Scale/PDF uncertainties

Hello Bjoern,

Which process are you using? I get reasonable scale variations when I change scalefact, both with and without matching.

Regarding the PDF, as you can see from the run log of the madevent channels (click on the cross section number and then again on the integration channel cross section number), the implemented PDFs are:
mrs02nl,mrs02nn,mrs0119,mrs0117,mrs0121,mrs01_j,mrs99_1,mrs99_2,mrs99_3,mrs99_4,mrs99_5,mrs99_6,mrs99_7,mrs99_8,mrs99_9,mrs9910,mrs9911,mrs9912,mrs98z1,mrs98z2,mrs98z3,mrs98z4,mrs98z5,mrs98ht,mrs98l1,mrs98l2,mrs98l3,mrs98l4,mrs98l5,cteq3_m,cteq3_l,cteq3_d,cteq4_m,cteq4_d,cteq4_l,cteq4a1,cteq4a2,cteq4a3,cteq4a4,cteq4a5,cteq4hj,cteq4lq,cteq5_m,cteq5_d,cteq5_l,cteq5hj,cteq5hq,cteq5f3,cteq5f4,cteq5m1,ctq5hq1,cteq5l1,cteq6_m,cteq6_d,cteq6_l,cteq6l1

So cteq6_m and cteq5_m need an underscore.

All the best,
Johan

Dear Johan

Thank you very much for your answer. The process I am using is:

generate p p > W+ b b~, W+ > l+ vl @0
add process p p > W- b b~, W- > l- vl~ @1
add process p p > W+ b b~ j, W+ > l+ vl @2
add process p p > W- b b~ j, W- > l- vl~ @3
add process p p > W+ b b~ j j, W+ > l+ vl @4
add process p p > W- b b~ j j, W- > l- vl~ @5

Regarding scale variation. When I use a scale variation of 4 I see a cross section change of about 6%. That is less than I expect... (?)
Thanks for the help with the PDFs.

Cheers
 Bjoern

Hello Bjoern,

Sorry for the delay in answer. This process is using 4-flavor matching (since the b quarks are here considered as massive final-state particles, not as partons in the proton), which is still at an experimental stage (it is fully functional, but scale choices still need to be treated with some care). To get a feeling for the full scale uncertainty, you need to also vary the alpsfact parameters, since this specifies the scale for gqq and ggg vertices in matched processes. Note that, for consistency, when you vary alpsfact, you should vary also the corresponding Pythia parameters (you can set them directly in the pythia_card.dat):
- PARP(64) (which sets the scale for ISR) should be set to the square of the alpsfact value.
- PARP(72), which gives Lambda_QCD for FSR. This value should be multiplied by 0.5 when alpsfact is 2, and by 2 when alpsfact is 0.5. The default value is 0.25 GeV.
Please see the Pythia manual for more information about these parameters.

All the best,
Johan

Dear Johan and Olivier,

I am experiencing the same problem, I vary the alpsfact and scalefact within the values 0.5, 1.0, 2.0 and I get the same xsec and events like when alpsfact=scalefact=1.

I am running in MG version 5.2.0.1 with this model https://cp3.irmp.ucl.ac.be/projects/madgraph/wiki/Models/ggHFullLoop, I also play with the dynamic scale, for this run test I modified manualy setscales.f like this http://www.desy.de/~stanescu/my-tmp/setscales.f (sum of m^2+pT^2 for the tops final decay products, the fermions, for both renormalization scale and factorization scale) and the following processes:

define vl = ve vm vt
define vl~ = ve~ vm~ vt~
define l+ = e+ mu+ ta+
define l- = e- mu- ta-
define p = g u c d s u~ c~ d~ s~ b b~
define j = p
generate g g > t t~ / h QED=99 QCD=99, ( t > b w+, w+ > l+ vl), ( t~ > b~ w-, w- > j j) @1
add process g g > t t~ / h QED=99 QCD=99, ( t > b w+, w+ > j j), ( t~ > b~ w-, w- > l- vl~) @2
add process g g > t t~ / h QED=99 QCD=99, ( t > b w+, w+ > j j), ( t~ > b~ w-, w- > j j) @3
add process g g > t t~ / h QED=99 QCD=99, ( t > b w+, w+ > l+ vl), ( t~ > b~ w-, w- > l- vl~) @4

I am running at parton level LO in Madgraph, no Pythia involved.

Some of the settings:

#*********************************************************************
# PDF CHOICE: this automatically fixes also alpha_s and its evol. *
#*********************************************************************
  lhapdf = pdlabel ! PDF set
#*********************************************************************
# Renormalization and factorization scales *
#*********************************************************************
  F = fixed_ren_scale ! if .true. use fixed ren scale
  F = fixed_fac_scale ! if .true. use fixed fac scale
  91.1880 = scale ! fixed ren scale
  91.1880 = dsqrt_q2fact1 ! fixed fact scale for pdf1
  91.1880 = dsqrt_q2fact2 ! fixed fact scale for pdf2
  1 = scalefact ! scale factor for event-by-event scales
#*********************************************************************
# Matching - Warning! ickkw > 1 is still beta
#*********************************************************************
  0 = ickkw ! 0 no matching, 1 MLM, 2 CKKW matching
  1 = highestmult ! for ickkw=2, highest mult group
  1 = ktscheme ! for ickkw=1, 1 Durham kT, 2 Pythia pTE
  1 = alpsfact ! scale factor for QCD emission vx
  F = chcluster ! cluster only according to channel diag
  .false. = pdfwgt ! for ickkw=1, perform pdf reweighting
  5 = asrwgtflavor ! highest quark flavor for a_s reweight
  .false. = clusinfo ! include clustering tag in output
#*********************************************************************
  .false. = cut_decays ! Cut decay products
  F = use_syst ! Enable systematics studies
# PDF sets and number of members (0 or none for all members).
  CT10nlo.LHgrid = sys_pdf # matching scales
  10800 = lhaid

Any idea why nohing changes in the output, if I vary the scalefact and alpsfact ?

Thank you,
Madalina

Btw, what I changed was:

initially:
scalefact=1 , alpsfact=1

then:
scalefact=0.5 , alpsfact=0.5
scalefact=0.5 , alpsfact=2.0
scalefact=2.0 , alpsfact=0.5
scalefact=2.0 , alpsfact=2.0

Hi,

You comment the multiplication by scalefact in your setscales.f file.
So this is normal that you do not have any dependence in that parameter.
Concerning alpsfact this parameter has only impact in presence of MLM merging.
So this is also normal that you do not observe any dependence in your case.

Cheers,

Olivier

> On Apr 18, 2016, at 12:08, Madalina Stanescu-Bellu <email address hidden> wrote:
>
> Question #206751 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/206751
>
> Madalina Stanescu-Bellu posted a new comment:
> Dear Johan and Olivier,
>
> I am experiencing the same problem, I vary the alpsfact and scalefact
> within the values 0.5, 1.0, 2.0 and I get the same xsec and events like
> when alpsfact=scalefact=1.
>
> I am running in MG version 5.2.0.1 with this model
> https://cp3.irmp.ucl.ac.be/projects/madgraph/wiki/Models/ggHFullLoop, I
> also play with the dynamic scale, for this run test I modified manualy
> setscales.f like this http://www.desy.de/~stanescu/my-tmp/setscales.f
> (sum of m^2+pT^2 for the tops final decay products, the fermions, for
> both renormalization scale and factorization scale) and the following
> processes:
>
> define vl = ve vm vt
> define vl~ = ve~ vm~ vt~
> define l+ = e+ mu+ ta+
> define l- = e- mu- ta-
> define p = g u c d s u~ c~ d~ s~ b b~
> define j = p
> generate g g > t t~ / h QED=99 QCD=99, ( t > b w+, w+ > l+ vl), ( t~ > b~ w-, w- > j j) @1
> add process g g > t t~ / h QED=99 QCD=99, ( t > b w+, w+ > j j), ( t~ > b~ w-, w- > l- vl~) @2
> add process g g > t t~ / h QED=99 QCD=99, ( t > b w+, w+ > j j), ( t~ > b~ w-, w- > j j) @3
> add process g g > t t~ / h QED=99 QCD=99, ( t > b w+, w+ > l+ vl), ( t~ > b~ w-, w- > l- vl~) @4
>
> I am running at parton level LO in Madgraph, no Pythia involved.
>
> Some of the settings:
>
> #*********************************************************************
> # PDF CHOICE: this automatically fixes also alpha_s and its evol. *
> #*********************************************************************
> lhapdf = pdlabel ! PDF set
> #*********************************************************************
> # Renormalization and factorization scales *
> #*********************************************************************
> F = fixed_ren_scale ! if .true. use fixed ren scale
> F = fixed_fac_scale ! if .true. use fixed fac scale
> 91.1880 = scale ! fixed ren scale
> 91.1880 = dsqrt_q2fact1 ! fixed fact scale for pdf1
> 91.1880 = dsqrt_q2fact2 ! fixed fact scale for pdf2
> 1 = scalefact ! scale factor for event-by-event scales
> #*********************************************************************
> # Matching - Warning! ickkw > 1 is still beta
> #*********************************************************************
> 0 = ickkw ! 0 no matching, 1 MLM, 2 CKKW matching
> 1 = highestmult ! for ickkw=2, highest mult group
> 1 = ktscheme ! for ickkw=1, 1 Durham kT, 2 Pythia pTE
> 1 = alpsfact ! scale factor for QCD emission vx
> F = chcluster ! cluster only according to channel diag
> .false. = pdfwgt ! for ickkw=1, perform pdf reweighting
> 5 = asrwgtflavor ! highest quark flavor for a_s reweight
> .false. = clusinfo ! include clustering tag in output
> #*********************************************************************
> .false. = cut_decays ! Cut decay products
> F = use_syst ! Enable systematics studies
> # PDF sets and number of members (0 or none for all members).
> CT10nlo.LHgrid = sys_pdf # matching scales
> 10800 = lhaid
>
> Any idea why nohing changes in the output, if I vary the scalefact and
> alpsfact ?
>
> Thank you,
> Madalina
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Hi Olivier,

Ah, good, thanks ! And I have to apply the scalefact for both the renormalization and factorization scales, right ? like this:

      rscale=0d0
      do i=3,nexternal
          rscale=rscale+pt(P(0,i))**2+dot(p(0,i),p(0,i))
      enddo
      rscale=dsqrt(rscale)
      rscale=rscale*scalefact

      q2fact(1)=0d0
      do i=3,nexternal
          q2fact(1)=q2fact(1)+pt(P(0,i))**2+dot(p(0,i),p(0,i))
      enddo
      q2fact(1)=q2fact(1)*scalefact
      q2fact(2)=q2fact(1)

Cheers,
Madalina

Hi,

I think so, yes.

Note that you do not need to perform any of this modification if you are using MG5_v2.3.0 (or higher)
since you have a switch in the run_card (dynamical_scale_choice) which allow to use the most common
scale choice:
https://answers.launchpad.net/mg5amcnlo/+faq/2014

This is much less error prone.

Cheers,

Olivier

> On Apr 18, 2016, at 13:51, Madalina Stanescu-Bellu <email address hidden> wrote:
>
> Question #206751 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/206751
>
> Madalina Stanescu-Bellu posted a new comment:
> Hi Olivier,
>
> Ah, good, thanks ! And I have to apply the scalefact for both the
> renormalization and factorization scales, right ? like this:
>
> rscale=0d0
> do i=3,nexternal
> rscale=rscale+pt(P(0,i))**2+dot(p(0,i),p(0,i))
> enddo
> rscale=dsqrt(rscale)
> rscale=rscale*scalefact
>
>
> q2fact(1)=0d0
> do i=3,nexternal
> q2fact(1)=q2fact(1)+pt(P(0,i))**2+dot(p(0,i),p(0,i))
> enddo
> q2fact(1)=q2fact(1)*scalefact
> q2fact(2)=q2fact(1)
>
> Cheers,
> Madalina
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.