MATLAB: For loop Optimization

for loop

Hi everybody, i want to optmize(make faster loop) this function:
for k=1:2001
integrando=exp(i*2*pi*var_spaziale(k)*var_spettrale).*funzionefor k=1:N
   integrando=exp(i*2*pi*var_spaziale(k)*var_spettrale).*funzione;
   f(k)=du*sum(integrando);
end
How is it possible to do please?
Many thanks.

Best Answer

The exp() function is expensive. So start with avoiding repeated calls with the same argument:
C = exp((2i * pi * t) .* f);    % Auto-expanding, >= R2016b
for k = 1:NPF
  v  = datif2(:,k);
  g  = zeros(2001,1);
  du = f(2)-f(1);
  for k1 = 1:N
     integrando = C(:, k1) .* v;
     g(k1)      = du * sum(integrando);
  end
  datit(:, k) = 2 * real(g);
end
Is datit pre-allocated before the loop? If not, add this.
For Matlab < 2016b:
C = exp(2i * pi * bsxfun(@times, t, f));
Try if this is faster:
C = exp((2i * pi) * t.' .* f.');    % Auto-expaning, >= R2016b
for k = 1:NPF
  du = f(2)-f(1);
  g  = du * C * datif2(:,k);
  datit(:, k) = 2 * real(g);
end
Unfortunately I cannot try this by my own, because your posted code misses the function "interplin". What is it?
Related Question