- aMC@NLO-MadGraph. Take V 2.6.1
- folders for yb^2 and ybyt terms
- message from Marius
I have already set them up according to the YR4 with LHCHXSWG recommendations.
In particular muR=muF=H_T/4 is set as default scales. Most importantly MG5_aMC
has changed its default shower scale from sqrt(s)-like to H_T/2. For bbH usually
a lower shower scale has proven useful. Since H_T/2 is already lower than the old
default we have decided to use H_T/2/Sqrt(2) as a central value, and then you can
do variations by Sqrt(2) up and down to get systematics of the matching uncertainties.
Since our extensive studies have been done with the old s-like shower scale, you
may also consider a different central value. A possible measure to find it is to
consider NLO+PS and NLO pT_bbH distribution and choose an upper bound for shower
scale variations which shows a satisfactory matching between them at high pT. In
any case I reckon that the provided recommendation fullfils this.
It would be very interesting to monitor whether with this choice the large differences
to POWHEG persist, or whether they are reduced for the acceptance and some observables.
(It shall be noted that, as before, the ybyt folder supports only running matched
to the shower no fixed order alone.)
* instrictions from Marius
You find there two folders ("bbH4FS_yb2","bbH4FS_ybyt") which determine the yb^2
and the yb*yt terms. They have to be run separately, but the results can be simply
added in the end (either the final distributions or already the event files).
Extract the .tar file in your MG5_aMC main folder (Use MG5_aMC 2.6.1). The bbH folders
created with v2.6.1 should be taken. First you compile MG5_aMC 2.6.1 with some dummy process:
Setup slc6_amd64_gcc530 compiler going to CMSSW_8_0_3 area and issue "cmsenv"
Go to MG5_aMC_v2_6_1 directory and run dummy process:
./bin/mg5
generate p p > h b b~ [QCD]
output bbH_test
exit
Delete bbH_test folder, copy bbH folders under main MG5 directory and go to bbH4FS_yb2 or bbH4FS_ybyt folder.
The following applies to both
folders bbH4FS_yb2 and bbH4FS_ybyt, go inside one of the two and:
First:
a) adjust mg5_path path in Cards/amcatnlo_configuration.txt
b) if want to use lhapdf (to access PDF4LHC_nlo_nf4 (92000)) add in amcatnlo_configuration.txt
lhapdf = /cvmfs/cms.cern.ch/slc6_amd64_gcc530/external/lhapdf/6.1.6-giojec7/bin/lhapdf-config
(see location of lhapdf with scram tool info lhapdf; setenv SCRAM_ARCH slc6_amd64_gcc530 - setup compiler used for 6.1.6. CMSSW_8_0_3)
c) comment fastjet setting
Run the process with
./bin/generate_events
Results (and event files) will be found in: Events/run_*
Inputs can be set in:
Cards/param_card.dat
Cards/run_card.dat
(Cards/shower_card.dat -- for shower inputs in case you run the shower via MG5_aMC)
The bottom Yukawa is MSbar renormalized, the internal bottom mass is on-shell.
mb(mb) input for Yukawa can be adjusted in param_card.dat. The particular choice of
mb(muR) can only be changed in the code (SubProcesses/reweight_xsec.f function
rwgt_muR_dep_fac, variable mbmuR) since it involves the running from mb(mb)->mb(muR).
The given default values were chosen after a comparison with Michael Spira and
should correspond to what he uses for his 4FS prediction. With these values we obtain
agreement with his results at the level of 1% (which is the quoted numerical
uncertainty of these numbers) when using a fixed scale (muF=muR=(mH+2mb)/4).
By default, I have set all the parameters in run_card.dat and param_card.dat such
that they should correspond to what we have chosen in our publication. The default
total inclusive results (muF=muR=H_T/4) should be (modulo ~<1% numerical uncertainty):
yb^2: 0.448 pb
yb*yt: -0.0365 pb
as can be read off table 1 of our paper.
Changing shower scale:
In the newer MG5_aMC versions there is no need anymore to hack the code. You can directly set the shower scale via the input param_card.dat:
0.3535534 = shower_scale_factor ! multiply default shower starting
! scale by this factor
I have set the default already to 1/Sqrt[2]/2 so that variations of it would imply setting it to 1/2 and 1/4.