Fast matrix multiplication in loop

Dear All,

I have two matrices with dimensions 3x3 and E6x3. I need to multiply each row of the latter with the former. It's like

a=rand(3,3);
b=(1000000,3);
for i=1:size(b,1)
   c=a*b(i,:)';
end

However, this step takes hours to be done. I wonder if there is any faster way to do this.

Cheers.

[Jos (10584)]

element-by-element or matrix multiplications?

Answer 1

[Azzi Abdelmalek]

a=rand(3,3);
b=rand(100,3);
n=size(a,2);
m=size(b,1);
c=zeros(m,n);
for i=1:size(b,1)
 c(i,:)=a*b(i,:)';
end

%or simply

c=(a*b')'

Answer 2

[Jos (10584)]

Two options:

1. pre-allocate C to avoid memory allocation in each iteration

C = zeros(N, ..) % pre-allocation 
for k = 1:N,
   C(k,:) = ..
end

2. Use BSXFUN

[Batuhan]

Thank you. But I'm not sure if bsxfun is the right one, since

C=bsxfun(@times,a*b) results in

Error using bsxfun Non-singleton dimensions of the two input arrays must match each other.

Answer 3

[Batuhan]

Thank you all. Really, preallocation was the issue and fixes it.

[Matt J]

No, it's crazy to do this with a loop, pre-allocated or otherwise. Just do c=b*a', as Azzi noted.