vectorizing 3 nested loops

Assuming I have the following matrices/vectors

A=[1 3 0
   2 0 0
   4 5 6];

B=[111
   222
   333];

C=[10 5 20 25 3 7
   1 2 4 6 8 12
   11 22 33 44 55 66] ;

I=[0 1 0 0 1 1
   1 1 1 1 1 0
   0 0 0 0 0 0] ;

D and E are also initialized to 0 before the loops

I have to following 3 nested loops that I want to replace/vectorize. How can I do it?

    [nrows, ncols]= size(A);
    E=zeros(nnz(A),1);
    D=zeros(nnz(A),nrows);
    for i = 1:nrows
        for j = 1:ncols
              if A(i,j)~=0
                  Id=A(i,j);
                  E(Id)=C(i,Id)+ B(i);
                  [p,~] =find(I(:,Id)~=0);
                  for k=p
                      D(Id,k)=C(k,Id)+(B(k)+1);
                  end
              else
                  break
              end
         end
    end

[Stephen23]

When I use those matrices and that code I get this error:

Subscripted assignment dimension mismatch.
Error in Untitled3 (line 38)
                D(Id,m)=C(k,Id)+(B(k)+1);

[etudiant_is]

You are right, I had a mistake in my example. I edited the question, now the loops run fine.

[Matt J]

Are the non-zero values of A always distinct?

[etudiant_is]

yes, they can not be present more than once in A.

Answer 1

[Matt J]

        [ma,na]=size(A);
            nnza=nnz(A);
        [mc,nc]=size(C);
        [mi,ni]=size(I);

            [i,j,Id]=find(A);
             k=sub2ind([ma,nc],i,Id);

        E=sum(accumarray([i,Id],C(k)+B(i)) ,1).',

           IA=bsxfun(@times,I,nonzeros(A).');
           [i,j,Id]=find(IA);
           k=sub2ind([mi,nc],i,Id);

        D=accumarray([i,Id],C(k)+B(i)+1,[ma,nnza]).'