From: Vasilis Vlachoudis (Vasilis.Vlachoudis@cern.ch)
Date: Mon Dec 10 2007 - 18:12:16 CET
Hi jamshid,
welcome to the wonderful world of monte carlo.
This zig-zag has a name (and a family-name :) and is called statistical
fluctuations and you will have to learn to live with it. Actually is a
very useful information that tells you how well you should trust your
result.
However a better quantity to plot, is the statistical mean of each bin
together with the error on the mean. In principle MC is a counting
experiment and should follow a Poisson distribution, therefore the error
is the square root of the counts in each bin, which is no longer true if
you are using biasing or other techniques. Therefore FLUKA exploits the
Central Limit Theorem to get the correct mean and error on the mean. For
this reason you need to perform a few cycles and based on the CLT the
final distribution on the average will tend to be Normal (for infinite
cycles) no matter what was the original distribution. Actually it
converges very quickly and 5 cycles are enough.
Please give a look on the FLUKA course material and especially the
presentations about the monte carlo and estimators.
Regards
Vasilis
Jamshid Soltani wrote:
> Dear fluka Users,
>
> I made a graph for interaction of Photon with 1.3 Mev with water as
> attach input file.there is my graph in attach file too.there is more
> widspreding(Zig-zag) in the graph. I changed DEFAULTS to PRECISIOn and
> EM-CASCAde but there is zig-zag yet.Can I omit it or not? what is the
> reasons of these event?
>
> thanks
> jamshid Soltani
>
>
>
>
This archive was generated by hypermail 2.1.6 : Mon Dec 10 2007 - 23:50:36 CET