Improve overlap percent estimation for low-density ranges in StatisticRange

Author: Anton Kutuzov

core/trino-main/src/main/java/io/trino/cost/StatisticRange.java

@@ -32,6 +32,7 @@ public class StatisticRange
 {
     private static final double INFINITE_TO_FINITE_RANGE_INTERSECT_OVERLAP_HEURISTIC_FACTOR = 0.25;
     private static final double INFINITE_TO_INFINITE_RANGE_INTERSECT_OVERLAP_HEURISTIC_FACTOR = 0.5;
+    private static final double DENSITY_HEURISTIC_THRESHOLD = 1e-3;
 
     // TODO unify field and method names with SymbolStatsEstimate
     /**
@@ -122,7 +123,15 @@ public double overlapPercentWith(StatisticRange other)
         if (isInfinite(length()) && isFinite(lengthOfIntersect)) {
             return INFINITE_TO_FINITE_RANGE_INTERSECT_OVERLAP_HEURISTIC_FACTOR;
         }
+
         if (lengthOfIntersect > 0) {
+            double thisDensity = this.distinctValues / length();
+            double otherDensity = other.distinctValues / other.length();
+            double minDensity = minExcludeNaN(thisDensity, otherDensity);
+
+            if (!isNaN(minDensity) && minDensity < DENSITY_HEURISTIC_THRESHOLD && isFinite(length())) {
+                return minExcludeNaN(this.distinctValues, other.distinctValues) / this.distinctValues;
+            }
             return lengthOfIntersect / length();
         }
 

core/trino-main/src/test/java/io/trino/cost/TestStatisticRange.java

@@ -20,6 +20,7 @@
 import static java.lang.Double.NaN;
 import static java.lang.Double.POSITIVE_INFINITY;
 import static org.assertj.core.api.Assertions.assertThat;
+import static org.assertj.core.api.AssertionsForClassTypes.within;
 
 public class TestStatisticRange
 {
@@ -59,6 +60,55 @@ public void testOverlapPercentWith()
         assertOverlap(unboundedRange(0.0), unboundedRange(0), 0);
     }
 
+    @Test
+    public void testLowDensityOverlap()
+    {
+        StatisticRange sparseRange = range(1, 3662098119.0, 14);
+        StatisticRange filterRange = range(1, 4, 4);
+
+        double expectedOverlap = 4.0 / 14.0;
+        assertOverlap(sparseRange, filterRange, expectedOverlap);
+    }
+
+    @Test
+    public void testDensityThresholdBoundary()
+    {
+        StatisticRange boundaryRange = range(0, 10000, 10);
+        StatisticRange smallFilter = range(0, 100, 5);
+
+        double overlap = boundaryRange.overlapPercentWith(smallFilter);
+        assertThat(overlap).isBetween(0.01, 0.5);
+    }
+
+    @Test
+    public void testHighDensityOverlap()
+    {
+        StatisticRange denseRange = range(0, 100, 50);
+        StatisticRange filterRange = range(20, 30, 5);
+
+        assertOverlap(denseRange, filterRange, 0.1);
+    }
+
+    @Test
+    public void testVeryLowDensity()
+    {
+        StatisticRange verySparse = range(0, 1e9, 10);
+        StatisticRange filterRange = range(100, 200, 5);
+
+        double expected = 5.0 / 10.0;
+        double actual = verySparse.overlapPercentWith(filterRange);
+        assertThat(actual).isCloseTo(expected, within(0.1));
+    }
+
+    @Test
+    public void testDensityWithZeroDistinctValues()
+    {
+        StatisticRange zeroDistinct = range(0, 1000, 0);
+        StatisticRange filterRange = range(100, 200, 5);
+
+        assertOverlap(zeroDistinct, filterRange, 0);
+    }
+
     @Test
     public void testIntersect()
     {