Slight cross section difference between Amegic and Comix in photon+2jet

[Frank Siegert]

The photon + 2 jet cross section is different by >10 sigma between Amegic and Comix when using this fairly standard Run.dat:

(run){
  EVENTS = 5000
  WRITE_MAPPING_FILE = 3
  FRAGMENTATION = Off
  ME_SIGNAL_GENERATOR = Comix
}(run)

(processes){
  Process 93 93 -> 22 93 93{1}
  Order_EW 1
  CKKW sqr(30.0/E_CMS)
  Integration_Error 0.001 {3}
  End process
}(processes)

(selector){
  PT             22  15.0  E_CMS
  PseudoRapidity 22  -6.0  6.0
  DeltaR         22  93  0.3  100.0
}(selector)

(beam){
  BEAM_1 = 2212
  BEAM_ENERGY_1 = 3500.0
  BEAM_2 = 2212
  BEAM_ENERGY_2 = 3500.0
}(beam)

I get for the photon + 2 jet process the following results:

Amegic: 2_3__j__j__P__j__j : 45528.3 pb +- ( 45.4327 pb = 0.0997901 % )  exp. eff: 0.104771 %
Comix:  2_3__j__j__P__j__j : 46642.9 pb +- ( 46.542 pb = 0.0997837 % )  exp. eff: 0.0226022 %

Also in both, Comix and Amegic, the longer I integrate by lowering the integration error, the lower the "exp. eff." value gets, which results in a significantly longer event generation time.

[Frank Siegert]

This can be explained by the different ways that the parton separation criterion is applied to Amegic and Comix: Comix produced colour sampled matrix elements such that the parton separation criterion can take colours into account when cutting on potentially logarithmically enhanced parton combinations. For Amegic this is not possible, it will cut on all combinations. So this is a feature and not a bug, and thus I will close it.

[Stefan Hoeche]

Milestone old deleted