src/test/java/org/apache/sysds/test/functions/aggregate/ColVariancesTest.java

/*
 * Licensed to the Apache Software Foundation (ASF) under one
 * or more contributor license agreements.  See the NOTICE file
 * distributed with this work for additional information
 * regarding copyright ownership.  The ASF licenses this file
 * to you under the Apache License, Version 2.0 (the
 * "License"); you may not use this file except in compliance
 * with the License.  You may obtain a copy of the License at
 *
 *   http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing,
 * software distributed under the License is distributed on an
 * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
 * KIND, either express or implied.  See the License for the
 * specific language governing permissions and limitations
 * under the License.
 */

package org.apache.sysds.test.functions.aggregate;

import org.junit.Assert;
import org.junit.Test;
import org.apache.sysds.api.DMLScript;
import org.apache.sysds.common.Types.ExecMode;
import org.apache.sysds.hops.OptimizerUtils;
import org.apache.sysds.lops.LopProperties.ExecType;
import org.apache.sysds.runtime.instructions.Instruction;
import org.apache.sysds.runtime.matrix.data.MatrixValue.CellIndex;
import org.apache.sysds.test.AutomatedTestBase;
import org.apache.sysds.test.TestConfiguration;
import org.apache.sysds.test.TestUtils;
import org.apache.sysds.utils.Statistics;

import java.util.HashMap;

/**
 * Test the column variances function, "colVars(X)".
 */
public class ColVariancesTest extends AutomatedTestBase {

    private static final String TEST_NAME = "ColVariances";
    private static final String TEST_DIR = "functions/aggregate/";
    private static final String TEST_CLASS_DIR =
            TEST_DIR + ColVariancesTest.class.getSimpleName() + "/";
    private static final String INPUT_NAME = "X";
    private static final String OUTPUT_NAME = "colVariances";

    private static final String colVarOp = "uacvar";
    private static final String varOp = "uavar";
    private static final int rows = 1234;
    private static final int cols = 1432;
    private static final double sparsitySparse = 0.2;
    private static final double sparsityDense = 0.7;
    private static final double eps = Math.pow(10, -10);

    private enum Sparsity {EMPTY, SPARSE, DENSE}
    private enum DataType {MATRIX, ROWVECTOR, COLUMNVECTOR}

    @Override
    public void setUp() {
        TestUtils.clearAssertionInformation();
        TestConfiguration config = new TestConfiguration(TEST_CLASS_DIR, TEST_NAME);
        addTestConfiguration(TEST_NAME, config);
    }

    // Dense matrix w/ rewrites
    @Test
    public void testColVariancesDenseMatrixRewritesCP() {
        testColVariances(TEST_NAME, Sparsity.DENSE, DataType.MATRIX, true, ExecType.CP);
    }

    @Test
    public void testColVariancesDenseMatrixRewritesSpark() {
        testColVariances(TEST_NAME, Sparsity.DENSE, DataType.MATRIX, true, ExecType.SPARK);
    }

    // Dense matrix w/o rewrites
    @Test
    public void testColVariancesDenseMatrixNoRewritesCP() {
        testColVariances(TEST_NAME, Sparsity.DENSE, DataType.MATRIX, false, ExecType.CP);
    }

    @Test
    public void testColVariancesDenseMatrixNoRewritesSpark() {
        testColVariances(TEST_NAME, Sparsity.DENSE, DataType.MATRIX, false, ExecType.SPARK);
    }

    // Dense row vector w/ rewrites
    @Test
    public void testColVariancesDenseRowVectorRewritesCP() {
        testColVariances(TEST_NAME, Sparsity.DENSE, DataType.ROWVECTOR, true, ExecType.CP);
    }

    @Test
    public void testColVariancesDenseRowVectorRewritesSpark() {
        testColVariances(TEST_NAME, Sparsity.DENSE, DataType.ROWVECTOR, true, ExecType.SPARK);
    }

    // Dense row vector w/o rewrites
    @Test
    public void testColVariancesDenseRowVectorNoRewritesCP() {
        testColVariances(TEST_NAME, Sparsity.DENSE, DataType.ROWVECTOR, false, ExecType.CP);
    }

    @Test
    public void testColVariancesDenseRowVectorNoRewritesSpark() {
        testColVariances(TEST_NAME, Sparsity.DENSE, DataType.ROWVECTOR, false, ExecType.SPARK);
    }

    // Dense column vector w/ rewrites
    @Test
    public void testColVariancesDenseColVectorRewritesCP() {
        testColVariances(TEST_NAME, Sparsity.DENSE, DataType.COLUMNVECTOR, true, ExecType.CP);
    }

    @Test
    public void testColVariancesDenseColVectorRewritesSpark() {
        testColVariances(TEST_NAME, Sparsity.DENSE, DataType.COLUMNVECTOR, true, ExecType.SPARK);
    }

    // Dense column vector w/o rewrites
    @Test
    public void testColVariancesDenseColVectorNoRewritesCP() {
        testColVariances(TEST_NAME, Sparsity.DENSE, DataType.COLUMNVECTOR, false, ExecType.CP);
    }

    @Test
    public void testColVariancesDenseColVectorNoRewritesSpark() {
        testColVariances(TEST_NAME, Sparsity.DENSE, DataType.COLUMNVECTOR, false, ExecType.SPARK);
    }

    // Sparse matrix w/ rewrites
    @Test
    public void testColVariancesSparseMatrixRewritesCP() {
        testColVariances(TEST_NAME, Sparsity.SPARSE, DataType.MATRIX, true, ExecType.CP);
    }

    @Test
    public void testColVariancesSparseMatrixRewritesSpark() {
        testColVariances(TEST_NAME, Sparsity.SPARSE, DataType.MATRIX, true, ExecType.SPARK);
    }

    // Sparse matrix w/o rewrites
    @Test
    public void testColVariancesSparseMatrixNoRewritesCP() {
        testColVariances(TEST_NAME, Sparsity.SPARSE, DataType.MATRIX, false, ExecType.CP);
    }

    @Test
    public void testColVariancesSparseMatrixNoRewritesSpark() {
        testColVariances(TEST_NAME, Sparsity.SPARSE, DataType.MATRIX, false, ExecType.SPARK);
    }

    // Sparse row vector w/ rewrites
    @Test
    public void testColVariancesSparseRowVectorRewritesCP() {
        testColVariances(TEST_NAME, Sparsity.SPARSE, DataType.ROWVECTOR, true, ExecType.CP);
    }

    @Test
    public void testColVariancesSparseRowVectorRewritesSpark() {
        testColVariances(TEST_NAME, Sparsity.SPARSE, DataType.ROWVECTOR, true, ExecType.SPARK);
    }

    // Sparse row vector w/o rewrites
    @Test
    public void testColVariancesSparseRowVectorNoRewritesCP() {
        testColVariances(TEST_NAME, Sparsity.SPARSE, DataType.ROWVECTOR, false, ExecType.CP);
    }

    @Test
    public void testColVariancesSparseRowVectorNoRewritesSpark() {
        testColVariances(TEST_NAME, Sparsity.SPARSE, DataType.ROWVECTOR, false, ExecType.SPARK);
    }

    // Sparse column vector w/ rewrites
    @Test
    public void testColVariancesSparseColVectorRewritesCP() {
        testColVariances(TEST_NAME, Sparsity.SPARSE, DataType.COLUMNVECTOR, true, ExecType.CP);
    }

    @Test
    public void testColVariancesSparseColVectorRewritesSpark() {
        testColVariances(TEST_NAME, Sparsity.SPARSE, DataType.COLUMNVECTOR, true, ExecType.SPARK);
    }

    // Sparse column vector w/o rewrites
    @Test
    public void testColVariancesSparseColVectorNoRewritesCP() {
        testColVariances(TEST_NAME, Sparsity.SPARSE, DataType.COLUMNVECTOR, false, ExecType.CP);
    }

    @Test
    public void testColVariancesSparseColVectorNoRewritesSpark() {
        testColVariances(TEST_NAME, Sparsity.SPARSE, DataType.COLUMNVECTOR, false, ExecType.SPARK);
    }
    
    // Empty matrix w/ rewrites
    @Test
    public void testColVariancesEmptyMatrixRewritesCP() {
        testColVariances(TEST_NAME, Sparsity.EMPTY, DataType.MATRIX, true, ExecType.CP);
    }

    @Test
    public void testColVariancesEmptyMatrixRewritesSpark() {
        testColVariances(TEST_NAME, Sparsity.EMPTY, DataType.MATRIX, true, ExecType.SPARK);
    }

    // Empty matrix w/o rewrites
    @Test
    public void testColVariancesEmptyMatrixNoRewritesCP() {
        testColVariances(TEST_NAME, Sparsity.EMPTY, DataType.MATRIX, false, ExecType.CP);
    }

    @Test
    public void testColVariancesEmptyMatrixNoRewritesSpark() {
        testColVariances(TEST_NAME, Sparsity.EMPTY, DataType.MATRIX, false, ExecType.SPARK);
    }

    // Empty row vector w/ rewrites
    @Test
    public void testColVariancesEmptyRowVectorRewritesCP() {
        testColVariances(TEST_NAME, Sparsity.EMPTY, DataType.ROWVECTOR, true, ExecType.CP);
    }

    @Test
    public void testColVariancesEmptyRowVectorRewritesSpark() {
        testColVariances(TEST_NAME, Sparsity.EMPTY, DataType.ROWVECTOR, true, ExecType.SPARK);
    }

    // Empty row vector w/o rewrites
    @Test
    public void testColVariancesEmptyRowVectorNoRewritesCP() {
        testColVariances(TEST_NAME, Sparsity.EMPTY, DataType.ROWVECTOR, false, ExecType.CP);
    }

    @Test
    public void testColVariancesEmptyRowVectorNoRewritesSpark() {
        testColVariances(TEST_NAME, Sparsity.EMPTY, DataType.ROWVECTOR, false, ExecType.SPARK);
    }

    // Empty column vector w/ rewrites
    @Test
    public void testColVariancesEmptyColVectorRewritesCP() {
        testColVariances(TEST_NAME, Sparsity.EMPTY, DataType.COLUMNVECTOR, true, ExecType.CP);
    }

    @Test
    public void testColVariancesEmptyColVectorRewritesSpark() {
        testColVariances(TEST_NAME, Sparsity.EMPTY, DataType.COLUMNVECTOR, true, ExecType.SPARK);
    }

    // Empty column vector w/o rewrites
    @Test
    public void testColVariancesEmptyColVectorNoRewritesCP() {
        testColVariances(TEST_NAME, Sparsity.EMPTY, DataType.COLUMNVECTOR, false, ExecType.CP);
    }

    @Test
    public void testColVariancesEmptyColVectorNoRewritesSpark() {
        testColVariances(TEST_NAME, Sparsity.EMPTY, DataType.COLUMNVECTOR, false, ExecType.SPARK);
    }

    /**
     * Test the column variances function, "colVars(X)", on
     * dense/sparse matrices/vectors on the CP/Spark/MR platforms.
     *
     * @param testName The name of this test case.
     * @param sparsity Selection between empty, sparse, and dense data.
     * @param dataType Selection between a matrix, a row vector, and a
     *                 column vector.
     * @param rewrites Whether or not to employ algebraic rewrites.
     * @param platform Selection between CP/Spark/MR platforms.
     */
    private void testColVariances(String testName, Sparsity sparsity, DataType dataType,
                                  boolean rewrites, ExecType platform) {
        // Configure settings for this test case
        boolean rewritesOld = OptimizerUtils.ALLOW_ALGEBRAIC_SIMPLIFICATION;
        OptimizerUtils.ALLOW_ALGEBRAIC_SIMPLIFICATION = rewrites;

        ExecMode platformOld = rtplatform;
        switch (platform) {
            case SPARK:
                rtplatform = ExecMode.SPARK;
                break;
            default:
                rtplatform = ExecMode.SINGLE_NODE;
                break;
        }

        boolean sparkConfigOld = DMLScript.USE_LOCAL_SPARK_CONFIG;
        if (rtplatform == ExecMode.SPARK)
            DMLScript.USE_LOCAL_SPARK_CONFIG = true;

        try {
            // Create and load test configuration
            getAndLoadTestConfiguration(testName);
            String HOME = SCRIPT_DIR + TEST_DIR;
            fullDMLScriptName = HOME + testName + ".dml";
            programArgs = new String[]{"-explain", "-stats", "-args",
                    input(INPUT_NAME), output(OUTPUT_NAME)};
            fullRScriptName = HOME + testName + ".R";
            rCmd = "Rscript" + " " + fullRScriptName + " " + inputDir() + " " + expectedDir();

            // Generate data
            // - sparsity
            double sparsityVal;
            switch (sparsity) {
                case EMPTY:
                    sparsityVal = 0;
                    break;
                case SPARSE:
                    sparsityVal = sparsitySparse;
                    break;
                case DENSE:
                default:
                    sparsityVal = sparsityDense;
            }
            // - size
            int r;
            int c;
            switch (dataType) {
                case ROWVECTOR:
                    r = 1;
                    c = cols;
                    break;
                case COLUMNVECTOR:
                    r = rows;
                    c = 1;
                    break;
                case MATRIX:
                default:
                    r = rows;
                    c = cols;
            }
            // - generation
            double[][] X = getRandomMatrix(r, c, -1, 1, sparsityVal, 7);
            writeInputMatrixWithMTD(INPUT_NAME, X, true);

            // Run DML and R scripts
            runTest(true, false, null, -1);
            runRScript(true);

            // Compare output matrices
            HashMap<CellIndex, Double> dmlfile = readDMLMatrixFromOutputDir(OUTPUT_NAME);
            HashMap<CellIndex, Double> rfile  = readRMatrixFromExpectedDir(OUTPUT_NAME);
            TestUtils.compareMatrices(dmlfile, rfile, eps, "Stat-DML", "Stat-R");

            // On CP and Spark modes, check that the following
            // rewrites occurred:
            //  - colVars(X), where X is a row vector, should be
            //    rewritten to an empty row vector of zeros.
            //  - colVars(X), where X is a column vector, should be
            //    rewritten to var(X).
            if (rewrites && (platform == ExecType.SPARK || platform == ExecType.CP)) {
                String prefix = (platform == ExecType.SPARK) ? Instruction.SP_INST_PREFIX : "";
                if (dataType == DataType.ROWVECTOR) {
                    String opcode = prefix + colVarOp;
                    boolean rewriteApplied = !Statistics.getCPHeavyHitterOpCodes().contains(opcode);
                    Assert.assertTrue("Rewrite not applied to row vector case.", rewriteApplied);

                } else if (dataType == DataType.COLUMNVECTOR) {
                    String opcode = prefix + varOp;
                    boolean rewriteApplied = Statistics.getCPHeavyHitterOpCodes().contains(opcode);
                    Assert.assertTrue("Rewrite not applied to column vector case.", rewriteApplied);
                }
            }
        }
        finally {
            // Reset settings
            OptimizerUtils.ALLOW_ALGEBRAIC_SIMPLIFICATION = rewritesOld;
            rtplatform = platformOld;
            DMLScript.USE_LOCAL_SPARK_CONFIG = sparkConfigOld;
        }
    }
}