Break square matrix into pairs

Question

I have a large square matrix 12128 x 12128. Example

      A    B    C    D     E
A    0.5  0.4  0.1  0.02  0.4
B    0.1  0.3  0.07 0.03  0.9
C    0.8  0.04 0.5  0.4   0.4
D    3.4  5.6  9.5  2     2.1
E    0.6  9.1  0.3  0.5   1.2

I have a list of pairs (# 18000), which exist in all the possible pairs from the matrix

 A   B
 B   E
 C   E

I need the values for only the pairs in my list from the matrix. Is there a easier way to do it rather than breaking down the whole matrix (melt(matrix))? Thanks

Answer 1

Using m and ind from the other post as test data try this:

m[ind]

EDIT:

If the row and column names can have duplicates (as suggested in a later comment) and if we want all possible values from these then try this:

merge(as.data.frame.table(m), ind, by = 1:2)

Answer 2

I make no promises about the performance characteristics of this method, since you actual data is fairly large, but this may at least give you the necessary tools to work with:

#Some example data like yours
m <- matrix(1:25,5,5)
rownames(m) <- letters[1:5]
colnames(m) <- letters[1:5]

#Matrix of row/col indices to select
ind <- cbind(c('a','e','b'),c('d','a','d'))

#Select elements of m matching ind
diag(m[match(ind[,1],rownames(m)),match(ind[,2],colnames(m))])