SoftwareReview October 25

From GlueXWiki
Revision as of 11:17, 28 October 2013 by Cmeyer (Talk | contribs) (Data Flow and Data Challenges by Mark)

Jump to: navigation, search

Software Review Planning Meeting Friday October 25, 2013.

2013 Software Review Page

Connecting

Phone:

  • +1-866-740-1260 : US+Canada
  • +1-303-248-0285 : International
  • then enter participant code: 3421244# (remember the "#").
  • or www.readytalk.com (and code without the #)

Outlines of Talks

We agreed that we would post a list of slides for the talks at the 12-Gev Software Review, but that we would most likely not have a meeting this week.

Morning Session (30 minutes)

Overview Talk by Curtis

  • [1-2] Title + Outline
  • [3-5] GlueX Physics an Amplitude Analysis
  • [6-12] Walk through major accomplishments. This includes
    • (i) Defined Data Formats: EVIO, REST, PART (root) data format.
    • (ii) December 2012 data challenge producing about 70% of the data we expect in 2016 with 10**7 running. What did we learn?
    • iii) Job management/ Work flow exercises.
    • (iv) Implementation of CCDB into analysis framework.
    • (v) development of analysis framework to initially facilitate the PID upgrade work, and ultimately develop the proposal in June.
    • (vi) High-level analysis tools: kinematic fitting, multivariate analysis, ....
    • (vii) July 2013 Software Workshop.
    • (viii) Code and algorithm improvements: Particle Identification, BCAL simulation and reconstruction, new tracking for spiraling tracks.
    • (ix) Start work on Level-III trigger and algorithms
    • (x) Online Data Challenge with level-III trigger.
    • (xi) Conversion to GEANT4.
  • [13] GlueX Running { Phase I,II,III,IV, ...
  • [14-17] Timeline to Physics with tasks
    • (i) Data Challenges
    • (ii) Calibration and Commissioning
    • (iii) Physics Analysis Planning
    • (iv) Level-III Trigger Development
    • (v) Testing of Work flow Developments.
  • [18] Manpower and Resources
  • [19-20] Summary/Conclusions

Afternoon Session (90 minutes)

Data Flow and Data Challenges by Mark

  • Data Flow
    • [1] Raw data taking -- write to buffer disk, EVIO format, transfer to tape library
    • [2] Reconstruction -- on JLab farm, JANA framework, REST formatted output
    • [3] Analysis -- REST to Root Tree option, REST to JANA option, application of Analysis Tools
    • [4] Simulation -- on JLab Farm, on the Grid
  • Data Challenges
    • Data Challenge I
      • [5] Goals -- produce and handle large data set, look for issues in scaling
      • [6] Results -- 5 Gevents, massive data set (2 weeks of data rate, 3 months of physics data).
      • [7] Lessons Learned -- tools need development (JLab CC may help here), need meta-data catalog
    • Online Data Challenge
      • [8] Goals -- look for bandwidth bottlenecks, measure performance of monitoring farm, test level 3 algorithm, transfer to tape library
      • [9] Results -- to be researched
      • [10] Lessons Learned -- to be researched
    • Data Challenge II
      • [10] Goals -- 10 Gevents, EM background, modern reconstruction
      • [11] Status -- ongoing mini-challenges

Reconstruction Development Highlights Since Last Review

  • Spiral Tracks
    • [1] Structure of tracking software
    • [2-3] Performance of tracking software
      • Resolutions, pulls
      • Efficiency, rates
    • [4] Motivation for new tracking code: Spiraling tracks
    • [5] Alternate algorithm to handle spiral tracks
    • [6] Performance of new algorithm
  • BCAL Reconstruction
    • [7] Barrel calorimeter specs and performance goals
    • [8] Brief history of BCAL Reconstruction
    • [9-10] Motivation for new code
    • [11] performance improvement
    • [12] Summary and outlook


Analysis Talk by Justin

  • [1] Overview (phrase as response to previous report requesting end user experience)
  • [2] Introduce software workshop as an example of a "complete" physics analysis
  • [3-4] Event generator frameworks
    • AmpTools
    • bggen
  • [5-7] Analysis Framework
    • DANA and reconstructed objects
    • Reaction setup example
    • Kinematic fitting
  • [8-9] "mini-DST" ROOT tree output
    • Format of "PART" ROOT trees
    • Simple analysis example with a TSelector
  • [10-12] Multivariate Analysis
    • Introduction
    • BDT description
    • BDT performance
  • [13-17] Amplitude analysis
    • Introduction
    • sample fit for n3pi
    • n3pi fit results
  • [18] Summary