[R] Matching vector order by indicators in a matrix

[Bert Gunter]

Folks:

This is sufficiently "tricky" that a word of explanation might be
helpful to reveal the magic.

>
> Rates <- c(0.006280048, 0.330934659)
> Index.Matrix <- matrix(c(1, 2, 1, 2), 2, 2)
>
>> Rates
> [1] 0.006280048 0.330934659
>
>> Index.Matrix
>     [,1] [,2]
> [1,]    1    1
> [2,]    2    2
>
>
> Just index 'Rates' by the values in Index.Matrix:
>
>> Rates[Index.Matrix]
> [1] 0.006280048 0.330934659 0.006280048 0.330934659
>
> NewMatrix <- matrix(Rates[Index.Matrix], dim(Index.Matrix))
>
>> NewMatrix
>            [,1]        [,2]
> [1,] 0.006280048 0.006280048
> [2,] 0.330934659 0.330934659
>

The key here is to understand that matrices (arrays of any dimension
actually) are just vectors stored in column major order (column 1 on
top of column 2 ....) . So Rates[Index.Matrix] is just a length 4
vector concatenating two copies of Rates when you think of
Index.matrix as a vector, which it must be here to index a vector
(There's also matrix indexing of arrays, but that's something else
again). The matrix() function then just makes it a matrix with the
desired dimension.

So no longer tricky, right?

Cheers,
Bert

Bert Gunter
Genentech