001/* 002 * Licensed to the Apache Software Foundation (ASF) under one or more 003 * contributor license agreements. See the NOTICE file distributed with 004 * this work for additional information regarding copyright ownership. 005 * The ASF licenses this file to You under the Apache License, Version 2.0 006 * (the "License"); you may not use this file except in compliance with 007 * the License. You may obtain a copy of the License at 008 * 009 * http://www.apache.org/licenses/LICENSE-2.0 010 * 011 * Unless required by applicable law or agreed to in writing, software 012 * distributed under the License is distributed on an "AS IS" BASIS, 013 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 014 * See the License for the specific language governing permissions and 015 * limitations under the License. 016 */ 017package org.apache.commons.statistics.distribution; 018 019import org.apache.commons.rng.UniformRandomProvider; 020import org.apache.commons.rng.sampling.distribution.StableSampler; 021 022/** 023 * Implementation of the Cauchy distribution. 024 * 025 * <p>The probability density function of \( X \) is: 026 * 027 * <p>\[ f(x; x_0, \gamma) = { 1 \over \pi \gamma } \left[ { \gamma^2 \over (x - x_0)^2 + \gamma^2 } \right] \] 028 * 029 * <p>for \( x_0 \) the location, 030 * \( \gamma > 0 \) the scale, and 031 * \( x \in (-\infty, \infty) \). 032 * 033 * @see <a href="https://en.wikipedia.org/wiki/Cauchy_distribution">Cauchy distribution (Wikipedia)</a> 034 * @see <a href="https://mathworld.wolfram.com/CauchyDistribution.html">Cauchy distribution (MathWorld)</a> 035 */ 036public final class CauchyDistribution extends AbstractContinuousDistribution { 037 /** The location of this distribution. */ 038 private final double location; 039 /** The scale of this distribution. */ 040 private final double scale; 041 /** Density factor (scale / pi). */ 042 private final double scaleOverPi; 043 /** Density factor (scale^2). */ 044 private final double scale2; 045 046 /** 047 * @param location Location parameter. 048 * @param scale Scale parameter. 049 */ 050 private CauchyDistribution(double location, 051 double scale) { 052 this.scale = scale; 053 this.location = location; 054 scaleOverPi = scale / Math.PI; 055 scale2 = scale * scale; 056 } 057 058 /** 059 * Creates a Cauchy distribution. 060 * 061 * @param location Location parameter. 062 * @param scale Scale parameter. 063 * @return the distribution 064 * @throws IllegalArgumentException if {@code scale <= 0}. 065 */ 066 public static CauchyDistribution of(double location, 067 double scale) { 068 if (scale <= 0) { 069 throw new DistributionException(DistributionException.NOT_STRICTLY_POSITIVE, scale); 070 } 071 return new CauchyDistribution(location, scale); 072 } 073 074 /** 075 * Gets the location parameter of this distribution. 076 * 077 * @return the location parameter. 078 */ 079 public double getLocation() { 080 return location; 081 } 082 083 /** 084 * Gets the scale parameter of this distribution. 085 * 086 * @return the scale parameter. 087 */ 088 public double getScale() { 089 return scale; 090 } 091 092 /** {@inheritDoc} */ 093 @Override 094 public double density(double x) { 095 final double dev = x - location; 096 return scaleOverPi / (dev * dev + scale2); 097 } 098 099 /** {@inheritDoc} */ 100 @Override 101 public double cumulativeProbability(double x) { 102 return cdf((x - location) / scale); 103 } 104 105 /** {@inheritDoc} */ 106 @Override 107 public double survivalProbability(double x) { 108 return cdf(-(x - location) / scale); 109 } 110 111 /** 112 * Compute the CDF of the Cauchy distribution with location 0 and scale 1. 113 * @param x Point at which the CDF is evaluated 114 * @return CDF(x) 115 */ 116 private static double cdf(double x) { 117 return 0.5 + (Math.atan(x) / Math.PI); 118 } 119 120 /** 121 * {@inheritDoc} 122 * 123 * <p>Returns {@link Double#NEGATIVE_INFINITY} when {@code p == 0} 124 * and {@link Double#POSITIVE_INFINITY} when {@code p == 1}. 125 */ 126 @Override 127 public double inverseCumulativeProbability(double p) { 128 ArgumentUtils.checkProbability(p); 129 if (p == 0) { 130 return Double.NEGATIVE_INFINITY; 131 } else if (p == 1) { 132 return Double.POSITIVE_INFINITY; 133 } 134 return location + scale * Math.tan(Math.PI * (p - 0.5)); 135 } 136 137 /** 138 * {@inheritDoc} 139 * 140 * <p>Returns {@link Double#NEGATIVE_INFINITY} when {@code p == 1} 141 * and {@link Double#POSITIVE_INFINITY} when {@code p == 0}. 142 */ 143 @Override 144 public double inverseSurvivalProbability(double p) { 145 ArgumentUtils.checkProbability(p); 146 if (p == 1) { 147 return Double.NEGATIVE_INFINITY; 148 } else if (p == 0) { 149 return Double.POSITIVE_INFINITY; 150 } 151 return location - scale * Math.tan(Math.PI * (p - 0.5)); 152 } 153 154 /** 155 * {@inheritDoc} 156 * 157 * <p>The mean is always undefined. 158 * 159 * @return {@link Double#NaN NaN}. 160 */ 161 @Override 162 public double getMean() { 163 return Double.NaN; 164 } 165 166 /** 167 * {@inheritDoc} 168 * 169 * <p>The variance is always undefined. 170 * 171 * @return {@link Double#NaN NaN}. 172 */ 173 @Override 174 public double getVariance() { 175 return Double.NaN; 176 } 177 178 /** 179 * {@inheritDoc} 180 * 181 * <p>The lower bound of the support is always negative infinity. 182 * 183 * @return {@linkplain Double#NEGATIVE_INFINITY negative infinity}. 184 */ 185 @Override 186 public double getSupportLowerBound() { 187 return Double.NEGATIVE_INFINITY; 188 } 189 190 /** 191 * {@inheritDoc} 192 * 193 * <p>The upper bound of the support is always positive infinity. 194 * 195 * @return {@linkplain Double#POSITIVE_INFINITY positive infinity}. 196 */ 197 @Override 198 public double getSupportUpperBound() { 199 return Double.POSITIVE_INFINITY; 200 } 201 202 /** {@inheritDoc} */ 203 @Override 204 double getMedian() { 205 // Overridden for the probability(double, double) method. 206 // This is intentionally not a public method. 207 return location; 208 } 209 210 /** {@inheritDoc} */ 211 @Override 212 public ContinuousDistribution.Sampler createSampler(final UniformRandomProvider rng) { 213 // Cauchy distribution = 214 // Stable distribution with alpha=1, beta=0, gamma=scale, delta=location 215 return StableSampler.of(rng, 1, 0, getScale(), getLocation())::sample; 216 } 217}