[Octave-bug-tracker] [bug #54293] freqz plot differs from MATLAB

[Roland Baudin]

URL:
  <http://savannah.gnu.org/bugs/?54293>

                 Summary: freqz plot differs from MATLAB
                 Project: GNU Octave
            Submitted by: roland65
            Submitted on: Thu 12 Jul 2018 01:32:32 PM UTC
                Category: Octave Function
                Severity: 3 - Normal
                Priority: 5 - Normal
              Item Group: Matlab Compatibility
                  Status: None
             Assigned to: None
         Originator Name: 
        Originator Email: 
             Open/Closed: Open
         Discussion Lock: Any
                 Release: 4.2.2
        Operating System: GNU/Linux

    _______________________________________________________

Details:

Plotting the frequency response of a filter gives different plots in MATLAB
and in Octave.

To test:

n = [0 : 127];
h = sinc(0.2*(n-128/2)); % Simple test filter
freqz(h, 1, 128, 'whole', 1); % Frequency response on the whole Nyquist

Plots in MATLAB (tested in versions 2014a and 2017b) and in Octave differs
because in this case Octave doubles the number of points used in the FFT,
while MATLAB does not. Thus MATLAB uses a 128 points FFT (as requested) while
Octave uses a 256 points FFT.

The same happens if we plot the frequency response on half the Nyquist band,
like this:
freqz(h, 1, 128, 'half', 1);

I found that commenting out the following lines in freqz.m gives the expected
result:

lines 154 to 157

    %~ if (k > n/2 && nargout == 0)
      %~ ## Ensure a causal phase response.
      %~ n *= 2 .^ ceil (log2 (2*k/n));
    %~ endif

lines 168 to 170

      %~ if (plot_output)
        %~ n += 1;
      %~ endif





    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?54293>

_______________________________________________
  Message sent via Savannah
  https://savannah.gnu.org/