Hi Costas,
TGenPhaseSpace::SetDecay is overestimating the maximum weigth for your process. This slows down the rejection technique used in TGenPhaseSpace::Generate, but the results should be correct.
If somebody provides an optimisation, I will be happy to include
it in CVS.
Note that you can slightly speed up your script by running it
with ACLIC. This is explained in your script with some comments
in the attachement
Rene Brun
On Wed,
10
May 2006, Costas Andreopoulos wrote:
> hello,
>
> at the GENIE neutrino generator http://www.genie-mc.org/ we are using
> TGenPhaseSpace as part of a KNO-based hadronization model, eg see:
> http://hepunx.rl.ac.uk/~candreop/generators/GENIE/devel/src/Fragmentation/KNOHadronization.cxx
>
> When I run the generator in a mode producing *unweighted* events
> (after selecting multiplicity and final state particles)
> I use the rejection method and based on the weight reported by each
> TGenPhaseSpace::Generate() I try to select an 'unweighted decay'.
>
> The problem is that for events with:
> (invariant mass) - (sum of final state particle masses) = (small)
> TGenPhaseSpace reports a *huge* maximum weight:
> many orders of magnitude away from the maximum weight one would
> *ever* get by running the TGenPhaseSpace::Generate().
>
> This can be seen in
> http://hepunx.rl.ac.uk/~candreop/outbox/phaseSpace.gif
> The red line corresponds to the max weight as reported by TGenPhaseSpace
> *A macro to reproduce this plot is attached.*
>
> For our generator this means that very few events (in the edges of
> our 'phase space' but still physical ones) take disproportionally
> long time to be generated - screwing the overall performance.
>
> I can always search for the maximum weight myself, but I would like
> to understand better why this happens: Do I use TGenPhaseSpace
> incorrectly? a bug? a limitation?
>
> cheers,
> Costas
>
>
This archive was generated by hypermail 2.2.0 : Mon Jan 01 2007 - 16:31:58 MET