[Users] Meeting Minutes 2022-9-15

[Cupp, Samuel D.]

Present: Sam, Steve, Keith, Gabriele, Peter, Leo, Zach, Bruno

New ET modules:
* no news on Canuda review
* Haley has addressed reviewers' comments for FLRW solver
* Sam has addressed reviewers' comments for selfforce-1d;
  only remaining task is to regenerate the FORD documentation
* Leo has addressed reviewers' comments for NRPyEllipticET;
  notebook which generates the code is not yet ready, but this
  is currently being worked on
* Baikal update: Zach hasn't heard back from Helvi. Since the
  new Baikal is waiting for inclusion, the old Baikal's notebook
  isn't being maintained in favor of the new Baikal


Other questions:
* For the particle tracer Leo is working on, he needs to access the
  cctk_timefac for the finest level, but cctk_timefac only gives the
  value for the current time level. He does this by accessing the
  Carpet parameter time_refinement_factors using CCTK_ParameterGet()
  but wanted to know if there was a better/more 'proper' way of
  accessing the parameters. The function is necessary because the
  parameter is private and cannot be accessed via the interface.

  Steve said that this is how he would do it, but Leo should send the
  question to the mailing list to see if Roland or Erik have other
  suggestions.


* We would like to shut down the svn servers which aren't being used
  anymore and have been maintained for preservation of old thornlists
  for a long time. By shutting them down, we can see if anyone
  complains about it and also track down anywhere else on the website
  that might reference them.

* There is no established policy on deprecating or renaming
  thorns/parameters. Zach proposed that we should set a clear policy
  for the timescale on which these things can be done, and the
  procedure for doing so.

* Someone asked about whether there will be recordings of the European
  ET meeting, but Peter said there were not. However, they do intend
  to upload the slides at some point.

* Next European meeting was set and will be in Aveiro, Portugal.

Questions on the mailing list:
No one answered 'Running with SLURM' from last week; Steve will respond to it.
http://lists.einsteintoolkit.org/pipermail/users/2022-August/008667.html

Open tickets:
Gabriele brought up ticket issue #2629
https://bitbucket.org/einsteintoolkit/tickets/issues/2629/mol_pseudoevolution-vs-analysis
about when one should schedule in CCTK_ANALYSIS vs MoL_PseudoEvolution.
The particular concern is how this will interact with parameters such
as the calculate_constraints_every in LeanBSSNMoL, as some choices
of output_every type parameters will be invalid depending on the
number of active refinement levels. However, since these calculations
are expensive, it is important to be able to control the calculation
to only occur when it will be output.

Zach said that one solution is to just leave it up to the user to
decide if it is sane to update these quantities every n timesteps.

This is similar to the issue Leo has been facing with the particle
tracer, as it must compute and output only so many iterations.
However, Cactus counts iterations based on the smallest possible
refinement level, even if it isn't active. This leads to the need
to determine the dtfac for the finest possible level, which is the
topic of the question raised earlier. He then errors out if the
parameters will produce incorrect output.

Since several thorns need this feature, discussion arose on whether
this should be provided as an aliased function by Carpet to easily
provide access to the needed information to validate the given values
of these types of output_every parameters. Carpet could set some
out_every_safe that provides the minimal safest out_every.

The chair for next week (9/22) is Leo, and the minutes taker is Bruno.

   Samuel Cupp
   Postdoctoral Researcher
   Department of Physics
   University of Idaho
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.einsteintoolkit.org/pipermail/users/attachments/20220915/8e5243f6/attachment.html