MCMC visualization from sublime text

metropolis-hasting-method.r

Sample <- c(4,3,4,7,7,
			6,3,7,8,0,
			1,7,8,6,7,
			4,4,7,3,4)
n <- 8
Step <- 0.001
mcmcStep <- 20000
mcmcTimes <- 20
 
Z <- 1:mcmcStep
 
LE <- function(p){
	prod(dbinom(Sample,n,p))
}
 
 
Culc.metro <- function(ini){
 
	X <- 1:mcmcStep
 
	q <- ini
	for(i in 1:mcmcStep){
	
		d <- Step *(2 * rbinom(1,1,0.5) - 1) 
		q1 <- q + d
		
		if(LE(q) < LE(q1)){
			q <- q1
		}else if(LE(q) > LE(q1)){
 
			if(rbinom(1,1,LE(q1)/LE(q)) == 1){
				q <- q1
			}
		}
		X[i] <- q1
  	}#1:100
 
  	X	#return
 
}
 
Init <- round(runif(mcmcTimes, min=0, max=1),2)
 
plot(Z,Culc.metro(Init[1]),ylim = c(0,1),type = "l")
 
 
for(i in 2:mcmcTimes){
	par(new = TRUE)
	plot(Z,Culc.metro(Init[i]),ylim = c(0,1),type = "l")
}