Failed to reach target for SMEFT interference generation

Asked by Archie Crossland

I am trying to generate 10k events for an interference term in SMEFT using the SMEFTsim_U35_MwScheme_UFO model. I have added a restriction to the model to improve generation time which uses massless quarks and sets all Wilson coefficients to zero except for Cw~. I process I am generating is
p p > mu+ mu- e+ e- j j QCD=0 NP^2==1
which gives around 15,000 diagrams. Madgraph will generate the events but I keep hitting a failed to reach target error after a second refine and instead of 10k events I am getting closer to 100. I have tried changing hard_survey to 1 &2 but with no improvement in event number, just increased runtime. I have also tried changing the dynamical scale choice but again with no improvement. I have yet to test all values of each with each other but considering my process this will take about 2-3 days per run. Interestingly for CP-even operators there is no such problem and I can generate events for CW and CHB with no problems, the problems seems to only be for odd operators.
Here is the restriction file,

######################################################################
## PARAM_CARD AUTOMATICALY GENERATED BY THE UFO #####################
######################################################################

###################################
## INFORMATION FOR SMINPUTS
###################################
Block SMINPUTS
    1 8.038700e+01 # MW
    2 1.166379e-05 # Gf
    3 1.179000e-01 # aS

###################################
## INFORMATION FOR MASS
###################################
Block MASS
    1 0.000000e+00 # MD
    2 0.000000e+00 # MU
    3 0.000000e+00 # MS
    4 0.000000e+00 # MC
    5 4.180000e+00 # MB
    6 1.727600e+02 # MT
   11 0.000000e+00 # Me
   13 0.000000e+00 # MMU
   15 0.000000e+00 # MTA
   23 9.118760e+01 # MZ
   25 1.250900e+02 # MH
## Not dependent paramater.
## Those values should be edited following analytical the
## analytical expression. Some generator could simply ignore
## those values and use the analytical expression
  22 0.000000 # a : 0.0
  21 0.000000 # g : 0.0
  9000005 91.187600 # Z1 : MZ
  9000006 80.387000 # W1+ : MWsm
  12 0.000000 # ve : 0.0
  14 0.000000 # vm : 0.0
  16 0.000000 # vt : 0.0
  9000007 172.760000 # t1 : MT
  9000008 125.090000 # H1 : MH
  24 80.387000 # W+ : MW

###################################
## INFORMATION FOR DECAY
###################################
DECAY 6 1.330000e+00
DECAY 23 2.495200e+00
DECAY 24 2.085000e+00
DECAY 25 4.070000e-03
## Not dependent paramater.
## Those values should be edited following analytical the
## analytical expression. Some generator could simply ignore
## those values and use the analytical expression
DECAY 22 0.000000 # a : 0.0
DECAY 21 0.000000 # g : 0.0
DECAY 9000005 2.495200 # Z1 : WZ
DECAY 9000006 2.085000 # W1+ : WW
DECAY 12 0.000000 # ve : 0.0
DECAY 14 0.000000 # vm : 0.0
DECAY 16 0.000000 # vt : 0.0
DECAY 11 0.000000 # e- : 0.0
DECAY 13 0.000000 # mu- : 0.0
DECAY 15 0.000000 # ta- : 0.0
DECAY 2 0.000000 # u : 0.0
DECAY 4 0.000000 # c : 0.0
DECAY 1 0.000000 # d : 0.0
DECAY 3 0.000000 # s : 0.0
DECAY 5 0.000000 # b : 0.0
DECAY 9000007 1.330000 # t1 : WT
DECAY 9000008 0.004070 # H1 : WH

###################################
## INFORMATION FOR SWITCHES
###################################
Block SWITCHES
    1 0.000000e+00 # linearPropCorrections

###################################
## INFORMATION FOR SMEFTCUTOFF
###################################
Block SMEFTcutoff
    1 1.000000e+03 # LambdaSMEFT

###################################
## INFORMATION FOR SMEFT
###################################
Block SMEFT
    1 0.000000e+00 # cG
    2 0.000000e+00 # cW
    3 0.000000e+00 # cH
    4 0.000000e+00 # cHbox
    5 0.000000e+00 # cHDD
    6 0.000000e+00 # cHG
    7 0.000000e+00 # cHW
    8 0.000000e+00 # cHB
    9 0.000000e+00 # cHWB
   10 0.000000e+00 # ceHRe
   11 0.000000e+00 # cuHRe
   12 0.000000e+00 # cdHRe
   13 0.000000e+00 # ceWRe
   14 0.000000e+00 # ceBRe
   15 0.000000e+00 # cuGRe
   16 0.000000e+00 # cuWRe
   17 0.000000e+00 # cuBRe
   18 0.000000e+00 # cdGRe
   19 0.000000e+00 # cdWRe
   20 0.000000e+00 # cdBRe
   21 0.000000e+00 # cHl1
   22 0.000000e+00 # cHl3
   23 0.000000e+00 # cHe
   24 0.000000e+00 # cHq1
   25 0.000000e+00 # cHq3
   26 0.000000e+00 # cHu
   27 0.000000e+00 # cHd
   28 0.000000e+00 # cHudRe
   29 0.000000e+00 # cll
   30 0.000000e+00 # cll1
   31 0.000000e+00 # cqq1
   32 0.000000e+00 # cqq11
   33 0.000000e+00 # cqq3
   34 0.000000e+00 # cqq31
   35 0.000000e+00 # clq1
   36 0.000000e+00 # clq3
   37 0.000000e+00 # cee
   38 0.000000e+00 # cuu
   39 0.000000e+00 # cuu1
   40 0.000000e+00 # cdd
   41 0.000000e+00 # cdd1
   42 0.000000e+00 # ceu
   43 0.000000e+00 # ced
   44 0.000000e+00 # cud1
   45 0.000000e+00 # cud8
   46 0.000000e+00 # cle
   47 0.000000e+00 # clu
   48 0.000000e+00 # cld
   49 0.000000e+00 # cqe
   50 0.000000e+00 # cqu1
   51 0.000000e+00 # cqu8
   52 0.000000e+00 # cqd1
   53 0.000000e+00 # cqd8
   54 0.000000e+00 # cledqRe
   55 0.000000e+00 # cquqd1Re
   56 0.000000e+00 # cquqd11Re
   57 0.000000e+00 # cquqd8Re
   58 0.000000e+00 # cquqd81Re
   59 0.000000e+00 # clequ1Re
   60 0.000000e+00 # clequ3Re

###################################
## INFORMATION FOR SMEFTCPV
###################################
Block SMEFTcpv
    1 0.000000e+00 # cGtil
    2 1.000000e+00 # cWtil
    3 0.000000e+00 # cHGtil
    4 0.000000e+00 # cHWtil
    5 0.000000e+00 # cHBtil
    6 0.000000e+00 # cHWBtil
    7 0.000000e+00 # ceWIm
    8 0.000000e+00 # ceBIm
    9 0.000000e+00 # cuGIm
   10 0.000000e+00 # cuWIm
   11 0.000000e+00 # cuBIm
   12 0.000000e+00 # cdGIm
   13 0.000000e+00 # cdWIm
   14 0.000000e+00 # cdBIm
   15 0.000000e+00 # cHudIm
   16 0.000000e+00 # ceHIm
   17 0.000000e+00 # cuHIm
   18 0.000000e+00 # cdHIm
   19 0.000000e+00 # cledqIm
   20 0.000000e+00 # cquqd1Im
   21 0.000000e+00 # cquqd8Im
   22 0.000000e+00 # cquqd11Im
   23 0.000000e+00 # cquqd81Im
   24 0.000000e+00 # clequ1Im
   25 0.000000e+00 # clequ3Im

###################################
## INFORMATION FOR CKMBLOCK
###################################
Block CKMBLOCK
    2 0.000000e+00 # CKMlambda
    3 0.000000e+00 # CKMA
    4 0.000000e+00 # CKMrho
    5 0.000000e+00 # CKMeta

###################################
## INFORMATION FOR YUKAWA
###################################
Block YUKAWA
    1 0.000000e+00 # ymdo
    2 0.000000e+00 # ymup
    3 0.000000e+00 # yms
    4 0.000000e+00 # ymc
    5 4.180000e+00 # ymb
    6 1.727600e+02 # ymt
   11 0.000000e+00 # yme
   13 0.000000e+00 # ymm
   15 0.000000e+00 # ymtau
#===========================================================
# QUANTUM NUMBERS OF NEW STATE(S) (NON SM PDG CODE)
#===========================================================

Block QNUMBERS 9000005 # Z1
        1 0 # 3 times electric charge
        2 3 # number of spin states (2S+1)
        3 1 # colour rep (1: singlet, 3: triplet, 8: octet)
        4 0 # Particle/Antiparticle distinction (0=own anti)
Block QNUMBERS 9000006 # W1+
        1 3 # 3 times electric charge
        2 3 # number of spin states (2S+1)
        3 1 # colour rep (1: singlet, 3: triplet, 8: octet)
        4 1 # Particle/Antiparticle distinction (0=own anti)
Block QNUMBERS 9000007 # t1
        1 2 # 3 times electric charge
        2 2 # number of spin states (2S+1)
        3 3 # colour rep (1: singlet, 3: triplet, 8: octet)
        4 1 # Particle/Antiparticle distinction (0=own anti)
Block QNUMBERS 9000008 # H1
        1 0 # 3 times electric charge
        2 1 # number of spin states (2S+1)
        3 1 # colour rep (1: singlet, 3: triplet, 8: octet)
        4 0 # Particle/Antiparticle distinction (0=own anti)

Why am I unable to generate events for specific Wilson coefficients, but have no issues with others, and is there a way to get my 10k events without increasing the runtime to a point that makes it logistically impractical.

Question information

Language:
English Edit question
Status:
Solved
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Solved by:
Olivier Mattelaer
Solved:
Last query:
Last reply:
Revision history for this message
Best Olivier Mattelaer (olivier-mattelaer) said :
#1

Hi,

My advise would be to try to use sde_strategy=2 if this is not the default. (if it is you can try sde_strategy=1 but I would expect no gain in that case).

Actually, the phase-space integrator is not designed to handle interference. This is a very complex topic and I have actually no clue how to design one phase-space integrator which is efficient in general.
The code that we provide for such integration is the one designed for definite positive matrix-element and we mainly hope that the machine learning layer used in our framework will be enough to adapt to the integrand.
For many case, it seems to be working fine but indeed in some other case the machine learning layer is not enough.

I hope that in the future we will have even more complex machine learning layer which should be able to adapt in more complex way that should allow us to better handle those kind of complex case, but for the short term if it does not work they will be no solution.

Cheers,

Olivier

Revision history for this message
Archie Crossland (acrossland) said :
#2
Revision history for this message
Archie Crossland (acrossland) said :
#3

Thanks Olivier Mattelaer, that solved my question.