VBFNLO's processes have been interfaced via the BLHA accord to the Herwig 7 event generator. The Matchbox module of Herwig 7 is able to perform the NLO matching using the subtractive (MC@NLO-type) and multiplicative (Powheg-type) matching to NLO.
Further to this, all processes can also be used for the (N)LO multi-jet merging as outlined in. Herwig 7 can also be used to perform fixed-order calculations with the VBFNLO matrix elements, if this eases comparison to the matched or merged simulations. Matching and merging systematics can then be addressed. All VBF and VBS processes including leptonic decays, but not the semileptonic modes, have been included.
Note that all the VBF processes are computed in the VBF approximation only, i.e. \(s\)-channel contributions and interference between \(t\)- and \(u\)-channel diagrams, which both are tiny in the VBF region, are neglected. BLHAv1 stipulates that all Born matrix elements are calculated by the Monte Carlo program. Therefore, to prevent inconsistencies, care must be taken that the VBF approximation is also present in the Born matrix elements. In version 2 of the interfaceall matrix elements are provided by VBFNLO, so in that case this is ensured automatically.
We further note that the parameters such as couplings, masses and widths are supplied as set in the Herwig input files and that cuts are applied by the Matchbox module, as well. Wihtout additional adjustments these different setup modes might result in small differences in between the results obtained from VBFNLO standalone as compared to those produced from Herwig. An integrator module resembling VBFNLO's Monaco algorithm is also available for Matchbox.
Explicit tests of the interface, using a 'tuned' comparison for which all parameters have been made equal for the VBFNLO and Herwig runs have been performed with Herwig 7. Example input files are provided in the directory regress/runs/BLHA of the source code. Using the BLHA interface and a corresponding Monte Carlo program also allows one to write out VBFNLO events at NLO. The corresponding setup for Herwig is also shown in the example input files mentioned above, and further examples are provided online within the Herwig tutorials.
If Herwig 7 is installed using the herwig-bootstrap, VBFNLO will be
directly available. To use a local build of VBFNLO instead, the flag
--with-VBFNLO=PATH can be passed to the
herwig-bootstrap or during the configuration step in the Herwig
compilation.
To use VBFNLO as an amplitude provider, the line
read Matchbox/VBFNLO.in has to be included in the
Herwig input file. The cuts for the process need to be set in Herwig
and are automatically passed to VBFNLO. If a Frixione Isolation cut
has to be set, the line
set PhotonIsolationCut:CutType VBFNLO will need to be
provided. The VBF processes inside VBFNLO are calculated in the VBF
approximation. Telling Herwig to not include \(s\)-channel
contributions, which are ignored in the VBF approximation, can be
done with read Matchbox/VBFDiagramsOnly.in.
The BLHA states that the MC has to drive the process generation
while the OLP has to give the amplitudes asked. In the interface
between VBFNLO and Herwig, an extension allowing the use of the
VBFNLO phasespace has been implemented. To use the VBFNLO phasespace
generator, which improves the convergence, we have to add
read Matchbox/VBFNLOPhasespace.in to the Herwig input
card.
A summary of the relevant lines concerning VBFNLO with a brief
description of their effect may be found at the end of this page. A
Herwig input card, ready to use VBFNLO as an amplitude provider, can
be found in the share/VBFNLO folder. It is recommended to
start from that input card, which can be copied to the run folder,
and make the appropriate changes for the required process. For
Herwig to be able to access the libraries required and to be able to
call the Herwig command from any folder, the pertinent environment
variables must be set. If the installation was done with the
herwig-bootstrap, it can be done with
source $INSTALLDIR/bin/activate.
The process is built with the command line
Herwig build {inputcardname}, which creates a
{runfilename} file that is used to generate the events.
Before the generation step, the phasespace grid has to be optimized
with Herwig integrate {runfilename}. This step can be
parallelized using the flag
--maxjobs={numberofjobs} which will create integration
jobs that can be integrated separately with
Herwig integrate --jobid=n {runfilename}, where
n is the identification number assigned to the
integration job, and merged afterwards with
Herwig mergegrids. Then, the event generation is
started with Herwig run {runfilename}.
As an example, if our input card is named LHC-Matchbox.in and creates a runfile named LHC.run, the following commands
Herwig build --maxjobs=10 LHC-Matchbox.in
for i in $(seq 0 9); do Herwig integrate --jobid=\$i LHC.run &
done
Herwig mergegrids
Herwig run LHC.run
will create 10 different integration jobs, merge the grids after the integration has finished and make an event generation run.
The comparison with VBFNLO standalone can be performed by turning off the parton shower, hadronization and MPI in Herwig, and using the same cuts and parameters in both programs. To help with the possible difficulties with the last step, curated input cards for VBFNLO and Herwig can be found inside VBFNLO/regress/runs and VBFNLO/regress/runs/BLHA respectively.
read Matchbox/VBFNLO.inset PhotonIsolationCut:CutType VBFNLOread Matchbox/VBFDiagramsOnly.inread Matchbox/VBFNLOPhasespace.in