octave-bug-tracker
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Octave-bug-tracker] [bug #51966] psi function wrong when built with 64-


From: Mike Miller
Subject: [Octave-bug-tracker] [bug #51966] psi function wrong when built with 64-bit Fortran indexing
Date: Fri, 8 Sep 2017 17:45:16 -0400 (EDT)
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:55.0) Gecko/20100101 Firefox/55.0

URL:
  <http://savannah.gnu.org/bugs/?51966>

                 Summary: psi function wrong when built with 64-bit Fortran
indexing
                 Project: GNU Octave
            Submitted by: mtmiller
            Submitted on: Fri 08 Sep 2017 02:45:15 PM PDT
                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: Any
                 Release: dev
        Operating System: GNU/Linux

    _______________________________________________________

Details:

When Octave is built with support for 64-bit indexed Fortran arrays, the psi
function returns a lot of wrong results on real inputs and fails its tests.


>> psi (1, 0.5)
ans = -Inf


The problem occurs in the Fortran routines DPSIFN and PSIFN (the initial value
of ierr doesn't matter):


Breakpoint 1, dpsifn (x=0.5, n=1, kode=1, m=1, ans=..., nz=0,
ierr=140289603415448)
(gdb) finish
3366          F77_XFCN (dpsifn, DPSIFN, (z, n, 1, 1, ans, flag, t_ierr));
(gdb) p t_ierr
$3 = 2


IERR = 2 indicates an overflow occurred in the routine. So even though the
only thing different is the size of the INTEGER variable type, something in
these routines is calculating an overflow differently than when built with
32-bit integers.




    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?51966>

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




reply via email to

[Prev in Thread] Current Thread [Next in Thread]