R - "row-wise" sampling from Matrix

Question

Let's say I have 2x matrices:

x <- matrix(rep(1:5, 2), ncol=5, byrow=T)
prob <- matrix(rep(0, each=5, 2), ncol=5, byrow=T)
prob[1, 3] <- 1
prob[2, 4] <- 1

> x
     [,1] [,2] [,3] [,4] [,5]
[1,]    1    2    3    4    5
[2,]    1    2    3    4    5

> prob
     [,1] [,2] [,3] [,4] [,5]
[1,]    0    0    1    0    0
[2,]    0    0    0    1    0

I want to perform "row-wise" sampling from the matrices with sample() . For example, sample will select one value from each row of x using the probabilities in the same row of prob . In the above example, I want sample to return c(3,4) every time because the probability has been set to 3=100% for row 1 and 4=100% for row 2.

However, when I run sample(x, prob=prob, size=nrow(x)) , I receive 3 and 4 is any order:

> sample(x, prob=prob, size=nrow(x))
[1] 4 3
> sample(x, prob=prob, size=nrow(x))
[1] 3 4

How can I perform "row-wise" sampling of the matrix x ?

Answer 1

One option with mapply and asplit

mapply(sample, asplit(x, 1 ), prob = asplit(prob, 1), size = 1)
#[1] 3 4

Answer 2

You can use vapply (safer by specifying what you're expecting) to loop over each row:

x <- matrix(rep(1:5, 2), ncol=5, byrow=T)
prob <- matrix(rep(0, each=5, 2), ncol=5, byrow=T)
prob[1, 3] <- 1
prob[2, 4] <- 1


vapply(seq_len(nrow(x)), function(i) {
  sample(x[i, ], prob = prob[i, ], size = 1)
}, FUN.VALUE = numeric(1)
)
#> [1] 3 4