRowAgg14CP() { dmlTestCase.testCodegenRowAgg14CP(); }\n+ @Test public void testCodegenRowAgg15CP() { dmlTestCase.testCodegenRowAgg15CP(); }\n+ @Test public void testCodegenRowAgg16CP() { dmlTestCase.testCodegenRowAgg16CP(); }\n+ @Test public void testCodegenRowAgg17CP() { dmlTestCase.testCodegenRowAgg17CP(); }\n+ @Test public void testCodegenRowAgg18CP() { dmlTestCase.testCodegenRowAgg18CP(); }\n+ @Test public void testCodegenRowAgg19CP() { dmlTestCase.testCodegenRowAgg19CP(); }\n+ @Test public void testCodegenRowAgg20CP() { dmlTestCase.testCodegenRowAgg20CP(); }\n+ @Test public void testCodegenRowAgg21CP() { dmlTestCase.testCodegenRowAgg21CP(); }\n+ @Test public void testCodegenRowAgg22CP() { dmlTestCase.testCodegenRowAgg22CP(); }\n+ @Test public void testCodegenRowAgg23CP() { dmlTestCase.testCodegenRowAgg23CP(); }\n+ @Test public void testCodegenRowAgg24CP() { dmlTestCase.testCodegenRowAgg24CP(); }\n+ @Test public void testCodegenRowAgg25CP() { dmlTestCase.testCodegenRowAgg25CP(); }\n+ @Test public void testCodegenRowAgg26CP() { dmlTestCase.testCodegenRowAgg26CP(); }\n+ @Test public void testCodegenRowAgg27CP() { dmlTestCase.testCodegenRowAgg27CP(); }\n+ @Test public void testCodegenRowAgg28CP() { dmlTestCase.testCodegenRowAgg28CP(); }\n+ @Test public void testCodegenRowAgg29CP() { dmlTestCase.testCodegenRowAgg29CP(); }\n+ @Test public void testCodegenRowAgg30CP() { dmlTestCase.testCodegenRowAgg30CP(); }\n+ @Test public void testCodegenRowAgg31CP() { dmlTestCase.testCodegenRowAgg31CP(); }\n+ @Test public void testCodegenRowAgg32CP() { dmlTestCase.testCodegenRowAgg32CP(); }\n+ @Test public void testCodegenRowAgg33CP() { dmlTestCase.testCodegenRowAgg33CP(); }\n+ @Test public void testCodegenRowAgg34CP() { dmlTestCase.testCodegenRowAgg34CP(); }\n+ @Test public void testCodegenRowAgg35CP() { dmlTestCase.testCodegenRowAgg35CP(); }\n+ @Test public void testCodegenRowAgg36CP() { dmlTestCase.testCodegenRowAgg36CP(); }\n+ @Test public void testCodegenRowAgg37CP() { dmlTestCase.testCodegenRowAgg37CP(); }\n+ @Test public void testCodegenRowAgg38CP() { dmlTestCase.testCodegenRowAgg38CP(); }\n+ @Test public void testCodegenRowAgg39CP() { dmlTestCase.testCodegenRowAgg39CP(); }\n+ @Test public void testCodegenRowAgg40CP() { dmlTestCase.testCodegenRowAgg40CP(); }\n+ @Test public void testCodegenRowAgg41CP() { dmlTestCase.testCodegenRowAgg41CP(); }\n+ @Test public void testCodegenRowAgg42CP() { dmlTestCase.testCodegenRowAgg42CP(); }\n+ @Test public void testCodegenRowAgg43CP() { dmlTestCase.testCodegenRowAgg43CP(); }\n+ @Test public void testCodegenRowAgg44CP() { dmlTestCase.testCodegenRowAgg44CP(); }\n+ @Test public void testCodegenRowAgg45CP() { dmlTestCase.testCodegenRowAgg45CP(); }\n+ @Test public void testCodegenRowAgg46CP() { dmlTestCase.testCodegenRowAgg46CP(); }\n+\n+ @Test public void testCodegenRowAggRewrite1CP() { dmlTestCase.testCodegenRowAggRewrite1CP(); }\n+ @Test public void testCodegenRowAggRewrite2CP() { dmlTestCase.testCodegenRowAggRewrite2CP(); }\n+ @Test public void testCodegenRowAggRewrite3CP() { dmlTestCase.testCodegenRowAggRewrite3CP(); }\n+ @Test public void testCodegenRowAggRewrite4CP() { dmlTestCase.testCodegenRowAggRewrite4CP(); }\n+ @Test public void testCodegenRowAggRewrite5CP() { dmlTestCase.testCodegenRowAggRewrite5CP(); }\n+ @Test public void testCodegenRowAggRewrite6CP() { dmlTestCase.testCodegenRowAggRewrite6CP(); }\n+ @Test public void testCodegenRowAggRewrite7CP() { dmlTestCase.testCodegenRowAggRewrite7CP(); }\n+ @Test public void testCodegenRowAggRewrite8CP() { dmlTestCase.testCodegenRowAggRewrite8CP(); }\n+ @Test public void testCodegenRowAggRewrite9CP() { dmlTestCase.testCodegenRowAggRewrite9CP(); }\n+ @Test public void testCodegenRowAggRewrite10CP() { dmlTestCase.testCodegenRowAggRewrite10CP(); }\n+ @Test public void testCodegenRowAggRewrite11CP() { dmlTestCase.testCodegenRowAggRewrite11CP(); }\n+ @Test public void testCodegenRowAggRewrite12CP() { dmlTestCase.testCodegenRowAggRewrite12CP(); }\n+ @Test public void testCodegenRowAggRewrite13CP() { dmlTestCase.testCodegenRowAggRewrite13CP(); }\n+ @Test public void testCodegenRowAggRewrite14CP() { dmlTestCase.testCodegenRowAggRewrite14CP(); }\n+ @Test public void testCodegenRowAggRewrite15CP() { dmlTestCase.testCodegenRowAggRewrite15CP(); }\n+ @Test public void testCodegenRowAggRewrite16CP() { dmlTestCase.testCodegenRowAggRewrite16CP(); }\n+ @Test public void testCodegenRowAggRewrite17CP() { dmlTestCase.testCodegenRowAggRewrite17CP(); }\n+ @Test public void testCodegenRowAggRewrite18CP() { dmlTestCase.testCodegenRowAggRewrite18CP(); }\n+ @Test public void testCodegenRowAggRewrite19CP() { dmlTestCase.testCodegenRowAggRewrite19CP(); }\n+ @Test public void testCodegenRowAggRewrite20CP() { dmlTestCase.testCodegenRowAggRewrite20CP(); }\n+ @Test public void testCodegenRowAggRewrite21CP() { dmlTestCase.testCodegenRowAggRewrite21CP(); }\n+ @Test public void testCodegenRowAggRewrite22CP() { dmlTestCase.testCodegenRowAggRewrite22CP(); }\n+ @Test public void testCodegenRowAggRewrite23CP() { dmlTestCase.testCodegenRowAggRewrite23CP(); }\n+ @Test public void testCodegenRowAggRewrite24CP() { dmlTestCase.testCodegenRowAggRewrite24CP(); }\n+ @Test public void testCodegenRowAggRewrite25CP() { dmlTestCase.testCodegenRowAggRewrite25CP(); }\n+ @Test public void testCodegenRowAggRewrite26CP() { dmlTestCase.testCodegenRowAggRewrite26CP(); }\n+ @Test public void testCodegenRowAggRewrite27CP() { dmlTestCase.testCodegenRowAggRewrite27CP(); }\n+ @Test public void testCodegenRowAggRewrite28CP() { dmlTestCase.testCodegenRowAggRewrite28CP(); }\n+ @Test public void testCodegenRowAggRewrite29CP() { dmlTestCase.testCodegenRowAggRewrite29CP(); }\n+ @Test public void testCodegenRowAggRewrite30CP() { dmlTestCase.testCodegenRowAggRewrite30CP(); }\n+ @Test public void testCodegenRowAggRewrite31CP() { dmlTestCase.testCodegenRowAggRewrite31CP(); }\n+ @Test public void testCodegenRowAggRewrite32CP() { dmlTestCase.testCodegenRowAggRewrite32CP(); }\n+ @Test public void testCodegenRowAggRewrite33CP() { dmlTestCase.testCodegenRowAggRewrite33CP(); }\n+ @Test public void testCodegenRowAggRewrite34CP() { dmlTestCase.testCodegenRowAggRewrite34CP(); }\n+ @Test public void testCodegenRowAggRewrite35CP() { dmlTestCase.testCodegenRowAggRewrite35CP(); }\n+ @Test public void testCodegenRowAggRewrite36CP() { dmlTestCase.testCodegenRowAggRewrite36CP(); }\n+ @Test public void testCodegenRowAggRewrite37CP() { dmlTestCase.testCodegenRowAggRewrite37CP(); }\n+ @Test public void testCodegenRowAggRewrite38CP() { dmlTestCase.testCodegenRowAggRewrite38CP(); }\n+ @Test public void testCodegenRowAggRewrite39CP() { dmlTestCase.testCodegenRowAggRewrite39CP(); }\n+ @Test public void testCodegenRowAggRewrite40CP() { dmlTestCase.testCodegenRowAggRewrite40CP(); }\n+ @Test public void testCodegenRowAggRewrite41CP() { dmlTestCase.testCodegenRowAggRewrite41CP(); }\n+ @Test public void testCodegenRowAggRewrite42CP() { dmlTestCase.testCodegenRowAggRewrite42CP(); }\n+ @Test public void testCodegenRowAggRewrite43CP() { dmlTestCase.testCodegenRowAggRewrite43CP(); }\n+ @Test public void testCodegenRowAggRewrite44CP() { dmlTestCase.testCodegenRowAggRewrite44CP(); }\n+ @Test public void testCodegenRowAggRewrite45CP() { dmlTestCase.testCodegenRowAggRewrite45CP(); }\n+ @Test public void testCodegenRowAggRewrite46CP() { dmlTestCase.testCodegenRowAggRewrite46CP(); }\n+\n+}\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/test/java/org/apache/sysds/test/gpu/BuiltinUnaryGPUInstructionTest.java", "new_path": "src/test/java/org/apache/sysds/test/gpu/BuiltinUnaryGPUInstructionTest.java", "diff": "package org.apache.sysds.test.gpu;\nimport org.apache.sysds.test.AutomatedTestBase;\n-import org.apache.sysds.test.functions.builtin.BuiltinSigmoidTest;\n+import org.apache.sysds.test.functions.builtin.part2.BuiltinSigmoidTest;\nimport org.apache.sysds.test.functions.unary.matrix.ACosTest;\nimport org.apache.sysds.test.functions.unary.matrix.ASinTest;\nimport org.apache.sysds.test.functions.unary.matrix.ATanTest;\n@@ -308,7 +308,7 @@ public class BuiltinUnaryGPUInstructionTest extends AutomatedTestBase {\n}\n@Test public void SigmoidTest() {\n- BuiltinSigmoidTest dmlTestCase = new org.apache.sysds.test.functions.builtin.BuiltinSigmoidTest();\n+ BuiltinSigmoidTest dmlTestCase = new BuiltinSigmoidTest();\ndmlTestCase.setUpBase();\ndmlTestCase.setUp();\ndmlTestCase.testSigmoidMatrixDenseCP();\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/python/docs/source/guide/federated.rst", "new_path": "src/main/python/docs/source/guide/federated.rst", "diff": "@@ -89,9 +89,9 @@ The print should look like\n.. code-block::\n- [[ 1. 4. 9. 1. 4. 9.]\n- [16. 25. 36. 16. 25. 36.]\n- [49. 64. 81. 49. 64. 81.]]\n+ [[198. 243. 288.]\n+ [198. 243. 288.]\n+ [198. 243. 288.]]\n.. note::\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/hops/rewrite/HopRewriteUtils.java", "new_path": "src/main/java/org/apache/sysds/hops/rewrite/HopRewriteUtils.java", "diff": "package org.apache.sysds.hops.rewrite;\nimport org.apache.commons.lang.ArrayUtils;\n+import org.apache.commons.logging.Log;\n+import org.apache.commons.logging.LogFactory;\nimport org.apache.sysds.api.DMLScript;\nimport org.apache.sysds.common.Types.DataType;\nimport org.apache.sysds.common.Types.ExecMode;\n@@ -87,8 +89,8 @@ import java.util.HashSet;\nimport java.util.LinkedHashMap;\nimport java.util.List;\n-public class HopRewriteUtils\n-{\n+public class HopRewriteUtils {\n+ private static final Log LOG = LogFactory.getLog(HopRewriteUtils.class.getName());\npublic static boolean isValueTypeCast( OpOp1 op ) {\nreturn op == OpOp1.CAST_AS_BOOLEAN\n@@ -1637,7 +1639,10 @@ public class HopRewriteUtils\n&& prog.getFunctionStatementBlock(sagg) != null;\n}\ncatch(Exception ex) {\n- //robustness invalid function keys\n+ // If the function keys are incorrect this exception is caught for robustness in error messages for users.\n+ // Intensionally only catching the exception!\n+ // For debugging if for some reason the error we encountered was something else we LOG the error.\n+ LOG.error(ex);\nreturn false;\n}\n}\n" } ]

[ { "change_type": "MODIFY", "old_path": "docs/_config.yml", "new_path": "docs/_config.yml", "diff": "@@ -41,7 +41,3 @@ exclude:\n# These allow the documentation to be updated with newer releases\nSYSTEMDS_VERSION: 2.3.0-SNAPSHOT\n-# if 'analytics_on' is true, analytics section will be rendered on the HTML pages\n-analytics_on: true\n-analytics_provider: google_universal\n-analytics_google_universal_tracking_id : UA-71553733-1\n" } ]

[ { "change_type": "MODIFY", "old_path": "dev/release/pypi-upload.sh", "new_path": "dev/release/pypi-upload.sh", "diff": "@@ -61,8 +61,8 @@ printf \"Is this RC voted and approved by PMC? [Yes/No]: \\n\"\n# Docs: https://www.gnu.org/software/bash/manual/bash.html#index-case\nselect yn in \"Yes\" \"No\"; do\ncase $yn in\n- Yes ) sed -i \"s/$RELEASE_VERSION-SNAPSHOT/$RELEASE_VERSION/\" systemds/project_info.py; break ;;\n- No ) sed -i \"s/$RELEASE_VERSION-SNAPSHOT/$RELEASE_TAG/\" systemds/project_info.py; break ;;\n+ Yes ) sed -i \"s/$RELEASE_VERSION-dev/$RELEASE_VERSION/\" systemds/project_info.py; break ;;\n+ No ) sed -i \"s/$RELEASE_VERSION-dev/$RELEASE_TAG/\" systemds/project_info.py; break ;;\n* ) echo \"Yes or No response is required.\";;\nesac\ndone\n@@ -79,15 +79,14 @@ python3 -m twine check dist/*\n# use Edit->paste to paste the API token (https://pypi.org/help/#invalid-auth)\n# else, use `right click` for paste in the terminal.\n-# Dev:\n+if [[ $dry_run_flag != 1 ]]; then\n+ python -m twine upload dist/*\n+else\n+ # Development test:\n# Test upload to test.pypi.org\n# Credentials are\n# username: __token__\n# password: pypi-DU5y...\n-\n-if [[ $dry_run_flag != 1 ]]; then\n- python -m twine upload dist/*\n-else\npython -m twine upload --repository testpypi dist/*\nfi\n" } ]

[ { "change_type": "MODIFY", "old_path": "scripts/builtin/gmmPredict.dml", "new_path": "scripts/builtin/gmmPredict.dml", "diff": "# compute posterior probabilities for new instances given the variance and mean of fitted data\nm_gmmPredict = function(Matrix[Double] X, Matrix[Double] weight,\n- Matrix[Double] mu, Matrix[Double] precisions_cholesky, String model)\n+ Matrix[Double] mu, Matrix[Double] precisions_cholesky, String model = \"VVV\")\nreturn(Matrix[Double] predict, Matrix[Double] posterior_prob)\n{\n# compute the posterior probabilities for new instances\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/systemds/operator/algorithm/builtin/gmmPredict.py", "new_path": "src/main/python/systemds/operator/algorithm/builtin/gmmPredict.py", "diff": "@@ -33,7 +33,7 @@ def gmmPredict(X: Matrix,\nweight: Matrix,\nmu: Matrix,\nprecisions_cholesky: Matrix,\n- model: str):\n+ **kwargs: Dict[str, VALID_INPUT_TYPES]):\n\"\"\"\n:param X: Matrix X (instances to be clustered)\n:param weight: Weight of learned model\n@@ -42,7 +42,8 @@ def gmmPredict(X: Matrix,\n:param model: fitted model\n:return: 'OperationNode' containing predicted cluster labels & probabilities of belongingness & for new instances given the variance and mean of fitted data\n\"\"\"\n- params_dict = {'X': X, 'weight': weight, 'mu': mu, 'precisions_cholesky': precisions_cholesky, 'model': model}\n+ params_dict = {'X': X, 'weight': weight, 'mu': mu, 'precisions_cholesky': precisions_cholesky}\n+ params_dict.update(kwargs)\nvX_0 = Matrix(X.sds_context, '')\nvX_1 = Matrix(X.sds_context, '')\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/python/pre_setup.py", "new_path": "src/main/python/pre_setup.py", "diff": "@@ -43,9 +43,19 @@ LIB_DIR = os.path.join(this_path, PYTHON_DIR, 'lib')\nif os.path.exists(LIB_DIR):\nshutil.rmtree(LIB_DIR, True)\nSYSTEMDS_BIN = 'systemds-*-bin.zip'\n+found_bin = False\nfor file in os.listdir(os.path.join(SYSTEMDS_ROOT, 'target')):\n# Take jar files from bin release file\nif fnmatch.fnmatch(file, SYSTEMDS_BIN):\n+ if found_bin:\n+ print(\"invalid install found multiple bin files, please package systemds with clean flag\")\n+ exit(-1)\n+ found_bin = True\n+\n+for file in os.listdir(os.path.join(SYSTEMDS_ROOT, 'target')):\n+ # Take jar files from bin release file\n+ if fnmatch.fnmatch(file, SYSTEMDS_BIN):\n+ print(\"Using java files from : \" + file )\nsystemds_bin_zip = os.path.join(SYSTEMDS_ROOT, 'target', file)\nextract_dir = os.path.join(TMP_DIR)\n@@ -56,6 +66,7 @@ for file in os.listdir(os.path.join(SYSTEMDS_ROOT, 'target')):\nzip.extract(f, TMP_DIR)\nunzipped_dir_name = file.rsplit('.', 1)[0]\nshutil.copytree(os.path.join(TMP_DIR, unzipped_dir_name, 'lib'), LIB_DIR)\n+ break\n# Take hadoop binaries.\nHADOOP_DIR_SRC = os.path.join(SYSTEMDS_ROOT, 'target', 'lib', 'hadoop')\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoder.java", "new_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoder.java", "diff": "@@ -116,6 +116,8 @@ public abstract class ColumnEncoder implements Encoder, Comparable<ColumnEncoder\nprotected abstract double getCode(CacheBlock in, int row);\n+ protected abstract double[] getCodeCol(CacheBlock in, int startInd, int blkSize);\n+\nprotected void applySparse(CacheBlock in, MatrixBlock out, int outputCol, int rowStart, int blk){\nint index = _colID - 1;\n@@ -126,10 +128,22 @@ public abstract class ColumnEncoder implements Encoder, Comparable<ColumnEncoder\n}\n}\n- protected void applyDense(CacheBlock in, MatrixBlock out, int outputCol, int rowStart, int blk){\n+ /*protected void applyDense(CacheBlock in, MatrixBlock out, int outputCol, int rowStart, int blk){\nfor(int i = rowStart; i < getEndIndex(in.getNumRows(), rowStart, blk); i++) {\nout.quickSetValue(i, outputCol, getCode(in, i));\n}\n+ }*/\n+\n+ protected void applyDense(CacheBlock in, MatrixBlock out, int outputCol, int rowStart, int blk){\n+ // Apply loop tiling to exploit CPU caches\n+ double[] codes = getCodeCol(in, rowStart, blk);\n+ int rowEnd = getEndIndex(in.getNumRows(), rowStart, blk);\n+ int B = 32; //tile size\n+ for(int i = rowStart; i < rowEnd; i+=B) {\n+ int lim = Math.min(i+B, rowEnd);\n+ for (int ii=i; ii<lim; ii++)\n+ out.quickSetValue(ii, outputCol, codes[ii-rowStart]);\n+ }\n}\nprotected abstract TransformType getTransformType();\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderBin.java", "new_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderBin.java", "diff": "@@ -95,6 +95,7 @@ public class ColumnEncoderBin extends ColumnEncoder {\n}\nprotected double getCode(CacheBlock in, int row){\n+ // find the right bucket for a single row\nif( _binMins.length == 0 || _binMaxs.length == 0 ) {\nLOG.warn(\"ColumnEncoderBin: applyValue without bucket boundaries, assign 1\");\nreturn 1; //robustness in case of missing bins\n@@ -108,6 +109,28 @@ public class ColumnEncoderBin extends ColumnEncoder {\nreturn ((ix < 0) ? Math.abs(ix + 1) : ix) + 1;\n}\n+ @Override\n+ protected double[] getCodeCol(CacheBlock in, int startInd, int blkSize) {\n+ // find the right bucket for a block of rows\n+ int endInd = getEndIndex(in.getNumRows(), startInd, blkSize);\n+ double codes[] = new double[endInd-startInd];\n+ for (int i=startInd; i<endInd; i++) {\n+ if (_binMins.length == 0 || _binMaxs.length == 0) {\n+ LOG.warn(\"ColumnEncoderBin: applyValue without bucket boundaries, assign 1\");\n+ codes[i-startInd] = 1; //robustness in case of missing bins\n+ continue;\n+ }\n+ double inVal = in.getDoubleNaN(i, _colID - 1);\n+ if (Double.isNaN(inVal) || inVal < _binMins[0] || inVal > _binMaxs[_binMaxs.length-1]) {\n+ codes[i-startInd] = Double.NaN;\n+ continue;\n+ }\n+ int ix = Arrays.binarySearch(_binMaxs, inVal);\n+ codes[i-startInd] = ((ix < 0) ? Math.abs(ix + 1) : ix) + 1;\n+ }\n+ return codes;\n+ }\n+\n@Override\nprotected TransformType getTransformType() {\nreturn TransformType.BIN;\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderComposite.java", "new_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderComposite.java", "diff": "@@ -207,6 +207,11 @@ public class ColumnEncoderComposite extends ColumnEncoder {\nthrow new DMLRuntimeException(\"CompositeEncoder does not have a Code\");\n}\n+ @Override\n+ protected double[] getCodeCol(CacheBlock in, int startInd, int blkSize) {\n+ throw new DMLRuntimeException(\"CompositeEncoder does not have a Code\");\n+ }\n+\n@Override\nprotected TransformType getTransformType() {\nreturn TransformType.N_A;\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderDummycode.java", "new_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderDummycode.java", "diff": "@@ -75,6 +75,11 @@ public class ColumnEncoderDummycode extends ColumnEncoder {\nthrow new DMLRuntimeException(\"DummyCoder does not have a code\");\n}\n+ @Override\n+ protected double[] getCodeCol(CacheBlock in, int startInd, int blkSize) {\n+ throw new DMLRuntimeException(\"DummyCoder does not have a code\");\n+ }\n+\nprotected void applySparse(CacheBlock in, MatrixBlock out, int outputCol, int rowStart, int blk){\nif (!(in instanceof MatrixBlock)){\nthrow new DMLRuntimeException(\"ColumnEncoderDummycode called with: \" + in.getClass().getSimpleName() +\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderFeatureHash.java", "new_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderFeatureHash.java", "diff": "package org.apache.sysds.runtime.transform.encode;\n+import static org.apache.sysds.runtime.util.UtilFunctions.getEndIndex;\nimport java.io.IOException;\nimport java.io.ObjectInput;\nimport java.io.ObjectOutput;\n@@ -65,12 +66,27 @@ public class ColumnEncoderFeatureHash extends ColumnEncoder {\n@Override\nprotected double getCode(CacheBlock in, int row) {\n+ // hash a single row\nString key = in.getString(row, _colID - 1);\nif(key == null)\nreturn Double.NaN;\nreturn (key.hashCode() % _K) + 1;\n}\n+ protected double[] getCodeCol(CacheBlock in, int startInd, int blkSize) {\n+ // hash a block of rows\n+ int endInd = getEndIndex(in.getNumRows(), startInd, blkSize);\n+ double codes[] = new double[endInd-startInd];\n+ for (int i=startInd; i<endInd; i++) {\n+ String key = in.getString(i, _colID - 1);\n+ if(key == null || key.isEmpty())\n+ codes[i-startInd] = Double.NaN;\n+ else\n+ codes[i-startInd] = (key.hashCode() % _K) + 1;\n+ }\n+ return codes;\n+ }\n+\n@Override\npublic void build(CacheBlock in) {\n// do nothing (no meta data other than K)\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderPassThrough.java", "new_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderPassThrough.java", "diff": "@@ -65,6 +65,15 @@ public class ColumnEncoderPassThrough extends ColumnEncoder {\nreturn in.getDoubleNaN(row, _colID - 1);\n}\n+ @Override\n+ protected double[] getCodeCol(CacheBlock in, int startInd, int blkSize) {\n+ int endInd = getEndIndex(in.getNumRows(), startInd, blkSize);\n+ double codes[] = new double[endInd-startInd];\n+ for (int i=startInd; i<endInd; i++) {\n+ codes[i-startInd] = in.getDoubleNaN(i, _colID-1);\n+ }\n+ return codes;\n+ }\nprotected void applySparse(CacheBlock in, MatrixBlock out, int outputCol, int rowStart, int blk){\nSet<Integer> sparseRowsWZeros = null;\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderRecode.java", "new_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderRecode.java", "diff": "@@ -171,6 +171,7 @@ public class ColumnEncoderRecode extends ColumnEncoder {\n}\nprotected double getCode(CacheBlock in, int r){\n+ // lookup for a single row\nObject okey = in.getString(r, _colID - 1);\nString key = (okey != null) ? okey.toString() : null;\nif(key == null || key.isEmpty())\n@@ -179,16 +180,19 @@ public class ColumnEncoderRecode extends ColumnEncoder {\nreturn (code < 0) ? Double.NaN : code;\n}\n- protected double[] getCodeCol(CacheBlock in) {\n- Object[] coldata = (Object[]) ((FrameBlock)in).getColumnData(_colID-1);\n- double codes[] = new double[in.getNumRows()];\n- for (int i=0; i<coldata.length; i++) {\n- Object okey = coldata[i];\n- String key = (okey != null) ? okey.toString() : null;\n- if(key == null || key.isEmpty())\n- codes[i] = Double.NaN;\n+ @Override\n+ protected double[] getCodeCol(CacheBlock in, int startInd, int blkSize) {\n+ // lookup for a block of rows\n+ int endInd = getEndIndex(in.getNumRows(), startInd, blkSize);\n+ double codes[] = new double[endInd-startInd];\n+ for (int i=startInd; i<endInd; i++) {\n+ String key = in.getString(i, _colID-1);\n+ if(key == null || key.isEmpty()) {\n+ codes[i-startInd] = Double.NaN;\n+ continue;\n+ }\nlong code = lookupRCDMap(key);\n- codes[i] = code;\n+ codes[i-startInd] = (code < 0) ? Double.NaN : code;\n}\nreturn codes;\n}\n" } ]

[ { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoder.java", "new_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoder.java", "diff": "@@ -119,13 +119,29 @@ public abstract class ColumnEncoder implements Encoder, Comparable<ColumnEncoder\nprotected abstract double[] getCodeCol(CacheBlock in, int startInd, int blkSize);\n- protected void applySparse(CacheBlock in, MatrixBlock out, int outputCol, int rowStart, int blk){\n+ /*protected void applySparse(CacheBlock in, MatrixBlock out, int outputCol, int rowStart, int blk){\nint index = _colID - 1;\nfor(int r = rowStart; r < getEndIndex(in.getNumRows(), rowStart, blk); r++) {\nSparseRowVector row = (SparseRowVector) out.getSparseBlock().get(r);\nrow.values()[index] = getCode(in, r);\nrow.indexes()[index] = outputCol;\n}\n+ }*/\n+\n+ protected void applySparse(CacheBlock in, MatrixBlock out, int outputCol, int rowStart, int blk){\n+ int index = _colID - 1;\n+ // Apply loop tiling to exploit CPU caches\n+ double[] codes = getCodeCol(in, rowStart, blk);\n+ int rowEnd = getEndIndex(in.getNumRows(), rowStart, blk);\n+ int B = 32; //tile size\n+ for(int i = rowStart; i < rowEnd; i+=B) {\n+ int lim = Math.min(i+B, rowEnd);\n+ for (int ii=i; ii<lim; ii++) {\n+ SparseRowVector row = (SparseRowVector) out.getSparseBlock().get(ii);\n+ row.values()[index] = codes[ii-rowStart];\n+ row.indexes()[index] = outputCol;\n+ }\n+ }\n}\n/*protected void applyDense(CacheBlock in, MatrixBlock out, int outputCol, int rowStart, int blk){\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderDummycode.java", "new_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderDummycode.java", "diff": "@@ -125,19 +125,28 @@ public class ColumnEncoderDummycode extends ColumnEncoder {\nthrow new DMLRuntimeException(\"ColumnEncoderDummycode called with: \" + in.getClass().getSimpleName() +\n\" and not MatrixBlock\");\n}\n- for(int i = rowStart; i < getEndIndex(in.getNumRows(), rowStart, blk); i++) {\n- // Using outputCol here as index since we have a MatrixBlock as input where dummycoding could have been\n- // applied in a previous encoder\n- double val = in.getDouble(i, outputCol);\n+ int rowEnd = getEndIndex(in.getNumRows(), rowStart, blk);\n+ double vals[] = new double[rowEnd -rowStart];\n+ for (int i=rowStart; i<rowEnd; i++)\n+ vals[i-rowStart] = in.getDouble(i, outputCol);\n+\n+ // Using outputCol here as index since we have a MatrixBlock as input where\n+ // dummycoding might have been applied in a previous encoder\n+ int B = 32;\n+ for(int i=rowStart; i<rowEnd; i+=B) {\n+ // Apply loop tiling to exploit CPU caches\n+ int lim = Math.min(i+B, rowEnd);\n+ for (int ii=i; ii<lim; ii++) {\n+ double val = vals[ii-rowStart];\nif(Double.isNaN(val)) {\n- // 0 if NaN\n- out.quickSetValue(i, outputCol, 0);\n+ out.quickSetValue(ii, outputCol, 0); //0 if NaN\ncontinue;\n}\nint nCol = outputCol + (int) val - 1;\nif(nCol != outputCol)\n- out.quickSetValue(i, outputCol, 0);\n- out.quickSetValue(i, nCol, 1);\n+ out.quickSetValue(ii, outputCol, 0);\n+ out.quickSetValue(ii, nCol, 1);\n+ }\n}\n}\n" }, { "change_type": "MODIFY", "old_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderPassThrough.java", "new_path": "src/main/java/org/apache/sysds/runtime/transform/encode/ColumnEncoderPassThrough.java", "diff": "@@ -78,17 +78,24 @@ public class ColumnEncoderPassThrough extends ColumnEncoder {\nprotected void applySparse(CacheBlock in, MatrixBlock out, int outputCol, int rowStart, int blk){\nSet<Integer> sparseRowsWZeros = null;\nint index = _colID - 1;\n- for(int r = rowStart; r < getEndIndex(in.getNumRows(), rowStart, blk); r++) {\n- double v = getCode(in, r);\n- SparseRowVector row = (SparseRowVector) out.getSparseBlock().get(r);\n+ // Apply loop tiling to exploit CPU caches\n+ double[] codes = getCodeCol(in, rowStart, blk);\n+ int rowEnd = getEndIndex(in.getNumRows(), rowStart, blk);\n+ int B = 32; //tile size\n+ for(int i = rowStart; i < rowEnd; i+=B) {\n+ int lim = Math.min(i+B, rowEnd);\n+ for (int ii=i; ii<lim; ii++) {\n+ double v = codes[ii-rowStart];\n+ SparseRowVector row = (SparseRowVector) out.getSparseBlock().get(ii);\nif(v == 0) {\nif(sparseRowsWZeros == null)\nsparseRowsWZeros = new HashSet<>();\n- sparseRowsWZeros.add(r);\n+ sparseRowsWZeros.add(ii);\n}\nrow.values()[index] = v;\nrow.indexes()[index] = outputCol;\n}\n+ }\nif(sparseRowsWZeros != null){\naddSparseRowsWZeros(sparseRowsWZeros);\n}\n" } ]

[ { "change_type": "MODIFY", "old_path": "scripts/staging/sklearn/mappers/transformations.py", "new_path": "scripts/staging/sklearn/mappers/transformations.py", "diff": "@@ -61,7 +61,9 @@ class SimpleImputerMapper(Mapper):\nelse:\nself.name = 'imputeByMean'\n- self.mapped_params = []\n+ self.mapped_params = [\n+ 'matrix(1, 1, ncol(X))'\n+ ]\nclass PCAMapper(Mapper):\n" }, { "change_type": "MODIFY", "old_path": "scripts/staging/sklearn/run_tests.py", "new_path": "scripts/staging/sklearn/run_tests.py", "diff": "@@ -74,13 +74,13 @@ if __name__ == '__main__':\n#TODO: Tests which use PCA or DBSCAN, trigger a NullPointerException during parsing for some reason\nmake_pipeline(StandardScaler(), DBSCAN()),\nmake_pipeline(Normalizer(), DBSCAN()),\n- make_pipeline(SimpleImputer(strategy='mean'), DBSCAN()),\n- make_pipeline(SimpleImputer(strategy='median'), DBSCAN()),\n+ # make_pipeline(SimpleImputer(strategy='mean'), DBSCAN()),\n+ # make_pipeline(SimpleImputer(strategy='median'), DBSCAN()),\nmake_pipeline(PCA(), KMeans()),\nmake_pipeline(PCA(), DBSCAN()),\n# TODO: GaussianMixtureModel results in LanguageException -- ERROR: [line 0:0] -- Function get_sample_maps() is undefined.\n- make_pipeline(StandardScaler(), GaussianMixture()),\n- make_pipeline(Normalizer(), GaussianMixture())\n+ # make_pipeline(StandardScaler(), GaussianMixture()),\n+ # make_pipeline(Normalizer(), GaussianMixture())\n]\nvalid_results = []\n" } ]

[ { "change_type": "MODIFY", "old_path": "scripts/staging/sklearn/mappers/unsupervised.py", "new_path": "scripts/staging/sklearn/mappers/unsupervised.py", "diff": "@@ -76,8 +76,10 @@ class GaussianMixtureMapper(Mapper):\nself.mapped_params = [\nself.params.get('n_components', 3),\nf'\"{self.model_map.get(self.params.get(\"covariance_type\", \"VVV\"))}\"',\n- self.params.get('init_params', '\"kmeans\"'),\n+ f'\"{self.params.get(\"init_params\", \"kmeans\")}\"',\nself.params.get('max_iter', 100),\nself.params.get('reg_covar', 1e-6),\n- self.params.get('tol', 0.000001)\n+ self.params.get('tol', 0.000001),\n+ self.params.get('seed', -1),\n+ self.params.get('verbose', 'FALSE')\n]\n" }, { "change_type": "MODIFY", "old_path": "scripts/staging/sklearn/run_tests.py", "new_path": "scripts/staging/sklearn/run_tests.py", "diff": "@@ -71,16 +71,14 @@ if __name__ == '__main__':\nmake_pipeline(Normalizer(), TweedieRegressor()),\nmake_pipeline(StandardScaler(), LogisticRegression()),\nmake_pipeline(Normalizer(), LogisticRegression()),\n- #TODO: Tests which use PCA or DBSCAN, trigger a NullPointerException during parsing for some reason\nmake_pipeline(StandardScaler(), DBSCAN()),\nmake_pipeline(Normalizer(), DBSCAN()),\n- # make_pipeline(SimpleImputer(strategy='mean'), DBSCAN()),\n- # make_pipeline(SimpleImputer(strategy='median'), DBSCAN()),\n+ make_pipeline(SimpleImputer(strategy='mean'), DBSCAN()),\n+ make_pipeline(SimpleImputer(strategy='median'), DBSCAN()),\nmake_pipeline(PCA(), KMeans()),\nmake_pipeline(PCA(), DBSCAN()),\n- # TODO: GaussianMixtureModel results in LanguageException -- ERROR: [line 0:0] -- Function get_sample_maps() is undefined.\n- # make_pipeline(StandardScaler(), GaussianMixture()),\n- # make_pipeline(Normalizer(), GaussianMixture())\n+ make_pipeline(StandardScaler(), GaussianMixture()),\n+ make_pipeline(Normalizer(), GaussianMixture())\n]\nvalid_results = []\n" } ]

[ { "change_type": "ADD", "old_path": null, "new_path": "src/main/python/tests/algorithms/test_gmm.py", "diff": "+# -------------------------------------------------------------\n+#\n+# Licensed to the Apache Software Foundation (ASF) under one\n+# or more contributor license agreements. See the NOTICE file\n+# distributed with this work for additional information\n+# regarding copyright ownership. The ASF licenses this file\n+# to you under the Apache License, Version 2.0 (the\n+# \"License\"); you may not use this file except in compliance\n+# with the License. You may obtain a copy of the License at\n+#\n+# http://www.apache.org/licenses/LICENSE-2.0\n+#\n+# Unless required by applicable law or agreed to in writing,\n+# software distributed under the License is distributed on an\n+# \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY\n+# KIND, either express or implied. See the License for the\n+# specific language governing permissions and limitations\n+# under the License.\n+#\n+# -------------------------------------------------------------\n+\n+import unittest\n+\n+from systemds.context import SystemDSContext\n+from systemds.operator.algorithm import gmm, gmmPredict\n+\n+\n+class TestGMM(unittest.TestCase):\n+\n+ sds: SystemDSContext = None\n+\n+ @classmethod\n+ def setUpClass(cls):\n+ cls.sds = SystemDSContext()\n+\n+ @classmethod\n+ def tearDownClass(cls):\n+ cls.sds.close()\n+\n+ def test_lm_simple(self):\n+ a = self.sds.rand(500, 10, -100, 100, pdf=\"normal\", seed=10)\n+ features = a # training data all not outliers\n+\n+ notOutliers = self.sds.rand(10, 10, -1, 1, seed=10) # inside a\n+ outliers = self.sds.rand(10, 10, 1150, 1200, seed=10) # outliers\n+\n+ test = outliers.rbind(notOutliers) # testing data half outliers\n+\n+ n_gaussian = 4\n+\n+ [_, _, _, _, mu, precision_cholesky, wight] = gmm(\n+ features, False, n_components=n_gaussian, seed=10)\n+\n+ [_, pp] = gmmPredict(\n+ test, wight, mu, precision_cholesky, model=self.sds.scalar(\"VVV\"))\n+\n+ outliers = pp.max(axis=1) < 0.99\n+ ret = outliers.compute()\n+\n+ self.assertTrue(ret.sum() == 10)\n+\n+\n+if __name__ == \"__main__\":\n+ unittest.main(exit=False)\n" }, { "change_type": "MODIFY", "old_path": "src/main/python/tests/source/test_source_list.py", "new_path": "src/main/python/tests/source/test_source_list.py", "diff": "@@ -50,7 +50,7 @@ class TestSource_01(unittest.TestCase):\n[a, b, c] = scale(m, True, True)\narr = self.sds.array(a, b, c)\nc = self.sds.source(self.source_path, \"test\").func(arr)\n- res = c.sum().compute(verbose=True)\n+ res = c.sum().compute()\nself.assertTrue(res == 0)\n# [SYSTEMDS-3224] https://issues.apache.org/jira/browse/SYSTEMDS-3224\n" } ]

[ { "change_type": "MODIFY", "old_path": "scripts/builtin/gmm.dml", "new_path": "scripts/builtin/gmm.dml", "diff": "@@ -114,7 +114,7 @@ return (Matrix[Double] weight, Matrix[Double] mean, Matrix[Double] sigma, Matrix\nresp = matrix(0, nrow(X), n_components)\nif(init_params == \"kmeans\") {\n[C, Y] = kmeans(X=X, k=n_components, runs=10,\n- eps=tol, is_verbose=FALSE, avg_sample_size_per_centroid=100)\n+ eps=tol, is_verbose=FALSE, avg_sample_size_per_centroid=100, seed=seed)\nresp = ((resp + t(seq(1, n_components))) == Y)\n}\nelse if(init_params == \"random\") {\n" } ]