How to optimize this program using a for loop? i want to repeat the part between percentage symbols n-times, not only 4, i use the E1 to build the E2, and then, E2 to build the E3, and so on, Ifrog is just a matrix of 4 columns and 2000 rows.

load Ifrog

dimension = 2000;

x = linspace(-5,5,dimension);

L = length(x);

n = 2^nextpow2(L);

FG1 = exp((-((x-2.5)./(1)).^2));

FG2 = exp((-((-x-2.5)./(1)).^2));

TFG1 = fftshift(FG1,n);

TFG2 = fftshift(FG2,n);

x1=linspace(0,.8,dimension);

w=2*pi./x1;

Fexpo = (1+exp(-2*w));

Convol = conv(TFG1,TFG2,'same');

kernel = 2*(Convol.*Fexpo);

a = led01(1:dimension);

b = led02(1:dimension);

c = led03(1:dimension);

d = led04(1:dimension);

mod1 = [a, a, a, a, a];

mod2 = [b, b, b, b, b];

mod3 = [c, c, c, c, c];

mod4 = [d, d, d, d, d];

traza = [mod1, mod2, mod3, mod4];

%%%%%%%%%%%

%%%%%%%%%%%

%%%%%%%%%%%

for m=1:20;

      if m==3|8|13|18

          E1(:,m) = ifft(fft(kernel')/norm(fft(kernel)).*traza(:,m));

      else

          E1(:,m) = traza(:,m);

      end

end

for m=1:20;

      if m==3|8|13|18

          E2(:,m) = ifft(fft(E1(:,m))/norm(fft(E1(:,m))).*traza(:,m));

      else

          E2(:,m) = ifft(fft(E1(:,m))/norm(fft(E1(:,m))));

      end

end

for m=1:20;

      if m==3|8|13|18

          E3(:,m) = ifft(fft(E2(:,m))/norm(fft(E2(:,m))).*traza(:,m));

      else

          E3(:,m) = ifft(fft(E2(:,m))/norm(fft(E2(:,m))));

      end

end

for m=1:20;

      if m==3|8|13|18

          E4(:,m) = ifft(fft(E3(:,m))/norm(fft(E3(:,m))).*traza(:,m));

      else

          E4(:,m) = ifft(fft(E3(:,m))/norm(fft(E3(:,m))));

      end

end

%%%%%%%%%%%

%%%%%%%%%%%

%%%%%%%%%%%

sumren = sum(abs(E4),1);

sumcol = sum(abs(E4),2);

figure()

subplot(2,2,1)

surf(abs(E4))

subplot(2,2,2)

contour(abs(E4))

subplot(2,2,3)

plot(abs(sumren))

subplot(2,2,4)

plot(abs(sumcol))

Answer 1

[Walter Roberson]

Use cell arrays.

E = cell{n, 1};

for m=1:20;
    if ismember(m, [3, 8, 13, 18])
        E{1}(:,m) = ifft(fft(kernel')/norm(fft(kernel)).*traza(:,m));
    else
        E{1}(:,m) = traza(:,m);
    end
end

for idx = 2 : n
  for m=1:20;
    if ismember(m, [3, 8, 13, 18])
        E{idx}(:,m) = ifft(fft(E{idx-1}(:,m))/norm(fft(E{idx-1}(:,m))).*traza(:,m));
    else
        E{idx}(:,m) = ifft(fft(E{idx-1}(:,m))/norm(fft(E{idx-1}(:,m))));
    end
 end

Please take note of the correction for your test on m

[David Zarate Villegas]

thanks for the answer, but when i tried to implement this, i got things that i don't spect in the program.

[Stephen23]

@David Zarate Villegas: and these things are so secret that you are not going to tell us so that we can help you to fix them.

Walter Roberson gave you a very good solution: using cell arrays is a very neat and robust way to solve your problem. We are also happy to help you fix any bugs making it work properly. However when you write "i got things" this does not really tell us anything useful: do you get error messages, warnings, or unexpected values ? How are you checking them, and what code are you using?

When you actually tell us information so that we can replicate your results and you tell us what you want the code to be be doing then we can help you much more easily. At the moment you are expecting us rely on our magical glass balls and predict what is shown on your computer screen...