Build huge matrix with vector components

1 visualización (últimos 30 días)
Kevin
Kevin el 10 de Jul. de 2014
Comentada: Kevin el 11 de Jul. de 2014
how can I avoid double for in matlab in order to build matrices like this code do:
pos=[0 0];
for i=1:m;
for j=1:n;
pos=[pos; i j];
end
end
m and n are numbers like 500 and 900.
I have to find a better solution in order improve computation time . Thank you so much
Vins

Respuesta aceptada

Jos (10584)
Jos (10584) el 11 de Jul. de 2014
n = 4 % a small example
m = 3
[a, b] = ndgrid(1:n,1:m)
pos = [0 0 ; b(:) a(:)]

Más respuestas (2)

James Tursa
James Tursa el 11 de Jul. de 2014
Editada: James Tursa el 11 de Jul. de 2014
One way:
c1 = ones(n,1)*(1:m);
c2 = (1:n)'*ones(1,m);
pos = [0 0; c1(:) c2(:)];
Or a similar method using repmat:
c1 = repmat(1:m,n,1);
c2 = repmat((1:n)',1,m);
pos = [0 0; c1(:) c2(:)];
You could have dramatically decreased the run time of your double loop simply by pre-allocating pos. E.g.,
pos = zeros(m*n+1,2);
k = 1;
for i=1:m;
for j=1:n;
k = k + 1;
pos(k,:) = [i j];
end
end

Image Analyst
Image Analyst el 11 de Jul. de 2014
You don't have to use double. You can use int16 if you want
for i = uint16(1:500)
and so on. It will be one 8th the size, though a 450,000 by 2 array is far from "huge". That's like 7 MB if it's double and less than 2 MB if uint16. Do you know how big an ordinary run of the mill point and shoot camera's digital image is? It's way bigger than that!

Categorías

Más información sobre Matrix Indexing en Help Center y File Exchange.

Community Treasure Hunt

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

Start Hunting!

Translated by