private static void generateSketches() throws Exception { Path path = FileSystems.getDefault().getPath("array_of_doubles_sketch_data.tsv"); try (BufferedWriter out = Files.newBufferedWriter(path, StandardCharsets.UTF_8)) { Random rand = ThreadLocalRandom.current(); int key = 0; for (int i = 0; i < 20; i++) { ArrayOfDoublesUpdatableSketch sketch = new ArrayOfDoublesUpdatableSketchBuilder().setNominalEntries(1024) .build(); sketch.update(key++, new double[] {1}); sketch.update(key++, new double[] {1}); out.write("2015010101"); out.write('\t'); out.write("product_" + (rand.nextInt(10) + 1)); out.write('\t'); out.write(StringUtils.encodeBase64String(sketch.compact().toByteArray())); out.newLine(); } } }
@Test(expectedExceptions = SketchesArgumentException.class) public void deserializeWithWrongSeed() { ArrayOfDoublesUpdatableSketch us = new ArrayOfDoublesUpdatableSketchBuilder().build(); for (int i = 0; i < 8192; i++) { us.update(i, new double[] {1.0}); } ArrayOfDoublesCompactSketch sketch1 = us.compact(); ArrayOfDoublesSketches.heapifySketch(Memory.wrap(sketch1.toByteArray()), 123); } }
@Test public void serializeDeserializeEstimation() { ArrayOfDoublesUpdatableSketch us = new ArrayOfDoublesUpdatableSketchBuilder().build(WritableMemory.wrap(new byte[1000000])); for (int i = 0; i < 8192; i++) { us.update(i, new double[] {1.0}); } ArrayOfDoublesCompactSketch sketch1 = us.compact(WritableMemory.wrap(new byte[1000000])); ArrayOfDoublesSketch sketch2 = ArrayOfDoublesSketches.wrapSketch(WritableMemory.wrap(sketch1.toByteArray())); Assert.assertFalse(sketch2.isEmpty()); Assert.assertTrue(sketch2.isEstimationMode()); Assert.assertEquals(sketch2.getEstimate(), sketch1.getEstimate()); Assert.assertEquals(sketch2.getThetaLong(), sketch1.getThetaLong()); }
@Test public void serializeDeserializeSmallExact() { ArrayOfDoublesUpdatableSketch us = new ArrayOfDoublesUpdatableSketchBuilder().build(); us.update("a", new double[] {1.0}); us.update("b", new double[] {1.0}); us.update("c", new double[] {1.0}); ArrayOfDoublesCompactSketch sketch1 = us.compact(); ArrayOfDoublesSketch sketch2 = ArrayOfDoublesSketches.heapifySketch(Memory.wrap(sketch1.toByteArray())); Assert.assertFalse(sketch2.isEmpty()); Assert.assertFalse(sketch2.isEstimationMode()); Assert.assertEquals(sketch2.getEstimate(), 3.0); Assert.assertEquals(sketch2.getLowerBound(1), 3.0); Assert.assertEquals(sketch2.getUpperBound(1), 3.0); Assert.assertEquals(sketch2.getRetainedEntries(), 3); Assert.assertEquals(sketch2.getThetaLong(), Long.MAX_VALUE); Assert.assertEquals(sketch2.getTheta(), 1.0); double[][] values = sketch2.getValues(); Assert.assertEquals(values.length, 3); for (double[] array: values) { Assert.assertEquals(array[0], 1.0); } }
@Test public void serializeDeserializeSmallExact() { ArrayOfDoublesUpdatableSketch us = new ArrayOfDoublesUpdatableSketchBuilder().build(WritableMemory.wrap(new byte[1000000])); us.update("a", new double[] {1.0}); us.update("b", new double[] {1.0}); us.update("c", new double[] {1.0}); ArrayOfDoublesCompactSketch sketch1 = us.compact(WritableMemory.wrap(new byte[1000000])); ArrayOfDoublesSketch sketch2 = ArrayOfDoublesSketches.wrapSketch(WritableMemory.wrap(sketch1.toByteArray())); Assert.assertFalse(sketch2.isEmpty()); Assert.assertFalse(sketch2.isEstimationMode()); Assert.assertEquals(sketch2.getEstimate(), 3.0); Assert.assertEquals(sketch2.getLowerBound(1), 3.0); Assert.assertEquals(sketch2.getUpperBound(1), 3.0); Assert.assertEquals(sketch2.getRetainedEntries(), 3); Assert.assertEquals(sketch2.getThetaLong(), Long.MAX_VALUE); Assert.assertEquals(sketch2.getTheta(), 1.0); double[][] values = sketch2.getValues(); Assert.assertEquals(values.length, 3); for (double[] array: values) { Assert.assertEquals(array[0], 1.0); } }
@Test(expectedExceptions = SketchesArgumentException.class) public void deserializeWithWrongSeed() { ArrayOfDoublesUpdatableSketch us = new ArrayOfDoublesUpdatableSketchBuilder().build(WritableMemory.wrap(new byte[1000000])); for (int i = 0; i < 8192; i++) { us.update(i, new double[] {1.0}); } ArrayOfDoublesCompactSketch sketch1 = us.compact(WritableMemory.wrap(new byte[1000000])); ArrayOfDoublesSketches.wrapSketch(WritableMemory.wrap(sketch1.toByteArray()), 123); }
@Test public void heapToDirectExactTwoDoubles() { ArrayOfDoublesUpdatableSketch sketch1 = new ArrayOfDoublesUpdatableSketchBuilder().setNumberOfValues(2).build(); sketch1.update("a", new double[] {1, 2}); sketch1.update("b", new double[] {1, 2}); sketch1.update("c", new double[] {1, 2}); sketch1.update("d", new double[] {1, 2}); sketch1.update("a", new double[] {1, 2}); sketch1.update("b", new double[] {1, 2}); sketch1.update("c", new double[] {1, 2}); sketch1.update("d", new double[] {1, 2}); ArrayOfDoublesSketch sketch2 = new DirectArrayOfDoublesCompactSketch(Memory.wrap(sketch1.compact().toByteArray())); Assert.assertFalse(sketch2.isEmpty()); Assert.assertFalse(sketch2.isEstimationMode()); Assert.assertEquals(sketch2.getEstimate(), 4.0); Assert.assertEquals(sketch2.getUpperBound(1), 4.0); Assert.assertEquals(sketch2.getLowerBound(1), 4.0); Assert.assertEquals(sketch2.getThetaLong(), Long.MAX_VALUE); Assert.assertEquals(sketch2.getTheta(), 1.0); double[][] values = sketch2.getValues(); Assert.assertEquals(values.length, 4); for (double[] array: values) { Assert.assertEquals(array.length, 2); Assert.assertEquals(array[0], 2.0); Assert.assertEquals(array[1], 4.0); } }
@Test public void serializeDeserializeEstimation() { ArrayOfDoublesUpdatableSketch us = new ArrayOfDoublesUpdatableSketchBuilder().build(); for (int i = 0; i < 8192; i++) { us.update(i, new double[] {1.0}); } WritableMemory wmem = WritableMemory.wrap(us.toByteArray()); ArrayOfDoublesUpdatableSketch wrappedUS = ArrayOfDoublesSketches.wrapUpdatableSketch(wmem); Assert.assertFalse(wrappedUS.isEmpty()); Assert.assertTrue(wrappedUS.isEstimationMode()); Assert.assertEquals(wrappedUS.getEstimate(), us.getEstimate()); Assert.assertEquals(wrappedUS.getThetaLong(), us.getThetaLong()); ArrayOfDoublesUpdatableSketch heapUS = ArrayOfDoublesSketches.heapifyUpdatableSketch(wmem); Assert.assertFalse(heapUS.isEmpty()); Assert.assertTrue(heapUS.isEstimationMode()); Assert.assertEquals(heapUS.getEstimate(), us.getEstimate()); Assert.assertEquals(heapUS.getThetaLong(), us.getThetaLong()); ArrayOfDoublesCompactSketch sketch1 = us.compact(); ArrayOfDoublesSketch sketch2 = ArrayOfDoublesSketches.heapifySketch(Memory.wrap(sketch1.toByteArray())); Assert.assertFalse(sketch2.isEmpty()); Assert.assertTrue(sketch2.isEstimationMode()); Assert.assertEquals(sketch2.getEstimate(), sketch1.getEstimate()); Assert.assertEquals(sketch2.getThetaLong(), sketch1.getThetaLong()); }
@Test public void directToHeapExactTwoDoubles() { ArrayOfDoublesUpdatableSketch sketch1 = new ArrayOfDoublesUpdatableSketchBuilder().setNumberOfValues(2).build(WritableMemory.wrap(new byte[1000000])); sketch1.update("a", new double[] {1, 2}); sketch1.update("b", new double[] {1, 2}); sketch1.update("c", new double[] {1, 2}); sketch1.update("d", new double[] {1, 2}); sketch1.update("a", new double[] {1, 2}); sketch1.update("b", new double[] {1, 2}); sketch1.update("c", new double[] {1, 2}); sketch1.update("d", new double[] {1, 2}); ArrayOfDoublesSketch sketch2 = new HeapArrayOfDoublesCompactSketch(Memory.wrap(sketch1.compact(WritableMemory.wrap(new byte[1000000])).toByteArray())); Assert.assertFalse(sketch2.isEmpty()); Assert.assertFalse(sketch2.isEstimationMode()); Assert.assertEquals(sketch2.getEstimate(), 4.0); Assert.assertEquals(sketch2.getUpperBound(1), 4.0); Assert.assertEquals(sketch2.getLowerBound(1), 4.0); Assert.assertEquals(sketch2.getThetaLong(), Long.MAX_VALUE); Assert.assertEquals(sketch2.getTheta(), 1.0); double[][] values = sketch2.getValues(); Assert.assertEquals(values.length, 4); for (double[] array: values) { Assert.assertEquals(array.length, 2); Assert.assertEquals(array[0], 2.0); Assert.assertEquals(array[1], 4.0); } } }