Re: error after 5 cycles

From: Santana, Mario <msantana_at_SLAC.Stanford.EDU>
Date: Sat, 13 Apr 2013 21:19:27 -0700

Hi Leila,

I have not been able to reproduce the crash in short runs. I estimate that it would take me about 17-20 h CPU with my computer if it happens during cycle 6 for the current seed. In any case it is not related to running 10 cycles or 5. If you reduce the number of particles to e.g. 1E2 you can run 10 cycles, as expected.

This may be a rare crash (in which case you should not worry too much) or it may be related to a tiny error in the geometry which makes that only when particles are very unlucky the run crashes.
I looked at the input file and found the following:

  1. your 'shield' region has marginal geometry errors. You are subtracting two RPPs of different transverse dimensions but with the same length. Therefore the points belong to the end surfaces are not determinate. In such cases it is good practice to avoid abiguity by making the inner body slightly larger in one of the dimensions. In your case you could case Zmin of bigtw from –200.0 to –200.1, then you should also redefine the 'tunnel' region as "+bigo +bigtw –spher1"
  2. In your region definition you have sphere "5 +spher1-stripp" . Normally you should have a space separation between the two (although maybe FLUKA knows there are two definitions by checking there's a + or minus sign).


From: Leila Mokhtari <<>>
Date: Wednesday, April 10, 2013 6:48 AM
To: fluka-discuss <<>>
Subject: error after 5 cycles

Dear user

I am trying to run my input for 10 cycles but the program gives error after five cycles running.

I could not find the reason. I would be grateful if anyone can help me on this.

You may find the input file in attachment
Received on Sun Apr 14 2013 - 22:54:56 CEST

This archive was generated by hypermail 2.3.0 : Sun Apr 14 2013 - 22:54:59 CEST