problem using matrix indexing with bsxfun.

I'd like to calculate only A(a)*B(b), while keeping the original format of bsxfun(@times,A,B)

>>  A=[-1 2 -3;4 -5 6;-7 8 -9];

B(:,:,1)=[-1 2 -3;4 -5 6;-7 8 -9];
B(:,:,2)=[-1 2 -3;4 -5 6;-7 8 -9];

bsxfun(@times,A,B)

ans(:,:,1) =

       1     4     9
      16    25    36
      49    64    81

ans(:,:,2) =

       1     4     9
      16    25    36
      49    64    81

>> a=A<0;b=B<0;

>> bsxfun(@times,A(a),B(b))

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

Answer 1

[Matt J]

I'd like to calculate only A(a)*B(b)

Have you checked what A(a) and B(b) look like? They are both vectors of different sizes so A(a)*B(b) has no clear definition,

    >> A(a)

    ans =

        -1
        -7
        -5
        -3
        -9

    >> B(b)

    ans =

        -1
        -7
        -5
        -3
        -9
        -1
        -7
        -5
        -3
        -9

If you want all combinations of products A(a(i))*B(b(j)), you don't need bsxfun at all. It's just an outer product calculation,

 A(a)*B(b).'

If this is not what you want, then you need to clarify what the final result should look like.

[Tristan]

what I'd like is for the results of B(b) to be kept multidimensional, like this:

B(b)(:,:,1) =

    ans =

        -1
        -7
        -5
        -3
        -9

B(b)(:,:,2) =

    ans =

        -1
        -7
        -5
        -3
        -9

[Matt J]

I assume you know for your specific data that such a multi-dimensional reshaping will always be possible. If you're sure it will be, then you can do

 Aa=A(A<0);
 Bb=reshape(B(B<0),[],1,size(B,3));

 bsxfun(@times,Aa,Bb)

[Tristan]

perfect, thanks.