Building block diagonal matrix from a known matrix-CodePudding

I want to build a block diagonal matrix (A) from a known matrix (B) by only keeping the elements in the block diagonal from matrix (B). I would like to use R to do it.

For example, B is a 6 by 6 matrix.

B <- matrix(rnorm(36), nrow = 6)

I am looking for a function like this: function(B,3) (3 is just a random number, it represents the total numbers of the block) which returns matrix A like this:

[1,] B11 B12  .   .   .   .
[2,] B21 B22  .   .   .   .
[3,] .    .  B33 B34  .   .
[4,] .    .  B43 B44  .   .
[5,] .    .   .   .  B55 B56
[6,] .    .   .   .  B65 B66

Really appreciate any help

CodePudding user response：

We may use bdiag to construct a sparseMatrix of 1s and multiply with the 'B' matrix so that values that corresponds to 0 in the sparseMatrix becomes 0 and those corresponds to 1 remains the same

library(Matrix)
as.matrix(bdiag(rep(list(matrix(c(1, 1, 1, 1), 2, 2)), 3))) * B

-output

       [,1]      [,2]       [,3]       [,4]     [,5]       [,6]
[1,] -0.8049569 -1.360807  0.0000000  0.0000000 0.000000  0.0000000
[2,] -0.2881172  1.810878  0.0000000  0.0000000 0.000000  0.0000000
[3,]  0.0000000  0.000000 -0.7705610  2.1809346 0.000000  0.0000000
[4,]  0.0000000  0.000000  0.3041445 -0.5894346 0.000000  0.0000000
[5,]  0.0000000  0.000000  0.0000000  0.0000000 2.033442  0.2190562
[6,]  0.0000000  0.000000  0.0000000  0.0000000 2.283470 -0.6036186

If we need it to be sparse, then remove the as.matrix wrapper

> bdiag(rep(list(matrix(c(1, 1, 1, 1), 2, 2)), 3)) * B
6 x 6 sparse Matrix of class "dgCMatrix"
                                                                   
[1,] -0.8049569 -1.360807  .          .         .         .        
[2,] -0.2881172  1.810878  .          .         .         .        
[3,]  .          .        -0.7705610  2.1809346 .         .        
[4,]  .          .         0.3041445 -0.5894346 .         .        
[5,]  .          .         .          .         2.033442  0.2190562
[6,]  .          .         .          .         2.283470 -0.6036186