CODEWITHSUNDEEP
rsample
William Liu
William Liu

Reputation: 339

R Gibbs sampler error subscript out of bounds

I am building a small gibbs sampler but R returns: "Error in [<-(*tmp*, i, j, value = NA_real_) : subscript out of bounds".

Problem is with this line:

mu[i,j] <- alpha[i]+beta[i]*(x[j]-mean(x[]))

I just do not understand why?

GS <- function(x){
    alpha <- c()
    beta <- c()
    mu <- matrix()
    Y <- matrix ()
    for (i in 1:x){
        alpha[i] <- rnorm(1,0)
        beta[i] <- rnorm(1,0)
        for (j in 1:5){
            mu[i,j] <- alpha[i]+beta[i]*(x[j]-mean(x[]))
            Y[i,j] <- rnorm(1, mu[i,j], 1)
        }
        Y
    }
}

Upvotes: 0

Views: 679

Answers (1)

nist
nist

Reputation: 1721

The problem is that you don't initialize your vectors or matrices so the vectors will have lenght 0 and the matrices will only consist of one NA. So you can't write anything to position i since it doesn't exist.

Change 

    alpha <- c()
    beta <- c()
    mu <- matrix()
    Y <- matrix ()

to 

    alpha <- numeric(x)
    beta <- numeric(x)
    mu <- matrix(0,x,5)
    Y <- matrix (0,x,5)

Upvotes: 3

Related Questions