[Users] Meeting Minutes 2025-09-04

Thu Sep 4 14:06:44 CDT 2025

Minutes for the Einstein Toolkit Meeting, 20250904.
https://docs.einsteintoolkit.org/et-docs/Main_Page#Weekly_Users_Call
9am CDT Thursdays

Present:  Steve B, Deborah F, Peter D, Peyhan, Roland H, Lucas TS, 
Johnny T, Nikolai W, Zach E, Keith D, Bill G, Leo W
Chair: Peter D   Minutes: Bill G

* BBH gallery example (Zach) -- not yet done (2024-11 also missing)
Peter and Zach, no new work.

* AOB
** Next release is in May 2026 and Steve is the release manager. 
https://docs.einsteintoolkit.org/et-docs/Release_Process

* Unanswered emails https://www.einsteintoolkit.org/tools/unanswered.php
** No updates.

* Open tickets sorted 
https://bitbucket.org/einsteintoolkit/tickets/issues?status=open&status=new&sort=-updated_on

#2885, test for std::filesystem availability in Silo output likely broken
Roland, this should not affect you if you are on a modern system. You 
have to look at a predefined variable and you have to use it to include 
the version of a file, and it is declared in those files.  There is a 
version.h that was standardized in C++20...exists in older gcc but not 
standardized.  In the IO code of Carpet that writes out Silo, maybe 
check on a file existence. Not a major issue.  Was in CC_Export also, 
which has a recent pull request with the fix.  Some bug with a static 
variable that depends on a function argument so it is never changed.  On 
Deborah's to do list also.  Deborah, it may not need to change. Will 
take a look at it.

#2882, Running BBH with CarpetX with CPUs: High memory consumption and 
low performance
Alejandra will take a look at it.

#1847, FFTW3 fortran interface not working for system installation
A very old ticket from Frank Loeffler.  Yosef may use FFTW in his code(s).

#2858, SF files for MN5 (was: Compiling CarpetX: issues with PDESolvers)
Changed the name to the current one.  Lucas has begun testing, thinks 
files very specific to Alejandra's work, on Marenostrum 5 (Barcelona).  
Some external libraries seem to compile themselves and put them into a 
folder referencing the user.  Steve, not uncommon but should not had 
code the user directory.  Lucas, that is what is happening here.  On LSU 
machines there are directories hardcoded to Steve's directories; that is 
where the external libraries are, and this works.

Steve, setup a system to have future checkouts and builds refer to 
previous compiled libraries/tools.  Some external libraries take forever 
to compile and better to refer to some already installed library.  Good 
to have a "benevolent user" to install these for global use.

Lucas, (Marenostrum) the machine is organized into projects and 
allocations, and projects have an expiration date...like a DOE machine.  
Steve, that is awful.  Lucas agrees.  Lucas, think singularity would be 
the right thing to do, if the image is put together correctly.  Peter, 
can you chance the setup so that they use external libraries instead of 
their pre-compiled ones?  Lucas, have not tested with the bundles 
libraries.  Believe they tried that and failed.  Roland, you might 
succeed, when they tried some of the libraries did a git clone of a 
repositories.  Lucas, Marenostrum has limitations that annoy one.  Will 
try the bare minimum and see if that works.

#2866, Problem while building the ET on MN5: OpenCL
Steve disavows any knowledge of this one---name removed from ticket.  
This is for Marenostrum 5.    [European clusters with ET include 
Marenostrum 5 (Barcelona), Leonardo (Bologna), and Sunrise 
(Stockholm).]    Peter, maybe Lucas can take a look at OpenCL? Roland, 
we are not using OpenCL for anything now---once we had an example using 
it.  Lucas, will try the bundled version of OpenCL and if it fails will 
ask them to take it out.  Leaching off of Steve's Singularity image 
which works on LSU machines, might try for Marenostrum.

#2867, WaveToyX examples, HDF5 output error in all three examples.
Steve, we discuss this every wekk and the error message is considered 
harmless.  We should close this ticket.

#2773, make CarpetX-ThornDoc is confused in InterLatex.pl
Steve, I need to check on that.

#2878, CarpetX: Add multipatch output support to Silo
Lucas, sitting and waiting for review.  Not urgent.

#2764, PUGH tries to free memory not allocated by malloc
Beyhan, no update.

#2282, gallery examples use low-order integration n Multipole
Roland, part of updating.

#963, Improve McLachlan accuracy
Peter, will meet with Zach and work on this.

#2706, Update default TwoPunctures parameters, or at least default 
parameters in BBH gallery example
Zach, part of the BBH Gallery example and will be looked at.

#2877, ET_BHaHAHA Features Ticket
Zach, keep this one, I use it as a ticket to jot down ideas to improve 
the robustness of BHaHAHA.

#2855, NoiseX: Improve noise generation in CarpetX
Lucas, not worked on it for a while, intend to get back to it.

#2874, NRPyElliptic lacks regeneration instructions
Leo, on my list, will get back to it.  Zach, you should wait on this.  I 
plan to refactor NRPyElliptic in a library like BHaHAHA. The refactor 
will be a different NRPyElliptic.

#2862, Update `SpacetimeX/Z4c` robust stability paramete file
Lucas, that is just a parameter file update.  Will review desire to 
include.  The existing one did not work, but this one should.

#2860, Don't call the optimized 4th order second derivative operators.
Peter, will take a look at it.

#2052, piraha assumes that assert(false) always aborts
Steve, I will get to this one.

* Tickets for Review 
https://bitbucket.org/einsteintoolkit/tickets/issues?status=new&status=open&sort=-updated_on&q=Please%20review

Nothing new.  Roland, no longer a ticket to backport libraries that 
compile with GCC 15 to earlier or to update with cmake 4.0. Applied 
those changes to master and release branches.  Tickets #2880, #2883.

*AOB at end of meeting

Roland, have a CarpetX call next and Erik said he would give a 
presentation on mesh files.

Zach, should we advertise the CarpetX call?  Roland, we should list it.  
Will draft a description of CarpetX and give the call link.  Bill, maybe 
we should put it down at the ET Meeting link [wiki] and not on the front 
web page.
CarpetX wiki https://github.com/EinsteinToolkit/CarpetX/wiki
CarpetX zoom 
https://ubc.zoom.us/j/66033196685?pwd=zxXTpcrBY1L7nT2kNFITm71ZuoZEiO.1

Next Meeting, Thursday, 11 September 2025. 
https://docs.einsteintoolkit.org/et-docs/Meeting_agenda

FYI, bill e.g.

-- 
Home Page <https://gabella.github.io/index.html> LinkedIn 
<https://www.linkedin.com/in/bill-gabella-350b016/>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.einsteintoolkit.org/pipermail/users/attachments/20250904/c2d376a1/attachment-0001.htm>