https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121185
Bug ID: 121185
Summary: [16.0 regression] Numerics of Monte Carlo integrator
changed
Product: gcc
Version: 16.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: fortran
Assignee: unassigned at gcc dot gnu.org
Reporter: juergen.reuter at desy dot de
Target Milestone: ---
Hi,
let me open up an issue already. I believe there was a regression/change
introduced in gfortran between July 12 and July 20, as our CI with a git-clone
built gcc/gfortran worked last week, and fails this week. One out of our ca.
145 unit tests fail because they return different numerical results. The test
sets up an implementation of the stream RNG by L'Ecuyer and does a VEGAS-like
numerical Monte Carlo adaptive integration of a test function using these
random numbers. The numerical output of this test (which should be
deterministic) deviates:
63c63
< Result: 1.99985E+00 +/- 2.33751E-03
---
> Result: 1.99967E+00 +/- 2.64783E-04
73c73
< Result: 2.00068E+00 +/- 7.71566E-03
---
> Result: 1.99971E+00 +/- 5.68290E-04
I will try to boil this down to a smaller reproducer (fingers crossed), but if
you want to play around already, checkout our code from here:
https://gitlab.tp.nt.uni-siegen.de/whizard/public
Note that you need noweb and OCaml besides gcc/gfortran. Do in the main
directory ./build_master.sh and autoreconf, then in a build directory _build
do ../configure, make -j4, make -C tests/unit_tests -j4 check.
The failing test is vamp2.run .
Cheers,
Juergen
