Page MenuHomeHEPForge

EvtGen models with test output problems
Open, NormalPublic

Description

  • Lb2Baryonlnu has prob of -nan and forward density matrix full of -nan
  • EvtSSD_DirectCP has prob > probmax for B+ -> f_2 pi+ (always 18.581 instead of 10.0)
  • VTOSLL has prob > probmax for D*0 -> MyD0 e+ e- (various numbers)

Others:

BC_VNPI=VLL__Bc_psi2Spipipi_mumu.out:28
BHADRONIC__Bd_D2star0pi0.out:7
BHADRONIC__Bd_D2star0rho0.out:23
BHADRONIC__Bd_Dstar0rho0.out:16
BQTOLLLL__Bs_mumumumu.out:6255
BQTOLLLLHYPERCP__Bd_mumumumu.out:298
BSTOGLLISRFSR__Bs_gammamumu.out:2391
BToDiBaryonlnupQCD__Bu_Deltappbarmunu.out:3
BTOXELNU=VSS__Bu_rho0munu_pipi.out:3
SSD_DirectCP=TSS__Bu_f2pi+_pi+pi-.out:10000
VPHOTOVISRHI=VSS__psi4415_psi4160gamma_DD.out:1
VTOSLL=PHSP__Dst0_D0ee_Kpi.out:3343
X38722-+_PSI_GAMMA=VSS=VLL__X3872_omegaJpsi_pipi,mumu.out:11
X38722-+_PSI_GAMMA=VSS=VLL__X3872_rho0Jpsi_pipi,mumu.out:20

Event Timeline

tlatham created this object with visibility "Public (No Login Required)".
tlatham created this object with edit policy "Restricted Project (Project)".
jback renamed this task from Models with problems to EvtGen models with test output problems.Thu, Jan 26, 12:35 PM
jback triaged this task as Normal priority.
jback removed a subscriber: tlatham.

Result of the command grep "chiSq/nDof" *.out | grep -v "pValue = 0.000" > problem-histos.txt

  1. We should increase the tensor probability from 10 to 20 for EvtSSD_DirectCP (e.g. for B+ -> f_2 pi).
  1. The NaN values for EvtLb2Baryonlnu are occurring because it sets the spin density matrix components for the parent baryon using the momentum in the lab (mother) frame. If the parent baryon is the mother particle, then it has zero momentum and the cosTheta term (which divides by the momentum magnitude) becomes NaN. This could be resolved by only calculating cosTheta if the momentum is non zero, otherwise it will be initialised to zero:

    double cosTheta = pmag > 0.0 ? vector4P.get( 3 ) / pmag : 0.0;

    This removes the NaN values in the density matrix.
  1. The probability distribution for EvtVtoSll has a very long tail going beyond the defined maximum of 10.0 and so we get many max probability warnings when the k2 variable is very small (<0.001). We would need to add some compensation method to avoid this problem.

Many of the zero nDof histograms in problem-histos.txt occur when there is only 1 bin filled (the others are empty), e.g. a constant PDG mass value with essentially zero width or a fixed momentum in a 2-body decay. So we could add a check for non-zero entries within the histogram range if nDof = 0 in addition to the chiSq/nDof number.