[Octave-bug-tracker] [bug #36364] DERF.F: Inaccurate results for very small values of x

[anonymous]

URL:
  <http://savannah.gnu.org/bugs/?36364>

                 Summary: DERF.F: Inaccurate results for very small values of
x
                 Project: GNU Octave
            Submitted by: None
            Submitted on: Tue 01 May 2012 23:32:30 UTC
                Category: Libraries
                Severity: 3 - Normal
                Priority: 5 - Normal
              Item Group: Inaccurate Result
                  Status: None
             Assigned to: None
         Originator Name: jofriet
        Originator Email: 
             Open/Closed: Open
         Discussion Lock: Any
                 Release: 3.6.1
        Operating System: Any

    _______________________________________________________

Details:

In the file DERF.F, part of libcruft, the line: 

IF (Y.LE.SQEPS) DERF = 2.0D0*X*X/SQRTPI

should be:

IF (Y.LE.SQEPS) DERF = 2.0D0*X/SQRTPI

Note that this should be immediately obvious from the definition of the error
function (http://en.wikipedia.org/wiki/Error_function) and from the wrong sign
for small negative values of x. The bug is present in the SLATEC version of
FNLIB, but not in FN: see http://www.netlib.org/slatec/fnlib/derf.f and
http://www.netlib.org/fn/derf.f.

The function erf in octave itself seems to function fine though.




    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?36364>

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