[Carpetx-developers] Weekly call at 10:00AM Eastern
Roland Haas
rhaas at illinois.edu
Fri Feb 16 13:07:59 CST 2024
present: Roland, Erik, Alex, Helvi, Steve, Allen, Cheng-Hsin, Jay,
Liwei, Lorenzo, Lucas, Sam, Yosef
* discussed what Canuda codes to propose for inclusion
** NewRadX (Alex, Liwei and Jay will review)
** Multipole(X) (Lucas, Roland is reviewer)
** will defer LeanBSSNMoLX, NPScalarsX to next release
* Erik proposes Z4c for inclusion, will champion it. Steve reviewer.
* Allen may propose inclusion of PunctureTracker and
ApparentHorizonFinderDirect. Will leave out QuasiLocalMeasures for
now since it requires hacks due to Fortran and complex valued grid
variables.
** Roland mentioned that the driver specific changes to Multipole
should be isolated to a small section of the the code that constructs
the test data. Would be good to have single code that supports both
drivers (easier to maintain)
** Jay has been using NewRadX with Z4c on Frontier, this worked on GPUs
** Alex suggested that someone at RIT who used NewRadX may be a good
person to review,
** Roland mentioned that tests for release will start May 3rd so all
reviews must have (positively) finished by then
** Jay will focus on finish the AsterX paper and AsterX will be
proposed for inclusion afterwards so that documentation can be
written
CarpetX testing and development:
* Erik is currently testing a qc0 setup and would like to use a common
setup for the different codes to compare them more directly.
Cheng-Hsin will upload the parameter file for grazing collisions that
he used with LeanBSSNMoLX to the CarpetX wiki.
** Jay asked about a speed benchmark of Z4c vs. other codes, right now
this does not exist. Newer versions of CarpetX make this easier
though by outputting performance data.
* Steve and Sam have been looking at CarpetX + NRPy+
** Sam ran into hard-coded limit on number of grid functions per grid
function group, will update limit in CarpetX source code
* Lucas and Erik are working on Multipatch code, currently doing some
debugging.
* Lucas made CarpetX documentation TeX files compatible with online
documentation system, so they now show up on einsteintoolkit.org
AsterX:
* Jay is making steady progress on the paper
* working with Samuel Tootle on FUKA reader
* want paper out ASAP
CanudaX:
* Cheng-Hsin has been debugging NPScalarX and in doing so found some
race condition in the older NPScalar code
* currently still sees some differences but only in the boundaries so
could be due to different boundary location in Carpet / CarpetX
Runge Kutta prolongation:
* discussed Liwei's current status of code
* required changes can all do done in "user" code and do not require
changes to CarpetX itself
* suggestion is to implement a standalone RK4 time stepper code for
this and not use ODESolvers for now
* will need full grid functions for Y1, Y2, Y3, Y4 just for prolongation
--
My email is as private as my paper mail. I therefore support encrypting
and signing email messages. Get my PGP key from http://pgp.mit.edu .
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 833 bytes
Desc: OpenPGP digital signature
URL: <http://lists.einsteintoolkit.org/pipermail/carpetx-developers/attachments/20240216/3cf1f7f7/attachment.sig>
More information about the Carpetx-developers
mailing list