LCG MCDB STATUS REPORT FOR MC4LHC WORKSHOP

ABSTRACT

The adoption of a central database of MC events is motivated by simulation needs which are specific to the high energy physics domain. In general, the correct MC simulation of complex processes requires a rather sophisticated expertise. Often, different physics groups in various experimental collaborations approach the same experts and authors of MC generators, respectively, in order to generate MC samples for a particular physics process. Having these events stored in a public place along with the corresponding documentation, allows for direct cross checks of the performances on reference samples, and prevents possible waste of precious human and computing resources.

MAIN LCG MCDB FEATURES

• Powerful WEB interface with Content Management System for the authors of event samples and their users
  
• SQL structure of event sample documentation
  
• Power SQL/XML based search engine
  
• CASTOR as the native storage for the MC samples
  
• Direct uploading of multiple files from AFS / CASTOR / GRID to LCG MCDB
  
• Direct downloading of files from LCG MCDB with WEB / CASTOR / GRID paths (URI)
  
• Flexible and reliable authorization system based on CERN AFS/Kerberos logins or LCG GRID certificates
  
• BackUp of samples and SQL information
  
• API to collaboration software (in progress)
  
• HEPML. Unified events format and automatic parsing of event samples from different MC generators (in progress)

details are in hep-ph/0604120 pages 200-204 and in http://mcdb.cern.ch

PARAMETERS OF EVENT SAMPLE DESCRIPTION

• General information
  + Title
  + Abstract
  + Authors
  + Experiment and/or Group
• Physical process + Initial state
  + Final state
  + QCD scale
  + Process PDF
• Event files + Physical process
  + File name
  + Events number
  + cross section and uncertainty
• Used generator + Name and version
  + Description
  + Home page address
• Physical models + Name
  + Description
  + Set of parameters and their values with author's descriptions
• Applied cuts

CURRENT STATUS OF LCG MCDB (JULY 2006)

Nowadays LCG MCDB is properly operating and ready to use.
System is accessible via Web-site (http://mcdb.cern.ch). All MC event files
accommodated in LCG MCDB are available over GridFTP or directly from CASTOR.
Web-interface was tested and adopted to work correctly with the most popular
browsers as Mozilla, Firefox, Internet Explorer and Opera.
Migration of files and corresponding documentation from CMS MCDB is in good progress.
HEPML project is the main stiff point for the present development.

Current status of LCG MCDB subsystems:

• Web interface with Content Management System [done 95%]
  • Uploading event files or group of files (from Castor, AFS, local disk) [done]
    
  • Uploading event files (GridFTP) [in progress]
  • Articles handling interface [done]
  • Flexible SQL/XML based search engine [done]
  • Downloading MC samples (through the Web or GridFTP or directly from Castor)[done]
  • Simplified authors registration [done]
• Moderators interface [done]
  • Managing files, their descriptions and related physics information in the DB
  • Physical categories management
  • Managing user's profiles
  • Moderate user's comments on articles
• Database
  • SQL tables structure [done]
  • Internal modules to access DB objects [done]
  • Backup DB (to AFS) [done]
• Other subsystems
  • Authorization system (using CERN AFS password or LCG Grid certificate) [done]
  • Access to Castor for MCDB files [done]
  • Log and reports subsystem [50% done]
• Documentation and user support
  • Developers documentation [80% done]
  • End-users documentation [50% done]
  • Mailing system: based on CERN Listbox Services [done]
• API to Collaborations software [in progress]
  • direct access to LCG MCDB samples with CASTOR/!GridFTP/WEB paths [done]
  • automatic parsing of event samples documentation require HEPML agreement
• HEPML. Unified XML format of event samples and its documentation
  • conception and goals is under discussion and describes in https://twiki.cern.ch/twiki/bin/view/Main/HepML
  • Publication of the HEPML conception need for the discussion of the idea in wide community
  • Presentation of the LCG HEPML on MC4LHC workshop
  • Collaboration with CEDAR to combine effort of two groups on the development of HEPML
  • Weekly teleconference meetings of the HEPML developers (S.Belov, A.Gusev, A.Sherstnev, L.Dudko, A.Kryukov, M.Kirsanov)
  • Development of HEPML libraries and utilities (Interfaces and libraries to save, read and exchange event files in different formats by means of HEPML unification which is based on Les Houches Accord #1)

Collaborations feedback

• Involved experiments: ATLAS, CMS, LCG
• Requests
  • Grid access to MC event files [done]
  • Files replication [in progress]
  • Upload from remote locations [WEB, Castor, AFS - done; GridFTP - in progress]
  • Multiple files upload (from Castor and AFS) [done]
  • Web interface improvements [continuous]
  • API [in progress]
  • HEPML libraries and tools [in progress]
    

HepML LCG MCDB XML schema

The aim of this schema is to represent all the information from LCG Monte Carlo Data Base in XML form.

The main object of MCDB meta-information is so-called article. It is the document which describes a set of MC event files, generation parameters, applied cuts, theory model, etc.

Shema itself is located on MCDB main server http://mcdb.cern.ch/hepml/schemas/mcdb.xsd and available from the Wiki page of HepML: https://twiki.cern.ch/twiki/bin/view/Main/HepML Full schema structure, Logical schema structure