Add implementation for exponential histogram merging and percentiles

benchmarks/build.gradle

@@ -49,6 +49,7 @@ dependencies {
   api(project(':x-pack:plugin:esql:compute'))
   implementation project(path: ':libs:native')
   implementation project(path: ':libs:simdvec')
+  implementation project(path: ':libs:exponential-histogram')
   expression(project(path: ':modules:lang-expression', configuration: 'zip'))
   painless(project(path: ':modules:lang-painless', configuration: 'zip'))
   nativeLib(project(':libs:native'))

benchmarks/src/main/java/org/elasticsearch/benchmark/exponentialhistogram/ExponentialHistogramGenerationBench.java

@@ -0,0 +1,98 @@
+/*
+ * Copyright Elasticsearch B.V. and/or licensed to Elasticsearch B.V. under one
+ * or more contributor license agreements. Licensed under the "Elastic License
+ * 2.0", the "GNU Affero General Public License v3.0 only", and the "Server Side
+ * Public License v 1"; you may not use this file except in compliance with, at
+ * your election, the "Elastic License 2.0", the "GNU Affero General Public
+ * License v3.0 only", or the "Server Side Public License, v 1".
+ */
+
+package org.elasticsearch.benchmark.exponentialhistogram;
+
+import org.elasticsearch.exponentialhistogram.ExponentialHistogramGenerator;
+import org.openjdk.jmh.annotations.Benchmark;
+import org.openjdk.jmh.annotations.BenchmarkMode;
+import org.openjdk.jmh.annotations.Fork;
+import org.openjdk.jmh.annotations.Measurement;
+import org.openjdk.jmh.annotations.Mode;
+import org.openjdk.jmh.annotations.OutputTimeUnit;
+import org.openjdk.jmh.annotations.Param;
+import org.openjdk.jmh.annotations.Scope;
+import org.openjdk.jmh.annotations.Setup;
+import org.openjdk.jmh.annotations.State;
+import org.openjdk.jmh.annotations.Threads;
+import org.openjdk.jmh.annotations.Warmup;
+import org.openjdk.jmh.profile.GCProfiler;
+import org.openjdk.jmh.profile.StackProfiler;
+import org.openjdk.jmh.runner.Runner;
+import org.openjdk.jmh.runner.RunnerException;
+import org.openjdk.jmh.runner.options.Options;
+import org.openjdk.jmh.runner.options.OptionsBuilder;
+
+import java.util.Random;
+import java.util.concurrent.ThreadLocalRandom;
+import java.util.concurrent.TimeUnit;
+import java.util.function.DoubleSupplier;
+
+@BenchmarkMode(Mode.AverageTime)
+@OutputTimeUnit(TimeUnit.NANOSECONDS)
+@Warmup(iterations = 3, time = 3, timeUnit = TimeUnit.SECONDS)
+@Measurement(iterations = 5, time = 2, timeUnit = TimeUnit.SECONDS)
+@Fork(1)
+@Threads(1)
+@State(Scope.Thread)
+public class ExponentialHistogramGenerationBench {
+
+    @Param({ "100", "500", "1000", "5000", "10000", "20000" })
+    int bucketCount;
+
+    @Param({ "NORMAL", "GAUSSIAN" })
+    String distribution;
+
+    Random random;
+    ExponentialHistogramGenerator histoGenerator;
+
+    double[] data = new double[1000000];
+
+    int index;
+
+    @Setup
+    public void setUp() {
+        random = ThreadLocalRandom.current();
+        histoGenerator = new ExponentialHistogramGenerator(bucketCount);
+
+        DoubleSupplier nextRandom = () -> distribution.equals("GAUSSIAN") ? random.nextGaussian() : random.nextDouble();
+
+        // Make sure that we start with a non-empty histogram, as this distorts initial additions
+        for (int i = 0; i < 10000; ++i) {
+            histoGenerator.add(nextRandom.getAsDouble());
+        }
+
+        for (int i = 0; i < data.length; ++i) {
+            data[i] = nextRandom.getAsDouble();
+        }
+
+        index = 0;
+    }
+
+    @Benchmark
+    @BenchmarkMode(Mode.AverageTime)
+    @OutputTimeUnit(TimeUnit.MICROSECONDS)
+    public void add() {
+        if (index >= data.length) {
+            index = 0;
+        }
+        histoGenerator.add(data[index++]);
+    }
+
+    public static void main(String[] args) throws RunnerException {
+        Options opt = new OptionsBuilder().include(".*" + ExponentialHistogramGenerationBench.class.getSimpleName() + ".*")
+            .warmupIterations(5)
+            .measurementIterations(5)
+            .addProfiler(GCProfiler.class)
+            .addProfiler(StackProfiler.class)
+            .build();
+
+        new Runner(opt).run();
+    }
+}

benchmarks/src/main/java/org/elasticsearch/benchmark/exponentialhistogram/ExponentialHistogramMergeBench.java

@@ -0,0 +1,112 @@
+/*
+ * Copyright Elasticsearch B.V. and/or licensed to Elasticsearch B.V. under one
+ * or more contributor license agreements. Licensed under the "Elastic License
+ * 2.0", the "GNU Affero General Public License v3.0 only", and the "Server Side
+ * Public License v 1"; you may not use this file except in compliance with, at
+ * your election, the "Elastic License 2.0", the "GNU Affero General Public
+ * License v3.0 only", or the "Server Side Public License, v 1".
+ */
+
+package org.elasticsearch.benchmark.exponentialhistogram;
+
+import org.elasticsearch.exponentialhistogram.BucketIterator;
+import org.elasticsearch.exponentialhistogram.ExponentialHistogram;
+import org.elasticsearch.exponentialhistogram.ExponentialHistogramGenerator;
+import org.elasticsearch.exponentialhistogram.ExponentialHistogramMerger;
+import org.openjdk.jmh.annotations.Benchmark;
+import org.openjdk.jmh.annotations.BenchmarkMode;
+import org.openjdk.jmh.annotations.Fork;
+import org.openjdk.jmh.annotations.Measurement;
+import org.openjdk.jmh.annotations.Mode;
+import org.openjdk.jmh.annotations.OutputTimeUnit;
+import org.openjdk.jmh.annotations.Param;
+import org.openjdk.jmh.annotations.Scope;
+import org.openjdk.jmh.annotations.Setup;
+import org.openjdk.jmh.annotations.State;
+import org.openjdk.jmh.annotations.Threads;
+import org.openjdk.jmh.annotations.Warmup;
+
+import java.util.List;
+import java.util.Random;
+import java.util.concurrent.ThreadLocalRandom;
+import java.util.concurrent.TimeUnit;
+
+@BenchmarkMode(Mode.AverageTime)
+@OutputTimeUnit(TimeUnit.NANOSECONDS)
+@Warmup(iterations = 3, time = 3, timeUnit = TimeUnit.SECONDS)
+@Measurement(iterations = 5, time = 2, timeUnit = TimeUnit.SECONDS)
+@Fork(1)
+@Threads(1)
+@State(Scope.Thread)
+public class ExponentialHistogramMergeBench {
+
+    @Param({ "1000", "2000", "5000" })
+    int bucketCount;
+
+    @Param({ "0.01", "0.1", "0.25", "0.5", "1.0", "2.0" })
+    double mergedHistoSizeFactor;
+
+    Random random;
+    ExponentialHistogramMerger histoMerger;
+
+    ExponentialHistogram[] toMerge = new ExponentialHistogram[10_000];
+
+    int index;
+
+    @Setup
+    public void setUp() {
+        random = ThreadLocalRandom.current();
+        histoMerger = new ExponentialHistogramMerger(bucketCount);
+
+        ExponentialHistogramGenerator initial = new ExponentialHistogramGenerator(bucketCount);
+        for (int j = 0; j < bucketCount; j++) {
+            initial.add(Math.pow(1.001, j));
+        }
+        ExponentialHistogram initialHisto = initial.get();
+        int cnt = getBucketCount(initialHisto);
+        if (cnt < bucketCount) {
+            throw new IllegalArgumentException("Expected bucket count to be " + bucketCount + ", but was " + cnt);
+        }
+        histoMerger.add(initialHisto);
+
+        int dataPointSize = (int) Math.round(bucketCount * mergedHistoSizeFactor);
+
+        for (int i = 0; i < toMerge.length; i++) {
+            ExponentialHistogramGenerator generator = new ExponentialHistogramGenerator(dataPointSize);
+
+            int bucketIndex = 0;
+            for (int j = 0; j < dataPointSize; j++) {
+                bucketIndex += 1 + random.nextInt(bucketCount) % (Math.max(1, bucketCount / dataPointSize));
+                generator.add(Math.pow(1.001, bucketIndex));
+            }
+            toMerge[i] = generator.get();
+            cnt = getBucketCount(toMerge[i]);
+            if (cnt < dataPointSize) {
+                throw new IllegalArgumentException("Expected bucket count to be " + dataPointSize + ", but was " + cnt);
+            }
+        }
+
+        index = 0;
+    }
+
+    private static int getBucketCount(ExponentialHistogram histo) {
+        int cnt = 0;
+        for (BucketIterator it : List.of(histo.negativeBuckets().iterator(), histo.positiveBuckets().iterator())) {
+            while (it.hasNext()) {
+                cnt++;
+                it.advance();
+            }
+        }
+        return cnt;
+    }
+
+    @Benchmark
+    @BenchmarkMode(Mode.AverageTime)
+    @OutputTimeUnit(TimeUnit.MICROSECONDS)
+    public void add() {
+        if (index >= toMerge.length) {
+            index = 0;
+        }
+        histoMerger.add(toMerge[index++]);
+    }
+}

gradle/verification-metadata.xml

@@ -66,6 +66,11 @@
             <sha256 value="3366d2c88fb576e486d830f521184e8f1839f8c15dcd2151a3f6e1f62b0b37a0" origin="Generated by Gradle"/>
          </artifact>
       </component>
+      <component group="ch.obermuhlner" name="big-math" version="2.3.2">
+         <artifact name="big-math-2.3.2.jar">
+            <sha256 value="693e1bb7c7f5184b448f03c2a2c0c45d07d8e89e4641fdc31ab0a8057027f43d" origin="Generated by Gradle"/>
+         </artifact>
+      </component>
       <component group="ch.randelshofer" name="fastdoubleparser" version="0.8.0">
          <artifact name="fastdoubleparser-0.8.0.jar">
             <sha256 value="10fe288fd7a2cdaf5175332b73529f9abf8fd54dcfff317d6967c0c35ffb133b" origin="Generated by Gradle"/>