Logic indexing optimization of computational time

2 次查看（过去 30 天）

Rafael Félix Soriano 2023-8-29

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/2014386-logic-indexing-optimization-of-computational-time

评论： Torsten 2023-8-29

I am running a computationally expensive code, and according to MATLAB's profiler, this is the line of code that is taking the most time to run:

b = a(idx,:);

where a is 30000x3 matrix, and idx is a logical 30000x1 array, computed as follows:

idx = a(:,3) >= M + N;

where M and N are integers.

That line of code is run inside a for loop with many other lines with functions such as find. I need to understand why that line of code is taking longer to run than all other lines, and how I could optimize it. Sorry for the low amount of information on the code, for property reasons I cannot divulge much more.

Thank you in advance!

5 个评论
显示 3更早的评论隐藏 3更早的评论

Rafael Félix Soriano 2023-8-29

Thank you for your answer Dyuman. My point is that having several lines of code running find, none of them are taking long. It is the one that I posted (b = a(idx,:)) that is taking significantly longer than all others. That is the only part of the code giving me problems.

I am sorry but I am not allowed to divulge any more info.

Torsten 2023-8-29

Often, it is the interaction of many factors that makes a code slow. So concentrating on a single command is not the correct approach in my opinion.

That having said, b = a(idx,:) is the optimal way to extract the rows of the matrix "a" where the element of the 3rd column is >= M+N. Nothing to optimize or critisize here.

请先登录，再进行评论。

请先登录，再回答此问题。