vectorize a block matrix calculation

Dear;

I want calculate a big matrix composed of blocks, for that I use the following code:

k=length(a);                          % Long vector
n=size(xx,1);                         % 2D or 3D matrix always the same
TP=repmat(zeros(size(xx),k,1);        % Big matriz composed by blocks 
for i=1:k
    TP((i-1)*n+1:i*n,:)=a(i)*xx;      % Block i of TP matrix
end

Is there any way to vectorize this code and speed up the calculation Many thanks;

Answer 1

[Matt J]

TP=kron(a(:),xx);

[Luis Isaac]

Many thaks Matt

I wondered, if I would want to calculate the sum, what is the correct operation. I mean:

...   
TP((i-1)*n+1:i*n,:)=a(i)+xx;      % Block i of TP matrix
...

Thanks in advanced;

[Andrei Bobrov]

TP = reshape(bsxfun(@plus,xx',reshape(a,1,1,[])),size(x,2),[])'

[Matt J]

You could also use my attached tensorfun() utility.

TP=tensorfun(@plus,a(:),xx);

It allows you to do similar things for any block structure and any bi-operand operation (@plus, @minus, etc...) supported by bsxfun.