vectorise for loop with element wise multiplication

3 次查看(过去 30 天)
I have this code that I am trying to vectorize with for loop
the function is very complicated so a simplified version is
f=@(a,b) a.exp(-b);
a and b of the same size
(a=[1;2;3;4;5]; b=[6;7;8;0;0]
)
so the for loop that I am trying to vectorize is this:
z=zeros(5,5)
for i=1:5
z(: , i)=a(i)*exp(-b);
end
I am interested in having this matrix:
z=[a(1)*exp(-b) , a(2)*exp(-b) ,a(3)*exp(-b), a(4)*exp(-b) , a(5)*exp(-b) ];
The aim is to vectorise this code so faster computation is achieved
I tried using bsxfun;

采纳的回答

Ameer Hamza
Ameer Hamza 2020-9-21
编辑:Ameer Hamza 2020-9-21
Loops are not always necessarily slow. However, the vectorized code is much more cleaner and readable. For the simple example, you can use automatic array expansion introduced in some recent release of MATLAB. The following two code are equivalent.
a=[1;2;3;4;5]; b=[6;7;8;0;0];
z=zeros(5,5);
for i=1:5
z(: , i)=a(i)*exp(-b);
end
and
a=[1;2;3;4;5]; b=[6;7;8;0;0];
z = a.'.*exp(-b); % note a is used as a row vector and b is used as a column vector
  4 个评论
Ameer Hamza
Ameer Hamza 2020-9-22
Yes, on older versions, bxsfun will be used.
Muna, following is equivalent to your code
z = (a.'.*c.').*exp(-b)

请先登录,再进行评论。

更多回答(0 个)

类别

Help CenterFile Exchange 中查找有关 Matrix Indexing 的更多信息

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by