[Users] Meeting minutes for 2025-01-16

[Roland Haas]

Hello all, Bruno,

Before this takes on a live of its own:

I re-did my own tests and found no more differences after a single
Euler step.

The issue that caused me to think there were differences before were:

* I had looked at the wrong column in norms output and mistaken an
  "average" norm for a "maximum" norm (and the average will change due
  to different orders of values being summed up)

* I still think I see a difference in values for points that are at the
  same location on different refinement levels (which should also not
  be in a vertex centered run), but those seem the same no matter
  whether I use 4 or 8 MPI ranks (also a bug if true, but of different
  kind)

> Regarding the BNS issues, is it something due to when you recover from
> checkpoint?

Maybe, I don't know.

> In Whisky we had an issue related to that due to Con2Prim running after
> recovery from checkpoint.

Yes, I had pointed it out in the past to Michal.

For those one can set:

MoL::run_MoL_PostStep_in_Post_Recover_Variables = "no"

which as Bruno said, disables the extra con2prim after checkpoint
recovery.

Yours,
Roland

-- 
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/users/attachments/20250118/b85d1368/attachment.sig>