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 */
017 package org.apache.commons.math.distribution;
018
019 import java.io.Serializable;
020
021 import org.apache.commons.math.MathException;
022 import org.apache.commons.math.MathRuntimeException;
023 import org.apache.commons.math.exception.util.LocalizedFormats;
024 import org.apache.commons.math.random.RandomDataImpl;
025 import org.apache.commons.math.util.FastMath;
026
027
028 /**
029 * Base class for integer-valued discrete distributions. Default
030 * implementations are provided for some of the methods that do not vary
031 * from distribution to distribution.
032 *
033 * @version $Revision: 1067494 $ $Date: 2011-02-05 20:49:07 +0100 (sam. 05 f??vr. 2011) $
034 */
035 public abstract class AbstractIntegerDistribution extends AbstractDistribution
036 implements IntegerDistribution, Serializable {
037
038 /** Serializable version identifier */
039 private static final long serialVersionUID = -1146319659338487221L;
040
041 /**
042 * RandomData instance used to generate samples from the distribution
043 * @since 2.2
044 */
045 protected final RandomDataImpl randomData = new RandomDataImpl();
046
047 /**
048 * Default constructor.
049 */
050 protected AbstractIntegerDistribution() {
051 super();
052 }
053
054 /**
055 * For a random variable X whose values are distributed according
056 * to this distribution, this method returns P(X ≤ x). In other words,
057 * this method represents the (cumulative) distribution function, or
058 * CDF, for this distribution.
059 * <p>
060 * If <code>x</code> does not represent an integer value, the CDF is
061 * evaluated at the greatest integer less than x.
062 *
063 * @param x the value at which the distribution function is evaluated.
064 * @return cumulative probability that a random variable with this
065 * distribution takes a value less than or equal to <code>x</code>
066 * @throws MathException if the cumulative probability can not be
067 * computed due to convergence or other numerical errors.
068 */
069 public double cumulativeProbability(double x) throws MathException {
070 return cumulativeProbability((int) FastMath.floor(x));
071 }
072
073 /**
074 * For a random variable X whose values are distributed according
075 * to this distribution, this method returns P(x0 ≤ X ≤ x1).
076 *
077 * @param x0 the (inclusive) lower bound
078 * @param x1 the (inclusive) upper bound
079 * @return the probability that a random variable with this distribution
080 * will take a value between <code>x0</code> and <code>x1</code>,
081 * including the endpoints.
082 * @throws MathException if the cumulative probability can not be
083 * computed due to convergence or other numerical errors.
084 * @throws IllegalArgumentException if <code>x0 > x1</code>
085 */
086 @Override
087 public double cumulativeProbability(double x0, double x1)
088 throws MathException {
089 if (x0 > x1) {
090 throw MathRuntimeException.createIllegalArgumentException(
091 LocalizedFormats.LOWER_ENDPOINT_ABOVE_UPPER_ENDPOINT, x0, x1);
092 }
093 if (FastMath.floor(x0) < x0) {
094 return cumulativeProbability(((int) FastMath.floor(x0)) + 1,
095 (int) FastMath.floor(x1)); // don't want to count mass below x0
096 } else { // x0 is mathematical integer, so use as is
097 return cumulativeProbability((int) FastMath.floor(x0),
098 (int) FastMath.floor(x1));
099 }
100 }
101
102 /**
103 * For a random variable X whose values are distributed according
104 * to this distribution, this method returns P(X ≤ x). In other words,
105 * this method represents the probability distribution function, or PDF,
106 * for this distribution.
107 *
108 * @param x the value at which the PDF is evaluated.
109 * @return PDF for this distribution.
110 * @throws MathException if the cumulative probability can not be
111 * computed due to convergence or other numerical errors.
112 */
113 public abstract double cumulativeProbability(int x) throws MathException;
114
115 /**
116 * For a random variable X whose values are distributed according
117 * to this distribution, this method returns P(X = x). In other words, this
118 * method represents the probability mass function, or PMF, for the distribution.
119 * <p>
120 * If <code>x</code> does not represent an integer value, 0 is returned.
121 *
122 * @param x the value at which the probability density function is evaluated
123 * @return the value of the probability density function at x
124 */
125 public double probability(double x) {
126 double fl = FastMath.floor(x);
127 if (fl == x) {
128 return this.probability((int) x);
129 } else {
130 return 0;
131 }
132 }
133
134 /**
135 * For a random variable X whose values are distributed according
136 * to this distribution, this method returns P(x0 ≤ X ≤ x1).
137 *
138 * @param x0 the inclusive, lower bound
139 * @param x1 the inclusive, upper bound
140 * @return the cumulative probability.
141 * @throws MathException if the cumulative probability can not be
142 * computed due to convergence or other numerical errors.
143 * @throws IllegalArgumentException if x0 > x1
144 */
145 public double cumulativeProbability(int x0, int x1) throws MathException {
146 if (x0 > x1) {
147 throw MathRuntimeException.createIllegalArgumentException(
148 LocalizedFormats.LOWER_ENDPOINT_ABOVE_UPPER_ENDPOINT, x0, x1);
149 }
150 return cumulativeProbability(x1) - cumulativeProbability(x0 - 1);
151 }
152
153 /**
154 * For a random variable X whose values are distributed according
155 * to this distribution, this method returns the largest x, such
156 * that P(X ≤ x) ≤ <code>p</code>.
157 *
158 * @param p the desired probability
159 * @return the largest x such that P(X ≤ x) <= p
160 * @throws MathException if the inverse cumulative probability can not be
161 * computed due to convergence or other numerical errors.
162 * @throws IllegalArgumentException if p < 0 or p > 1
163 */
164 public int inverseCumulativeProbability(final double p) throws MathException{
165 if (p < 0.0 || p > 1.0) {
166 throw MathRuntimeException.createIllegalArgumentException(
167 LocalizedFormats.OUT_OF_RANGE_SIMPLE, p, 0.0, 1.0);
168 }
169
170 // by default, do simple bisection.
171 // subclasses can override if there is a better method.
172 int x0 = getDomainLowerBound(p);
173 int x1 = getDomainUpperBound(p);
174 double pm;
175 while (x0 < x1) {
176 int xm = x0 + (x1 - x0) / 2;
177 pm = checkedCumulativeProbability(xm);
178 if (pm > p) {
179 // update x1
180 if (xm == x1) {
181 // this can happen with integer division
182 // simply decrement x1
183 --x1;
184 } else {
185 // update x1 normally
186 x1 = xm;
187 }
188 } else {
189 // update x0
190 if (xm == x0) {
191 // this can happen with integer division
192 // simply increment x0
193 ++x0;
194 } else {
195 // update x0 normally
196 x0 = xm;
197 }
198 }
199 }
200
201 // insure x0 is the correct critical point
202 pm = checkedCumulativeProbability(x0);
203 while (pm > p) {
204 --x0;
205 pm = checkedCumulativeProbability(x0);
206 }
207
208 return x0;
209 }
210
211 /**
212 * Reseeds the random generator used to generate samples.
213 *
214 * @param seed the new seed
215 * @since 2.2
216 */
217 public void reseedRandomGenerator(long seed) {
218 randomData.reSeed(seed);
219 }
220
221 /**
222 * Generates a random value sampled from this distribution. The default
223 * implementation uses the
224 * <a href="http://en.wikipedia.org/wiki/Inverse_transform_sampling"> inversion method.</a>
225 *
226 * @return random value
227 * @since 2.2
228 * @throws MathException if an error occurs generating the random value
229 */
230 public int sample() throws MathException {
231 return randomData.nextInversionDeviate(this);
232 }
233
234 /**
235 * Generates a random sample from the distribution. The default implementation
236 * generates the sample by calling {@link #sample()} in a loop.
237 *
238 * @param sampleSize number of random values to generate
239 * @since 2.2
240 * @return an array representing the random sample
241 * @throws MathException if an error occurs generating the sample
242 * @throws IllegalArgumentException if sampleSize is not positive
243 */
244 public int[] sample(int sampleSize) throws MathException {
245 if (sampleSize <= 0) {
246 MathRuntimeException.createIllegalArgumentException(LocalizedFormats.NOT_POSITIVE_SAMPLE_SIZE, sampleSize);
247 }
248 int[] out = new int[sampleSize];
249 for (int i = 0; i < sampleSize; i++) {
250 out[i] = sample();
251 }
252 return out;
253 }
254
255 /**
256 * Computes the cumulative probability function and checks for NaN values returned.
257 * Throws MathException if the value is NaN. Rethrows any MathException encountered
258 * evaluating the cumulative probability function. Throws
259 * MathException if the cumulative probability function returns NaN.
260 *
261 * @param argument input value
262 * @return cumulative probability
263 * @throws MathException if the cumulative probability is NaN
264 */
265 private double checkedCumulativeProbability(int argument) throws MathException {
266 double result = Double.NaN;
267 result = cumulativeProbability(argument);
268 if (Double.isNaN(result)) {
269 throw new MathException(LocalizedFormats.DISCRETE_CUMULATIVE_PROBABILITY_RETURNED_NAN, argument);
270 }
271 return result;
272 }
273
274 /**
275 * Access the domain value lower bound, based on <code>p</code>, used to
276 * bracket a PDF root. This method is used by
277 * {@link #inverseCumulativeProbability(double)} to find critical values.
278 *
279 * @param p the desired probability for the critical value
280 * @return domain value lower bound, i.e.
281 * P(X < <i>lower bound</i>) < <code>p</code>
282 */
283 protected abstract int getDomainLowerBound(double p);
284
285 /**
286 * Access the domain value upper bound, based on <code>p</code>, used to
287 * bracket a PDF root. This method is used by
288 * {@link #inverseCumulativeProbability(double)} to find critical values.
289 *
290 * @param p the desired probability for the critical value
291 * @return domain value upper bound, i.e.
292 * P(X < <i>upper bound</i>) > <code>p</code>
293 */
294 protected abstract int getDomainUpperBound(double p);
295
296 /**
297 * Use this method to get information about whether the lower bound
298 * of the support is inclusive or not. For discrete support,
299 * only true here is meaningful.
300 *
301 * @return true (always but at Integer.MIN_VALUE because of the nature of discrete support)
302 * @since 2.2
303 */
304 public boolean isSupportLowerBoundInclusive() {
305 return true;
306 }
307
308 /**
309 * Use this method to get information about whether the upper bound
310 * of the support is inclusive or not. For discrete support,
311 * only true here is meaningful.
312 *
313 * @return true (always but at Integer.MAX_VALUE because of the nature of discrete support)
314 * @since 2.2
315 */
316 public boolean isSupportUpperBoundInclusive() {
317 return true;
318 }
319 }