[commons-statistics] 08/08: Use a single instance of the standard normal distribution

[ah...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

aherbert pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/commons-statistics.git

commit 81cbacd3f43fb3ec2c188143fb641c36e22a3853
Author: aherbert <ah...@apache.org>
AuthorDate: Tue Sep 14 13:07:36 2021 +0100

    Use a single instance of the standard normal distribution
    
    The distribution is immutable. The underlying computations use static
    methods and can be used across threads.
    
    Note: A micro-optimisation to eliminate the mean=0 and SD=1 from the
    computations used from NormalDistribution(0, 1) has not been performed
    to aid maintenance. It is presumed that multiplication by 1 or addition
    of 0 is insignificant compared to the calls to:
    
    - Math.exp (density)
    - Erf.value (cumulativeProbability)
    - InverseErf.value (inverseCumulativeProbability)
    
    Only the logDensity function would be expected to improve performance as
    it would eliminate 1 subtraction and 1 devision to leave the computation
    as 2 multiplications and a subtraction:
    
    return -0.5 * x * x - halfLog2Pi
---
 .../distribution/TruncatedNormalDistribution.java  | 23 +++++++++++-----------
 1 file changed, 12 insertions(+), 11 deletions(-)

diff --git a/commons-statistics-distribution/src/main/java/org/apache/commons/statistics/distribution/TruncatedNormalDistribution.java b/commons-statistics-distribution/src/main/java/org/apache/commons/statistics/distribution/TruncatedNormalDistribution.java
index 4f17d8b..08751ea 100644
--- a/commons-statistics-distribution/src/main/java/org/apache/commons/statistics/distribution/TruncatedNormalDistribution.java
+++ b/commons-statistics-distribution/src/main/java/org/apache/commons/statistics/distribution/TruncatedNormalDistribution.java
@@ -24,6 +24,10 @@ package org.apache.commons.statistics.distribution;
  * Truncated normal distribution (Wikipedia)</a>
  */
 public class TruncatedNormalDistribution extends AbstractContinuousDistribution {
+    /** A standard normal distribution used for calculations.
+     * This is immutable and thread-safe and can be used across instances. */
+    private static final NormalDistribution STANDARD_NORMAL = new NormalDistribution(0, 1);
+
     /** Mean of parent normal distribution. */
     private final double parentMean;
     /** Standard deviation of parent normal distribution. */
@@ -37,8 +41,6 @@ public class TruncatedNormalDistribution extends AbstractContinuousDistribution
     /** Upper bound of this distribution. */
     private final double upper;
 
-    /** A standard normal distribution used for calculations. */
-    private final NormalDistribution standardNormal;
     /** Stored value of @{code standardNormal.cumulativeProbability((lower - mean) / sd)} for faster computations. */
     private final double cdfAlpha;
     /**
@@ -75,21 +77,20 @@ public class TruncatedNormalDistribution extends AbstractContinuousDistribution
 
         parentMean = mean;
         parentSd = sd;
-        standardNormal = new NormalDistribution(0, 1);
 
         final double alpha = (lower - mean) / sd;
         final double beta = (upper - mean) / sd;
 
-        final double cdfBeta = standardNormal.cumulativeProbability(beta);
-        cdfAlpha = standardNormal.cumulativeProbability(alpha);
+        final double cdfBeta = STANDARD_NORMAL.cumulativeProbability(beta);
+        cdfAlpha = STANDARD_NORMAL.cumulativeProbability(alpha);
         cdfDelta = cdfBeta - cdfAlpha;
 
         parentSdByCdfDelta = parentSd * cdfDelta;
         logParentSdByCdfDelta = Math.log(parentSdByCdfDelta);
 
         // Calculation of variance and mean.
-        final double pdfAlpha = standardNormal.density(alpha);
-        final double pdfBeta = standardNormal.density(beta);
+        final double pdfAlpha = STANDARD_NORMAL.density(alpha);
+        final double pdfBeta = STANDARD_NORMAL.density(beta);
         final double pdfCdfDelta = (pdfAlpha - pdfBeta) / cdfDelta;
         final double alphaBetaDelta = (alpha * pdfAlpha - beta * pdfBeta) / cdfDelta;
 
@@ -124,7 +125,7 @@ public class TruncatedNormalDistribution extends AbstractContinuousDistribution
         if (x < lower || x > upper) {
             return 0;
         }
-        return standardNormal.density((x - parentMean) / parentSd) / parentSdByCdfDelta;
+        return STANDARD_NORMAL.density((x - parentMean) / parentSd) / parentSdByCdfDelta;
     }
 
     /** {@inheritDoc} */
@@ -133,7 +134,7 @@ public class TruncatedNormalDistribution extends AbstractContinuousDistribution
         if (x < lower || x > upper) {
             return Double.NEGATIVE_INFINITY;
         }
-        return standardNormal.logDensity((x - parentMean) / parentSd) - logParentSdByCdfDelta;
+        return STANDARD_NORMAL.logDensity((x - parentMean) / parentSd) - logParentSdByCdfDelta;
     }
 
     /** {@inheritDoc} */
@@ -144,7 +145,7 @@ public class TruncatedNormalDistribution extends AbstractContinuousDistribution
         } else if (x >= upper) {
             return 1;
         }
-        return (standardNormal.cumulativeProbability((x - parentMean) / parentSd) - cdfAlpha) / cdfDelta;
+        return (STANDARD_NORMAL.cumulativeProbability((x - parentMean) / parentSd) - cdfAlpha) / cdfDelta;
     }
 
     /** {@inheritDoc} */
@@ -159,7 +160,7 @@ public class TruncatedNormalDistribution extends AbstractContinuousDistribution
         } else if (p == 1) {
             return upper;
         }
-        final double x = standardNormal.inverseCumulativeProbability(cdfAlpha + p * cdfDelta) * parentSd + parentMean;
+        final double x = STANDARD_NORMAL.inverseCumulativeProbability(cdfAlpha + p * cdfDelta) * parentSd + parentMean;
         // Clip to support to handle floating-point error at the support bound
         if (x <= lower) {
             return lower;