Construct large sparse matrix by blocks

General Usage Performance
1

I’m looking for a function to construct a sparse matrix based on several blocks of smaller sparse matrices.
An simple example:
A11=A12=A21=A22=sprandn(4,2,0.6)
The final sparse matrix A should be a combination of these 4 blocks like:
A11 A12
A21 A22
blockdiag() doesn’t help directly here since it only returns block diagonal matrix. And the matrices I worked with are pretty large. So I do care about the performance of such operation.

2

Is normal concatenation not good/fast enough?

julia> A11 = A12 = A21 = A22 = sprand(4, 4, 0.6);

julia> A = [A11 A12;
            A21 A22];
3

That’s what I’m looking for! Thank you!!

4

What do I do when the number of blocks are dynamic?

5

[ ] is just helpful syntax for the functions vcat, hcat, and hvcat, so you can just call those yourself:

julia> blocks = [eye(2) for _ in 1:6]

julia> hcat(blocks...)
2×12 Array{Float64,2}:
 1.0  0.0  1.0  0.0  1.0  0.0  1.0  0.0  1.0  0.0  1.0  0.0
 0.0  1.0  0.0  1.0  0.0  1.0  0.0  1.0  0.0  1.0  0.0  1.0

julia> vcat(blocks...)
12×2 Array{Float64,2}:
 1.0  0.0
 0.0  1.0
 1.0  0.0
 0.0  1.0
 1.0  0.0
 0.0  1.0
 1.0  0.0
 0.0  1.0
 1.0  0.0
 0.0  1.0
 1.0  0.0
 0.0  1.0

julia> hvcat((3, 3), blocks...)
4×6 Array{Float64,2}:
 1.0  0.0  1.0  0.0  1.0  0.0
 0.0  1.0  0.0  1.0  0.0  1.0
 1.0  0.0  1.0  0.0  1.0  0.0
 0.0  1.0  0.0  1.0  0.0  1.0

julia> hvcat((2, 2, 2), blocks...)
6×4 Array{Float64,2}:
 1.0  0.0  1.0  0.0
 0.0  1.0  0.0  1.0
 1.0  0.0  1.0  0.0
 0.0  1.0  0.0  1.0
 1.0  0.0  1.0  0.0
 0.0  1.0  0.0  1.0
3 Likes