svtsim (offline) status and plans (WJA, 2001-02-06)

• Goals

• Given silicon and XFT raw data, make bit-for-bit prediction of all boards' output

• Initialize from Giovanni's standard high-level files

• Modular, fast, lean, ANSI C, minimal memory use, works online

• "Latest" version of svtsim should always be able to reproduce SVT behavior for all good physics runs to date, by reading databases

• Provide mechanism for using TF output to provide TF input or other quantities useful for display

• Provide access to other useful internal state information

• Get (me) out of SVT dog house ASAP!

• translation: I have been holding up progress on other work for a while now :(

• Status

• HF in good shape

• same code used to test boards

• AMS+AMB, HB in good shape

• low-resource evolutions of Alex's code

• TF simulator itself is good

• same code used to test boards (but emulate memories w/ funcs)

• but need to integrate w/ parts of svtramgen (emulate big memories from .gcon file, etc.)

• XTF, Merger are a bit simple so far, but fine for now

• reads GP's high-level files

• Reads real SIXD and XFLD (not yet XTRD)

• some messy things (e.g. magic numbers) scattered about code; need minor (a day or two) clean-up

• Recent validation effort

• Checking hit coordinates against OBSP prediction and SVTQWK

• Have function to predict hit coords, given helix, using offline geometry methods (useful?)

• HF rms 13um (good); -20um bias (!)

• XFT/XTF phi rms 1mrad, bias -0.2mrad

• XFT 1/pt rms 0.8%/GeV, bias -0.8%/GeV (about 3/4 bin)

• Toni Munar has made a validation ntuple: OBSP, XFT, SVT info

• This is a really big help!

• Minor fixes to decoding of XFT output

• SVXII geometry mixup fixed

• A half-strip offset in definition of cluster position remains

• Asked Matt Herndon to check it out

• Ran 2--4 GeV muons through CDF GEANT simulation (SVX, COT, XFT)

• Using .gcon directly on HB output, got ~30um dvx resolution

• Expect TF sim to do same; check in next day or two

• Plan:

• Sit down with GP to make sure I have fully working SVTQWK

• Update CVS so that Toni can run/check

• Check TF output OK (it was in May)

• Clean up clutter

• Validate with SVTQWK

• Integrate with Alex B's work (DB issues)

• Work with Un-Ki, Marco, Subir:

• validate against real boards

• run on crate CPUs

• Add hooks for access to intermediate results (for debug, display, histos, etc.)

• Generalize to handle L00, XTF general lookup, multiple TF gcon files, etc.