GelmanRubin: function for monitoring convergence of Markov chain

In Scopia/StatComp18053: generating and monitoring convergence of Markov chain and permutation test for equal distribution

Description

This function use the Gelman-Rubin method to monitor convergence of Markov chains.

Usage

GelmanRubin(psi)

Arguments

psi	the statistic for chain of X

Details

user should provide the statistic psi(X)

References

Statistic computing with R. Maria L. Rizzo

Examples

## Not run: 
lden=dcauchy
sigma=3
M4=cumsum((MC4=Metropolis(dist=lden,N=n,sigma=sigma,print_acc=F)))/1:length(MC4)
M3=cumsum((MC3=Metropolis(dist=lden,N=n,sigma=sigma,print_acc=F)))/1:length(MC3)
M2=cumsum((MC2=Metropolis(dist=lden,N=n,sigma=sigma,print_acc=F)))/1:length(MC2)
M1=cumsum((MC1=Metropolis(dist=lden,N=n,sigma=sigma,print_acc=F)))/1:length(MC1)
psi=rbind(M1,M2,M3,M4)
plot((R=sapply(1:2000,function(i)GelmanRubin(psi[,1:i]))),
      main="R value of Gelman-Rubin method",ylim=c(1,2))

## End(Not run)

Scopia/StatComp18053 documentation built on May 22, 2019, 2:44 p.m.