Roadmap for SiD Detector Simulation Development December 2007 – June 2008
Not the Originally Planned Talk The SiD Detector Simulation Project –Motivation –Tasks –Progress and Prognosis But Now What? –Solid Consequences of the Abrupt Halt Avoiding moralizing –Discussion for Planning Refocus if ILC work continues?
Roadmap Document Fairly solid plans distilled by MF based on work of RK, HW, and others. Focused on delivering soup-to-nuts package for scientists to investigate questions relevant for October 2008 LoI –Our take on what sort of studies are beneficial to this LoI – others feel the LoI could be OK without this level of detail Is in DocDB as 2353-v32353-v3
Motivation For the LoI –Make possible studies of the proposed SiD detector, at an honest level including pattern recognition –Allow physicists to have credible ideas of physics reach for benchmark and other processes –Provide ways to “tune” detector choices based on genuine and believable simulation –Allow for easy incorporation of modules developed by physics/algorithm experts to improve or extend what the software can do Post-LoI –Iterate and Improve
Deadlines/Deliverables January 28, 2008 – SLAC SiD Meeting –Kalman Filter fitter, hits and tracks model Late February first pass on finer detector decisions –Kalman Filter working on forward and barrel regions in a waferized detector –full simulated analyses, with no cheating steps April 2008 – UK meeting –Initial serious studies using mature code June 2008 – Tested turn-the-crank system –allowing people to get generated MC events, send them through as realistic a detector as is practical, do the cuts, mix in background, and so forth Every one of these is blown out of the water now…
Tasks Explore occupancies in the endcap and forward parts of the tracking system using simplified software Bookkeeping Infrastructure Completion –Upgrade the endcap and forward software, to model a waferized detector –Add the code to generate digis (RawTrackerHits) from hits (SimTrackerHits) –Define and implement a method to do the bookkeeping of which hits are used in previous iterations of pattern recognition –Define and implement proper classes for digis, clusters of digis, collections of clusters of digis –Define and implement an improved track class –Classes for collections of electrons/muons
Tasks Get a Kalman Filter working as a final fitter –By the January deliverable! Enhancement of org.lcsim Capabilities –Help to define the first official release of the waferized version of the detector –Incorporate the vertex pattern recognition code. –Incorporate the jet finding code –Incorporate the Linear Collider Flavor Initiative (LCFI) jet flavor classification code –Lay the groundwork for “layered” pattern recognition –Develop pattern recognition codes
Simulated Analyses Perform the estimate of sensitivity for Higgs to heavy quarkonium branching ratios, or some other useful physics benchmark –Depends somewhat on what others will have done and/or staked out as their turf –Can be done April-June as an exemplar of how to use system, or June-August to be part of LoI studies
An Inconvenient Truth In org.lcsim persistency is a straightjacket, not a tool Our “game plan” is –Provide Kalman Filter and hit codes that do useful work but without persistency –Use this as a wedge to re-open persistency discussions
Progress and Prognosis Explore occupancies in the endcap using simplified software Postponed in favor of doing more useful job once Tim Nelson’s pixel (2D) hit code is available Bookkeeping Infrastructure Completion –Upgrade the endcap and forward software, to model a waferized detector SLAC turf; we can complete if they don’t come through by late January –Add the code to generate digis (RawTrackerHits) from hits (SimTrackerHits) Done. SLAC improvements for 2D case expected soon. –Define and implement a method to do the bookkeeping of which hits are used in previous iterations of pattern recognition Well enough completed for January deadline; close to overall completion –Define and implement proper classes for digis, clusters of digis, collections of clusters of digis Done, but collaboration is not in agreement on using these outside our work –Define and implement an improved track class Learned “top-down” approach of getting early agreement on track class properties was uninteresting to SiD collaboration Class is in place and meeting needs; evolution expected –Classes for collections of electrons/muons De-scoped in light of collaboration dis-interest Get a Kalman Filter working as a final fitter –Essential by the January deliverable! –Lots of things are working now –Rob will have to focus full-time this month to get this into good shape
Progress and Prognosis Help to define the first official release of the waferized version of the detector –All parts are in place except extension of Tim’s hit making code to the endcap and pixel regions. Not of immediate concern but work needed in March-May: –Incorporate the vertex pattern recognition code. –Incorporate the jet finding code –Incorporate the Linear Collider Flavor Initiative (LCFI) jet flavor classification code Lay the groundwork for “layered” pattern recognition –Mostly done; examples needed Develop pattern recognition codes –Major focus for Hans through April and beyond Perform the estimate of sensitivity for some useful physics benchmark –Will be ready to start in April
But Now What? Solid Consequences of the Abrupt Halt –We cannot recover to deliver on this roadmap Unless the detector LoI schedule is delayed, even a detour of only two weeks invalidates our road map and plans –But there is some “scope contingency” Fewer iterations of detector design Less feedback from users to coders –Probably missing features and/or headaches Final pattern recognition code is less mature Fewer benchmark processes studied
But Now What? Solid Consequences of the Abrupt Halt –We can be decently positioned to resume work when ILC detector simulation is again sanctioned About a week now would put us in a recoverable state –To the degree that org.lcsim remains stable But loss of key personnel to other projects can negate that –Smart newcomers might have a 2-month learning curve This is not the sort of project that can be pursued with “after- thought” priority And resuming the same roadmap under the same “ground rules” may or may not be the most effective course
Discussion/Planning What is best for FNAL detector simulation to do if and when we resume ILC work? –How best to resuscitate or exploit this org.lcsim work Can’t answer now – depends on –Length of time we are missing –What work others have done in the meantime –Who we have available at restart –Should we plan to re-focus? Broader effort (GLD as well) –Physics simulations – absolutely –Computing infrastructure – possibly –Core simulation/reconstruction – awkward »Unrealistic to think same codes can work in both contexts »Separate efforts are realistic, if we have the manpower
Discussion/Planning Other ILC Detector Simulation and Reconstruction planning issues –Participate in setting benchmarking agenda –Other issues?