fixes #7

There were two problems.

1) Math.pow(2,32) == Math.pow(-2,32) in java, not good...
2) The long range correction algorithm reduces precision in some cases
so I've given the caller the option of disabling this function

Author: abramsm

src/main/java/com/clearspring/analytics/stream/cardinality/HyperLogLog.java

@@ -54,8 +54,8 @@
  */
 public class HyperLogLog implements ICardinality
 {
-    private final static int POW_2_32 = (int) Math.pow(2, 32);
-    private final static int NEGATIVE_POW_2_32 = (int) Math.pow(-2, 32);
+    private final static double POW_2_32 = Math.pow(2, 32);
+    private final static double NEGATIVE_POW_2_32 = - -4294967296.0;
 
     private final RegisterSet registerSet;
     private final int log2m;
@@ -180,6 +180,11 @@ public boolean offer(Object o)
 
     @Override
     public long cardinality()
+    {
+        return cardinality(true);
+    }
+
+    public long cardinality(boolean enableLongRangeCorrection)
     {
         double registerSum = 0;
         int count = registerSet.count;
@@ -211,7 +216,14 @@ else if (estimate <= (1.0 / 30.0) * POW_2_32)
         else if (estimate > (1.0 / 30.0) * POW_2_32)
         {
             // Large Range Estimate
-            return Math.round((NEGATIVE_POW_2_32 * Math.log(1 - (estimate / POW_2_32))));
+            if (enableLongRangeCorrection)
+            {
+                return Math.round((NEGATIVE_POW_2_32 * Math.log(1.0 - (estimate / POW_2_32))));
+            }
+            else
+            {
+                return Math.round(estimate);
+            }
         }
         return 0;
     }

src/test/java/com/clearspring/analytics/stream/cardinality/TestHyperLogLog.java

@@ -122,4 +122,26 @@ public void testMerge() throws CardinalityMergeException
         assertTrue(mergedEstimate >= expectedCardinality - (3 * se));
         assertTrue(mergedEstimate <= expectedCardinality + (3 * se));
     }
+
+    @Test
+    public void testPrecise_disableLongRangeCorrection() throws CardinalityMergeException
+    {
+        int cardinality = 150000000;
+
+        HyperLogLog baseline = new HyperLogLog(20);
+        for (int j = 0; j < cardinality; j++)
+        {
+            double val = Math.random();
+            baseline.offer(val);
+        }
+
+
+        long mergedEstimate = baseline.cardinality(false);
+        double se = cardinality * (1.04 / Math.sqrt(Math.pow(2, 20)));
+
+        System.out.println("Expect estimate: " + mergedEstimate + " is between " + (cardinality - (3 * se)) + " and " + (cardinality + (3 * se)));
+
+        assertTrue(mergedEstimate >= cardinality - (3 * se));
+        assertTrue(mergedEstimate <= cardinality + (3 * se));
+    }
 }