[GiNaC-devel] Bug fix polylogs

Stefan Weinzierl weinzierl at uni-mainz.de
Fri Jul 24 17:44:09 CEST 2015


Dear all,

this is a bug fix for the bug reported in
  http://www.cebix.net/pipermail/ginac-list/2015-July/002053.html

The bug occured because a call to ex_to<numeric>(y) inside the 
polylog routines would fail for
y=sqrt(2), resulting in a sub-sequent segmentation fault.
In GiNaC sqrt(2) is treated as an exact number, not as a 
floating number.
With the fix, the routine now returns unevaluated (as it should).
An explicit call to evalf() forces a conversion to floats.
In order to enable this, the option do_not_evalf_params() has been removed 
for the G-functions. This is safe for the G-functions, the purpose of the
do_not_evalf_params-option is to avoid a conversion of genuine integer 
arguments (as in Li, S, H, zeta1 or zeta2) to floats.

  Best wishes,

    Stefan
-------------- next part --------------
A non-text attachment was scrubbed...
Name: polylog8.patch
Type: text/x-diff
Size: 2633 bytes
Desc: 
URL: <http://www.cebix.net/pipermail/ginac-devel/attachments/20150724/73b511cc/attachment.bin>


More information about the GiNaC-devel mailing list