generator services review and planning meeting
DESCRIPTION
Generator Services review and planning meeting. Witek Pokorski 25 .04.2012. Outline. General news Progress reports and plans GENSER and validation collaboration with LPCC ( MCPLOTS, P. Skands ) HepMC Summary Discussion. General news. - PowerPoint PPT PresentationTRANSCRIPT
1
Generator Services review and planning meeting
Witek Pokorski
25.04.2012
2
Outline
General news Progress reports and plans
GENSER and validation collaboration with LPCC (MCPLOTS, P. Skands) HepMC
Summary Discussion
3
General news new mailing list for all Generator Services
announcements [email protected]
please subscribe
MCDB subproject stopped in the context of Generator Services CMS was the only user mid 2011 CMS decided to migrate away from MCDB with no production use from any of the experiment,
our support for MCDB has been dropped project still running but not in the context Generator
Services
4
GENSER - Progress report (1/2) moved to SVN completed (internal scripts) ~50 new generators/versions installed over the
last year experiments regularly requesting new generators
installation done according to the needs
platforms i686-slc5-gcc43-opt, x86_64-slc5-gcc43-opt x86_64-mac106-gcc42-opt, i386-mac106-gcc42-op
builds for MacOSX still a bit of a problem
5
GENSER - Progress report (2/2) new degree of freedom needed to be introduced: version
of HepMC MCGenerators_hepmc2.03.11, MCGenerators_hepmc2.06.05,
06, 07, 08 LCGCMT configuration with generator interfaces pointing
to MCGenerators_HepMC2.06.XX for CMT users configurable using $(HepMC_config_version) variable
rpath issue has been raised in the context of CVMFS problems seems to come from autotools (autolib)
Anton K. has a script which solves it (used successfully in MCPLOTS)
issue of _dbg platforms (links) recently raised
6
GENSER – Open questions
when do we drop 2.03.11? how many versions of HepMC to support? can we come up with an 'official' list of
generators we build on MacOSX? what do we do with rpath?
can we consider applying Anton's script (used in MCPLOTS) as a solution?
can we handle _dbg platforms at the level of CMT requirement files?
7
GENSER - Plans
continue installation of new generators/versions implement tests for new generators internal for GENSER:
organization of _hepmc trees separation in AFS volumes
finalization of MCGenerators_hepmcxxx installation tool
8
Validation - Progress report regression tests available for more than 90% of
generators tests results linked from the table of generators HepMC Analysis Tool based regression testing now done
for the new versions of several most important generators some improvements needed in the generating script need to review the dependency on MCPLOTS scripts
physics validation done very extensively (and very well) by MCPLOTS Rivet-based analysis for validation against LHC (and older) data
9
MCPLOTS (P. Skands)
very nice collaboration between Generator Services and LPCC
10
Validation - Plans implement missing regression tests for
recently added generators extend existing regression tests for better
coverage improve GENSER interface to HepMC
Analysis Tool continue testing and checking, but main
validation done by MCPLOTS (P.Skands) LPCC funding very appreciated
11
HepMC – StatusLynn Garren
2011-12-06 HepMC 2.06.06 Add the ability to build with cmake (preferred for MacOSX) Remove fio/PythiaWrapper.cc and inline getPythiaCrossSection (bug #88376) Move Pythia and Herwig examples to examples/fio example_BuildEventFromScratch does not use CLHEP vectors
2012-01-11 HepMC 2.06.07
Cmake is now the preferred build method for all platforms. fix a typo in the error report for GenEvent::use_length_unit (sr #125496)Local variables in some headers have been renamed to avoid warnings when compiling with -Wshadow: GenParticle.h, GenVertex.h, PdfInfo.h, SimpleVector.h, SimpleVector.icc (sr #125483) add examples/pythia8
2012-02-16 HepMC 2.06.08
because of requested changes in 2.06, IO_GenEvent::use_input_units() was not setting units for an input stream. This bug affected only 2.06.x and is now fixedGenEvent::define_units() can be used when the units in an event do not match the default units. Define_units() does not convert momenta or length.Add more testing of the Units functionality.
12
HepMC Outstanding requests
Reduce size of events by selecting only "interesting" particles. Have to make sure remaining particles are attached to the appropriate vertex. Some vertices may become superfluous. Otherwise, fairly straightforward
Rework HepMC data members so they play nicely with root. Definitely can be done. Exploring options.
Lynn Garren
13
Milestones from last meetingDONE
DONE
DONE
GENSER_1 01/06/2012 include new versions of supported generators
GENSER_2 01/08/2011 lightweight bootstrap script
GENSER_3 01/07/2011 migrate GENSER CVS to SVN
VALIDATION_1 01/12/2011extend HepMC analysis tests to other generators
VALIDATION_2 01/08/2011 add missing tests for new generators
HEPMC_1 01/06/2012 implement changes for ROOT I/O
MCDB_1 01/12/2011 (consider to) move to different storage system
MCDB_2 01/06/2012 (consider to) move to Oracle DB
ONGOING
DONE
DONE + RIVET/MCPLOTS
14
Proposed milestonesGENSER_1 01/06/2013 include new versions of supported generators
GENSER_2 01/06/2013finalize MCGenerators_hepmcxxx tree installation tool
GENSER_3 30/06/2012 organization of hepmc tree
VALIDATION_1 01/08/2012 review of HepMC Analysis GENSER interface
VALIDATION_2 01/08/2012implement missing tests for new generators
VALIDATION_3 01/12/2012 extension of existing regression tests
HEPMC_1 01/11/2012 implement changes for ROOT I/O
HEPMC_2 01/09/2012 implement mechanism for 'reduced size events'
15
Manpower situation technical student Anton P. finished his studentship end of
January 2012 current situation
LCG Russia: 16 months FTE per year WP: 0.1 FTE L. Garren – HepMC, external contribution from
FERMILAB this is the minimum needed to run routinely Generator
Services very valuable support from LPCC for the development of
MCPLOTS
16
Summary project running routinely
GENSER stable testing constantly extended
collaboration with MCPLOTS (P. Skands) testing of beta versions of generators comparison to LHC data
looking forward to definitely retire HepMC 2.03 ROOT I/O developments in HepMC ongoing