Re: [fluka-discuss]: How to set the lower and upper weight limit in ww-factor

From: Alberto Fasso' <fasso_at_SLAC.Stanford.EDU>
Date: Sun, 13 Oct 2013 05:57:58 -0700 (PDT)

All what Mario has written is correct, but I would like to add another way
to obtain what is the actual average weight of particles in a given region.
This other way is much easier than those suggested by Mario.
Put in your input a BIASING card with SDUM=PRINT (it doesn't matter here what
the WHATs are: do it on any one of the BIASING cards you already have inserted
to set up importances). On the manual, at command BIASING, you can read:
    SDUM = PRINT : importance biasing counters are printed (useful to tune
                   importances and weight windows)
These counters are printed on the main output near the end of the run (just
before the final statistics). Read on the manual, in Chapter 9.1 (Main Output)
what is said about them:

           Russian Roulette/Splitting counters (if requested, see BIASING)
If the BIASING option has been used with SDUM = PRINT, the following statistics
is printed for each region:

       "N. of RR" --> Number of Russian Roulette operations made on particles
                        ENTERING that region
       "<Wt> in" --> Average weight of particles submitted to Russian Roulette
                        when entering the region
       "<Wt> kil" --> Average weight of particles killed after being submitted
                        to Russian Roulette when entering the region
       "N. of Sp" --> Number of splitting operations made on particles ENTERING
                        the region
       "<Wt> in" --> Average weight of particles submitted to splitting when
                        entering the region
       "<Wt> out" --> Average weight of particles after being submitted to
                        splitting when entering the region

Separate counters are printed for hadrons/muons, electrons/photons and
low-energy neutrons (referring to importance biasing requested by BIASING
respectively with WHAT(1) = 1.0, 2.0 and 3.0, or = 0.0 for all). The number of
RR actually refers to äll particles which have not been splitte" (a particl
crossing a boundary between two regions of equal importance is submitted neither
to RR nor to splitting, but is counted as if it was a RR). Therefore, the
counters can be used to calculate the following quantities, useful as a guide to
set importances and weight windows:

    A = "N. of RR" + "N. of Sp" = total number of particles entering the region
    B = ("<Wt> in"_RR * "N. of RR") + ("<Wt> in"_Sp * "N. of Sp") = total
        weight of the particles entering the region
    B/A = average weight of the particles entering the region

By the way, these counters are useful not only to get the average
particle weights, but also to check the amount of splitting and of RR done
in each region: sometimes they may show that you too much splitting (or too
much RR) in some regions when compared to the neighbor ones.

A warning: in these counters the regions are indicated by their number, not
by their names. The correspondence between region numbers and names can be
found on the same main output, in the Geometry echo.


On Fri, 11 Oct 2013, Santana, Mario wrote:

> Yuanjie,
> Statistical weights are not always 1.0, otherwise WW would not be needed.
> Region based BIASING, but also any biased interaction, e.g. biased photo
> nuclear reaction, EMF biasing, LAM-BIAS, etc will alter the weight of the
> resulting particles and of its descendants. When you use WW you want to
> adjust the upper limit to about ~3-4 times the average weight and the
> lower limit to about 1/3-1/4 of the average weight in that given region,
> so that your window is not too narrow (excessive splitting/russian-rulete)
> or too wide (=useless). To obtain what is the ACTUAL average weight you
> could either use SCORE (just one card in input file, but somehow cryptic
> results), or mgdraw.f (in bxdraw). With the later option you can actually
> get the weight/energy distribution and whatever else you want to analyze
> how your weight is scattered, but it requires some programming.
> On a side note, you could also center your weight window around some
> INTENDED average weight (different from the actual one). By doing that you
> can spare having the region BIASING card.
> To conclude, you will see that you have a couple knobs to tune how the
> window acts as a function of energy (one of those is particle-dependant).
> Maybe you should not invest so much time getting the best optimized WW,
> but some basing filtering can help a lot.
> Mario
> On 10/11/13 12:04 AM, "???" <> wrote:
>> Dear fluka user
>> I want to use weight window to biasing particles. But I am confused by
>> the set of the lower and upper weight limit in ww-factor. I think the
>> normal paricle weight is 1.0, so the lower weight limit shoule be less
>> than 1.0 and the upper limit should be greater than 1.0. But in example 2
>> of the manual, the lower and upper weight limit are 13 and 120, which are
>> all above 1.0. Could someone explain to me how to set the lower and
>> upper weight limit in ww-factor?
>> Many thanks
>> Yuanjie

Alberto Fasso`
SLAC-RP, MS 48, 2575 Sand Hill Road, Menlo Park CA 94025
Phone: (1 650) 926 4762   Fax: (1 650) 926 3569
Received on Sun Oct 13 2013 - 15:48:37 CEST

This archive was generated by hypermail 2.3.0 : Sun Oct 13 2013 - 15:48:38 CEST