Hi all,
Here are the minutes from today’s meeting.
Chair: Sam Minutes: Leo
Attending: Sam, Leo, Peter, Cheng-Hsin, Johnny, Keith, Roland, Zach, Yosef
ET Release - Progress is being made, fixing failures in clusters that are not frequently used. - Odd compilation error in Summit cluster that hasn't been fixed yet. - Roland will contact contributors. - All galery examples are running. - Peter will re-run tests on Sunrise.
Extension of PreSync's Capabilities with Carpet - PreSync is used to fill in ghostzones and apply BCs automatically whenever needed. - Extend to fill in "auxiliary" quantities automatically when access is requested. For example, if a thorn evolves the BSSN variables, then it provides a function to compute the ADMBase variables from the BSSN variables and PreSync would schedule this function appropriately whenever another function needs to access the ADMase variables. - CarpetX has some support for gridfunction dependencies, which could potentially guide this process. - There are subtleties that could lead to circular logic, so care must be taken when doing this. - First idea is to not deal with more complicated cases where e.g., a function adds something to a gridfunction (like KO dissipation). Instead, the new feature would fill in gridfunctions for which a "READ" was requested but the gridfunctions are not valid anywhere.
Open Tickets - No warning printed by the Perl scripts for ambiguous READ/WRITES since it is unlikely users will attempt to use the old PreSync standard. Zach suggests: would be nice to have the option of being more verbose instead of completely removing the warning, if possible. Roland says: maybe we can give the Perl script another option, e.g. "Picky" (Verbose is already taken) so that it can print the warnings if requested. - There is a ticket ready for review which prevents PreSync from syncing coarser refinement levels if those have already been synced previously. - There's an interface error in some of the ADMBase functions where read-only gridfunctions are maked as "OUT". Sam says: changing "OUT" to "IN" fixes the interface. Fix is working on local machines but needs more testing. - cctk_time and cctk_deltas are wrong sometimes and a fix has been implemented that needs further review. - With PreSync, CarpetRegrid does not change the validity of the gridfunctions after a regrid, which leads to problems. The logic should be: after a regrid all gridfunctions are valid nowhere. Then the gridfunctions that support prolongation can be flagged as valid in the interior after prolongation. Everything else should remain invalid everywhere until something else sets it.
Unanswered Questions in the Mailing List None.
Next Chair: Leo Next Minutes: Johnny
Cheers, Leo
------ Leonardo R. Werneck, Ph.D. Postdoctoral researcher Office EP 314 | Department of Physics | University of Idaho 875 Perimeter Dr. MS 0903 Moscow, ID 83844-0903, USA leonardo@uidaho.edu mailto:leonardo@uidaho.edu https://leowerneck.github.io https://leowerneck.github.io/