Moving average - for loop vectorization

Question

Dawid Smolen 2016-4-21

1
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/280185-moving-average-for-loop-vectorization

评论： Stephen23 2018-12-5

Hello. I've got a problem that my loop doesn't execute fast enough, even though it is simple moving average. Is there any smart way to make it work faster? x is a huge array, and there is some indexing using : in every iteration, that's probably the reason.

len = length(x); % around 6000000
A = 1/200;
for n = 100+1:len-100
    MA = A*sum(x(n - 100: n + 100));
    % Do something with MA. However I know that the above part is the slow one
end

I was trying to find some information about vectorization, but I can't see how could I apply these methods.

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

Stephen23 2016-4-21

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/280185-moving-average-for-loop-vectorization#answer_218811

编辑：Stephen23 2016-4-21

在 MATLAB Online 中打开

To efficiently calculate a moving average you should use conv, something like this:

>> xi = 0:0.1:2*pi;
>> yi = sin(xi)+0.4*rand(size(xi))-0.2;
>> N = 4; % length of moving average
>> yo = conv(yi,ones(1,N),'same')/N;
>> plot(xi,yi, xi,yo)