simany {subrank}

Test statistic distribution under any hypothesis

Description

Simulates the test statistic, under independence

Usage

simany(sampsize,dimension,subsampsizes,sampnum,nbsafe=5,nthreads=2, fun=NULL, ...)

Arguments

sampsize

sample size

dimension

sample dimension

subsampsizes

vector of sub-sample sizes

sampnum

number of samples

nbsafe

the ratio between the number of sub-samples and the cardinality of the discretized copula.

nthreads

number of number of threads, assumed to be strictly positive. For "full throttle" computations, consider using parallel::detectCores()

fun

the function describing the dependence.

...

optional arguments to fun

Value

lrs

the distances with independent case

lrs2mean

the distances with theoretical value, given dependence fun

scarcities

the proportions of non-reached vector ranks

DistTypes

a recall of the list of the distance types: "KL","L2","L1","APE"

Author(s)

Jerome Collet

Examples

depquad <- function(lon,dd,a)
{
  x <- rnorm(lon)
  y0 <- a*x^2
  y <- y0 + rnorm(lon)
  reste=rnorm((dd-2)*lon)
  return(c(x,y,reste))
}
sims0=simany(101,3,8,50,nbsafe=1)
seuils=apply(sims0$lrs,3,quantile,0.95)
seuils=matrix(ncol=4,nrow=50,seuils,byrow=TRUE)
sims1=simany(101,3,8,50,nbsafe=1,fun=depquad,a=0.5)
apply(sims1$lrs[,1,]>seuils,2,mean)