How to optmized for loop

[pankaj kumarParticipant]

I have a vector “dd” having 25,000 element ,for which I like to check some property using heviside function as below but it goes out of memory. May I know any way to optimized the following code.

for i = 1:25000
for j = 1:25000
M(i,j)=heaviside(0.1-sqrt(abs(dd(i)-dd(j))));
end
end

[cKKeymaster]

For this big calcs you need huge amounts of memory. This one would normally need 25e3 x 25e3 x 8bytes (double data) = 4.7GB just to store the variable. A computer with 8GB memory will likely pull it off. But if precision can be traded off, you halve the above memory requirement by using single data rather than double. Initialise M as
M(25e3,25e3) = single(0);

Then, you may speed up by vectorising the calculations. Use only one loop and compute the formula for each row at a time as
M(:,j) = heaviside(0.1-sqrt(abs(dd-dd(j))));

• This reply was modified 9 years, 9 months ago by  cK.

[Author]

Posts