[bug #68605] Signed 0 affects sort result in sparse complex arrays

Please use the bug tracker to post updates to a bug report.  The mailing list is intended as a read-only notification stream.  Info posted to this mailing list address won't appear in the tracker database where it is most useful.

URL:
  <https://savannah.gnu.org/bugs/?68605>

                 Summary: Signed 0 affects sort result in sparse complex
arrays
                   Group: GNU Octave
               Submitter: dasergatskov
               Submitted: Mon 10 Aug 2026 01:18:21 PM UTC
                Category: Octave Function
                Severity: 3 - Normal
                Priority: 5 - Normal
              Item Group: Incorrect Result
                  Status: None
             Assigned to: None
         Originator Name:
        Originator Email:
             Open/Closed: Open
         Discussion Lock: Unlocked
                 Release: dev
        Operating System: Any
           Fixed Release: None
         Planned Release: None


    _______________________________________________________

Follow-up Comments:


-------------------------------------------------------
Date: Mon 10 Aug 2026 01:18:21 PM UTC By: Dmitri A. Sergatskov <dasergatskov>

octave:1> z = [complex(0, 2), complex(-2, -0.0)]
z =

   0 + 2i  -2 - 0i

octave:2> [y, idx] = sort(z) 
y =

   0 + 2i  -2 - 0i

idx =

   1   2

octave:3> [y_s, idx_s] = sort(sparse(z)) 
y_s =

Compressed Column Sparse (rows = 1, cols = 2, nnz = 2 [100%])

  (1, 1) -> -2 - 0i
  (1, 2) ->  0 + 2i

idx_s =

   2   1

octave:4> z1 = [complex(0, 2), complex(-2, 0.0)]
z1 =

   0 + 2i  -2 + 0i

octave:5> [y, idx] = sort(z1) 
y =

   0 + 2i  -2 + 0i

idx =

   1   2

octave:6> [y_s, idx_s] = sort(sparse(z1)) 
y_s =

Compressed Column Sparse (rows = 1, cols = 2, nnz = 2 [100%])

  (1, 1) ->  0 + 2i
  (1, 2) -> -2 + 0i

idx_s =

   1   2


In `liboctave/util/oct-cmplx.cc`:

32 // For complex-complex and complex-real comparisons, Octave uses the
following
 33 // ordering: compare absolute values first; if they match, compare phase
 34 // angles.  This is partially inconsistent with M*b, which compares
complex
 35 // numbers only by their real parts; OTOH, it uses the same definition
for
 36 // max/min and sort.  The abs/arg comparison is definitely more useful
(the
 37 // other one is emulated rather trivially), so let's be consistent and use
that
 38 // all over.
 39 
 40 // The standard C library function arg() returns [-pi,pi], which creates
a
 41 // non-unique representation for numbers along the negative real axis
branch
 42 // cut.  Change this to principal value (-pi,pi] by mapping -pi to pi.


The same should be done for sparse. 
I will provide a patch shortly.

Dmitri.
-- 









    _______________________________________________________

Reply to this item at:

  <https://savannah.gnu.org/bugs/?68605>

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

-----BEGIN PGP SIGNATURE-----

iHUEABYIAB0WIQQk97aszIMMAvLLwm6qLAuaBUf3TgUCannPogAKCRCqLAuaBUf3
TuVsAQCgE24SIY+vOwJR31icMK8MHs034XKcUhc/AwvKQ24QHgD+ISzl+G3kBwmc
yByoxqEbnaWAvg6LADMrICgjTmbBPwU=
=QX86
-----END PGP SIGNATURE-----