Question

Sorry for vague question title, i couldn’t figure out something more specific.

I have 3×2 matrix c:

> c
     [,1] [,2]
[1,]    1    2
[2,]    1    3
[3,]    2    3

It is important that ncol(c) == 2.

I also have matrix ind:

> ind
      [,1] [2] [,3] [,4]
[1,]    2    2    2    1
[2,]    1    1    2    2
[3,]    2    2    2    1

It is important that nrow(c) == nrow(ind), and that the values of matrix ind are 1 and 2 (like column indices for each row of c)

What i want to get is matrix a with same dim as ind such that a[i,j] == c[i,ind[i,j]]:

> a
      [,1] [2] [,3] [,4]
[1,]    2    2    2    1
[2,]    1    1    3    3
[3,]    3    3    3    2

I can do something similar in less comprehensive situations, for example if nrow(c) == 1 i’ll use apply:

 > apply(c,2,function(x){return(matrix(x[ind], nrow(ind)))})

I know there is a way to iterate by 2 lists using mapply, but

1) i don’t know what’s the best way to represent matrix as list of rows
2) i fing this solution ugly

What is the best way to achieve what i descibed here?

Answer 1

Matrix indexing to the rescue!

> c.mat <- matrix(c(1,1,2,2,3,3), ncol=2)
> ind <- matrix(c(2,1,2,2,1,2,2,2,2,1,2,1), ncol=4)
> matrix(c.mat[cbind(as.vector(row(ind)), as.vector(ind))], ncol=ncol(ind))
     [,1] [,2] [,3] [,4]
[1,]    2    2    2    1
[2,]    1    1    3    3
[3,]    3    3    3    2