Private sample generation for measurement studies |
|
|
Madgraph/Madevent configuration
To build the stand-alone code that will take care of generating the matrix element based events for the
processes one must register as a MG/ME user
here and, after that, upload the process files that describe the events to generate. You can find some examples below.
The server takes care of generating stand-alone code which you can download at anytime and install locally. The
MG/ME main web page grants access to a personal database with the processes you have generated. You can access it at anytime.
After generating the code download it and compile it following the instructions. To control the generation process a run card is needed. In this card the phase space for the event generation is defined along with other parameters. You can find examples for run cards below.
The output of the generators is a file in the Les Houches format which is used as in input to PYTHIA which will take care of showering the partons (fragmentation+hadronization), adding the initial and final state radiation and other concurrent interactions, decaying the unstable particles, etc. It also takes care of matching the partons generated by the matrix elements with the ones generated with the parton shower algorithm.
Control of the generated samples
The control of the matched samples (after PYTHIA generation) can be done using a tool called
MatchChecker provided
by the
Madgraph
team. To download it from
CVS
to your local
Madgraph
directory you can do the following:
export CVSROOT=:pserver:anonymous@cp3wks05.fynu.ucl.ac.be:/usr/local/CVS
cvs co MatchChecker
cd MatchChecker
make
You can run the
MatchChecker.sh
script giving a steering card as input.
An example of such steering card is the following
MatchCheckerCard.dat.
To run the tool just type
./MatchChecker.sh MatchCheckerCard.dat
This will produce the control plots under
MatchChecker/test
directory.
Expand the following sections to see a more detailed description of the samples we have generated and which configurations were chosen, depending on the CMSSW
version in use.
2_1_x Samples
More Less
Summary
Samples are stored at /lustre/lip.pt/data/cmslocal/pedros/Madgraph/
and were generated with CMSSW_2_1_9
.
The table below summarizes the main characteristics of the files.
SAMPLES SUMMARY |
File |
Process |
Matching |
Decays |
Cross section (pb) |
|
Obs. |
Madgraph-tt3j_inc_run1-200810241009-CMSSW_2_1_9.root |
inclusive |
at 40 GeV |
, , |
314.0 |
2958 |
Fastsim, no pile-up |
Madgraph-tt3j_inc_run1-200810241049-CMSSW_2_1_9.root |
inclusive |
at 40 GeV |
, , |
312.6 |
2944 |
Fastsim, pile-up=5.0 |
Madgraph-tt3j_inc_run1-200810241642-CMSSW_2_1_9.root |
inclusive |
at 30 GeV |
, , |
|
6000 |
Fastsim, no pile-up |
CMSSW
configuration
Validation
1_6_x samples
More Less
Summary
For our studies we have produced the parton samples for using Madgraph/Madevent. The table below shows the average time spent in generating the matrix elements for 20 kEvents with Madgraph running on a single processor.
From the parton level samples we generate 3 different scenarios for the top decay, in the context of the standard model:
- All b : 100%
- Mix : 90% , 10% (q=d,s)
- All q : 100% (q=d,s)
The 3 scenarios are simulated using PYTHIA6
+ FastSim
( CMSSW_1_6_9
) and are described below:
PRIVATE SAMPLES USED FOR Vtb STUDIES (GEN+SIM+DIGI+RECO) |
N(jets) |
Jet matching |
Cross section (pb) |
N(events) |
Luminosity (1/pb) |
Scenario |
Sample |
Total Size (GB) |
0 |
exclusive |
11.00 |
|
|
All b |
tt0j_allb_dil_mgme-pythia-CMSSW_1_6_9-FASTSIM-GEN-SIM-DIGI-RECO_{1,...,10} |
9.53 |
|
|
All q |
tt0j_allq_dil_mgme-pythia-CMSSW_1_6_9-FASTSIM-GEN-SIM-DIGI-RECO_{1,...,5} |
5.07 |
1 |
exclusive |
7.97 |
|
|
All b |
ttj_allb_dil_mgme-pythia-CMSSW_1_6_9-FASTSIM-GEN-SIM-DIGI-RECO_{1,...,6} |
7.16 |
|
|
All q |
ttj_allq_dil_mgme-pythia-CMSSW_1_6_9-FASTSIM-GEN-SIM-DIGI-RECO_{1,...,6} |
6.83 |
2 |
exclusive |
3.53 |
|
|
All b |
ttjj_allb_dil_mgme-pythia-CMSSW_1_6_9-FASTSIM-GEN-SIM-DIGI-RECO_{1,..,4} |
4.38 |
|
|
All q |
ttjj_allq_dil_mgme-pythia-CMSSW_1_6_9-FASTSIM-GEN-SIM-DIGI-RECO_{1,...,4} |
4.23 |
3 |
inclusive |
2.10 |
|
|
All b |
ttjjj_allb_dil_mgme-pythia-CMSSW_1_6_9-FASTSIM-GEN-SIM-DIGI-RECO_{1,...,3} |
4.95 |
|
|
All q |
ttjjj_allq_dil_mgme-pythia-CMSSW_1_6_9-FASTSIM-GEN-SIM-DIGI-RECO_{1,...,3} |
4.81 |
All these samples (table 1.1 and 1.2) are stored in CASTOR at CERN: /castor/cern.ch/user/p/psilva/ttbar/madgraph
.
The run card used to generate these samples can be retrieved from here.
PYTHIA 6 configuration
In order to work properly with the matrix elements generated for the hard process
the generator needs to activate the algorithm that matches the jets generated
by the matrix element with the ones generated by the parton showers.
PYTHIA6
provides such a mechanism and we make use of the version provided by the Madgraph
group.
The version can be downloaded from this link.
Note: it has to be installed under the Madgraph
directory.
Our samples consist of dilepton channel. As so we disable all the branching ratios for the W except those to electrons, muons or taus. The taus are forced to decay leptonically (either electron or muon plus neutrinos).
Taus are decayed with Tauola
. A simple solution to force the leptonic decays is to comment all widths except
the leptonic ones:
- open
pythia-pgs/libraries/pgs/src/tauola.f
- go to routine
tauola_init
(line 230)
- comment gamprt(3) to gamprt(8) (lines 335 to 342)
There are 2 ways to force the top quark to decay with different braching ratios in PYTHIA 6
(according to T. Sjostrand):
- "You change the
VCKM
matrix in commonblock PYDAT2
, which then takes effect in the internal calculation"
- "You set
MWID(6)=2
(in commonblock PYINT4
) to use the PYUPDA
input for top rather than the internal calculation."
The different decay scenarios were simulated using the second option. For that purpose we have created 1 private particle data tables which define the 3 different top decay scenarios in the standard model : PDTable.dat. In order to read this table from within PYTHIA 6
the following piece of code is needed in pythia-pgs/src/pythia.f
, just before calling PYINIT
.
C... Open and read Particle data file
PRINT *,'Opening particle data table'
OPEN(78,file="PDTable.dat")
CALL PYUPDA(2,78)
PRINT *,'Read particle data table from file, closing now'
CLOSE(78)
C... Force the top quark to decay according to the settings of the particle data table
CALL PYGIVE('MWID(6)=2')
After hacking tauola.f
and pythia.f
you can compile typing make
inside the pythia-pgs
directory.
To steer the simulation one has to take into account if the matrix elements were simulated for an inclusive or exclusive
jet multiplicity bin and also if the matching algorithm should use or ordered showers.
We use the following card to steer PYTHIA 6
: * PythiaCard.dat.
As the ouput of PYTHIA6
is in HEP
binary format file we make use of the HEP
to HepMC
converter provided in this link for HepMC V2.01.08
(more info in here).
To use plain ascci files in HepMC
format as input for CMSSW
we have followed the instructions given in here. However in order to be compliant with the input expected by CMSSW
, somepatches were needed in the IO_Ascii
class of the HepMC
library... The patches are the following:
- Change the following code
HepMC/IO_Ascii.h:124: int m_eventCtr; //private event counter
src/IO_Ascii.cc:16: : m_mode(mode), m_file(filename, mode), m_finished_first_event_io(0),m_eventCtr(1) //init also the event counter
src/IO_Ascii.cc:63: // m_file << "\n" << "HepMC::Version " << versionName() << "\n"; // disable version printing
src/IO_Ascii.cc:64: m_file << "HepMC::IO_Ascii-START_EVENT_LISTING\n"; // remove escape character before
src/IO_Ascii.cc:71: if(evt->vertices_size()==0) return; //ignore null events
src/IO_Ascii.cc:74: output( m_eventCtr++ ); //use private event counter
src/IO_Ascii.cc:351: //output( p->momentum().m()); //do not print mass
- add the following symbolic links under the
HepMC
installation directory:
ln -s ./ include
ln -s src/ lib
- compile the patched version of
HepMC
- compile the converter tool.
You are now ready to convert HEP
files to HepMC
format and give it as input to CMSSW
. To use the converter just do:
convertStdHep -i input_HEP_file.hep -o output_HepMC_file.dat
CMSSW
configuration
The samples produced with the fast simulation were produced using CMSSW_1_6_9
and the template configuration file included in the TopQuarkAnalysisFramework
to produce fast simulation samples from scratch (TopQuarkAnalysis/Examples/test/TtbarRECO_fromScratch_fast.cfg
). You can pick our configuration file from here.
In order to generate HLT information from the Madgraph samples we make use of two configuration files that are parte of the HLTrigger/Configuration
package. These files take care of generating the RAW data from the Digis and of generating the L1 and HLT information from that:
- to generate the raw:
RelVal_Digi_DigiToRaw.cfg
- to generate the L1 and HLT info:
RelVal_HLTFromRaw.cfg
For more details look at: HLT tutorial and HLT table.
Validation
--
PedroSilva - 27 Mar 2008