create big matrix with only 0s and 1s

3 次查看（过去 30 天）
freebil
freebil 2013-11-6
评论： freebil 2013-11-7
I want to run the following
n = 10^6;
m = n/2;
H = sparse(m,n);
parfor jj=1:n
pos = sample(cumulative);
H(:,jj) = (rand(m,1) > 1 - degrees(pos)/m);
end
but the memory is over.
degrees(pos)/m is a different number every time. Is there any way to run it quickly?

请先登录，再回答此问题。

回答（1 个）

the cyclist
the cyclist 2013-11-6
Unless I did the math in my head wrong, you are trying to create a matrix that is 4 terabytes of data (and is not actually sparse). That's too big.
  2 个评论
Image Analyst
Image Analyst 2013-11-6
theodor, why do you need it that big? You'll just have to get by without the whole thing in memory at once. You'll have to figure out how to process the thing in smaller chunks.
freebil
freebil 2013-11-7
thanks for the answers.. I actually don't need zeros but only 1s.. Is there any way to create a sparse matrix with only ones? I thought something like that
finalcolpos=0;
parfor jj=1:n
pos = sample(cumulative);
col = (rand(m,1) > 1 - degrees(pos)/m);
colpos = find(col);
finalcolpos = [finalcolpos; colpos];
end
finalcolpos(1)=[];
H(finalcolpos)=1;
but it is slow for n>10^5

请先登录，再进行评论。

类别

在 Help Center 和 File Exchange 中查找有关 Creating and Concatenating Matrices 的更多信息

标签

Community Treasure Hunt

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

Start Hunting!

Translated by