How can vectorize the below codes?

4 views (last 30 days)
Nx=512;
Ny=Nx
x0 = Nx/2;
y0 = Ny/2;
isolve = 2;
tic
for i=1:512
for j=1:512
ii=(i-1)*Nx+j;
if(isolve == 2)
etas(ii,1)=1.0;
phis(ii,1) =0;
phis(ii,2)=0;
etas(ii,2)=0.0;
else
etas(i,j,1) =1.0;
etas(i,j,2) =0.0;
phis(ii,1) =0;
phis(ii,2)=0;
end
xlength =sqrt((i-x0)^2+(j-y0)^2);
if(xlength <= 14)
if(isolve == 2)
etas(ii,1)=0.0;
etas(ii,2)=1.0;
phis(ii,1) =0;
phis(ii,2)=0;
else
phis(ii,1) =0;
phis(ii,2)=0;
end
end %if
end %j
end %i

Accepted Answer

the cyclist
the cyclist on 1 Dec 2019
Well, not vectorizing it, but if you preallocate etas and phis with these line:
etas = zeros(262144,2);
phis = zeros(262144,2);
you will speed up the code by a huge factor. (For me, it was a factor of over 1000.)
  5 Comments
the cyclist
the cyclist on 2 Dec 2019
The following is a vectorized version. It is not that much faster, and actually more difficult to understand what is going on.
Nx = 512;
Ny = Nx;
x0 = Nx/2;
y0 = Ny/2;
isolve = 2;
NxNy = Nx*Ny;
etas = [ones(NxNy,1) zeros(NxNy,1)];
phis = zeros(NxNy,2);
tic
xvec = 1:Nx;
yvec = 1:Ny;
xlength = sqrt((xvec-x0).^2+(yvec'-y0).^2);
idx = isolve==2 & xlength(:) <= 14;
etas(idx,:) = repmat([0 1],sum(idx),1);
toc
Amir Torabi
Amir Torabi on 2 Dec 2019
very grateful. It really works great!.

Sign in to comment.

More Answers (0)

Categories

Find more on Loops and Conditional Statements in Help Center and File Exchange

Community Treasure Hunt

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

Start Hunting!