kscience/kmath
9
0
Fork 1
Code Issues Pull Requests 6 Packages Projects Releases 16 Wiki Activity

Implement Commons RNG-like samplers in kmath-prob module for Multiplatform #164

Merged
CommanderTvis merged 44 commits from feature/mp-samplers into dev 2021-03-31 09:25:44 +03:00
Conversation 0 Commits 44 Files Changed 41 +21 -16
15 changed files with 21 additions and 16 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files
Showing only changes of commit 5a3fccb455 - Show all commits

5
examples/src/main/kotlin/kscience/kmath/commons/prob/DistributionBenchmark.kt → examples/src/main/kotlin/kscience/kmath/stat/DistributionBenchmark.kt
View File

@ -1,11 +1,8 @@
package kscience.kmath.commons.prob
package kscience.kmath.stat
import kotlinx.coroutines.Dispatchers
import kotlinx.coroutines.async
import kotlinx.coroutines.runBlocking
import kscience.kmath.stat.RandomGenerator
import kscience.kmath.stat.blocking
import kscience.kmath.stat.fromSource
import kscience.kmath.stat.samplers.GaussianSampler
import org.apache.commons.rng.simple.RandomSource
import java.time.Duration

2
examples/src/main/kotlin/kscience/kmath/commons/prob/DistributionDemo.kt → examples/src/main/kotlin/kscience/kmath/stat/DistributionDemo.kt
View File

@ -1,4 +1,4 @@
package kscience.kmath.commons.prob
package kscience.kmath.stat
import kotlinx.coroutines.runBlocking
import kscience.kmath.chains.Chain

3
kmath-stat/src/commonMain/kotlin/kscience/kmath/stat/distributions/NormalDistribution.kt
View File

@ -9,6 +9,9 @@ import kscience.kmath.stat.samplers.NormalizedGaussianSampler
import kscience.kmath.stat.samplers.ZigguratNormalizedGaussianSampler
import kotlin.math.*
/**
* Implements [UnivariateDistribution] for the normal (gaussian) distribution.
*/
public inline class NormalDistribution(public val sampler: GaussianSampler) : UnivariateDistribution<Double> {
public constructor(
mean: Double,

4
kmath-stat/src/commonMain/kotlin/kscience/kmath/stat/internal/InternalErf.kt
View File

@ -2,6 +2,10 @@ package kscience.kmath.stat.internal
import kotlin.math.abs
/**
* Based on Commons Math implementation.
* See [https://commons.apache.org/proper/commons-math/javadocs/api-3.3/org/apache/commons/math3/special/Erf.html].
*/
internal object InternalErf {
fun erfc(x: Double): Double {
if (abs(x) > 40) return if (x > 0) 0.0 else 2.0

2
kmath-stat/src/commonMain/kotlin/kscience/kmath/stat/samplers/AhrensDieterExponentialSampler.kt
View File

@ -12,7 +12,7 @@ import kotlin.math.pow
* Sampling from an [exponential distribution](http://mathworld.wolfram.com/ExponentialDistribution.html).
*
* Based on Commons RNG implementation.
* See https://commons.apache.org/proper/commons-rng/commons-rng-sampling/apidocs/org/apache/commons/rng/sampling/distribution/AhrensDieterExponentialSampler.html
* See [https://commons.apache.org/proper/commons-rng/commons-rng-sampling/apidocs/org/apache/commons/rng/sampling/distribution/AhrensDieterExponentialSampler.html].
*/
public class AhrensDieterExponentialSampler private constructor(public val mean: Double) : Sampler<Double> {
public override fun sample(generator: RandomGenerator): Chain<Double> = generator.chain {

3
kmath-stat/src/commonMain/kotlin/kscience/kmath/stat/samplers/AhrensDieterMarsagliaTsangGammaSampler.kt
View File

@ -15,7 +15,8 @@ import kotlin.math.*
* Marsaglia and Tsang, A Simple Method for Generating Gamma Variables. ACM Transactions on Mathematical Software, Volume 26 Issue 3, September, 2000.
*
* Based on Commons RNG implementation.
* See https://commons.apache.org/proper/commons-rng/commons-rng-sampling/apidocs/org/apache/commons/rng/sampling/distribution/AhrensDieterMarsagliaTsangGammaSampler.html
*
* See [https://commons.apache.org/proper/commons-rng/commons-rng-sampling/apidocs/org/apache/commons/rng/sampling/distribution/AhrensDieterMarsagliaTsangGammaSampler.html].
*/
public class AhrensDieterMarsagliaTsangGammaSampler private constructor(
alpha: Double,

2
kmath-stat/src/commonMain/kotlin/kscience/kmath/stat/samplers/AliasMethodDiscreteSampler.kt
View File

@ -34,7 +34,7 @@ import kotlin.math.min
* that exploit the power of 2.
*
* Based on Commons RNG implementation.
* See https://commons.apache.org/proper/commons-rng/commons-rng-sampling/apidocs/org/apache/commons/rng/sampling/distribution/AliasMethodDiscreteSampler.html
* See [https://commons.apache.org/proper/commons-rng/commons-rng-sampling/apidocs/org/apache/commons/rng/sampling/distribution/AliasMethodDiscreteSampler.html].
*/
public open class AliasMethodDiscreteSampler private constructor(
// Deliberate direct storage of input arrays

2
kmath-stat/src/commonMain/kotlin/kscience/kmath/stat/samplers/BoxMullerNormalizedGaussianSampler.kt
View File

@ -11,7 +11,7 @@ import kotlin.math.*
* distribution.
*
* Based on Commons RNG implementation.
* See https://commons.apache.org/proper/commons-rng/commons-rng-sampling/apidocs/org/apache/commons/rng/sampling/distribution/BoxMullerNormalizedGaussianSampler.html
* See [https://commons.apache.org/proper/commons-rng/commons-rng-sampling/apidocs/org/apache/commons/rng/sampling/distribution/BoxMullerNormalizedGaussianSampler.html].
*/
public class BoxMullerNormalizedGaussianSampler private constructor() : NormalizedGaussianSampler, Sampler<Double> {
private var nextGaussian: Double = Double.NaN

2
kmath-stat/src/commonMain/kotlin/kscience/kmath/stat/samplers/GaussianSampler.kt
View File

@ -9,7 +9,7 @@ import kscience.kmath.stat.Sampler
* Sampling from a Gaussian distribution with given mean and standard deviation.
*
* Based on Commons RNG implementation.
* See https://commons.apache.org/proper/commons-rng/commons-rng-sampling/apidocs/org/apache/commons/rng/sampling/distribution/GaussianSampler.html
* See [https://commons.apache.org/proper/commons-rng/commons-rng-sampling/apidocs/org/apache/commons/rng/sampling/distribution/GaussianSampler.html].
*
* @property mean the mean of the distribution.
* @property standardDeviation the variance of the distribution.

2
kmath-stat/src/commonMain/kotlin/kscience/kmath/stat/samplers/KempSmallMeanPoissonSampler.kt
View File

@ -16,7 +16,7 @@ import kotlin.math.exp
* Sampling uses 1 call to UniformRandomProvider.nextDouble(). This method provides an alternative to the SmallMeanPoissonSampler for slow generators of double.
*
* Based on Commons RNG implementation.
* See https://commons.apache.org/proper/commons-rng/commons-rng-sampling/apidocs/org/apache/commons/rng/sampling/distribution/KempSmallMeanPoissonSampler.html
* See [https://commons.apache.org/proper/commons-rng/commons-rng-sampling/apidocs/org/apache/commons/rng/sampling/distribution/KempSmallMeanPoissonSampler.html].
*/
public class KempSmallMeanPoissonSampler private constructor(
private val p0: Double,

2
kmath-stat/src/commonMain/kotlin/kscience/kmath/stat/samplers/LargeMeanPoissonSampler.kt
View File

@ -18,7 +18,7 @@ import kotlin.math.*
* This sampler is suitable for mean >= 40.
*
* Based on Commons RNG implementation.
* See https://commons.apache.org/proper/commons-rng/commons-rng-sampling/apidocs/org/apache/commons/rng/sampling/distribution/LargeMeanPoissonSampler.html
* See [https://commons.apache.org/proper/commons-rng/commons-rng-sampling/apidocs/org/apache/commons/rng/sampling/distribution/LargeMeanPoissonSampler.html].
*/
public class LargeMeanPoissonSampler private constructor(public val mean: Double) : Sampler<Int> {
private val exponential: Sampler<Double> = AhrensDieterExponentialSampler.of(1.0)

2
kmath-stat/src/commonMain/kotlin/kscience/kmath/stat/samplers/MarsagliaNormalizedGaussianSampler.kt
View File

@ -13,7 +13,7 @@ import kotlin.math.sqrt
* [BoxMullerNormalizedGaussianSampler].
*
* Based on Commons RNG implementation.
* See https://commons.apache.org/proper/commons-rng/commons-rng-sampling/apidocs/org/apache/commons/rng/sampling/distribution/MarsagliaNormalizedGaussianSampler.html
* See [https://commons.apache.org/proper/commons-rng/commons-rng-sampling/apidocs/org/apache/commons/rng/sampling/distribution/MarsagliaNormalizedGaussianSampler.html]
*/
public class MarsagliaNormalizedGaussianSampler private constructor() : NormalizedGaussianSampler, Sampler<Double> {
private var nextGaussian = Double.NaN

2
kmath-stat/src/commonMain/kotlin/kscience/kmath/stat/samplers/PoissonSampler.kt
View File

@ -14,7 +14,7 @@ import kscience.kmath.stat.Sampler
* Devroye, Luc. (1981). The Computer Generation of Poisson Random Variables Computing vol. 26 pp. 197-207.
*
* Based on Commons RNG implementation.
* See https://commons.apache.org/proper/commons-rng/commons-rng-sampling/apidocs/org/apache/commons/rng/sampling/distribution/PoissonSampler.html
* See [https://commons.apache.org/proper/commons-rng/commons-rng-sampling/apidocs/org/apache/commons/rng/sampling/distribution/PoissonSampler.html].
*/
public class PoissonSampler private constructor(mean: Double) : Sampler<Int> {
private val poissonSamplerDelegate: Sampler<Int> = of(mean)

2
kmath-stat/src/commonMain/kotlin/kscience/kmath/stat/samplers/SmallMeanPoissonSampler.kt
View File

@ -17,7 +17,7 @@ import kotlin.math.exp
*
* Based on Commons RNG implementation.
*
* See https://commons.apache.org/proper/commons-rng/commons-rng-sampling/apidocs/org/apache/commons/rng/sampling/distribution/SmallMeanPoissonSampler.html
* See [https://commons.apache.org/proper/commons-rng/commons-rng-sampling/apidocs/org/apache/commons/rng/sampling/distribution/SmallMeanPoissonSampler.html].
*/
public class SmallMeanPoissonSampler private constructor(mean: Double) : Sampler<Int> {
private val p0: Double = exp(-mean)

2
kmath-stat/src/commonMain/kotlin/kscience/kmath/stat/samplers/ZigguratNormalizedGaussianSampler.kt
View File

@ -12,7 +12,7 @@ import kotlin.math.*
* implementation has been adapted from the C code provided therein.
*
* Based on Commons RNG implementation.
* See https://commons.apache.org/proper/commons-rng/commons-rng-sampling/apidocs/org/apache/commons/rng/sampling/distribution/ZigguratNormalizedGaussianSampler.html
* See [https://commons.apache.org/proper/commons-rng/commons-rng-sampling/apidocs/org/apache/commons/rng/sampling/distribution/ZigguratNormalizedGaussianSampler.html].
*/
public class ZigguratNormalizedGaussianSampler private constructor() :
NormalizedGaussianSampler, Sampler<Double> {