RE: missing room in FLUKA stack

From: Fasso, Alberto <>
Date: Wed, 11 Nov 2009 12:26:58 -0800

The reason for biasing the photon hadronic interaction length is that
the hadronic cross section of a photon is much smaller than its
electromagnetic cross section. The ratio between the two cross sections
is of the order of 0.01 (the fine structure constant). That means that
without biasing, when sampling the next photon interaction point, only
in about one case out of 100 the interaction will be photonuclear, and
in all other cases it will be a Compton or a pair production interaction
(the only two possible electromagnetic interactions at energies higher
than the photonuclear threshold).
Using LAM-BIAS with a biasing parameter 0.01 makes the probability of
nuclear interaction similar to that of electromagnetic interaction: so
in about one case out of two we will get a nuclear interaction, and with
a probability of same order an electromagnetic one.

Giving a number as you did (2e-6) you would get only one electromagnetic
interaction out of 5000, so practically ALL interactions would be
nuclear! As you have found out, the number of nuclear interactions has
become so large that it has filled the stack: but this has been lucky
because eliminating all electromagnetic interactions would have
completely distorted the whole physics.

This can be a lesson about biasing in general. Biasing means modifying a
probability by using a weight so that the product weight x probability
remains constant. Mathematically, if the number of samplings were
infinite, the modifying factor and the corresponding weight could take
any value: but in reality this is not true, because the number of
samplings is finite. So, you can easily miss sampling a few rare
interactions with a very large weight, and the product weight x
probability would not be conserved.
It is very important to do biasing gently, without forcing. In many
cases, it can be recommended to try different amounts of biasing,
checking that the results don't change.

A few more comments on the LAM-BIAS issue.
At high photon energies, the stack gets filled faster because of
multiplication in the hadronic cascade produced by the photons. It is
then better to do less biasing, using a factor 0.02 or 0.03.
Your case seems to be the opposite: the photons produced by 6 MeV
electrons have energies mainly below the threshold for photonuclear
interactions, so maybe you could try a factor 0.005 or so (but checking
as explained above!)


-----Original Message-----
[] On Behalf Of Mr. Bhushankumar
Jagnnath Patil.
Sent: Friday, November 06, 2009 10:53 PM
Subject: missing room in FLUKA stack

Dear Fluka Experts and especially to Paola Sala, Francesco Cerutti,
Vasilis Vlachoudis,Florian Sommerer, Andrea Mairani,

As I ask in 9th FLUKA course at BARC, India that what to do if the
FLUKA stack is full ?
  When I was running the input for estimating the neutron fluence from
6MeV electron source. I faced the problem of FLUKA stack is full. While
running upto 6 cycle it runs and at 7th cycle after completing 480000
histories it shows the following error in .err file

because of missing room in FLUKA stack ***
  **** Photonuclear interaction not performed

Because of this the fluka run stops.

Before this also I have got the same error. May be I am wrong but I
thought that this cause because of I have given very low number (2e-6)
in biasing of inelastic hadronic interaction length of photon using
LAM-BIAS option. So at that time I changed the biasing option in lambias
card and carry out the run for next runs, then it runs. So I never sent
the files and error to FLUKA discuss. But as discussion in fluka course
I understand that I should share such things.

Please correct me if I am wrong somewhere. Also tell me what to do in
such case.

Thanking you.

Mr. Bhushankumar Jagnnath Patil.
Microtron Accelerator Lab.
Department of Physics,
University of Pune,
Ganeshkhind, Pune 411007
Ph. No. Off. 020-25692678 Ext. 421
Mob. 9823968377
Alternate e-mail
Received on Fri Nov 13 2009 - 18:27:44 CET

This archive was generated by hypermail 2.2.0 : Fri Nov 13 2009 - 18:28:14 CET