How to use Aster "the old-school way"

SadriMével

Hello everyone,
Is it still possible to launch a study with the old-fashion way :
/opt/aster/bin/as_run --run mon_etude.export
?
It was very useful and efficient to run multiple studies and, moreover, link them with other internal or professionnal tools I have on my computer.

I've just downloaded the "Scibian10 Singularity container" and the last sif with Salome_Meca. Thank you : it works ! I have a graphical interface, with all the tools (Geometry, Mesh, etc.). But I do not figure out how to write and launch my own script, in a bash-style.

I wanted to download Aster without Salome, but I havnt found Aster alone 🙁 I've tried to read :

https://gitlab.com/codeaster-opensource-documentation/opensource-installation-development => but I didn't find the answer ;
https://codeaster.readthedocs.io/en/latest/userguide.html => but the guide is empty, and the links are broken.

I would be soooo pleased to find a way to just write my scripts in text, and launch it without any graphical interface.
Can I ?

Thank you for your help !
Pardon if the answer is already written somewhere.
A+

PS : I'm not a EDF worker - I usse Aster both for personnal et professionnal use

mf

SadriMével
Hello,
there you go:

There's a shortcut too (only one command) but that I forgot. Your container may have a different filename,
Mario.

lIlIl

mf It doesn't work in WSL for me.

user@mypc:~/sm22> ls
salome_meca-lgpl-2022.1.0-1-20221225-scibian-9.sif
user@mypc:~/sm22> ./salome_meca-lgpl-2022.1.0-1-20221225-scibian-9 --shell
-bash: ./salome_meca-lgpl-2022.1.0-1-20221225-scibian-9: No such file or directory

jeanpierreaubry I get this:

user@mypc:~/sm22> singularity run ./salome_meca-lgpl-2022.1.0-1-20221225-scibian-9.sif shell
*****************************************************
INFO : Running salome_meca in software rendering mode
*****************************************************

Then nothing happens. If I add << END, I cannot run the command (it waits for me to end the command line).

Cherblanc This seems to work.

user@mypc:~/sm22> singularity run salome_meca-lgpl-2022.1.0-1-20221225-scibian-9.sif shell -- as_run --test forma01a
*****************************************************
INFO : Running salome_meca in software rendering mode
*****************************************************
<INFO> démarrage de forma01a

--------------------------------------------------------------------------------
 Exécution de Code_Aster

<INFO> préparation de l'environnement (...)

jeanpierreaubry

hello

this is the way i have proceeded for almost 16 years now on
and the way i keep proceeding to day with version 16.5 and 17

have a look at this post it explains how it is possible
https://forum.code-aster.org/public/d/27218-where-is-code-aster-website-going/8

jean pierre aubry

jacob

singularity run ~/salome_meca-lgpl-2021.0.0-0-20210601-scibian-9.sif shell << END

as_run Stage_0_PreProcess.export
as_run Stage_1_Assembly.export
as_run Stage_2_Load.export
as_run Stage_3_PostProcess.export
as_run Stage_4_Reactions.export

END

Cherblanc

I added the following alias in my .bashrc:
alias aster='singularity run /home/path_to_Salome/salome_meca-lgpl-2022.1.0-1-20221225-scibian-9.sif shell -- as_run --run'
And then I can launch a simulation from the command line with:
aster mon_etude.export

jeanpierreaubry

lIlIl
the command you display is not what my post is about
not at all
read it more carefully
it needs some installation before running code_aster
which means installing a stand alone version of code_aster within a conda environment
main advantage:
-you have full access to what is installed
-you have only code_aster stand alone, no salome-meca, no asterstudy, no paravis
the old school as was asked for in the tile of the post by SadriMével

mf

lIlIl
Hello,

yes, repeating half of the commands probably does not work :-)

Mario.

lIlIl

My reply went to show that short answers do not actually help beginners solve their problems. I share my own tests to save time for others.

mf Since your first command contains install I did not run it since I had already downloaded the SIF (Singularity Image File) when following the tutorial by Hassan Berro.
I tried to run your first command just now: nothing else was installed, a Python file was created in the folder (salome_meca-lgpl-2022.1.0-1-20221225-scibian-9) and now the second line worked. I looked up "--app install" in the documentation, but it does not explain what it does in details. singularity run -h explains that:

--app string             set an application to run inside a container

I do not understand what it means. No mention of "--app install".

k_zurawski I managed to do what you did in the PowerShell. When I tried:
singularity run ./salome_meca-lgpl-2022.1.0-1-20221225-scibian-9.sif shell
again today in the Windows terminal, it worked! So jeanpierreaubry's command also works (without the << END). The conda environment approach is interesting.

My problem now is that many of the tests don't work (forma01a, forma01b, forma02a, adlv100d):

 ╔════════════════════════════════════════════════════════════════════════════════════════════════╗
 ║ <F> <JEVEUX1_71>                                                                               ║
 ║                                                                                                ║
 ║ La mémoire totale de 256.00 Mo allouée à l'étude est insuffisante, il est nécessaire           ║
 ║  de disposer d'au moins 688.39 Mo uniquement pour démarrer l'exécution.                        ║
 ║                                                                                                ║
 ║                                                                                                ║
 ║ Cette erreur est fatale. Le code s'arrête.                                                     ║
 ╚════════════════════════════════════════════════════════════════════════════════════════════════╝

k_zurawski

I'm running WSL2 Debian 12 and Salome-Meca 2022 under Singularity 4.2.0. It works in terminal:

k_zurawski

lIlIl change amount of memory in export file to something bigger than required.

lIlIl

This solved the problem (as described in the error message). I mentioned it because I was surprised a test would fail in the official container. Dose the required amount of memory change depending on the hardware?..

Sound_Spinning

lIlIl I also found this part very odd in code_aster, a default small RAM & run time, dunno why....

I set in .export file:
P memory_limit 56640.625 ## Mb, 90% of my PC RAM
P time_limit 400000.0 ## in seconds, odd unit

With those 2 it always seems to work. Is there a config file somewhere to set these by default for all runs?

k_zurawski

Mega octets, not megabytes, and memory limit is max possible to use not assigned to a job. mem jeveux is.

Sound_Spinning

k_zurawski I don't quite understand, but look interesting, what do you exactly mean?

When I use the values I posted, code_aster writes in the .export file these:
P memjob 58000000 P memory_limit 56640.625 P time_limit 400000.0 P tpsjob 6667 A memjeveux 7080.078125 A tpmax 400000.0
What's the difference vs what you suggest?

k_zurawski

memjeveux controls the amount of memory (RAM) allocated to the Jeveux memory manager, which is used internally by Code_Aster for managing temporary data structures and computation buffers.

memory_limit is upper limit for the total memory that can be used by the analysis, including Jeveux memory and other memory used by the solver, libraries, and linked modules.

k_zurawski

In my opinion, and this is only opinion not based on facts, the same testcase created years ago for standalone Code_Aster might require less memory than run in the latest Salome_Meca in container.