Change default NDV to 1024

Signed-off-by: Chen Dai <[email protected]>
opensearch-project · Mar 8, 2024 · 11be2f9 · 11be2f9
1 parent c697cf3
commit 11be2f9
Show file tree

Hide file tree

Showing 3 changed files with 12 additions and 11 deletions.
diff --git a/...c/main/java/org/opensearch/flint/core/field/bloomfilter/adaptive/AdaptiveBloomFilter.java b/...c/main/java/org/opensearch/flint/core/field/bloomfilter/adaptive/AdaptiveBloomFilter.java
@@ -34,7 +34,7 @@ public class AdaptiveBloomFilter implements BloomFilter {
   public AdaptiveBloomFilter(int numCandidate, double fpp) {
     this.candidates = new BloomFilterCandidate[numCandidate];
 
-    int expectedNumItems = 128;
+    int expectedNumItems = 1024;
     for (int i = 0; i < candidates.length; i++) {
       candidates[i] =
           new BloomFilterCandidate(
@@ -48,7 +48,7 @@ public AdaptiveBloomFilter(int numCandidate, double fpp) {
     this.total = total;
     this.candidates = new BloomFilterCandidate[candidates.length];
 
-    int expectedNumItems = 128;
+    int expectedNumItems = 1024;
     for (int i = 0; i < candidates.length; i++) {
       this.candidates[i] = new BloomFilterCandidate(expectedNumItems, candidates[i]);
       expectedNumItems *= 2;

diff --git a/...st/java/org/opensearch/flint/core/field/bloomfilter/adaptive/AdaptiveBloomFilterTest.java b/...st/java/org/opensearch/flint/core/field/bloomfilter/adaptive/AdaptiveBloomFilterTest.java
@@ -24,26 +24,26 @@ public class AdaptiveBloomFilterTest {
 
   @Test
   public void shouldChooseBestCandidateAdaptively() {
-    // Insert 50 items
-    for (int i = 0; i < 50; i++) {
+    // Insert 500 items
+    for (int i = 0; i < 500; i++) {
       bloomFilter.put(i);
     }
     BloomFilterCandidate candidate1 = bloomFilter.bestCandidate();
-    assertEquals(128, candidate1.expectedNumItems);
+    assertEquals(1024, candidate1.expectedNumItems);
 
-    // Insert 100 (total 150)
-    for (int i = 50; i < 150; i++) {
+    // Insert 1000 (total 1500)
+    for (int i = 500; i < 1500; i++) {
       bloomFilter.put(i);
     }
     BloomFilterCandidate candidate2 = bloomFilter.bestCandidate();
-    assertEquals(256, candidate2.expectedNumItems);
+    assertEquals(2048, candidate2.expectedNumItems);
 
-    // Insert 400 (total 550)
-    for (int i = 150; i < 550; i++) {
+    // Insert 4000 (total 5500)
+    for (int i = 1500; i < 5500; i++) {
       bloomFilter.put(i);
     }
     BloomFilterCandidate candidate3 = bloomFilter.bestCandidate();
-    assertEquals(1024, candidate3.expectedNumItems);
+    assertEquals(8192, candidate3.expectedNumItems);
   }
 
   @Test

diff --git a/...la/org/opensearch/flint/spark/skipping/bloomfilter/BloomFilterSkippingStrategySuite.scala b/...la/org/opensearch/flint/spark/skipping/bloomfilter/BloomFilterSkippingStrategySuite.scala
@@ -23,6 +23,7 @@ class BloomFilterSkippingStrategySuite
   test("parameters") {
     strategy.parameters shouldBe Map(
       BLOOM_FILTER_ADAPTIVE_KEY -> DEFAULT_BLOOM_FILTER_ADAPTIVE.toString,
+      ADAPTIVE_NUMBER_CANDIDATE_KEY -> DEFAULT_ADAPTIVE_NUMBER_CANDIDATE.toString,
       CLASSIC_BLOOM_FILTER_FPP_KEY -> DEFAULT_CLASSIC_BLOOM_FILTER_FPP.toString)
   }
 }