How can I code this more effectively regarding running time?

I have this data: 'randoms' (cell of cells array size 4319*100) that is: (4319 columns)

my code runs, but it takes 7 min! is there any way to make it more efficient?

this is my code:

for j=1:length(randoms)
    all_100=randoms{j};
    for l=1:100
        randi=all_100{l};
        dimers = dimercount(randi);
        for k=1:16
            x=cell2mat(struct2cell(dimers));
            B{l}=x.';
            C=cell2mat(B);
            S(j)=mean(C(:,k));
        end
    end
end

Thank you!!!

[the cyclist]

Generic advice would be to use the profile command, to see which lines of code are taking the most time, and focusing on those.

Answer 1

[Jan]

It is hard to impossible to optimize code without having the complete code and the input. You have to guess too many details:

• Are B and S pre-allocated?
• Why do you overwrite S(j) 1600 times?
• What is the purpose of struct2cell->cell2mat->cell->cell2mat? This looks far to complicated. Why don't you work with numerical arrays directly?
• Why do you create the complete matrix C, if you need the k'th column only?