Back to: [BABAR] [Level 3 PDRR+CDR] [Level 3] [Trigger] [Electronics] [OEP] [Online] [Computing]
Level 3 Trigger PDRR + CDR Committee Report
- Dave Brown, LBL
- Bob Jacobsen, LBL, Chair of Review Committee
- Usha Mallik, Iowa
- Helmut Marsiske, SLAC
- Walter Toki, Colorado State
The Committee would like to thank the review organizers and presenters for their work. It is clear that a lot of thought and effort has gone into the preparations for this review.
The Committee was asked to review the Level Three (L3) related aspects of the trigger system at the "Preliminary Design Requirements Review" and "Conceptual Design Review" levels. The four main areas presented were the L3-related requirements, the
conceptual design of the L3 trigger, various interfaces, and prototype work to date on specific L3 algorithms.
The requirements as defined in version 4.71 of the Trigger requirements document seem largely complete and sufficient to guide the design and implementation phase of the L3 trigger. Clarification of the following few points would however still be
- The requirement that the subsystem groups implement software for converting digi format to raw data tagged-container format for L3 simulation, and vice versa for L3 algorithm use should be made formal and transmitted to the appropriate subsystem
- 'Changebars' indicating the L3-specific revisions of the requirements document would also be appreciated in any future reviews.
The conceptual design is appropriate to the task. It appears to be a powerful, flexible system with all the required capabilities. It is based on existing structure and implementations where possible. It should provide the needed infrastructure for
running and controlling multiple L3 trigger algorithms. Only a small amount of detail was presented on the L3 interfaces. This is perfectly appropriate to a predominantly-software effort at this stage in its development.
Only minimal detail was provided on plans for further development of the L3 trigger. The Committee recommends that the group develop a series of milestones to test feasibility of the L3 trigger over the next seven months. Particularly in view of the
need for large MC simulation runs, this should be coordinated with the other parts of the online and offline development. This recommendation is not intended to result in a lot of planning overhead, but rather to generate and record a concensus among the
group regarding what needs to be done, in what order. The committee would also like to point out to the overall trigger group that organizing the L3 trigger subgroup is a large effort, and should be done by somebody who can devote the majority of their
time to it.
Some physics channels are very closely coupled to the trigger, either due to difficulty in achieving high efficiency at sufficient rejection or because of stringent requirements on knowing the acceptance. Examples include the "unbalanced two-prong"
physics of tau pair and two photon interactions. It is likely that the L3 trigger will need specific algorithms to recognize these physics channels. It is important that people with expertise and interest in these channels become involved in developing
the trigger algorithms to select them soon. In the absence of this, suitable algorithms will have to be developed after the start of data-taking, with the result that the triggers for these channels will not be optimized for maximum efficiency at the
start of the data taking.
The algorithms presented in the meeting appear to be preliminary efforts recently developed. On the whole, they meet the required efficiencies for B physics but not low multiplicity physics. Also the x10 background rejection is not yet sufficient.
Nevertheless, the algorithms for the SVT, DC and EMC appear to be promising. In the B=>pipi modes, the SVT and DC achieve +99% and the EMC finds 72-98% efficiency depending on the minimum energy cut. The x10 backgrounds from photoproduction and beam
gas however can be large and saturate the L3 trigger (i.e. >100 Hz).
Although keeping triggers "orthogonal" is a conservative plan in order to permit separate detector checks on trigger efficiencies, combining triggers at a later stage will reduce the trigger rates and may be necessary in the case of larger than
expected backgrounds. The Committee encourages the L3 group to develop a full spectrum of tools for rejecting background and recognizing physics before spending significant additional time fine-tuning the existing algorithms. Several improvements might be
possible by looking directly at specific backgrounds, for example by hand scanning events to explore the use of other variables such as pt, pLong and multiplicity. These other variables could add useful flexibility to cut bigger than expected backgrounds
in order to retain B physics (possibly at the expense of tau and two photon physics). Further improvement of technical aspects (e.g. developing different track fitters) may be unnecessary until a better understanding of background characteristics is
achieved. Eventually, the algorithms should be tested for actual CPU time to check the feasibility of their use in a farm of workstations.
Original transmitted by
Bob Jacobsen (Bob_Jacobsen@lbl.gov)
Date: Tue, 16 Sep 1997 23:21:37 -0700
Minimal HTML conversion by Gregory Dubois-Felsmann.
Last modified: Wed Sep 17 17:00 PDT 1997