Running FLUKA on Cluster

Running and Runtime Errors
1

Dear FLUKA experts,
I’m running FLUKA on our cluster and I wanted to utilized the “spawn” option present in FLAIR. That is, run many cycles with one cycle/CPU.

So I wrote the Following slurm script “Example_FLUKA_Sbatch.sh” to do this. I ran the code with one of the examples that came with the code “example.inp”
BUT this only prints output files for one cycle even after creating the temporal directories just after the simulation is lunched. (see screenshots and zip folder)
Please kindly advice.
fluka_cluster.zip (1.4 MB)

cluster_runs_error-2
cluster_runs_error-2848×885 129 KB

cluster_runs_error-1
cluster_runs_error-1859×811 137 KB

Best,

Zavier

2

Dear Zavier,

the issue is that all your jobs run the same

rfluka -M 1 example.inp

command. Thus, overwriting the results. The solution is to create input files with unique names or run them in separate directories.

Also, it is important to change the random number seed for each input file, to get meaningful statistical errors.

Cheers,
David

3

Dear David,

Thank you for the response. I created about 6 different input files with unique seeds,

rad the simulation that got outputs.

But how can I process/visualize them? I tried loading them from flair BUT only those for a single cylce seem to open.

image
image1656×509 56.2 KB

I then tried adding manually but wouldn’t see the files

image
image1339×867 67.9 KB

I’ve attached a zip file containing the output of just two inputs due to uploading size limitations
fluka_cluster_out.zip (2.9 MB)

please advise.

Zavier

4

Dear Zavier,

please, have a look at this post explaining how to use FLUKA’s post-processing tools from the command line:

Cheers,
David