generateK: Generate Standard Gaussian samples with a Gaussian transiiton...

In mistral: Methods in Structural Reliability

Generate Standard Gaussian samples with a Gaussian transiiton kernel

Description

Generate Standard Gaussian samples with a Gaussian transiiton kernel

Usage

generateK(X, N = 100, thinning = 4, sigma = 1, lsf, burnin = 20)

Arguments

X	the seeds for the Markov Chain. There are as many MC drawn as given seeds
N	the number of desired samples"'
thinning	the proportion of kept samples, ie. 1 each thinning draw.
sigma	the exploration parameter for the transition kernel
lsf	a boolean limit-state function for definig a subdomain of the input space.
burnin	the burnin parameter, ie. the number of discarded samples before keeping one.

Details

This function generates standard Gaussian samples with a Markov Chain using a suitable transition kernel

Value

A matrix X with the number of desired samples

Author(s)

Clement WALTER [e-mail]

Examples

# Get a seed in dimension 2
X <- matrix(rnorm(2), nrow = 2)
X <- generateK(X, N = 1000)

library(ggplot2)
ggplot(as.data.frame(t(X)), aes(x_1,x_2)) + geom_point()

# One can also specify a limit-state function
lsf <- function(X){
     sqrt(colSums(X^2)) > 2
}
X <- matrix(c(2, 2), nrow = 2)
X <- generateK(X, N = 1000, lsf = lsf)

ggplot(as.data.frame(t(X)), aes(x_1,x_2)) + geom_point()

mistral documentation built on Aug. 29, 2025, 5:13 p.m.