How to preallocate to decrease run time?

Question

Bixius 2018-7-18

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/410922-how-to-preallocate-to-decrease-run-time

回答： Greg 2018-7-19

Re_1, Re_6, etc.. are have different dimensions i.e. Re_1 = 5000:20000, Re_6 = 200:10000, etc. I wrote this for loop to get the mean values at each value for Reynolds number. How can I preallocate this?

tot_Re = [Re_1 Re_6 Re_8 Re_9 Re_12 Re_13];
tot_Nu = [Nu_1 Nu_6 Nu_8 Nu_9 Nu_12 Nu_13];
Re_mean = 1:max(tot_Re); 
for i = 1:max(tot_Re)
      k = find(tot_Re == i);
      Nu_mean(i) = mean(tot_Nu(k));
end

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Greg 2018-7-19

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/410922-how-to-preallocate-to-decrease-run-time#answer_329394

在 MATLAB Online 中打开

You can use unique to make your loop more robust. Then, logical indexing is much better than the find you've included. In fact, the find is completely unnecessary.

tot_Re = [Re_1 Re_6 Re_8 Re_9 Re_12 Re_13];
tot_Nu = [Nu_1 Nu_6 Nu_8 Nu_9 Nu_12 Nu_13];
Re_mean = 1:max(tot_Re); % You never use this...
[~,ind,~] = unique(tot_Re); % Using unique allows for gaps in values of tot_Re
Nu_mean = NaN(size(ind)); % Pre-allocate Nu_mean
for lp = 1:length(ind)
    k = tot_Re == tot_Re(ind(lp)); % Leave indices as logicals
    Nu_mean(lp) = mean(tot_Nu(k));
end