KSLCore/ksl.simopt.solvers.algorithms/CrossEntropySolver/Companion

Companion

object Companion(source)

Properties

ceDefaultMaxIterations

@JvmStatic

var ceDefaultMaxIterations: Int

The maximum number of iterations permitted for the main loop for the Cross-Entropy method. This must be greater than 0.

defaultCEQuantileConfidenceLevel

@JvmStatic

var defaultCEQuantileConfidenceLevel: Double

A value between 0 and 1 that represents the default approximate confidence level for estimating the quantile from the cross-entropy sample. By default, this is 0.95.

defaultElitePct

@JvmStatic

var defaultElitePct: Double

A value between 0 and 1 that represents the default proportion of the CE sample that determines the elite sample. By default, this is 0.1.

defaultMaxCESampleSize

@JvmStatic

var defaultMaxCESampleSize: Int

This value is used to help determine the cross-entropy sample size (population) that is processed to determine the elite sample. By default, this value is 100. This value represents the default maximum CE population size that is permissible.

defaultMaxProportionHalfWidth

@JvmStatic

var defaultMaxProportionHalfWidth: Double

A value between 0 and max(0,defaultElitePct) that represents the default confidence level half-width bound on the proportion estimate for determining the quantile from the cross-entropy sample. By default, this is 0.1. See Chen and Kelton (1999)

defaultMinCESampleSize

@JvmStatic

var defaultMinCESampleSize: Int

This value is used as the default minimum size for the cross-entropy population sample. By default, set to 10. The size of elite sample must be 2 or more.

defaultMinEliteSize

@JvmStatic

var defaultMinEliteSize: Int

This value is used as the default minimum size of the elite sample. By default, set to 5. The size of elite sample must be at least 1 or more.

defaultNoImproveThreshold

@JvmStatic

var defaultNoImproveThreshold: Int

This value is used as the default termination threshold for the largest number of iterations during which no improvement of the best function value is found. By default, set to 5.

Functions

recommendCESampleSize

fun recommendCESampleSize(elitePct: Double = defaultElitePct, ceQuantileConfidenceLevel: Double = defaultCEQuantileConfidenceLevel, maxProportionHalfWidth: Double = defaultMaxProportionHalfWidth): Int

Based on the theory in See Chen and Kelton (1999), this function computes a recommended cross-entropy sample size. The sample size may be sufficient to adequately estimate the quantile associated with the desired elite percentage using on a desired quantile confidence level and an adjusted half-width bound for the associated quantile proportion.