Find positions of elements in rows of one matrix in rows of another matrix (efficiency question)

Hello.

Im trying to process 2 big matrices in the following way:

a = perms(1:9);
b = perms(1:9);
c = zeros(size(a));

[~,m] = size(a);

tic;
for i = 1:m
    c(bsxfun(@eq, a, b(:,i))) = i;
end
toc

Note that a and b are generated with perms just for example purposes.

Can this code be improved to run faster?

Thanks!

Answer 1

[Matt J]

N=500;
a = randi(N,N);
b = randi(N,N);
c1 = zeros(size(a));

[~,m] = size(a);

tic;
for i = 1:m
    c1(bsxfun(@eq, a, b(:,i))) = i;
end
toc

Elapsed time is 0.399945 seconds.

tic;
 map=reshape(fliplr(b),[],1,m)==a;
 [map,c]=max(map,[],3);
 c=(m+1-c).*map;
toc

Elapsed time is 0.178260 seconds.

isequal(c1,c)

ans = logical
   1

[Somebody]

Hello.

This doesnt work because element, say, b(1,1), can be on any position in a(1,:). And this position must be flagged as 1 in c. So loc calculated incorrectly.

[Matt J]

OK. I fixed it.

[Somebody]

Thanks alot!

It realy works faster.