Skip to content

Commit c2b9f27

Browse files
cicirellocicirello
authored
Merge pull request #304 from cicirello/next-sample-p
Add SequenceSampler.nextSample methods with probability of sampling element
2 parents 113c2b6 + e1cf463 commit c2b9f27

10 files changed

+442
-27
lines changed

‎src/main/java/org/cicirello/sequences/SequenceSampler.java‎

Lines changed: 118 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -142,6 +142,124 @@ public interface SequenceSampler {
142142
*/
143143
<T> T[] nextSample(T[] source, int k, T[] target);
144144

145+
/**
146+
* Generates a random sample, without replacement, from a given source array with a specified
147+
* probability of an element's inclusion in the sample.
148+
*
149+
* @param source The array from which to sample.
150+
* @param p The probability that element is included in the sample. The expected sample size is
151+
* source.length * p.
152+
* @return An array containing the sample, whose sample size is simply the length of the array.
153+
*/
154+
default int[] nextSample(int[] source, double p) {
155+
return nextSample(source, RandomVariates.nextBinomial(source.length, p), null);
156+
}
157+
158+
/**
159+
* Generates a random sample, without replacement, from a given source array with a specified
160+
* probability of an element's inclusion in the sample.
161+
*
162+
* @param source The array from which to sample.
163+
* @param p The probability that element is included in the sample. The expected sample size is
164+
* source.length * p.
165+
* @return An array containing the sample, whose sample size is simply the length of the array.
166+
*/
167+
default long[] nextSample(long[] source, double p) {
168+
return nextSample(source, RandomVariates.nextBinomial(source.length, p), null);
169+
}
170+
171+
/**
172+
* Generates a random sample, without replacement, from a given source array with a specified
173+
* probability of an element's inclusion in the sample.
174+
*
175+
* @param source The array from which to sample.
176+
* @param p The probability that element is included in the sample. The expected sample size is
177+
* source.length * p.
178+
* @return An array containing the sample, whose sample size is simply the length of the array.
179+
*/
180+
default short[] nextSample(short[] source, double p) {
181+
return nextSample(source, RandomVariates.nextBinomial(source.length, p), null);
182+
}
183+
184+
/**
185+
* Generates a random sample, without replacement, from a given source array with a specified
186+
* probability of an element's inclusion in the sample.
187+
*
188+
* @param source The array from which to sample.
189+
* @param p The probability that element is included in the sample. The expected sample size is
190+
* source.length * p.
191+
* @return An array containing the sample, whose sample size is simply the length of the array.
192+
*/
193+
default byte[] nextSample(byte[] source, double p) {
194+
return nextSample(source, RandomVariates.nextBinomial(source.length, p), null);
195+
}
196+
197+
/**
198+
* Generates a random sample, without replacement, from a given source array with a specified
199+
* probability of an element's inclusion in the sample.
200+
*
201+
* @param source The array from which to sample.
202+
* @param p The probability that element is included in the sample. The expected sample size is
203+
* source.length * p.
204+
* @return An array containing the sample, whose sample size is simply the length of the array.
205+
*/
206+
default double[] nextSample(double[] source, double p) {
207+
return nextSample(source, RandomVariates.nextBinomial(source.length, p), null);
208+
}
209+
210+
/**
211+
* Generates a random sample, without replacement, from a given source array with a specified
212+
* probability of an element's inclusion in the sample.
213+
*
214+
* @param source The array from which to sample.
215+
* @param p The probability that element is included in the sample. The expected sample size is
216+
* source.length * p.
217+
* @return An array containing the sample, whose sample size is simply the length of the array.
218+
*/
219+
default float[] nextSample(float[] source, double p) {
220+
return nextSample(source, RandomVariates.nextBinomial(source.length, p), null);
221+
}
222+
223+
/**
224+
* Generates a random sample, without replacement, from a given source array with a specified
225+
* probability of an element's inclusion in the sample.
226+
*
227+
* @param source The array from which to sample.
228+
* @param p The probability that element is included in the sample. The expected sample size is
229+
* source.length * p.
230+
* @return An array containing the sample, whose sample size is simply the length of the array.
231+
*/
232+
default char[] nextSample(char[] source, double p) {
233+
return nextSample(source, RandomVariates.nextBinomial(source.length, p), null);
234+
}
235+
236+
/**
237+
* Generates a random sample, without replacement, from a given source String with a specified
238+
* probability of an element's inclusion in the sample.
239+
*
240+
* @param source The String from which to sample.
241+
* @param p The probability that element is included in the sample. The expected sample size is
242+
* source.length() * p.
243+
* @return An array containing the sample, whose sample size is simply the length of the array.
244+
*/
245+
default char[] nextSample(String source, double p) {
246+
return nextSample(source, RandomVariates.nextBinomial(source.length(), p), null);
247+
}
248+
249+
/**
250+
* Generates a random sample, without replacement, from a given source array with a specified
251+
* probability of an element's inclusion in the sample.
252+
*
253+
* @param <T> The type of array elements.
254+
* @param source The array from which to sample.
255+
* @param p The probability that element is included in the sample. The expected sample size is
256+
* source.length * p.
257+
* @return An array containing the sample, whose sample size is simply the length of the array.
258+
*/
259+
default <T> T[] nextSample(T[] source, double p) {
260+
return nextSample(source, RandomVariates.nextBinomial(source.length, p), null);
261+
}
262+
145263
/**
146264
* Generates a random sample, without replacement, from a given source array with a specified
147265
* probability of an element's inclusion in the sample.

‎src/test/java/org/cicirello/sequences/SequenceSamplerByteTests.java‎

Lines changed: 36 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -53,13 +53,41 @@ public void testSampleInsertion() {
5353
validateSamples(r::nextSample);
5454
}
5555

56+
@Test
57+
public void testSampleCompositeP() {
58+
SequenceCompositeSampler r = new SequenceCompositeSampler(new SplittableRandom(42));
59+
validateWithP(r::nextSample);
60+
}
61+
62+
@Test
63+
public void testSampleReservoirP() {
64+
SequenceReservoirSampler r = new SequenceReservoirSampler(new SplittableRandom(42));
65+
validateWithP(r::nextSample);
66+
}
67+
68+
@Test
69+
public void testSamplePoolP() {
70+
SequencePoolSampler r = new SequencePoolSampler(new SplittableRandom(42));
71+
validateWithP(r::nextSample);
72+
}
73+
74+
@Test
75+
public void testSampleInsertionP() {
76+
SequenceInsertionSampler r = new SequenceInsertionSampler(new SplittableRandom(42));
77+
validateWithP(r::nextSample);
78+
}
79+
5680
@Test
5781
public void testSample() {
5882
validateSamples(SequenceSampler::sample);
5983
}
6084

6185
@Test
6286
public void testSampleByteP() {
87+
validateWithP(SequenceSampler::sample);
88+
}
89+
90+
private void validateWithP(PSampler sampler) {
6391
for (int n = 1; n <= 10; n++) {
6492
byte[] allDiff = new byte[n];
6593
byte[] mixedQuantities = new byte[n];
@@ -78,13 +106,18 @@ public void testSampleByteP() {
78106
allSame[i] = 100;
79107
}
80108
for (double p = 0.25; p <= 0.8; p += 0.25) {
81-
validateSample(allDiff, SequenceSampler.sample(allDiff, p));
82-
validateSample(mixedQuantities, SequenceSampler.sample(mixedQuantities, p));
83-
validateSample(allSame, SequenceSampler.sample(allSame, p));
109+
validateSample(allDiff, sampler.sample(allDiff, p));
110+
validateSample(mixedQuantities, sampler.sample(mixedQuantities, p));
111+
validateSample(allSame, sampler.sample(allSame, p));
84112
}
85113
}
86114
}
87115

116+
@FunctionalInterface
117+
interface PSampler {
118+
byte[] sample(byte[] source, double p);
119+
}
120+
88121
@FunctionalInterface
89122
interface Sampler {
90123
byte[] sample(byte[] source, int k, byte[] target);

‎src/test/java/org/cicirello/sequences/SequenceSamplerCharTests.java‎

Lines changed: 36 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -53,13 +53,41 @@ public void testSampleInsertion() {
5353
validateSamples(r::nextSample);
5454
}
5555

56+
@Test
57+
public void testSampleCompositeP() {
58+
SequenceCompositeSampler r = new SequenceCompositeSampler(new SplittableRandom(42));
59+
validateWithP(r::nextSample);
60+
}
61+
62+
@Test
63+
public void testSampleReservoirP() {
64+
SequenceReservoirSampler r = new SequenceReservoirSampler(new SplittableRandom(42));
65+
validateWithP(r::nextSample);
66+
}
67+
68+
@Test
69+
public void testSamplePoolP() {
70+
SequencePoolSampler r = new SequencePoolSampler(new SplittableRandom(42));
71+
validateWithP(r::nextSample);
72+
}
73+
74+
@Test
75+
public void testSampleInsertionP() {
76+
SequenceInsertionSampler r = new SequenceInsertionSampler(new SplittableRandom(42));
77+
validateWithP(r::nextSample);
78+
}
79+
5680
@Test
5781
public void testSample() {
5882
validateSamples(SequenceSampler::sample);
5983
}
6084

6185
@Test
6286
public void testSampleCharP() {
87+
validateWithP(SequenceSampler::sample);
88+
}
89+
90+
private void validateWithP(PSampler sampler) {
6391
for (int n = 1; n <= 10; n++) {
6492
char[] allDiff = new char[n];
6593
char[] mixedQuantities = new char[n];
@@ -78,13 +106,18 @@ public void testSampleCharP() {
78106
allSame[i] = 100;
79107
}
80108
for (double p = 0.25; p <= 0.8; p += 0.25) {
81-
validateSample(allDiff, SequenceSampler.sample(allDiff, p));
82-
validateSample(mixedQuantities, SequenceSampler.sample(mixedQuantities, p));
83-
validateSample(allSame, SequenceSampler.sample(allSame, p));
109+
validateSample(allDiff, sampler.sample(allDiff, p));
110+
validateSample(mixedQuantities, sampler.sample(mixedQuantities, p));
111+
validateSample(allSame, sampler.sample(allSame, p));
84112
}
85113
}
86114
}
87115

116+
@FunctionalInterface
117+
interface PSampler {
118+
char[] sample(char[] source, double p);
119+
}
120+
88121
@FunctionalInterface
89122
interface Sampler {
90123
char[] sample(char[] source, int k, char[] target);

‎src/test/java/org/cicirello/sequences/SequenceSamplerDoubleTests.java‎

Lines changed: 36 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -53,13 +53,41 @@ public void testSampleInsertion() {
5353
validateSamples(r::nextSample);
5454
}
5555

56+
@Test
57+
public void testSampleCompositeP() {
58+
SequenceCompositeSampler r = new SequenceCompositeSampler(new SplittableRandom(42));
59+
validateWithP(r::nextSample);
60+
}
61+
62+
@Test
63+
public void testSampleReservoirP() {
64+
SequenceReservoirSampler r = new SequenceReservoirSampler(new SplittableRandom(42));
65+
validateWithP(r::nextSample);
66+
}
67+
68+
@Test
69+
public void testSamplePoolP() {
70+
SequencePoolSampler r = new SequencePoolSampler(new SplittableRandom(42));
71+
validateWithP(r::nextSample);
72+
}
73+
74+
@Test
75+
public void testSampleInsertionP() {
76+
SequenceInsertionSampler r = new SequenceInsertionSampler(new SplittableRandom(42));
77+
validateWithP(r::nextSample);
78+
}
79+
5680
@Test
5781
public void testSample() {
5882
validateSamples(SequenceSampler::sample);
5983
}
6084

6185
@Test
6286
public void testSampleDoubleP() {
87+
validateWithP(SequenceSampler::sample);
88+
}
89+
90+
private void validateWithP(PSampler sampler) {
6391
for (int n = 1; n <= 10; n++) {
6492
double[] allDiff = new double[n];
6593
double[] mixedQuantities = new double[n];
@@ -78,13 +106,18 @@ public void testSampleDoubleP() {
78106
allSame[i] = 100;
79107
}
80108
for (double p = 0.25; p <= 0.8; p += 0.25) {
81-
validateSample(allDiff, SequenceSampler.sample(allDiff, p));
82-
validateSample(mixedQuantities, SequenceSampler.sample(mixedQuantities, p));
83-
validateSample(allSame, SequenceSampler.sample(allSame, p));
109+
validateSample(allDiff, sampler.sample(allDiff, p));
110+
validateSample(mixedQuantities, sampler.sample(mixedQuantities, p));
111+
validateSample(allSame, sampler.sample(allSame, p));
84112
}
85113
}
86114
}
87115

116+
@FunctionalInterface
117+
interface PSampler {
118+
double[] sample(double[] source, double p);
119+
}
120+
88121
@FunctionalInterface
89122
interface Sampler {
90123
double[] sample(double[] source, int k, double[] target);

‎src/test/java/org/cicirello/sequences/SequenceSamplerFloatTests.java‎

Lines changed: 36 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -53,13 +53,41 @@ public void testSampleInsertion() {
5353
validateSamples(r::nextSample);
5454
}
5555

56+
@Test
57+
public void testSampleCompositeP() {
58+
SequenceCompositeSampler r = new SequenceCompositeSampler(new SplittableRandom(42));
59+
validateWithP(r::nextSample);
60+
}
61+
62+
@Test
63+
public void testSampleReservoirP() {
64+
SequenceReservoirSampler r = new SequenceReservoirSampler(new SplittableRandom(42));
65+
validateWithP(r::nextSample);
66+
}
67+
68+
@Test
69+
public void testSamplePoolP() {
70+
SequencePoolSampler r = new SequencePoolSampler(new SplittableRandom(42));
71+
validateWithP(r::nextSample);
72+
}
73+
74+
@Test
75+
public void testSampleInsertionP() {
76+
SequenceInsertionSampler r = new SequenceInsertionSampler(new SplittableRandom(42));
77+
validateWithP(r::nextSample);
78+
}
79+
5680
@Test
5781
public void testSample() {
5882
validateSamples(SequenceSampler::sample);
5983
}
6084

6185
@Test
6286
public void testSampleFloatP() {
87+
validateWithP(SequenceSampler::sample);
88+
}
89+
90+
private void validateWithP(PSampler sampler) {
6391
for (int n = 1; n <= 10; n++) {
6492
float[] allDiff = new float[n];
6593
float[] mixedQuantities = new float[n];
@@ -78,13 +106,18 @@ public void testSampleFloatP() {
78106
allSame[i] = 100;
79107
}
80108
for (double p = 0.25; p <= 0.8; p += 0.25) {
81-
validateSample(allDiff, SequenceSampler.sample(allDiff, p));
82-
validateSample(mixedQuantities, SequenceSampler.sample(mixedQuantities, p));
83-
validateSample(allSame, SequenceSampler.sample(allSame, p));
109+
validateSample(allDiff, sampler.sample(allDiff, p));
110+
validateSample(mixedQuantities, sampler.sample(mixedQuantities, p));
111+
validateSample(allSame, sampler.sample(allSame, p));
84112
}
85113
}
86114
}
87115

116+
@FunctionalInterface
117+
interface PSampler {
118+
float[] sample(float[] source, double p);
119+
}
120+
88121
@FunctionalInterface
89122
interface Sampler {
90123
float[] sample(float[] source, int k, float[] target);

0 commit comments

Comments
 (0)