[Octave-bug-tracker] [bug #62186] [Octave Forge](Control) 'bode' produces plots with inappropriate frequency ranges

[Torsten Lilge]

Update of bug #62186 (project octave):

        Operating System:       Microsoft Windows => Any                    
                 Summary: Function 'bode' produces possibly correct but
mis-presented result => [Octave Forge](Control) 'bode' produces plots with
inappropriate frequency ranges

    _______________________________________________________

Follow-up Comment #1:

Thank you for the report. I think the cause for this behavior is that, due to
numerical issues, the zeros of system(3,1) are not located at the origin of
the complex plane but at

-7.8505e-16 + 5.6037e-08i
-7.8505e-16 - 5.6037e-08i
 
which causes bode() to compute a strange frequency range in order to include
these zeros.

Please find below a patch for __frequency_vector__.m, which can be edited by
"edit __frequency_vector__":


@@ -138,8 +138,8 @@
     endif
   else
     ## continuous
-    iip = find (abs(pol) > norm(pol)*eps);
-    iiz = find (abs(zer) > norm(zer)*eps);
+    iip = find ((abs(pol) > eps) && (abs(pol) > norm(pol)*eps));
+    iiz = find ((abs(zer) > eps) && (abs(zer) > norm(zer)*eps));
 
     if (! isempty (zer))
       czer = zer(iiz);


With this patch applied, zeros and poles are included in the frequency range
if they are relevant relative to the norm of the vector of all zeros and poles
resp. *and*, which is new,  if the zeros and poles are absolutely relevant,
too. I am not yet totally sure if there might be unwanted side effects in
other cases.

cc-ing Doug since he is currently working on a bug related to the bode
function, too.


    _______________________________________________________

Reply to this item at:

  <https://savannah.gnu.org/bugs/?62186>

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