Re: Biasing particle type

Hi Mitja

I would still suggest that you first explore BIASING before using usimbs
and weight-windows, also because of the CPU penalty coming with usimbs
(called at each step during transport!).

For documentation on variance reduction techniques you may want to check
out the FLUKA course material which can be found on the FLUKA Web-site
under "Quick launch - Courses".

Cheers
Stefan

On Wed, 15 Oct 2008, Mitja Majerle wrote:

> Hi,
>
> at the time I was doing that I did not notice that BIASING can be used for
> different particle types unfortunately. Now I have found the alternative, I
> use usimbs.f routine to increase weights for all particles as they travel
> along my z direction and then the combination of WW-THRES and WW-FACTOR to set
> the weight windows, so that when the proton weights reaches certain value it
> is splitted. But I do not know if it really works, at least the simulation
> runs slower, so there must be something going on. Could you recommend me a
> good literature about different variance reduction techniques (besides MCNP
> manual) ?
>
> Thank you,
> M
>
> Stefan Roesler wrote:
> > Hi Mitja
> >
> > The BIASING card allows you to tune region importances for individual
> > particle types by combining cards with What(1)>=0 and What(1)<0. See Note
> > 2) in the Manual. Did you consider this possibility? Of course, you will
> > have to split regions in order to divert CPU-time towards your detector.
> >
> > Cheers
> > Stefan
> >
> >
> > On Mon, 13 Oct 2008, Mitja Majerle wrote:
> >
> > > Dear fluka users,
> > >
> > > during the spallation process only 10% of produced particles are protons,
> > > and
> > > most of them are stopped in the target, but I am more interested at
> > > protons
> > > than at neutrons. How do I increase the statistics of my proton USRTRACK
> > > detectors ?
> > > I tried to use BIASING with usimbs.f routine, where I set the weight for
> > > protons to values higher than one:
> > > IF (JTRACK .EQ. 1) THEN
> > > FIMP = TWOTWO
> > > ELSE
> > > FIMP = ONEONE
> > > END IF
> > > RETURN
> > > But it seems that fluka hangs while initializing the problem. What are the
> > > alternatives, which is the most effective variance reduction for my
> > > example ?
> > > WW-FACTOR and WW-THRESH could be used ? I know only the simplest method
> > > with
> > > assigning different importances to regions, and it is not useful.
> > >
> > > Thank you,
> > > M
> > >
> > >
> >
>

-- 
___________________________________
Stefan Roesler
CERN, SC/RP
CH-1211 Geneva 23
Switzerland
Phone:  +41-22-7679891
Fax:    +41-22-7669639
E-mail: Stefan.Roesler_at_cern.ch