better method for evaluating matrix

For given vector such as following, I want to make square matrix s such that

%constant, original form
N = 500;
x = 2*pi*linspace(0,1,N);
for i = 1:N
 for j = 1:N
  s(i,j) = sin(x(i)-x(j))
 end
end

But it was too slow, so recently I edited it to following, but it's still too slow!!

Can anybody please help me??

 %constant
 N = 500;
 x = 2*pi*linspace(0,1,N);

 for i = 1: N;
   s(:,i) = x-x(i);
 end

 s=sin(s);

Answer 1

[Stephen23]

You really need to learn how to write vectorized code. Solving every task using lots of ugly loops is not an efficient way to write MATLAB code. Try this:

N = 500;
vec = 2*pi*linspace(0,1,N);
mat = sin(bsxfun(@minus,vec(:),vec))

Answer 2

[Andrei Bobrov]

N = 500;
x = 2*pi*linspace(0,1,N);

s = sin(x(:)' - x(:));

for old versions of MATLAB:

s = sin(  bsxfun(@minus, x(:)',x(:)) );

[Stephen23]

Note that the output is transposed compared to the code given in the question.