Fluka output

[grzanka]

We need to be sure that Fluka output is properly presented by the UI:

• pymchelper should properly digest all Fluka outputs
• units should be properly presented (1-D and 2-D plots)
• normalisation should be correct (when running on multiple tasks)

For all of the quantities present in the examples we should check, by comparison with SHIELD-HIT12A is:

• axis and quantities are named correctly
• units are present
• if the quantity is of the same order if running on 1 or 10 tasks (check if averaging works correctly)
• check if differential scorer works

[grzanka]

Most probably that needs to be implemented in pymchelper

[hendzeld]

PR DataMedSci/pymchelper#691

Update in USRBIN for cartesian mesh

[hendzeld]

PR DataMedSci/pymchelper#693

Update in USRBIN for 3D cylindrical scoring

[hendzeld]

[grzanka]

What do we see on the plots ? If you aim at comparison SHIELD-HIT12A vs Fluka, then I'd start with 1D, not with 2D plots.

[hendzeld]

We should unify units for LET (KeV/um, GeV/cm MeV/cm, etc.)

[hendzeld]

Unifying output units: DataMedSci/pymchelper#702

[MatoKnap]

I conducted several tests to compare the outputs of FLUKA and SHIELD-HIT12A using 1 and 10 tasks. I used the common example, 'Beam of protons with lead collimator,' which is compatible with both simulators, but modified the number of particles to 10^4. I ran the simulations on PLGrid using a direct run.

Here are the results:

Dose	1 task	10 tasks
Fluka
ShieldHit

Fluence	1 task	10 tasks
Fluka
ShieldHit

Both normalization and averaging seem to work correctly, but the SHIELD-HIT12A result shows some anomalies.

[grzanka]

Interesting. What kind of medium if placed in the region where SHIELD-HIT12A dose is ~0.5 ? Is it air ?
There are couple of definition of dose: "dose-to-medium" or "dose-to-water". There may be a difference between medium air and water. I dont remember what is implemented as default one for Fluka and SHIELD-HIT12A

[grzanka]

Also funny to see that SHIELD-HIT12A protons seems to have shorter range than in Fluka

[MatoKnap]

After resolving input file issues for SHIELD-HIT12A, I re-ran the simulations with 10^4 particles on PLGrid using corrected input files. The updated results now closely align. However, the dose peaks differ slightly: FLUKA reports ~1.8, while SHIELD-HIT12A shows ~1.6. Does this difference matter? If not, it suggests both simulators use the same units.

Dose	1 task	10 tasks
Fluka
ShieldHit

Fluence	1 task	10 tasks
Fluka
ShieldHit

[grzanka]

Can you reran with 10^6 or 10^7 particles ? It may be on 100 or 200 tasks

[grzanka]

Indeed it seems that the units are correct and the averaging works well. I am bit puzzled by the bump on the dose vs depth (Z) plot between-5 and -4 cm. After rerunning the simulation with much higher statistics, can you upload the results to a GitHub gist ? I usually downloaded JSON into my local drive, opened it in reasonable text editor (i.e. notepad++ under Windows), then copied the content into the clipboard (Ctrl-C), then opened in the web browser GitHub gist site and pasted there the content of the file.

You could then point me to the results, like I did here: https://github.com/grzanka/plotly_lgad_dosimetry/blob/main/montecarlo/README.md

[MatoKnap]

Here is a quick comparison using 10^5 particles. There are a few minor discrepancies in my tests:

1. In Fluka with 200 tasks, only 84,000 primaries are reported as requested. This seems to be a visual bug, as the task bars are correct (I counted them).
2. Shield-HIT with 200 tasks was run in batch mode, while the rest of the simulations were run directly.

This is an early preview; I’ll upload a detailed version on GitHub Gist later. I might run the next batch of tests tomorrow, as one of my simulations has been pending for three hours.

The differences between the simulators are real, particularly in the peak and the -5.5 to -4 cm region. Below are the current results:

Dose	100 tasks	200 tasks
Fluka
ShieldHit

Fluence	100 tasks	200 tasks
Fluka
ShieldHit

[MatoKnap]

Here is a GitHub Gist with simulation results for Fluk and Shieldhit with 10^7 particles.
https://gist.github.com/MatoKnap/a69d508bd5b57ae2f5b8517c42ff3db5