Hwanbae Park         303-492-2031       S
Patricia Rankin      703-306-1898       S
David Quarrie        510-486-4868       S
Bob Jacobsen         510-642-3618       S
Torre Wenaus         011 41 22 767 7239 S
Fred Kral            510-649-9189       S
Ed Frank             617-496-2362       S
Moe Foucher          301-405-5908       S
Maria Grazia Pia     011 39 10 353 6328 S
Riccardo de Sangro   011 396 9403-2788  S
Fabrizio Bianchi    
David Kirkby         415 725 2346       S
Tony Gabriel         423-574-6082       S
Mike Ronan           dial-in            S
Doug Wright          510 426-1403       S
Dave Aston
Paul Harrison

Minutes

by T. Wenaus

Trigger fast control system

See notes from Fred. The trigger group will provide access via AbsEvent to trigger information as documented in the note; see 'fast data for simulation'. Times will be stored relative to 'Geant Physics Time (GPT)', the T=0 in the Geant simulated event. Wall clock at GPT=0, L1A beam crossing time, the 32 L1A trigger decision lines, and ORed state of L1A after prescaling of 32 lines (for simu data (only) may be false) will be provided. See Fred's writeup for details. A first version using a fast, simple L1A model will be made available in a few weeks at which time it will first be exercised by Fabrizio Bianchi in SVT digitizations which he will modify to use the L1A trigger clock. Over time, various ways of filling the data (via a package FcsData) at various levels of detail (and speed) will be provided. C++ access to the clocks and various conversion functions is provided via a header file defining functions that call up routines in the (Fortran) trgFrame/FC* code (may migrate to a TrgFC package).

Status of Framework-based digitization codes

SVT: In addition to the incorporation of the L1A clock mentioned, following recent discussions with Bob Fabrizio has implemented Digi-based accessors to associated GHits following a proposal put forward by Bob (linked here). This proposal was adopted for uniform use by the subsystem digitization codes. It efficiently provides the flexibility required to provide access to 1 or N GHits associated with a Digi (and vice versa, the interface is symmetrical). Timescale for the new SVT digi implementation being ready is ~Oct 7.
DCH: Moe Foucher is beginning work on implementing Framework-based digitizations and will draw on work already done by Art. Tricia will organize a small meeting to start the effort.
DRC: Dave Aston is working on moving DIRC digitization -- more involved than other subsystems because charged track hits in the DIRC are far removed from the eventual Cerenkov signal in the PMTs -- to the Framework, but a review of this code suggests it can be moved out of BBSIM. In fact Brian has told the production team that generation of digis (DrcGHits) in BBSIM can be turned off in production (which we'd like to do, as the code has been a source of crashes, albeit at a much reduced rate in the current production version). Timescale for functional Framework digis is the end of the month.
EMC: Not represented in the meeting. In good shape with Phil Strother's waveform code in EmcSim, now in SRT. Probably next to move to using the trigger clock info.
IFR: Luca Lista (not at the meeting) has an implementation in the framework, albeit not including the inner RPC or fully reflecting the readout topology (eg. regions of 1-D readout). The IFR and particularly inner IFR need model improvements in BBSIM, some of which are in the head and which Doug will shortly release. An IFR meeting to discuss in more detail is being organized.

Bob asked that a uniform convention on the implementation of the time quantity in digis be adopted, in particular that the time be ns (real) relative to the L1A time (ie. what can ultimately be expected from real data, thus isolating reco code from the evolution of how this number is filled). This was agreed and adopted. Implementation can be addressed once the L1A time is available as discussed.

Rationalization of the digi code's package organization was discussed briefly and it was agreed that, while necessary, this is best deferred, probably to around the end of the year.

The DIRC problem noted in radiative bhabha production runs was briefly discussed; Brian believes he has found the solution in the storage of _all_ Geant steps as drcGTrkHits which in these events occur in very large numbers due to electron curl-ups. A new version is out with this fixed, but another possible problem (associated with ascribing user-level meaning to VOUT, a Geant internal array) needs to be checked by Brian or someone else (Gautier?) in advance of his return.

Documentation inventory

The documentation list linked here has been updated in light of the discussion. Thanks for the universal readiness of people to contribute to the documentation effort. Note that regarding subsystem simulation documentation, the level of documentation required is suggested by the relevant item in the review's charge:

1. Current status and adequacy of the detailed and fast simulations of the detector as the principal tools for evaluating detector and physics performance.

That is, beyond describing the basic mechanics of the code the detail level of the model needs to be described and related to the current engineering design and the requirements coming from application of the simulation in detector and physics performance studies.

In the course of the documentation discussion Gerry Lynch commented that Gautier is said to have ported DIRC modelling code from BBSIM into Aslund. Would like to learn more about this!

Simu review agenda review and feedback

The preliminary agenda is linked here; please review it and give feedback. The review fills two days with (in its current form) a little over 8 hours of presentation. Subsystems are broken out into individual talks, so subsystem coordinators please plan on being at SLAC Oct 28-29.

Geant4 status, prototype development

Torre presented a brief Geant4 update linked here. On prototype development, Gerard Bonneaud and Marc Verderi will meet with Torre this Fri at CERN to take up work.

More data on the XDR file: status, usage, and issues raised

Geometry, run control from .db files:
- All code in place with 1.2.1 but not activated; needs further test
- AND, if you thought the versioning problem was bad with the hit structures, just wait... the geometry I/O will be sensitive to any geometry database structural change.
- Possible approach to survivability:
  - don't delete, rearrange or resize structure elements
  - add elements at end of structure
  - add word count,name prefixing structures on the XDR file [allow ignore or abort if mismatch and potential overflow]
- Contemplating storing streamed ascii and building the geometry using the .db file reader
G3 geometry representation for G4: see Geant4 status notes.
Derived data: Like geometry, waiting to be turned on.
Event-level info decoupled from existing event record: To record monitoring and quality assessment info (timing information, error information, states, flags) a new McFio record type, written each event, will be added. Independent of event record in order to decouple versioning from the event record.

Commissioning detectors

Defer discussion to a dedicated meeting.

geant321 package

Package is in the works, based on CERNLIB 96a (and aided by 96a's adoption of unix-style code distribution). Build/run of BBSIM with 96a cernlibs including Geant works OK, except one unresolved reference 'fpclassifyf' in packlib (dummied). Tested on HP only. A curiosity is that a program (part of G3toG4 convertor) that reads pre-95a and 96a Geant RZ files (including BBSIM) fine can't read BBSIM 95a RZ files. Hmm.

Next Meeting

Tue Sep 17 8:30am PDT. Note that the meeting schedule and other review-prep info are available on the simu review page linked here.