How to vectorize a "For Loop"

Question

Ahmad Jamialahmadi 2013-10-30

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/104348-how-to-vectorize-a-for-loop

评论： Ahmad Jamialahmadi 2013-10-30

I have function "" f(t,z)=cos(t).*z "" which for every value of t must be integrated wrt z and output the result as a matrix. I use the following code which takes approximately 15 sec for my real long formulas while the function f which has to be integrated executes almost instantly for any value of its variables.

f=@(t,z) cos(t) .* z
t= linspace(0,3,100);
z= linspace(0,10,1000);
out=zeros(1,numel(t));
for i = 1:numel(t)
     Y=f(t(i),z);
     out(i)=trapz(z,Y);
end

Any help on how to vectorize this would be greatly appreciated :-)

2 个评论
显示无隐藏无

David Sanchez 2013-10-30

what is zz?

Ahmad Jamialahmadi 2013-10-30

oh ! sorry , typo !!! it's z

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Andrei Bobrov 2013-10-30

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/104348-how-to-vectorize-a-for-loop#answer_113673

在 MATLAB Online 中打开

f=@(t,z) cos(t) .* z;
t= linspace(0,3,100);
z= linspace(0,10,1000);
out = trapz(z(:),bsxfun(@(t,z)f(t,z),t(:)',z(:)));

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Ahmad Jamialahmadi 2013-10-30

在 MATLAB Online 中打开

Thank you Andrei for your response but unfortunately this vectorization method doesn't seem to optimize the loop !

Using for loop :

Elapsed time is 15.282242 seconds.

Vectorization using bsxfun :

Elapsed time is 15.361217 seconds.

However the above answer was what I was looking for :-)

Although switching from using "Integral" to "trapz" in order to optimize the integration saved me almost 60 sec, but yet it is desirable time for my calculations. Any other idea on how to approach this kinda questions ?

请先登录，再进行评论。