From wohn@iastate.edu Fri Dec 4 10:32:54 MST 1998 Message-Id: <199812041732.LAA22019@pv7453.vincent.iastate.edu> To: young@orph01.phy.ornl.gov (Glenn Young), mjbennett@lanl.gov (Mike Bennett), schlei@lanl.gov (Bernd Schlei), jsimon@lanl.gov (Jehanne Simon-Gillo), hubert@lanl.gov (Hubert van Hecke), sullivan@p2hp4.lanl.gov Cc: wohn@iastate.edu (Fred Wohn), lebedev@iastate.edu (Sasha Lebedev), mrosati@iastate.edu (marzia Rosati), jhill@iastate.edu (John Hill), lwood@iastate.edu (Lynn Wood), lajoie@iastate.edu (John Lajoie), petridis@iastate.edu (Athan Petridis) Subject: more comments on MVD multiplicity at L1 Date: Fri, 04 Dec 1998 11:32:52 CST From: Fred K Wohn Status: RO Dear Glenn, It is not clear to me how the intrinisic event-to-event fluctuations (which we had always thought would be about 10%) can be overcome at L1. It may just be necessary, as we were discussing here in our local MVD meetings, to cut at 20% at L1 to ensure high efficiency of the 10% most central. Extracting the true top 10% from the L1 top 20% seems to me to be what L2 is well suited to do (or offline until luminosoty gets high enough to require L2). I agree that we should try to separate measurement rms from fluctuations, but I don't see how we can do this at L1 with the data available for L1 decisions. You have raised a topic needing discussion, with Soren or whomever else can contribute, but the time scale is very short. As you know, the overall latency of L1 is greatly affected by the use of BBC LL1 output into the MVD LL1 input. What makes it worse is the the divide logic (step 5 of the new BBC block diagram) could take more than a few clock ticks to get a truncated 8-bit output (longer if not truncated, so John L and I already told Harold to truncate). This was the main motivator for me to consider using MVD itself for a vertex correction, so MVD and BBC can go in parallel rather than in serial. It seems an organized meeting on this topic next week at BNL seems to be called for. Cheers, Fred ------- Forwarded Message Date: Fri, 4 Dec 1998 02:25:51 -0500 (EST) From: glenn young To: young@orph01.phy.ornl.gov (Glenn Young), mjbennett@lanl.gov (Mike Bennett), schlei@lanl.gov (Bernd Schlei), jsimon@lanl.gov (Jehanne Simon-Gillo), hubert@lanl.gov (Hubert van Hecke), sullivan@p2hp4.lanl.gov, Fred K Wohn Subject: Re: comments on MVD multiplicity at L1 Cc: wohn@iastate.edu (Fred Wohn), lebedev@iastate.edu (Sasha Lebedev), mrosati@iastate.edu (marzia Rosati), jhill@iastate.edu (John Hill), lwood@iastate.edu (Lynn Wood), lajoie@iastate.edu (John Lajoie), petridis@iastate.edu (Athan Petridis) Dear Fred, We maybe would benefit from bringing in also people like Soren to this discussion, as we are getting out of purely instrumentation and into physics discussions. I pick Soren not at random but because he's worked a lot on this (fluctuations) from an EMCal and also event-generator point of view. Similarly, Evert Stenlund might be queried. I won't argue about the rms Athan finds since I'd expect it, since we deal with finite multiplicities. And you could argue that is still one step (at least) away from physics since the connection between multiplicity and reaction centrality has yet another rms term entering when you go from multiplicity (or Sum.Et) to impact parameter/centrality. My concern if we don't make the vertex correction is then we are mixing measurement rms, at LVl-1, with a different rms, due to multiplicity fluctuations. Shouldn't we try to keep those separate? I'm sticking my neck out for somebody to take pot-shots using HIJING plus PISA, etc., and show me this isn't a concern. Glenn ------- End of Forwarded Message