Public minutes of the Geant4 Steering Board meeting 63

3rd February 2011

Version 0.5***

Attending: John Allison (representing the Visualization WG and OB), Makoto, Dennis, Gunter, Hisaya, Koichi, Sebastien, Marc, Gabriele, Michel, Daniel

Excuses: Vladimir, Takashi, John Ap., Joseph, Christina, Peter

Note about OB representative: OB provides now a representative to SB meetings, John Allison undertakes this role.

1. Regular items

a. Approval of minutes on SB meeting of December 9th, 2010

Minutes are approved.

b. Reviewed pending Actions from December 2010

2. Following-up known technical issues

a. Issue on visible energy (V.Ivantchenko)

During the most recent Technical Forum LHCb reported about a 30% change of EM scale with the new Geant4 version. It was mentioned during the report that these results were very preliminary.

The results for all calorimeters under the Geant4 test suite are monitored to be stable within 1% for many years. However, there is a known significant difference of visible energy between the default Physics Lists QGSP_BERT and the “fast” list QGSP_BERT_EMV. For the LHCb-type calorimeter this difference is about 30% and is entirely due to the choice of multiple scattering parameters.

Geant4 has requested more information on how the LHC-b results were obtained. No feedback was yet given, and communication is kept ongoing.

b. Issue on 3% performance penalty (G.Cosmo/A.Dotti)

The 9.4 release exhibits a performance penalty of about 3%. This measurement is made simulating a 10 interaction length thick calorimeter, with 50 GeV incident pions and electrons. The investigation done by gradually including the various Geant4 modules (global, particles, materials, geometry, event, run, process/management, process/cross_sections, e.m.) in their version used for release 9.4 could not spot the origin of the problem. The interpretation for this loss at this point is a concomitant effect of several modules.

ATLAS could not conclude about this performance penalty as their test was statistically not precise enough, but they reported a 10% increase of memory usage. This could be a hypothesis to explain the performance penalty.

The investigation of both the performance penalty and memory consumption is kept ongoing, and action 62.3 is kept open.

3. Architecture revision

A draft document for the items proposed to revision in 2011 in the context of architecture revision in 2011 has been submitted to the SB by Hisaya.

The SB members are invited to comment in the next two weeks and consider proper allocation of the proposed items affecting their Category in their working group plan for this year.

4. 2011 development plans

a. Collaboration-wide development

It is proposed that a list of highlight items is distributed well in advance for the next TF on March 3rd and to interact with Geant4 customers at this TF. The development plan for year 2011 will then be finalized for March 17th.

The granular development item plan will be referenced to this document:

- Gabriele reported that at the last CERN/SFT architecture meeting, the CMS experiment requested to access the architecture review document. Given this document was conceived and written for a collaboration internal use, reshaping it to a form that would be appropriate for a public distribution is a concern.

- How to proceed will be further discussed and agreed by e-mail.

b. Releasing Geant4MT 9.4-beta

- In 2010, a multi-threading (MT) capable prototype was successfully implemented and provided promising performance improvements, with both a mostly optimal scaling of the performance with the number of cores, and a good memory sharing between the threads.

- It is proposed, and agreed, to have a task force to release the MT prototype. This task force will have to address the following issues:

  1. Document on performance evaluation on a “realistic” setup

2. Step-by-step user’s guide, including clear user code samples demonstrating how to make proper use of the prototype

  1. Clear definition of distribution model, maintenance & support

5. 2011 release schedule

At the last TF in November 2010, the LHC experiments requested to have the major release of Geant4 delivered one month earlier, to allow for early validation, and production start close to the Christmas break (SB 62).

A release schedule with a reference tag submitted to the experiments by mid-November and the release on the 2nd of December is proposed.

A beta release is proposed by 30th of June. A beta release is simply a reference tag, therefore, scheduling is much simpler than a release. Tags provided by mid-may could go into the beta release.

6. ** **Migration to SVN

a. The report on completion of the migration to SVN has been provided.

b. The OB representative congratulates with Gunter and collaborators from the smooth transition to SVN.

7. 2011 Collaboration processes

A report of the collaboration ongoing membership registration is provided, and a status of the current situation is given, including registrations done, pending and future possible suspensions.

WG census - target Feb. 16 (M.Asai):

It is requested to continue census, to collect granular items, and send to Makoto by February, 16th.

8. ** **Collaboration matters

a. New Documentation Coordinator

- Katsuya is stepping down as documentation coordinator, by end of March 2011.

- Several people have been contacted, but no positive response at this point.

- This is an urgent issue: volunteers or suggestions are welcome!***

b. Creation of Physics validation Task

- The creation of a physics validation task is proposed; the goal of this task is to ensure adequate physics performance of components critical to the leading application areas. A draft mandate is circulated; and if no objections are raised within one week, the task will be approved.

- The chair of the task will have a seat at the SB.

c. Proposal for creation of a novice + extended WG:

- The creation of a novice + extended WG is proposed; and a draft mandate is circulated. This WG group will have to renew the existing novice + extended examples suite, caring about appropriate map of the Geant4 functionalities into the examples, ensure code quality, ensure cross-reference between examples and documentation, etc.

A one week period for reviewing the draft is given. If the draft mandate is agreed, the mandate and the WG creation will be accepted and an ad-interim coordinator will be nominated.

- It is expected that each WG coordinator will belong to the (yet to be approved) N\&E examples WG.

d. Prototype publication editor system (D.Wright)

- No report.

e. Requirements tracking system (K.Murakami)

- A prototype has been provided

- It was noted that it would be highly desirable to announce the creation of the tracking system at next TF and have all open requirements migrated in the new system.

9. AOB

a. New collaborators

No new member was proposed.

b. Near future conferences and events

- Proposal to create a web page for tracking these events.

D.W. and the publication board will add this into the yet-to-be-made publication tracking system.

c. 2011 Collaboration Workshop

- It will be held at SLAC

- Date proposed is September 19th to 23rd, 2011:

d. Tutorials

- After PTcog, in may 2011

J.P. will inform SB of more details as soon as they are fixed.

e. Requests from LowEM group:

- For restructuring of lowEM directory and split it in LowEM + DNA

  1. For code and for data

  2. This changes are not anticipated to have impact on users’ and other categories’ code

  3. Approved if above conditions are satisfied

- For removing obsolete classes

  1. The so-called “Livermore” classes

  2. Approved.

Meeting Date

Thursday, February 3, 2011 - 12:00