How to produce the CBNT
Due to move from CMZ to CVS/SRT, it is not possible
to make uptodate cbnt.g file from cvs=> CBNT with ATLSIM is not supported
as long as I do not have an automatic cvs -> g tool
The combined ntuple can be produced when running the
reconstruction (in atlsim or atrecon environment) or from the RECB banks
saved from a previously run reconstruction job.
CBNT is now fully integrated into Atrecon. In general,
one should use the official executable $SRT_DIST/current/installed/$SRT_TARGET/bin/main_atrecon
or, if atrecon needs to be relink with different reconstruction packages
or personal code, the official installation procedure (with SRT/CVS) described
here,
(to be used in directory atrecon) will create a tree work/ XXX with the
executable. In both cases you need to have SRT variables defined (instructions
here).
Only if one wants the latest bug fixes, Atlfast, or run
within atlsim, or some more elaborate examples, one needs to use files
in ~droussea/public/cbnt/pro and below.
The full tree can be copied in the user own directory
by cp -R ~droussea/public/cbnt/pro pro.
Atlsim can be used in directory pro/atlsim, atrecon can
be used in pro/atrecon
Example input files are available in directory pro/event:
H100bb.O containing fully simulated H(100)->bb events ,
H100bb_RECB.O points to a handfull of these events reconstructed
(full detector reconstruction but no ipatrec, no pixlrec)
If a file seem to have disappear, the corresponding tape
must be staged again
pro/atrecon/atlas_h100bb.his contains the cbnt from a few
events (ignore first event which is empty)
Before creating a new ntuple, ZEBRA.P should be linked onto
one of the above files e.g (ln -s ../event/H100bb_RECB.O
ZEBRA.P)
Datacards example are available in /work/AtreconMain/datacard/
(from CVS)
The content of the ntuple can be adapted to the user
using switches and parameters in data cards DETP CBNT. A user routine has
also been provided.
Driving datacards and variables are described here
With Atlsim (WARNING obsolete as long
as no recipe to rebuilt cbnt.g from cvs)
In directory pro/atlsim
Make sure ZEBRA.P points to the right datafile
If ZEBRA.P points to an unreconstructed file, switch reconstruction
on in cbnt.kumac (DORECO=1, only xkalman will be run for time being)
Set the datacards DETP CBNT according to your requirements
(be careful : no space before the =, one space after the =, and numbers
should respect their declared types (no "." for integers))
then type:
atlsim -G 12
exe cbnt
(nt/plo 3333.xxx, nt/pri 3333, nt/scan 3333 ! ! ! ! xxx)
exit
histogram cbnt.his contains the ntuple number 3333 in directory
cbnt
"-G 12" is used to enlarge ZEBRA memory in Atlsim.
This maybe necessary when running on fully simulated events with calorimeters
and muon spectrometer.
With Atrecon
In directory pro/atrecon
Make sure ZEBRA.P points to the right datafile
if no cbntuser routine was used, the standard atrecon executable
can be used: ln -fs atrecon_standard atrecon
if atrecon was relinked (see below),
create a link to this executable
if ZEBRA.P points to an unreconstructed file, use datacard
reco_cbnt.databack (which switches on the full reconstruction (except
pixlrec and ipatrec))
if ZEBRA.P points to a reconstructed file, switch off all
reconstruction in the datacards (set all RECO to 1, except the one from
CBNT, set OUTP to 0, make sure you have the datacards *BKIO 'P' 'RECB')
Set the datacards *DETP 'CBNT' according to your requirements
(be careful: no space after the keyword, and all numbers are real ("."
is mandatory, even for integers)). Switch on the necessary reconstruction
packages.
type:
atrecon <reco_cbnt.databack >atrecon.out
file atlas.his contain the ntuple
to store RECB banks in output ZEBRA.O, change OUTP
0 to OUTP 1 in reco_cbnt.databack
The special case of atlfast
To use atlfast, the usual file atlfast.dat specifying atlfast
options should be present, and also a pointer to a "data" directory where
it will pick all the smearing parameters:
cp /afs/cern.ch/atlas/www/GROUPS/PHYSICS/HIGGS/ATLFAST-dir/dev/atlfast/data/atlfast.dat
atlfast.dat
ln -s /afs/cern.ch/atlas/www/GROUPS/PHYSICS/HIGGS/ATLFAST-dir/dev/atlfast/data
data
CBNTATLF the interface to atlfast is available both for atlsim
and atrecon.
Since release 0.0.37, the interface is available in
the official releases. Only one executable is available with atlfast :
atrecon_CXF (with calorimeter and xkalman.).
To have atlfast available in your own executable (see below),
GNUmakefile.in should be modified so that $(ATLFAST_LIB) appears between
$(ATLAS_LIB1) and $(ATLAS_LIB2) in the definition of main_atrecon_LIBS
(where "main_atrecon" can be replaced by any standard atrecon executable)
(as done for atrecon_CXF_LIBS)
For atlsim, it has to be compiled a special way to load dynamically
atlfast library. This is done with a makefile in directory cbnt/pro/atlsim/cbntatlf.
The directory where atlsim is run should contain a link to cbntatlf.sl
with the name cbntatlff.sl (Note the additional f to avoid untimely initialisation).
Modifying the user routine cbntuser
Cbntuser can be updated to perform personal analysis: driving datacards can be added, more variables added to the ntuple. All the banks are acessible (RECB, GENZ, KINE), the truth common (in $SRT_DIST/current/atrecon/include/atrecon/cbntcdes/cbntttrack.inc , which can be used by adding #include "atrecon/cbntcdes/cbntttrack.inc" ) allows easy access to truth information,
and also ATLFAST commons are available if switched on.
for atlsim pro/src/g/cbntuser.g should be edited
Switch on the user block in cbnt.kumac (USER.BASIC=
1) or one of the .databack file
Run ATLSIM as above
Note that in directory pro/atlsim cbntuser.g is linked with
the name cbntuserf.g to avoid untimely initialisation.
To relink atrecon executable with modified user routine:
- cd work/AtreconMain/src ; cp $SRT_DIST/current/atrecon/cbnt/cbntuser.age
.
- in work/GNUmakefile.in, add cbntuser.age to the list
of SRC files (e.g. atrecon_CX_SRC if you are using atrecon_CX)
- edit/modify cbntuser.age
- delete main_atrecon, rerun gmake as in installation
instructions
Read the Guidelines
for contributors for more information on the code itself
Comments/suggestions/assistance on this page or the CBNT itself David.Rousseau@cern.ch