Repmat with pattern

Does anyone know an efficient way of replicating a matrix in a predetermined pattern(such as diagonal or triangular)?

Some examples: 1. We have A=[1,2;3,4] and we want B=[A,zeros(2);zeros(2),A] 2. We have A=[1,2;3,4] and we want B=[A,zeros(2);A,A] 3. We have a=[1;2;3;4] and we want B=[a,zeros(4,1);zeros(4,1),a]

Answer 1

[Walter Roberson]

1. kron([1 0;0 1],A)
2. kron([1 0;1 0],A)
3. kron([1 0 0;0 0 1],A)

That is, put a 1 for each copy of A that should appear, and put a 0 for each place 0's the same size as A should appear.

[Andrew Newell]

Wow! I wish I could vote for this one twice!

[Andrew Newell]

However, I think 2 and 3 should be

kron([1 0; 1 1],A)

kron([1 0; 1 0],A)

[Mohammad Tabesh]

Just Awesome!

[Walter Roberson]

2 should be [1 0;1 1], but for #3 notice that Mohammad wanted 4 zeros rather than 2.

[Andrew Newell]

Right, but lowercase a is a 4 x 1 vector.

[Walter Roberson]

Ah, I missed that different variables were being used.

Answer 2

[Andrew Newell]

For 1. you could use

B = blkdiag(A,A);

For 2, why not just use the command you gave?

B=[A,zeros(2);A,A] ;

For 3:

B=[a' zeros(1,4); zeros(1,4) a']'

EDIT: Here is a generalization of 1:

n = 3; % e.g.
Amat = repmat(A,n,1);
Acell = mat2cell(Amat,2*ones(1,n),2);
B = blkdiag(Acell{:})

[Mohammad Tabesh]

Thanks for the answer but the point is that I don't want to hard code the number of times I need to replicate the A matrix (Imagine that I need 10 replications of A for example 2)