[Scipy-tickets] [SciPy] #648: Additional functions for scipy.signal

SciPy Trac scipy-tickets@scipy....
Wed Feb 20 22:30:16 CST 2013


#648: Additional functions for scipy.signal
-----------------------------+----------------------------------------------
  Reporter:  bow_der_kleine  |       Owner:  cdavid     
      Type:  enhancement     |      Status:  reopened   
  Priority:  normal          |   Milestone:  Unscheduled
 Component:  scipy.signal    |     Version:             
Resolution:                  |    Keywords:             
-----------------------------+----------------------------------------------

Comment(by endolith):

 [https://gist.github.com/endolith/4525003 I've been looking at
 implementing some of these], but I'm not sure what's the best way to do
 things or where to put things:

 cplxpair and cplxreal are more general than signals.  What package should
 they be in?  Maybe in Numpy?

 The second-order sections stuff doesn't help much even when implemented,
 since the filter design tools internally convert to tf format and back
 again, causing numerical problems before it even gets to the SOS stage.
 So the filter design needs to be rewritten to use zpk or ss format.
 Matlab uses ss format, but zpk is probably simpler and the prototypes are
 created in zpk.

 But then what?  Should lp2lp() and bilinear() be rewritten to take only
 zpk instead of tf?  Should they be rewritten to take both formats like
 Matlab/Octave?  (Though I don't know how *args would be backwards-
 compatible)  Should new functions like zpklp2lp() and zpkbilinear() be
 created instead?  These would work on analog filters, though, and Matlab
 has a zpklp2lp that works on digital filters.  zpklp2lp could have an
 option for analog or digital?

-- 
Ticket URL: <http://projects.scipy.org/scipy/ticket/648#comment:9>
SciPy <http://www.scipy.org>
SciPy is open-source software for mathematics, science, and engineering.


More information about the Scipy-tickets mailing list