Warning when generating photon+1/2j with aMC@NLO

Hi,

I am generating the following process:

generate p p > a j [QCD] @0
add process p p > a j j [QCD] @1

and adding a cut on the photon pt, in order to have binned samples: 100 - 250 GeV, 250 - 400 GeV, 400 - 650 GeV, and > 650 GeV.

Everything seems to run fine, but at the end of the output I see the following:

...
INFO: Idle: 0, Running: 1, Completed: 260 [ 4m 20s ]
INFO: Idle: 0, Running: 0, Completed: 261 [ 4m 23s ]
INFO: Collecting events
INFO:
      Summary:
      Process p p > a j [QCD] @0 ; p p > a j j [QCD] @1
      Run at p-p collider (6500 + 6500 GeV)
      Total cross-section: 2.544e+03 +- 7.6e+00 pb
      Ren. and fac. scale uncertainty: +9.9% -9.6%
      PDF uncertainty: +1.2% -1.2%
      Number of events generated: 100000
      Parton shower to be used: PYTHIA8
      Fraction of negative weights: 0.33
      Total running time : 9h 29m
  Number of loop ME evaluations (by MadLoop): 1745915
    Stability unknown: 0
    Stable PS point: 1744130
    Unstable PS point (and rescued): 1781
    Unstable PS point (and not rescued): 4
    Only double precision used: 1744130
    Quadruple precision used: 1785
    Initialization phase-space points: 0
    Reduction methods used:
      > CutTools (double precision) 1739702
      > IREGI 4428
      > CutTools (quadruple precision) 1781
      > Not identified (CTModeRun != -1) 4
  Total number of unstable PS point detected: 4 (0.00%)
    Maximum fraction of UPS points in channel process/SubProcesses/P1_gu_aug/GF2 (0.03%)
    Please report this to the authors while providing the file
    /afs/cern.ch/work/i/isdebruy/MonojetNLO/CMSSW_7_1_19/src/GJets/100To250/process/SubProcesses/P1_gu_aug/GF2/UPS.log

INFO: The /afs/cern.ch/work/i/isdebruy/MonojetNLO/CMSSW_7_1_19/src/GJets/100To250/process/Events/cmsgrid/events.lhe.gz file has been$

INFO: Events generated
...

The content of the log file is:

 ===== EPS # 1 =====
 mu_r = 128.41287500947337
 alpha_S = 0.12685260064356785
 MadLoop return code, pole check and accuracy reported 420 F 8.6590707492361765E-002
 helicity (MadLoop only) 0 1
 1/eps**2 expected from MadFKS= -1.4762372668860205E-005
 1/eps**2 obtained in MadLoop = -6.4369865664907168E-005
 1/eps expected from MadFKS= -1.8104945341383236E-005
 1/eps obtained in MadLoop = -9.9017904302286513E-005
 finite obtained in MadLoop = 5.4967384969100865E-005
 Accuracy estimated by MadLop = 8.6590707492361765E-002
 1 0.135571520462559E+03 0.000000000000000E+00 0.000000000000000E+00 0.135571520462559E+03 0.000000000000000E+00
 2 0.135571520462559E+03 -0.000000000000000E+00 -0.000000000000000E+00 -0.135571520462559E+03 0.000000000000000E+00
 3 0.129336928880558E+03 0.762839241948448E+02 -0.100258904959084E+03 0.292738118135472E+02 0.000000000000000E+00
 4 0.101870999727799E+03 -0.725281631163831E+02 0.715129191216263E+02 -0.180791017703410E+01 0.000000000000000E+00
 5 0.399351123167624E+02 -0.375576107846171E+01 0.287459858374578E+02 -0.274659016365131E+02 0.000000000000000E+00

Something similar (different subprocess) happens for all pt bins, except for the 250 - 400 bin. Do you have an idea what the problem is and if this means I can not use the generated samples? Thanks.

Cheers,
Isa