Dear All, here are the minutes from today's meeting. Sorry if I missed someone or something.
Present: Bruno Giacomazzo, Steven Brandt, Roland Haas, Peter Diener, Beyhan, Artectek, Keith Dow, Zachariah Etienne, Gabriele Bozzola, Leo Rosa Werneck, Allen Wen, Yosef Zlochower
Chair: Leo Minutes: Bruno
*ET release recap Roland provided a recap. The delay was due to the addition of an Intel Compiler 19 workaround for GRHydro (it seems that there is a compiler bug that cannot recognize properly that arrays were initialized). In the future we also need a release co-chair because it is becoming too much work for just one person. It was a very large release with 3 new contributions and a large list of contributors with many students contributing to the testing.
Steve thinks we need to better recognize the work done by the release chair since it is a lot of work. Moreover there is always some missing point in the documentation regarding how to manage a new release. Zach agrees with what Steve said. We all thank Roland for his work for this release.
**missing tests in the new release Roland pointed out that Canuda does not have a testsuite. We did not notice it during the release.
*Next ET release chair search The search has begun. Any volunteers? Leo is thinking about it, but he is looking for a co-chair
*Open tickets: #2670 https://bitbucket.org/einsteintoolkit/tickets/issues/2670/refinement-prolongation-all-points-must (opened by Artectek): it seems to be a failure in the Carpet grid division algorithm happening after regridding, but Roland does not have an idea of what is causing this. Nobody has seen this error before. Roland pointed out that the sum of dens reported in INFO is ~1e-4 while the maximum of dens in ~1e-3, but this may be OK since the sum is the sum of the values weighted by the grid spacing (to get the integral one would need to multiply the sum by the coarsest grid spacing). Roland suggests having a look at the output for dens. Artectek says that he did have a look at the output using the script in the gallery example and the NSs are still there. Artectek's simulation is based on the BNS gallery example. Gabriele suggests starting from the gallery example and changing only one parameter at a time. Artectek said that he only changed the EOS and the grid setup for the initial separation of the NSs. Roland suggests to run with the same EOS of the gallery example (gamma law) and just change the grid setup. If this works then Gabriele suggests trying with a piecewise polytropic EOS. Roland suggests producing a checkpoint at the iteration before the error appears so that by looking at the checkpoint maybe we can understand what is happening. Gabriele also noted that the memory requirement for the run is 11 GB, which seems to be very low for a BNS. It could be that there are too few points maybe shared through too many MPI processes (and therefore each MPI may have too small grids). Gabriele and Roland therefore suggest running it on only one node as a first test. Leo also suggests checking that the NS is fully inside the finest grid (but this would probably give errors in C2P and not in regridding, since regridding is done also if the NS was fully contained).
#2671 https://bitbucket.org/einsteintoolkit/tickets/issues/2671/add-instuctions-for-what-information-to: Roland says that it would be useful to make it more clear to users which information to share when encountering problems with the Einstein Toolkit.
#2665 https://bitbucket.org/einsteintoolkit/tickets/issues/2665/two-open-pull-requests-in: it was opened by Gabriele. There seems to be a couple of typos in the documentation of the TOV Solver and there was a pull request to fix it.
*Call for new contribution proposals Fishbone-Moncrief gallery example Elliptica reader (and solver timeline) ticket FUKA reader announcement
Roland said that Elliptica and FUKA should be included in the next release and we need to ping their authors. The Fishbone-Moncrief gallery example was done recently and it should be included.
*simfactory website repo server migration Steve said it is going to be moved from svn to git.
*ET Jenkins replacement status update Roland said that we used to run tests on Jenkins, but there was the idea to move to git actions. Some students worked on this and it seems to be working fine. See einsteintoolkit.github.io/tests/ (it runs on github resources that we can access for free since the einstein toolkit has an academic license and therefore we do not need to manage virtual machines by ourselves as we do now). Github action also builds documentation every day.
*ET seminar and hackathon search Roland says that we should start these again. Any volunteers for seminars? What about a tutorial on NRPY Elliptic?
*NRPy+ plans Zach said that Steve and Zach's group would like to get together for future plans since Steve has a very nice fork of NRPY that improves users' interaction. Zach also has his own fork with some infrastructure improvements. Steve is also looking for a student to work on NRPY (see Steve's announcement on the einstein toolkit mailing list).
*EOS plans call recap postponed to next call since we ran out of time
*SPECv8 update postponed to next call since we ran out of time
*Unanswered questions postponed to next call since we ran out of time
*tickets ready for review: postponed to next call since we ran out of time
Next call (December 8): Chair: Peter Minutes: Steve
Cheers, Bruno