Re: Timeout error

From: mir inamul haq jeri <mirinamulhaq_at_gmail.com>
Date: Fri, 24 Jun 2011 11:42:42 +0200

Hello,
I beleive that you are right if the final deviation would be calculated
from the the scores made by each single particle.then if you make the
distribution of these scores from each individual particle it should be same
whether you run the particles in 5 independent runs or all particles in one
single run because the scores recorded by each individual particle have no
co-realtion.
However what happens here is that you run a cycle of n particles and obtain
an average score from this cycle, then you run another cycle and obtain
other averaged score and in the same way you have many cycles and results
from many runs.now you make a distributuin of all these averages to get as
near as possible to actual (physical) value.the code calculates the STD of
this distribution and prints it out.(It is also obivious that results from
each independent run need not be same).
Now suppose if we run only one cycle then in fact we dont have enough data
to calculte the STD hence it is printed as 99 percent and 0 percent in case
we record no score or if all the scores in a given bin are exactly same.
This is at least what i have learned from last few days while using
Fluka.there are anyway people who know it much better and we will be waiting
for their answers.

On Thu, Jun 23, 2011 at 4:52 PM, Nicholas Bolibruch<nboli_at_acanac.net>wrote:

> Sorry to take this topic on a slight detour, but I feel that I need to
> ask this question to improve the results (and confidence) of my
> simulations.
>
> My understanding of Monte-Carlo simulation is that if I run a simulation
> with 5E6 particles in a single run, it would have the same statistical
> result as 5 independent runs at 1E6. Is there something else I'm
> missing here?
>
>
> On Wed, 2011-06-22 at 12:56 -0700, Alberto Fasso' wrote:
> > 99% is not an error! It is a flag that an error could not be calculated.
> > It needs at least 4 (better 5) independent runs to calculate a standard
> > deviation. If the number of runs is less than 4, the FLUKA analysis
> > tools print 99%. That number has no meaning, it is only a code for
> > "error not calculable"
> >
> > Alberto
> >
> >
> > On Tue, 21 Jun 2011, mir inamul haq jeri wrote:
> >
> > > Hello,
> > > I got the reason of the 99% error.This is because i was running only
> one
> > > batch this time while as i used to run 5 batches previously.So i
> would say
> > > that each run serves as a batch of particles and for better
> statistics we
> > > need to run as many batches as possible.I suppose that the
> statistical
> > > deviation as calculated by code results from the distribution of
> average
> > > values of each batch.I would appreciate any comments and suggestions
> on
> > > this.
> > > Thank you
> > >
> > > On Tue, Jun 21, 2011 at 3:16 PM, mir inamul haq jeri<
> mirinamulhaq_at_gmail.com
> > >> wrote:
> > >
> > >> Thank you for your suggestion.
> > >> Now the problem is that i am getting the usrbin errors only as 99%
> and 0%
> > >> ir-respective of the number of particles that i run.i have run the
> same file
> > >> previously when i was getting only few percent of errors.Although i
> tried it
> > >> for the same number of particles as well as i had done previously
> when i was
> > >> getting only 6-10% error, but now i am getting only 99% and 0%.since
> the
> > >> values of yield/stars are different and the particle fluences are
> also
> > >> different in different regions(as seen from usrbin) i think there is
> some
> > >> error done by the code in calculating the errors.
> > >> I will be looking forward for your suggestions
> > >> Thank you
> >
>
>

-- 
MIR INAMUL HAQ JERI
INSTN SACLAY.
UNIVERSITY OF PARIS SUD 11.
Received on Mon Jun 27 2011 - 16:22:54 CEST

This archive was generated by hypermail 2.2.0 : Mon Jun 27 2011 - 16:22:55 CEST