[Commits] commit/WVUThorns: zach_etienne: IllinoisGRMHD, ID_converter_ILGRMHD: Fix the double-added T_munu at t=0; also fix a conservative-to-primitive scheduling typo
Bitbucket
commits-noreply at bitbucket.org
Tue Jun 19 07:11:16 CDT 2018
1 new commit in WVUThorns:
https://bitbucket.org/zach_etienne/wvuthorns/commits/212ff43d0e89/
Changeset: 212ff43d0e89
Branch: ET_2018_02
User: zach_etienne
Date: 2018-06-19 11:54:02+00:00
Summary: IllinoisGRMHD,ID_converter_ILGRMHD: Fix the double-added T_munu at t=0; also fix a conservative-to-primitive scheduling typo
Two bugs in IllinoisGRMHD were reported by Elias Most, who also suggested fixes.
1) The HydroBase eT[mu][nu] variables were being double-added at t=0 due to the fact that
eT[mu][nu] were being set via
eT[mu][nu] += TDNMUNU
in IllinoisGRMHD's conservative-to-primitive solver. While this is a cute solution given that
the solver is often called within the *Add*toTmunu scheduling bin (emphasis mine), it is
inconsistent with how eT[mu][nu] is set within the outer_boundary routine (i.e.,
non-additively), and generally inconsistent with how IllinoisGRMHD interacts with other thorns.
(IllinoisGRMHD should set eT[mu][nu], and if other thorns want to add, they are welcome to do
so. Due to an additional call to IllinoisGRMHD's conservative-to-primitive solver in
CCTK_POSTPOSTINITIAL, TDNMUNU was being added *twice* at t=0, resulting in bad data being
added to the RHSs of Einstein's equations at t=0.
The IllinoisGRMHD/test results were updated (difference at the 4th or 5th significant digit)
due to this bug fix.
2) The outer boundary routine was scheduled "after conserv_to_prims", except no such function
or scheduling alias exists anymore. The correct scheduling is
"after IllinoisGRMHD_conserv_to_prims". The scheduler had already scheduled this correctly, so
fixing this bug has no effect on results. Regardless, the typo was fixed.
Affected #: 4 files
Repository URL: https://bitbucket.org/zach_etienne/wvuthorns/
--
This is a commit notification from bitbucket.org. You are receiving
this because you have the service enabled, addressing the recipient of
this email.
More information about the Commits
mailing list