CMS OO Pilot Projects an LHC Point of view

1
CMS OO Pilot Projectsan LHC Point of view

• Vincenzo Innocente
• CERN EP-CMC

2
CMS OO Pilot Projects

• Test-Beam DAQ and Analysis
• persistent focus on Objectivity and Irix
  Explorer
• Goal complete OO analysis
  chain
• Reconstruction and Analysis of GEANT3 events
• transient focus on user transition from
  FORTRAN to OO
• Goal all reconstruction
  software in C end 98
• GEANT4 simulation
• include persistency Goal full CMS simulation
  in G4 - 1.0
• Current effort is in consolidating the work done
• Not much real RD is going on at the moment

3
Licenses and Distribution

• Licensing scheme seems working
• The CMS computing group would like feedback from
  IT about CMS users
• Distribution could be made easier if tar files
  and/or installation scripts were provided
• Documentation is improving

4
Components in use

• CMS usage of LHC
• -Standard C Library Objectspace for STL and
  STD Helper and Time classes
• (We still have some RW, migrating to Ospace,
  ACE for Thread and communication)
• - ODBMS (Objectivity/DB) for the test-beam
  prototypes (version 4.0.2, version 5??)
• - HEPODBMS not yet used (CMS Solution)
• - CLHEP Geometry, Vector and Random
• - GEANT4 full CMS Barrel geometry implemented
  several test-beam set-ups
• working on Hits and persistency
• - HISTOOgrams used for the Test-beam prototypes
• - eventTags used in the Test-beam prototypes
• - HEPExplorer is used for Test-beams prototypes
• - Nag C library start to be used (binomial
  random number generator)
• - Gemini not used yet
• - HepVis and HepInventor for the event
  visualization

5
(Un)Portability

• Struggling with
• Operating Systems (Solaris, HP-UX, Linux,
  W/NT)
• Compilers (CC-4.2, aCC,
  egcs, VC45)
• Third-party Software (Objy, OSpace, GL, OIV, IE
  HEP-LHC)
• Test-beam software is working under Solaris using
  LHC 1.0 configuration (Objy-4.0.2) which is
  essentially not supported on any of the other 3
  platforms.

6
LHC (Geant4, Objectivity and HepExplorer) in CMS

• A simple
• program is able
• to read the hits
• information from
• the ODBMS and
• produces some
• first plots
• Next stepHits to
• digi (ready the
• transient part).
• We are waiting
• Objectivity V5
• for HP (aCC)

7
LHC (Objectivity and HepExplorer) in CMS

• Components of the Test-Beams Analysis Chain
• Online (event data recording )
• DAQ (H2, X5B and PS) / Conversion of Zebra files
  (X5B)
• Objectivity formatter (It creates Databases and
  Containers and the event structure)
• Central Data Recording (CDR)
• Data Quality/Online Monitoring
• Offline (event data processing)
• Reconstruction Framework
• Interface to Geant4 simulation
• Interactive analysis environment
• Data Browser
• Histogrammer and Histogram visualizer

8
Online Analysis/Monitoring Tools
Raw Data Distribution for a single Silicon
Detector. This is produced using HISTOO and
HepExplorer for the visualization
9
Scripting requirements from online monitoring

• Display a set of histograms in a predefined order
  from a given source
• example
• produce the previous plot for the current run
• superimpose it to a reference one (with proper
  normalization)
• add some statistical/monitoring info text
  (dead/noisy channels etc.)
• do it for all other channels
• Repeat it, without restarting the application, on
  the same or another source
• Display another set of predefined histogram
  without restarting the application

10
Interactive Analysis
Goal Use Reconstruction and Analysis software as
a plug-in for the interactive framework
Requires full support of shared libraries for
any of the used software

• Collection of type understood by the Framework
  ? Tag
• cast to a CMS type
  ? CMSEvent
• Return a type understood by the Framework
  ? double

11
Next Milestone

• Objectivity-only DAQ in a test-beam at PS
• Focus on Reliability
• Use Objectivity 4.0.2
• No new development in current software
• Upgrade to Objectivity 5 after data-taking

12
Conclusions

• Technology seems available
• in the Graphics and GUI area is moving fast (as
  usual)
• BUT Standards are not always such
• CMS medium term strategy
• focus on current needs
• stay on mainstream
• follow the evolution of the market
• continue RD on leading edge technologies and
  challenging areas (distributed databases,
  interactive analysis, event display)