Vectorization of nested for loop

Hello !

I am trying to vectorize this nested for loop but I cannot think of any solution.

L=zeros(4,200,3);

for i=1:3

for p=4:200

for q=1:4

L(q,p,i)=G(p-q+1,i);

end

end

end

What I tried (but doesnt work) :

L = zeros(4,200,3);

[p,q] = ndgrid([4:200],[1:4]);

p = p(:);

q = q(:);

L(q,p,:)=G(p-q+1,:);

Can someone please help me with what is wrong with my logic ?

Answer 1

[Voss]

% random matrix G:
G = rand(200,3);
[r,m] = size(G);

% parameter:
n = 4;

% for loop method (generalized):
L=zeros(n,r,m);
for i=1:m
    for p=n:r
        for q=1:n
            L(q,p,i)=G(p-q+1,i);
        end
    end
end

% vectorized method:
p = n:r;
q = 1:n;
idx = p(:).'-q(:)+1 + reshape(0:m-1,[1 1 m])*r;
L_vectorized = [zeros(n,n-1,m) G(idx)];

% check that the two methods give the same result:
isequal(L,L_vectorized)

ans = logical
   1