Variable appears to change size on every loop iteration

Hi to everybody i have a problem running these codes .its a linear segment price in power generation economic dispatch matlab code

in line 17 AND SOME OTHER LINES THIS ERROR COMES: Variable F appears to change size on every loop iteration I DONT KNOW HOW TO SOLVE THAT!

[Walter Roberson]

Note that this is a warning, not an error. It hints that your code is less efficient than it could be. For large arrays the cost of growing the array can end up being a major slowdown.

Answer 1

[José-Luis]

The solution is to pre-allocate:

Do

results = ones(1,10)
for ii = 1:10
    results(ii) = ii;
end

instead of:

for ii = 1:10
   begin_for_error(ii) = ii;
end

I didn't go through your code so you'll have to modify to fit your needs.

[Jan]

And replace

q=1;
for i=1:N % converts the column matrix "x" to a rectangular shape
          % for easy row addition on next loop
  for j=1:N
      X(i,j)=x(q);
      q=q+1;
  end
end

by:

X = reshape(x, N, N);

Learn to use vectorized expressions. The fact that they are faster might not matter in your case, but they are cleaner and leaner: Less chances for typos and easier to write and maintain.

[Sultan Al-Hammadi]

What if you don't want to allocate specific values to the enteries of the array? I got the same one while I was trying to plot an animation. Is there a way of optimising it as the exceution of my code is noticably slow.

[Rik]

Since the goal is to overwrite them anyway, does it matter what value it starts out with?

You could consider pre-allocating with NaN, as many functions will ignore those.

If you want more specific advice, you should post this as a separate question.

[Walter Roberson]

https://www.mathworks.com/matlabcentral/fileexchange/8334-incremental-growth-of-an-array-revisited

[Kelvin Ling Chia Hiik]

thanks Jose Luis

Answer 2

[Sohan]

function waterfallspect(s, fs, spectsize, spectnum)
%WATERFALLSPECT Display spectrum of signal s as a 3-D plot.
%   s - input signal, use wavload to load a WAV file
%   fs - Sampling frequency (Hz).
%   spectsize - FFT window size
%   spectnum - number of windows to analyze

frequencies = [0:fs/spectsize:fs/2];
offset = floor((length(s)-spectsize)/spectnum);
for i=0:(spectnum-1)
    start = i*offset;
    A = abs(fft(s((1+start):(start+spectsize))));
    magnitude(:,(i+1)) = A(1:spectnum/2+1);
end
waterfall(frequencies, 0:(spectnum-1), magnitude');
xlabel('frequency');
ylabel('time');
zlabel('magnitude');

[Walter Roberson]

No, this would have the same warning as the original code, about a variable changing size in every iteration, so this code would not fix the problem that the original poster had.

Is there a reason you did not pre-allocate the matrix magnitude ?