Reducing cpu time in cosmic ray simulations

Dear Experts,
I am using the cosmic ray I/P file with atmol.f for my local geometry. Now, as I want to study variation of thermal neutrons with the composition of the soil, I have changed the material composition of the EARTH region to soils from the vacuum(default input material in the example). But it increases the simulation time to many many folds even for moderate particle history ( more than 7 days for 10 k particle history). Is there any effective way to modify the material composition of the ground EARTH or any variance reduction technique which can reduce the CPU time. (149.0 KB)
I have attached my I/P for reference

Thanks in advance for the kind help


Dear @sandipan,

It looks to me that the your Earth region is quite large and when changing the material from vacuum to soil the code has to track particles for several thousands of km. If you are interested in the neutron variation through the Earth soil, I would suggest slicing the Earth region and scoring only in the slices you are interested to (not sure if, for you project, you really need to score in such large region). You might use importance biasing to help the simulation to converge in the slices of interest, by increasing importance to neutrons: that might save some time.

Please, make sure that assignment of the materials is properly done. I have tested your input switching the Earth material from vacuum to soil and I see an increment of the average CPU time (as expected). However, the increment is such that 10k particle can be still simulated in a few hours.

Hope this can help.



Dear Sir,
Thanks for the kind reply. Actually I am trying to find neutron variation at ground ( basically AIR100 to EARTH boundary). But when I am trying to modify the geometry through flairs, its not allowing to modify the geometry with the message " geometry cannot be modified outside atmloc.geo and If you can you kindly guide me on how I can effectively make a earth slice at my region of interest? Any demo input will be great help for me.

Secondly sir, as far as CPU time is concerned, I am not able to use “spawn” parallelization of FLAIR for the cosmic ray INPUT. Use of more than 1 spawn shows error and stops the run. But other inputs are running smoothly with spawning. Is it possible to spawn cosmic ray input?

Thank you @sandipan for reporting.
Indeed there was an error in writing the spawned file when there was a filename in the geometry and it had #include cards.
We fixed it in the dev version of flair, and we will publish quite soon flair together with some other fixes.

Dear @sandipan,

In addition to @vasilis’ comment, I was able to modify the geometry directly from flair. I attach an example where I split the Earth region in two sub-regions. You an edit new regions directly from the Geometry tab (geoviewer), where you can see in real-time the zone/region you are editing. The changes will be stored directly at the level of your atmoloc.geo file.

Please, find here a zip with the modified .inp and .geo: (5.2 KB)

Hope it can help.



1 Like