View Javadoc
1   /*
2    * Licensed to the Apache Software Foundation (ASF) under one or more
3    * contributor license agreements.  See the NOTICE file distributed with
4    * this work for additional information regarding copyright ownership.
5    * The ASF licenses this file to You under the Apache License, Version 2.0
6    * (the "License"); you may not use this file except in compliance with
7    * the License.  You may obtain a copy of the License at
8    *
9    *      https://www.apache.org/licenses/LICENSE-2.0
10   *
11   * Unless required by applicable law or agreed to in writing, software
12   * distributed under the License is distributed on an "AS IS" BASIS,
13   * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
14   * See the License for the specific language governing permissions and
15   * limitations under the License.
16   */
17  
18  /*
19   * This is not the original file distributed by the Apache Software Foundation
20   * It has been modified by the Hipparchus project
21   */
22  
23  package org.hipparchus.distribution.continuous;
24  
25  import java.io.BufferedReader;
26  import java.io.IOException;
27  import java.io.InputStream;
28  import java.io.InputStreamReader;
29  
30  import org.hipparchus.UnitTestUtils;
31  import org.hipparchus.exception.MathIllegalArgumentException;
32  import org.hipparchus.special.Gamma;
33  import org.hipparchus.util.FastMath;
34  import org.junit.Assert;
35  import org.junit.Test;
36  
37  /**
38   * Test cases for GammaDistribution.
39   */
40  public class GammaDistributionTest extends RealDistributionAbstractTest {
41  
42      //-------------- Implementations for abstract methods -----------------------
43  
44      /** Creates the default continuous distribution instance to use in tests. */
45      @Override
46      public GammaDistribution makeDistribution() {
47          return new GammaDistribution(4d, 2d);
48      }
49  
50      /** Creates the default cumulative probability distribution test input values */
51      @Override
52      public double[] makeCumulativeTestPoints() {
53          // quantiles computed using R version 2.9.2
54          return new double[] {0.857104827257, 1.64649737269, 2.17973074725, 2.7326367935, 3.48953912565,
55                  26.1244815584, 20.0902350297, 17.5345461395, 15.5073130559, 13.3615661365};
56      }
57  
58      /** Creates the default cumulative probability density test expected values */
59      @Override
60      public double[] makeCumulativeTestValues() {
61          return new double[] {0.001, 0.01, 0.025, 0.05, 0.1, 0.999, 0.990, 0.975, 0.950, 0.900};
62      }
63  
64      /** Creates the default probability density test expected values */
65      @Override
66      public double[] makeDensityTestValues() {
67          return new double[] {0.00427280075546, 0.0204117166709, 0.0362756163658, 0.0542113174239, 0.0773195272491,
68                  0.000394468852816, 0.00366559696761, 0.00874649473311, 0.0166712508128, 0.0311798227954};
69      }
70  
71      // --------------------- Override tolerance  --------------
72      @Override
73      public void setUp() {
74          super.setUp();
75          setTolerance(1e-9);
76      }
77  
78      //---------------------------- Additional test cases -------------------------
79      @Test
80      public void testParameterAccessors() {
81          GammaDistribution distribution = (GammaDistribution) getDistribution();
82          Assert.assertEquals(4d, distribution.getShape(), 0);
83          Assert.assertEquals(2d, distribution.getScale(), 0);
84      }
85  
86      @Test
87      public void testPreconditions() {
88          try {
89              new GammaDistribution(0, 1);
90              Assert.fail("Expecting MathIllegalArgumentException for alpha = 0");
91          } catch (MathIllegalArgumentException ex) {
92              // Expected.
93          }
94          try {
95              new GammaDistribution(1, 0);
96              Assert.fail("Expecting MathIllegalArgumentException for alpha = 0");
97          } catch (MathIllegalArgumentException ex) {
98              // Expected.
99          }
100     }
101 
102     @Test
103     public void testProbabilities() {
104         testProbability(-1.000, 4.0, 2.0, .0000);
105         testProbability(15.501, 4.0, 2.0, .9499);
106         testProbability(0.504, 4.0, 1.0, .0018);
107         testProbability(10.011, 1.0, 2.0, .9933);
108         testProbability(5.000, 2.0, 2.0, .7127);
109     }
110 
111     @Test
112     public void testValues() {
113         testValue(15.501, 4.0, 2.0, .9499);
114         testValue(0.504, 4.0, 1.0, .0018);
115         testValue(10.011, 1.0, 2.0, .9933);
116         testValue(5.000, 2.0, 2.0, .7127);
117     }
118 
119     private void testProbability(double x, double a, double b, double expected) {
120         GammaDistribution distribution = new GammaDistribution( a, b );
121         double actual = distribution.cumulativeProbability(x);
122         Assert.assertEquals("probability for " + x, expected, actual, 10e-4);
123     }
124 
125     private void testValue(double expected, double a, double b, double p) {
126         GammaDistribution distribution = new GammaDistribution( a, b );
127         double actual = distribution.inverseCumulativeProbability(p);
128         Assert.assertEquals("critical value for " + p, expected, actual, 10e-4);
129     }
130 
131     @Test
132     public void testDensity() {
133         double[] x = new double[]{-0.1, 1e-6, 0.5, 1, 2, 5};
134         // R2.5: print(dgamma(x, shape=1, rate=1), digits=10)
135         checkDensity(1, 1, x, new double[]{0.000000000000, 0.999999000001, 0.606530659713, 0.367879441171, 0.135335283237, 0.006737946999});
136         // R2.5: print(dgamma(x, shape=2, rate=1), digits=10)
137         checkDensity(2, 1, x, new double[]{0.000000000000, 0.000000999999, 0.303265329856, 0.367879441171, 0.270670566473, 0.033689734995});
138         // R2.5: print(dgamma(x, shape=4, rate=1), digits=10)
139         checkDensity(4, 1, x, new double[]{0.000000000e+00, 1.666665000e-19, 1.263605541e-02, 6.131324020e-02, 1.804470443e-01, 1.403738958e-01});
140         // R2.5: print(dgamma(x, shape=4, rate=10), digits=10)
141         checkDensity(4, 10, x, new double[]{0.000000000e+00, 1.666650000e-15, 1.403738958e+00, 7.566654960e-02, 2.748204830e-05, 4.018228850e-17});
142         // R2.5: print(dgamma(x, shape=.1, rate=10), digits=10)
143         checkDensity(0.1, 10, x, new double[]{0.000000000e+00, 3.323953832e+04, 1.663849010e-03, 6.007786726e-06, 1.461647647e-10, 5.996008322e-24});
144         // R2.5: print(dgamma(x, shape=.1, rate=20), digits=10)
145         checkDensity(0.1, 20, x, new double[]{0.000000000e+00, 3.562489883e+04, 1.201557345e-05, 2.923295295e-10, 3.228910843e-19, 1.239484589e-45});
146         // R2.5: print(dgamma(x, shape=.1, rate=4), digits=10)
147         checkDensity(0.1, 4, x, new double[]{0.000000000e+00, 3.032938388e+04, 3.049322494e-02, 2.211502311e-03, 2.170613371e-05, 5.846590589e-11});
148         // R2.5: print(dgamma(x, shape=.1, rate=1), digits=10)
149         checkDensity(0.1, 1, x, new double[]{0.000000000e+00, 2.640334143e+04, 1.189704437e-01, 3.866916944e-02, 7.623306235e-03, 1.663849010e-04});
150     }
151 
152     private void checkDensity(double alpha, double rate, double[] x, double[] expected) {
153         GammaDistribution d = new GammaDistribution(alpha, 1 / rate);
154         for (int i = 0; i < x.length; i++) {
155             Assert.assertEquals(expected[i], d.density(x[i]), 1e-5);
156         }
157     }
158 
159     @Test
160     public void testInverseCumulativeProbabilityExtremes() {
161         setInverseCumulativeTestPoints(new double[] {0, 1});
162         setInverseCumulativeTestValues(new double[] {0, Double.POSITIVE_INFINITY});
163         verifyInverseCumulativeProbabilities();
164     }
165 
166     @Test
167     public void testMoments() {
168         final double tol = 1e-9;
169         GammaDistribution dist;
170 
171         dist = new GammaDistribution(1, 2);
172         Assert.assertEquals(dist.getNumericalMean(), 2, tol);
173         Assert.assertEquals(dist.getNumericalVariance(), 4, tol);
174 
175         dist = new GammaDistribution(1.1, 4.2);
176         Assert.assertEquals(dist.getNumericalMean(), 1.1d * 4.2d, tol);
177         Assert.assertEquals(dist.getNumericalVariance(), 1.1d * 4.2d * 4.2d, tol);
178     }
179 
180     private static final double HALF_LOG_2_PI = 0.5 * FastMath.log(2.0 * FastMath.PI);
181 
182     public static double logGamma(double x) {
183         /*
184          * This is a copy of
185          * double Gamma.logGamma(double)
186          * prior to MATH-849
187          */
188         double ret;
189 
190         if (Double.isNaN(x) || (x <= 0.0)) {
191             ret = Double.NaN;
192         } else {
193             double sum = Gamma.lanczos(x);
194             double tmp = x + Gamma.LANCZOS_G + .5;
195             ret = ((x + .5) * FastMath.log(tmp)) - tmp +
196                 HALF_LOG_2_PI + FastMath.log(sum / x);
197         }
198 
199         return ret;
200     }
201 
202     public static double density(final double x, final double shape,
203                                  final double scale) {
204         /*
205          * This is a copy of
206          * double GammaDistribution.density(double)
207          * prior to MATH-753.
208          */
209         if (x < 0) {
210             return 0;
211         }
212         return FastMath.pow(x / scale, shape - 1) / scale *
213                FastMath.exp(-x / scale) / FastMath.exp(logGamma(shape));
214     }
215 
216     /*
217      * MATH-753: large values of x or shape parameter cause density(double) to
218      * overflow. Reference data is generated with the Maxima script
219      * gamma-distribution.mac, which can be found in
220      * src/test/resources/org.hipparchus/distribution.
221      */
222 
223     private void doTestMath753(final double shape,
224         final double meanNoOF, final double sdNoOF,
225         final double meanOF, final double sdOF,
226         final String resourceName) throws IOException {
227         final GammaDistribution distribution = new GammaDistribution(shape, 1.0);
228         final UnitTestUtils.SimpleStatistics statOld = new UnitTestUtils.SimpleStatistics();
229         final UnitTestUtils.SimpleStatistics statNewNoOF = new UnitTestUtils.SimpleStatistics();
230         final UnitTestUtils.SimpleStatistics statNewOF = new UnitTestUtils.SimpleStatistics();
231 
232         final InputStream resourceAsStream;
233         resourceAsStream = this.getClass().getResourceAsStream(resourceName);
234         Assert.assertNotNull("Could not find resource " + resourceName,
235                              resourceAsStream);
236         final BufferedReader in;
237         in = new BufferedReader(new InputStreamReader(resourceAsStream));
238 
239         try {
240             for (String line = in.readLine(); line != null; line = in.readLine()) {
241                 if (line.startsWith("#")) {
242                     continue;
243                 }
244                 final String[] tokens = line.split(", ");
245                 Assert.assertTrue("expected two floating-point values",
246                                   tokens.length == 2);
247                 final double x = Double.parseDouble(tokens[0]);
248                 final String msg = "x = " + x + ", shape = " + shape +
249                                    ", scale = 1.0";
250                 final double expected = Double.parseDouble(tokens[1]);
251                 final double ulp = FastMath.ulp(expected);
252                 final double actualOld = density(x, shape, 1.0);
253                 final double actualNew = distribution.density(x);
254                 final double errOld, errNew;
255                 errOld = FastMath.abs((actualOld - expected) / ulp);
256                 errNew = FastMath.abs((actualNew - expected) / ulp);
257 
258                 if (Double.isNaN(actualOld) || Double.isInfinite(actualOld)) {
259                     Assert.assertFalse(msg, Double.isNaN(actualNew));
260                     Assert.assertFalse(msg, Double.isInfinite(actualNew));
261                     statNewOF.addValue(errNew);
262                 } else {
263                     statOld.addValue(errOld);
264                     statNewNoOF.addValue(errNew);
265                 }
266             }
267             if (statOld.getN() != 0) {
268                 /*
269                  * If no overflow occurs, check that new implementation is
270                  * better than old one.
271                  */
272                 final StringBuilder sb = new StringBuilder("shape = ");
273                 sb.append(shape);
274                 sb.append(", scale = 1.0\n");
275                 sb.append("Old implementation\n");
276                 sb.append("------------------\n");
277                 sb.append(statOld.toString());
278                 sb.append("New implementation\n");
279                 sb.append("------------------\n");
280                 sb.append(statNewNoOF.toString());
281                 final String msg = sb.toString();
282 
283                 final double oldMin = statOld.getMin();
284                 final double newMin = statNewNoOF.getMin();
285                 Assert.assertTrue(msg, newMin <= oldMin);
286 
287                 final double oldMax = statOld.getMax();
288                 final double newMax = statNewNoOF.getMax();
289                 Assert.assertTrue(msg, newMax <= oldMax);
290 
291                 final double oldMean = statOld.getMean();
292                 final double newMean = statNewNoOF.getMean();
293                 Assert.assertTrue(msg, newMean <= oldMean);
294 
295                 final double oldSd = statOld.getStandardDeviation();
296                 final double newSd = statNewNoOF.getStandardDeviation();
297                 Assert.assertTrue(msg, newSd <= oldSd);
298 
299                 Assert.assertTrue(msg, newMean <= meanNoOF);
300                 Assert.assertTrue(msg, newSd <= sdNoOF);
301             }
302             if (statNewOF.getN() != 0) {
303                 final double newMean = statNewOF.getMean();
304                 final double newSd = statNewOF.getStandardDeviation();
305 
306                 final StringBuilder sb = new StringBuilder("shape = ");
307                 sb.append(shape);
308                 sb.append(", scale = 1.0");
309                 sb.append(", max. mean error (ulps) = ");
310                 sb.append(meanOF);
311                 sb.append(", actual mean error (ulps) = ");
312                 sb.append(newMean);
313                 sb.append(", max. sd of error (ulps) = ");
314                 sb.append(sdOF);
315                 sb.append(", actual sd of error (ulps) = ");
316                 sb.append(newSd);
317                 final String msg = sb.toString();
318 
319                 Assert.assertTrue(msg, newMean <= meanOF);
320                 Assert.assertTrue(msg, newSd <= sdOF);
321             }
322         } catch (IOException e) {
323             Assert.fail(e.getMessage());
324         } finally {
325             in.close();
326         }
327     }
328 
329 
330     @Test
331     public void testMath753Shape1() throws IOException {
332         doTestMath753(1.0, 1.5, 0.5, 0.0, 0.0, "gamma-distribution-shape-1.csv");
333     }
334 
335     @Test
336     public void testMath753Shape8() throws IOException {
337         doTestMath753(8.0, 1.5, 1.0, 0.0, 0.0, "gamma-distribution-shape-8.csv");
338     }
339 
340     @Test
341     public void testMath753Shape10() throws IOException {
342         doTestMath753(10.0, 1.0, 1.0, 0.0, 0.0, "gamma-distribution-shape-10.csv");
343     }
344 
345     @Test
346     public void testMath753Shape100() throws IOException {
347         doTestMath753(100.0, 1.5, 1.0, 0.0, 0.0, "gamma-distribution-shape-100.csv");
348     }
349 
350     @Test
351     public void testMath753Shape142() throws IOException {
352         doTestMath753(142.0, 3.3, 1.6, 40.0, 40.0, "gamma-distribution-shape-142.csv");
353     }
354 
355     @Test
356     public void testMath753Shape1000() throws IOException {
357         doTestMath753(1000.0, 1.0, 1.0, 160.0, 220.0, "gamma-distribution-shape-1000.csv");
358     }
359 }