kmath-stat/space.kscience.kmath.random

Package-level declarations

Types

class DefaultGenerator(val random: Random = Random) : RandomGenerator

Implements RandomGenerator by delegating all operations to Random.

class MCScope(val coroutineContext: CoroutineContext, val random: RandomGenerator)

A scope for a Monte-Carlo computations or multi-coroutine random number generation. The scope preserves the order of random generator calls as long as all concurrency calls is done via launch and async functions.

RandomChain

class RandomChain<out R>(val generator: RandomGenerator, gen: suspend RandomGenerator.() -> R) : Chain<R>

A possibly stateful chain producing random values.

RandomGenerator

interface RandomGenerator

An interface that is implemented by random number generator algorithms.

UniformDoubleChain

class UniformDoubleChain(val generator: RandomGenerator) : BlockingDoubleChain

A type-specific double chunk random chain

Functions

asCoroutineScope

fun MCScope.asCoroutineScope(): CoroutineScope

async

inline fun <T> MCScope.async(context: CoroutineContext = EmptyCoroutineContext, start: CoroutineStart = CoroutineStart.DEFAULT, crossinline block: suspend MCScope.() -> T): Deferred<T>

Specialized async for MCScope. Behaves the same way as regular CoroutineScope.async, but also stores the generator fork. The method itself is not thread safe.

chain

fun <R> RandomGenerator.chain(generator: suspend RandomGenerator.() -> R): RandomChain<R>

Create a generic random chain with provided generator

launch

inline fun MCScope.launch(context: CoroutineContext = EmptyCoroutineContext, start: CoroutineStart = CoroutineStart.DEFAULT, crossinline block: suspend MCScope.() -> Unit): Job

Specialized launch for MCScope. Behaves the same way as regular CoroutineScope.launch, but also stores the generator fork. The method itself is not thread safe.

mcScope

inline suspend fun <T> mcScope(seed: Long, block: MCScope.() -> T): T

Launch mc scope with a given seed

inline suspend fun <T> mcScope(generator: RandomGenerator, block: MCScope.() -> T): T

Launches a supervised Monte-Carlo scope