Need to improve efficiency in this code, which multiplies a 3D matrix by a 2D one

1 visualizzazione (ultimi 30 giorni)
Hi, I have this code, which uses for loops and multiplies a 3D matrix with a 2D one. It works, but I need f_N to be 3200, and once f_N is that high, it is just too big a calculation for my computer to do. The weird thing is, I have run this code successfully about 10 months ago - I know this because I have a plot generated from it. But I must have changed something maybe.
Anyway, here's the code:
clear all;
num_x = 275;
num_y = 275;
f_N = 100;
W_rsTpre = rand(25,num_x,num_y);
W_rsT(1,:,:,:) = W_rsTpre;
q_pre = rand(25,f_N);
q(:,1,:) = q_pre;
w_pre = zeros(num_x,num_y,1,f_N); % preallocate
for xx = 1:num_x
for yy = 1:num_y
for ff = 1:f_N
w_pre(xx,yy,1,ff) = W_rsT(1,:,xx,yy)*q(:,1,ff);
w = squeeze(w_pre);

Risposta accettata

Andrei Bobrov
Andrei Bobrov il 13 Lug 2015
Modificato: Andrei Bobrov il 13 Lug 2015
W = reshape(W_rsT,size(W_rsT,2),[]).';
q1 = squeeze(q);
w_pre = reshape(W*q1,[num_x, num_y,1,f_N]);
  1 Commento
Tom il 16 Lug 2015
Thank you. Strangely, when I used iequal to see if the two results matched, they did not. But when I plotted them both, they seemed to sit on top of one another - so they must have been pretty much the same.
Many thanks

Accedi per commentare.

Più risposte (0)


Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by