Event Reconstruction for LHCb

1
Event Reconstruction for LHCb

• What is the scope of the project?
• What are the goals (shortmedium term)?
• How do we organise ourselves?
• How do we start?

2
Scope of reconstruction

• Reconstruction is everything that takes raw data
  as input and produces new objects as output
• High level triggers
• Calibration and alignment
• Creation of event tags (c.f. event classes at
  LEP)
• Full event reconstruction
• Creation of analysis objects
• etc.
• At LHC we will use a database to store the data
• Implication is that reconstruction can be
  performed in several stages
• Full reconstruction may not be necessary for all
  events, or for all parts of an event

3
Some short term goals

• Build a team of programmers
• Bring together physicists working on software
• Learn by doing design reviews, CRC workshops etc.
• Define reconstruction strategies
• What can be done independently by subsystems
• What requires input from other subsystems
• (What are subsystems? - sub-detector, tracking,
  particle ID)
• What should be sequence of actions
• What should be reconstruction output
• What initialisations are required and when
• Identify and define external inputs
• Raw data event model
• Detector geometry, calibration, alignment

4
Some medium term goals

• Build a reconstruction skeleton within Gaudi
• Implementing reconstruction phases -e.g.
• common specific (time dependent?)
  initialisation
• high level trigger reconstruction pass
• subsystem standalone reconstruction pass
• global reconstruction pass
• Placeholders for each subsystem
• Make a reconstruction program independent of
  SICB
• Wrap existing SICB reconstruction code into
  algorithms
• Integrate new OO code
• Agree (and begin to enforce) conventions
• Units, co-ordinate system
• Error reporting and handling, monitoring
• Coding rules, metrics, documentation, look and
  feel

5
Project organisation (1)short term

• Team building
• Programming community is not large - do not
  isolate people
• not yet reconstruction vs. simulation vs.
  analysis
• never SICB vs. OO. Encourage same people to do
  both
• Regular meetings
• Start with a slot in the weekly Thursday
  computing meeting
• Should be used by all who wish to present or
  review designs,requirements,code etc.
• Presentations and reviews of ongoing work also at
  physics meetings
• Ensure that the physicist community is
  aware/involved in design choices
• Project leader
• Collects requirements from subsystems
• Responsible for development of reconstruction
  strategies
• Interacts with GAUDI team
• Ensure specific needs of reconstruction are
  addressed by GAUDI
• Ensures milestones are set and adhered to

6
Project organisation (2)medium term

• Identify a software contact in each subsystem
• Not necessarily a programmer BUT
• Has to be link person between project and
  subsystem
• Discusses new software within subsystem community
• Collects requirements of subsystem
• Delivers requested information by agreed
  deadlines
• e.g. what input does subsystem need from other
  subsystems?
• e.g. how does subsystem plan to provide output
  needed by other subsystems?
• e.g. verify that wrapped SICB algorithms work
  as expected
• Incremental approach to medium term goals
• Attainable milestones every 2-3 months
• Dedicated meeting at each milestone
• Review progress
• Agree next steps
• Attended by (at least) all contacts

7
How do we start?

• Repeat this presentation at software week
• Including a name for the project
• Take advantage of software week to
• Appoint a project leader
• Begin discussion of reconstruction strategies
• Begin collection of subsystem requirements
• Agree goals for end-November software week
• Already some requests for design reviews
• We can schedule these from next week
• Should invite relevant experts at each review

8
A name?

• For the architecture we have an architect
• We now start a (re)construction project
• We need a civil engineer
• BRUNEL
• (B Reconstruction, UNderstanding Events in Lhcb)