is there any way to replace this code part with an equivalent part that runs faster?

yy = 0.5*ones(300,500);

yy_abs = abs_coeff_mat.*yy;

yy_abs_cum=zeros(size(yy_abs));

for i = 2:size(yy_abs,1)

yy_abs_cum(i,:) = sum(yy_abs(1:i,:));

end

Answer 1

[Jos (10584)]

help cumsum

Answer 2

[Walter Roberson]

As you appear to be running those in a loop, move the

 ones(300,500)
 

to outside the loop.

You do not appear to be doing implicit expansion, so size(yy_abs)appears to be the same as that 300, 500, so you can probably move the zeros(size(yy_abs)) to outside the loop too.

You can probably replace the loop with a cumsum() followed by zeroing the first row of the result.

I have to say that it looks odd that you would want row 2 to be the sum of row 1 and 2 of yy_abs, and row 3 to be the sum of rows 1, 2, 3 of yy_abs, and so on, but that you want row 1 of the output to be left at zero from the initialization to 0, instead of being the same as the first row of yy_abs.