Based on last week's meeting and conservations with others, here's a proposal for a stream scheme that the physics groups should take seriously. I used global_CMT/8.0 runs. You can try it out yourself. See the xml web page (http://www-d0.fnal.gov/computing/analysis_tools/tools/evaluateStreaming /xml). Use proposal1.xml for the stream scheme, global_CMT80.xml for the data (or any other data with that trigger list) and you *must* use global_CMT80_ovPrimals_proposal.xml to override primal streams. For example, cut and paste the following into a d0mino or clued0 window: setup evaluateStreaming evaluateStreaming \ http://www-d0.fnal.gov/computing/analysis_tools/tools/evaluateStreaming/xml/proposal1.xml \ http://www-d0.fnal.gov/computing/analysis_tools/tools/evaluateStreaming/xml/global_CMT80.xml \ http://www-d0.fnal.gov/computing/analysis_tools/tools/evaluateStreaming/xml/global_CMT80_ovPrimals_proposals.xml If you're running on clued0, you may see an error message about an unsupported encoding 'UTF-8'; you can safely ignore it. Here are some details of the stream scheme---- Make your mail window at least as wide as the bar below ====================================================================================================== Read from left to right, going up means the event has that primal stream (YES), going down means the event does not have that primal stream (NO). HIGHPTELE+X electron? < HIGHPTELE+X ejet? < HIGHPTMU+X muon? < LOWPTLEP-HIGHPTLEP multiele? < LOWPTLEP-HIGHPTLEP elelo? < LOWPTLEP-HIGHPTLEP multimu? < JETS jet? < GAP gap? < TAU tau? < MONITOR There are 9 primal streams: electron = high Pt electron triggers (used by W/Z+top+higgs groups) ejet = high PT electron + jet triggers (used by higgs+top groups) muon = muon triggers (used by W/Z+top+higgs groups) multiele = low Pt multielectron triggers elelo = low pt electron triggers multimu = multimuon triggers (not used by W/Z+top+higgs) jet = jet triggers gap = JT_GAP triggers (and JT_15TT) tau = tau triggers (but these completely overlap the high pt lepton triggers The population of the physical streams are: HIGHPTMU+X gives 354744 events (33.1% of total) HIGHPTELE+X gives 353681 events (33.0% of total) LOWPTLEP-HIGHPTLEP gives 178275 events (16.7% of total) MONITOR gives 88569 events (8.3% of total) JETS gives 82346 events (7.7% of total) GAP gives 12708 events (1.2% of total) TAU gives 0 events -- Since TAU triggers always come in with a high pt lepton trigger, the TAU physical stream is empty. Some more work needs to be done on on that. MONTIOR is made up of zerobias, minbias and mark and pass triggers (lots for the runs I tried). -------------- The features of this stream scheme is: 1) W/Z analyses only have to look at two streams (HIGHPTELE+X and HIGHPTMU+X). The HIGHPTELE+X stream contains events with a high Pt electron AND a high Pt muon. Since this overlap is so tiny (about 0.3% of the data), it would be hard to give it a stream of its own. Heidi suggests that we could make a high Pt e + high Pt mu + X stream offline so the muon people wouldn't have to look at all of the electron data. 2) Top and Higgs high Pt physics are in two streams, same with the W's and Z's. So hopefully counting W's and scaling to other analyses (see minutes from last meeting) is still possible. New phenomena analyses with high pT leptons would also use this stream. 3) The LOWPTLEP-HIGHPTLEP (events that pass a low Pt lepton trigger but not a high Pt lepton trigger) stream is meant for the b physics group who does many analyses vetoing high Pt leptons. Such analyses would not have to look at the high pt analyses. 4) The JETS stream would contain only jet events (not passing any lepton triggers). QCD analyses would veto on leptons, so hopefully the other streams would not have to be looked at. 5) Andrew Brandt requested a GAP stream for their QCD-gap triggers. Since they are mostly exclusive with the other QCD triggers, this stream would be able to live on its own. Note that the size of 1.2% maybe too small. 6) Right now all of the triggers with a tau primal stream come in events that also have a high PT lepton. So there's nothing in the Tau physical stream. Some possible changes you should try: 1) Swap the order of LOWPTLEP-HIGHPTLEP and the JET physical streams (this is easy enough to do, though a little tedious, in the evaluateStreaming program). Then JET analysis that don't care about low PT leptons will only have to look at one stream. Since the b analyses probably don't care much about jet-only events, this would not be a problem for them. 2) Assign other triggers to the tau primal stream (from the View/Override Primal Streams menu option) and see if any events show up in the TAU physical stream. Disadvantages: 1) High Pt leptons are only split up into two streams. One could imagine having streams for two leptons, or a HighPT+lowPT lepton streams, but this makes for small streams and hurts the ease of W counting and scaling. 2) Nothing in this scheme is vary fine grained (except perhaps gap). The antithesis of the priority scheme is one where physical streams are determined by object type and number (e.g. a stream for events with only one electron, stream for events with two electrons, ... All permutations of objects ...). I think it was clear from the meeting that would make for too many small streams and "unprotect" the W/Z analyses. Since New Phenomena uses a wide variety of triggers, I'm hoping this scheme makes things work out naturally for them. Let me know if that's not the case. -- So to the physics groups: please use evaluateStreaming to see what happens to events with your analysis triggers. Don't be afraid to make changes to the scheme and experiment. There is no meeting next week. The week after, Wyatt is back, but I am away [for my wedding]. Wyatt will probably hold a meeting, but it may be about SBIR stuff. But perhaps some of you could report then on your findings. While I would like to propose more streams schemes, it appears that keeping high PT leptons in as few streams as possible is very important, thus a priority scheme like this one seems the most appropriate. Of course don't hesitate to let me know if you have other ideas. --- Adam