sherpa is hosted by Hepforge, IPPP Durham
close Warning: Can't synchronize with repository "(default)" (/hepforge/svn/sherpa does not appear to be a Subversion repository.). Look in the Trac log for more information.
Modify

Opened 7 years ago

Closed 7 years ago

#371 closed task (fixed)

high weights in on-the-fly variations

Reported by: Johannes Krause Owned by: Stefan Hoeche
Priority: major Milestone: rel-2.3.0
Component: ME Generator Version: 2.2.1
Keywords: Cc:

Description

I recently performed some on-the-fly scale variations with a MEPS@NLO setup and encountered several events with large weights in the variations but not in the nominal. This is related to the local k-factor in the PertubativeInterface, where large k-Factors are vetoed for the nominal but not for the variations:

if (K==0.0 || dabs(K)>m_maxkfac) continue;

I have isolated some events which have such a high k-Factor and found several places in PHASIC++/Process/Single_Process.C where large weights can emerge. The first issue are ratios of pdf values which give very low values. This happens both in AddISR() but also in CollinearCounterTerms(). Stefan had encountered similar problems in DIRE and found a solution which I have implemented here, too. The second issue is a missing scale check for quark masses when reading pdf values.

All fixes are documented in the attached diff file which refers to the PHASIC++/Process/Single_Process.C of the 2.2.2 release.

Here are some validation plots for a Z\gamma + jets setup:

  • differential cross sections:

https://wwwpub.zih.tu-dresden.de/~s0118321/z+gamma/pdf_weights/central_comparison/MC_Z_DIRECT_E/index.html

  • variations before fixes

https://wwwpub.zih.tu-dresden.de/~s0118321/z+gamma/pdf_weights/variation_old/MC_Z_DIRECT_E/index.html

  • variations after fixes

https://wwwpub.zih.tu-dresden.de/~s0118321/z+gamma/pdf_weights/variation_fixed/MC_Z_DIRECT_E/index.html

Some of these changes might also affect large weights in nominal runs since the AddISR() method is not exclusively called by the localKfactor. However, there are still events with large weights present (e.g. in the leadinglepton pT plot).

Attachments (1)

pdf_ratio.patch (4.2 KB) - added by Johannes Krause 7 years ago.

Download all attachments as: .zip

Change History (2)

Changed 7 years ago by Johannes Krause

Attachment: pdf_ratio.patch added

comment:1 Changed 7 years ago by Frank Siegert

Resolution: fixed
Status: newclosed

Thanks Johannes, this has been committed to trunk (r30501) and rel-2-2-4 (r30502). But I have kept the maximal local kfactor condition as an additional safeguard.

Modify Ticket

Change Properties
Action
as closed The owner will remain Stefan Hoeche.

Add Comment


E-mail address and name can be saved in the Preferences.

 
Note: See TracTickets for help on using tickets.