multiplying each non zero element of a sparse matrix by a certain number

Hello all,

I have a sparse matrix S with few non-zero elements in its row and columns. I have a vector P with the same number of elements as the total non-zero elements of S. I would like to multiply each non-zero element of S by a value in P. For example, let

S = [0 0 1 1;0 1 1 0;1 1 0 0; 1 0 0 1]
P = [0.7421 0.3751 1.9079 1.4141 1.4411 2.0796 2.4199 1.1002]

I would like to get

W = [0 0 0.7421 0.3751;0 1.9079 1.4141 0;1.4411  2.0796 0 0; 2.4199 0 0 1.1002]

Can somebody please help me with a fast way of doing this task? Thank you.

Answer 1

[Star Strider]

This works — and NO LOOPS:

S = [0 0 1 1;0 1 1 0;1 1 0 0; 1 0 0 1] ;
P = [0.7421 0.3751 1.9079 1.4141 1.4411 2.0796 2.4199 1.1002];

Snz = find(S == 1);
Wv = S(Snz).*P';
W = zeros(1,numel(S));
W(Snz) = Wv;
W = reshape(W, size(S))'

W =
            0            0       0.7421       0.3751
            0       1.9079       1.4141            0
       1.4411       2.0796            0            0
       2.4199            0            0       1.1002

[Mnr]

Thank you!

[Star Strider]

As always, my pleasure!

Answer 2

[James Tursa]

W = S;
G = logical(S);
W(G) = W(G) .* P(:);

Note: This preserves the sparse attribute of S. I.e., if S is sparse, then W will be sparse. If S is full, then W will be full.

[Mnr]

Thanks! but it does not give me the same answer as Star Strider's method?!

[James Tursa]

Ah ... I didn't notice that you wanted the multiplication by rows, not columns. So some transposes will be required for this method. E.g.,

W = S';
G = logical(W);
W(G) = W(G) .* P(:);
W = W';

[Mnr]

I really appreciate it!

Answer 3

[Stalin Samuel]

S = [0 0 1 1;0 1 1 0;1 1 0 0; 1 0 0 1] ;
 P =  [0.7421 0.3751 1.9079 1.4141 1.4411 2.0796 2.4199 1.1002] ;
[r c] = size(S);
W = zeros(r,c);
n = 1;
for r1 = 1:r
    for c1 = 1:c
        if S(r1,c1)~=0
            W(r1,c1) =  S(r1,c1)*P(n) ;
            n = n+1;
        end
    end
end