Present: Bill G, Helvi W, Steve B, Miguel G, Yosef Z, Roland H, Leonardo RW, Atul K, Peter D, Zach E, Maria H, Gabriele B
Chair: Helvi W, Minutes: Bill G
* SPEC benchmark contribution [https://bitbucket.org/einsteintoolkit/tickets/issues/2470 task] [RH, SB]
Roland, no real updates, do need to send an extended description and ask if they need more details. Steve would like to know why they asked for this detailed description.
https://www.spec.org/cpu2017/Docs/benchmarks/507.cactuBSSN_r.html
* next ET release manager [RH]
Next release is in November; recall feature freeze is about two weeks before release and hard freeze a few days before.
https://docs.einsteintoolkit.org/et-docs/Release_Details https://docs.einsteintoolkit.org/et-docs/Release_Process
Steve agreed to be the manager with help from Yosef. Helvi, it is useful to have more than one person to do this. Steve, Roland left off the conditional that Steve will do it if Roland does this "soon," need not be the next release after.
** New inclusions for next release
FLRWSolver/Cosmology by Hayley Macpherson and Helvi will review it. Ticket #2549, https://bitbucket.org/einsteintoolkit/tickets/issues/2549/inlcude-flrwsolver... .
Kuibit and possibly RePriMand for inclusion. RePriMand has ticket #2551, https://bitbucket.org/einsteintoolkit/tickets/issues/2551/include-reprimand-... . RePriMand does have a thorn and an external library for it. Unlikely to build out of the box...requires a couple of other packages/tools. If it requires external libraries not typically available on clusters, we cannot enable it by default. For example, Peter's 1D Force is a model of a stand-alone version that is not built by default and uses SCons for building.
Atul, wanted to volunteer for Kuibit review, and he is unable to review Kuibit this semester. Can do less intensive work for the release though. Yosef intends to try to review Kuibit. Review does not check for correctness, checks for running and documentation. Ticket #2438, https://bitbucket.org/einsteintoolkit/tickets/issues/2538/inclusion-of-kuibi... .
Yosef, build systems question, can RePriMand run on all clusters. Wolfgang says it is mostly Python and if you have some Pip version maybe can install/build that way. Builds using Meson which uses Ninja and not Make.
Gabriele, how will Kuibit is to be included in the ET? In the repo or installed by Pip? Roland, could use it in GetComponents and that gives best exposure and lets different branches be checked out. Kuibit was reviewed in the Journal of Open Science Software...there is a link in the ticket. JOSS paper at https://joss.theoj.org/papers/10.21105/joss.03099 .
Roland, If the software is independently hosted, we ask for a branch label with the ET name, and then if their Master branch for that software fails, we use the ET branch of their software.
Helvi, has a code in the ET and has one branch for the ET and there is another branch her group uses routinely.
Roland discusses a couple of options and how to get the dependencies for Kuibit, for example. Gabriele also uses Tags on the Branches that he can make very easily. He likes distributing the repo because of all the examples. Helvi, maybe include these links in the documentation.
** Gallery Examples
Helvi has two students that should do examples.
* Any Other Business
** Maria, would like documentation on thorns Multipole and QuasiLocalMeasures and SphericalSurface used to extract the GWs. Why/How do they depend on each other and how to tune the parameters. Helvi, Check the Readme for the thorn, some information there.
Roland, WeylScal4 everyhwere in 3D space, and Multipole calculates the projection and spin spherical harmonics. QuasiLocalMeasures is separate, concerned with calculations on horizons, like spin, etc, not concerned with waves but with horizons. There are papers cited there, in the README. An older thorn does the same thing. Some documentation was started but does not go very far. SphericalSurface does not do any analysis, it is a way to describe in the code star-shaped surface. Horizon finder will store the shape in the spherical surface to do the isolated horizon quantities. Waveform extraction does not use the SphericalSurface at all.
** Helvi, there is a new monthly call for the NumRel community started by Nils Fischer; the next call is 13th September. The idea is to have speakers from different code communities, next time SXS/Mitman and Meudon/Oertel are presenting. Structure is to describe the project, here is what we are working, and here is where we have difficulties and ask if there are any ideas in the community.
https://github.com/sxs-collaboration/nr-community-call/wiki
* [https://www.einsteintoolkit.org/tools/unanswered.php unanswered question on mailing list]
Perl script is failing, in many ways. It does not display unanswered entries on the mailing list right now. Roland is aware.
* [https://bitbucket.org/einsteintoolkit/tickets/issues?status=open&status=... open tickets sorted by update time]
Helvi goes through the tickets.
#2553, CPPFLAGS and CXXFLAGS trouble when you use a non-g++ compiler, like using the cuda compiler. Roland has a work around. Only affects Cuda. If doing Cuda this might affect you.
#2559, HDF5 library generates 2GB of files and we do not use them and causes trouble in some user quotas. Roland, this is cleanup.
#2558, link for Power incorrect, ticket is almost closed, need to wait one more week and then will backport the fix.
#2538, inclusion of Kuibit, discussed above.
* [https://bitbucket.org/einsteintoolkit/tickets/issues?status=new&status=o... tickets ready for review]
Nothing new or discussed.
* Any Other Business
Next weeks chair and minute taker:
Yosef will chair. Maria will take minutes.