[Octave-bug-tracker] [bug #52874] 3rd argument of lscov not compatible with MATLAB implementation of lscov

[Kenneth Hoste]

URL:
  <http://savannah.gnu.org/bugs/?52874>

                 Summary: 3rd argument of lscov not compatible with MATLAB
implementation of lscov
                 Project: GNU Octave
            Submitted by: kehoste
            Submitted on: Fri 12 Jan 2018 03:13:56 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: 4.2.1
        Operating System: GNU/Linux

    _______________________________________________________

Details:

When running StaMPS (http://homepages.see.leeds.ac.uk/~earahoo/stamps/) which
I'm porting Octave (a large part of it was written for MATLAB), I ran into the
following error when lscov is being called (from StaMPS's ps_weed.m):


error: lscov: V should be a square matrix or a vector with the same number of
rows as A


After a bit of digging around, I figured out that this seems to be an
incompatibility compared to MATLAB.

>From the MATLAB documentation @
https://nl.mathworks.com/help/matlab/ref/lscov.html for "x = lscov(A,B,w)":
"where w is a vector length m of real positive weights".

I guess the MATLAB documentation leaves some interpretation whether w should
be mx1 (which is the size that Octave asserts) or 1xm (which works just fine
with MATLAB).

I fixed this in the StaMPS code by calling lscov(A,B,w') rather than
lscov(A,B,w), but I think this could/should be handled in lscov itself.

This can probably be done in a backward-compatible way: if the 3rd argument (V
in the Octave implementation) is 1xm, lscov could convert that to mx1 and just
continue?




    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?52874>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/