This is an old revision of the document!
Table of Contents
Expert groups
Basics
Impact parameters
Analysis
- jet selection
- flavor identification of jets from truth level
- b-jet identification & selection
- measure efficiency of b-tagger on light jets, b-jets, charm jets
- tune b-tagger to give us expected phase 2 b-tagging performance
- muon selection
- electron selection
- photon selection
- tau selection
- MET selection
Layout
https://code.google.com/p/tkgeometry/wiki/ConfigFilesOverview
Comparison of different scenarios
https://docs.google.com/spreadsheets/d/1Xyob5gcT0jpA4U5QQ1Etb1BdoJ6AQKqdlawrNMs3uBM/edit#gid=0
Sample format for comparison - feel free to edit to get a consistent format
https://docs.google.com/spreadsheets/d/1ZiYucbYVhe7bd3DOS9xPOOqitBRbNkWTFtgWBsKKiiM/edit?usp=sharing
List from Ben
Object ID (Valere):
Technical Proposal: https://cms-docdb.cern.ch/cgi-bin/DocDB/ShowDocument?docid=12143
(e.g. based on example channel HH→WWbb→e nu mu nu b b)
1) Electron object ID - add simple electron ID, perhaps use Technical proposal analysis studies as a guide.
2) Muon object ID - …
3) Tau object ID - …
4) MET object ID - …
5) Higgs ID based on bb : Build up H→bb analysis finder that can be used in all HH→bb+X analysis
B-jet ID (Yong+Thea):
0) Are we treating the sign of the impact parameter in the tagger the correct way ? We know it is signed, and we should make sure that we are using the positive and negative tracks optimally (check how other b-taggers do it).
1) Make a b-tagger which uses 2 or 3 of the highest impact parameter tracks in a b-jet.
2) Add impact parameter smearing from tklayout to analysis macro (in bins of Pt and Eta)
3) Make a b-tagger that uses smeared impact parameter significance : smeared impact parameter divided by resolution (in bins of Pt and Eta)
Jet-ID (Andreas):
1) Is particle flow being used in the jet energy calculation ? If so, is it using smeared tracks that are input into DELPHES ?
2) Consider adding smearing to particle flow jets based on PT resolutions from tklayout.
Analysis- level (Thea+Deborah):
1) What is the eta distribution of the 4 jets from HH → bb decay ?
2) What is the eta distribution of the most central jet ? The second most central jet ? The third most central jet ? The most forward jet ?
3) From (2) above, can we define eta regions we are most interested in for track resolution numbers ?
4) What is the median track PT in the most central jet ? The second most central jet ? The third most central jet ? The most forward jet ?
5) From (4) above, can we define track PT regions that we are most interested in for each of the eta regions we defined in (3) above ?
6) From (3) and (5) above, we have the tklayout developers use these regions of Pt and Eta to gather the resolutions ?
tklayout (Annapaola+Daniel):
1) someone go in the tklayout code, find the resolution calculations for the “pixel” part and the “tracker” part and figure out for sure what is included.
We removed the pixel detector entirely and the tracker resolutions changed, confirming that the tracker resolutions are indeed pixel + tracker. The resolutions are compared in a table under “change tracker layout”.
2) remove an outer tracker layer (the 5th one or 2nd one) in one design and see the effect on resolutions
Removing one layer from the outer tracker did not affect in a relevant way the resolution. The effect on the impact parameter resolution is negligible. The pt resolution results to be degraded a bit.
3) what is the effect on resolution of doubling the material in the pixel material configuration by just doubling the mass of all the components ?
4) what is the effect if we instead approximate the doubling of the material by making the modules “carry” the extra mass from doubling the material ? Perhaps by making the sensors thicker or adding a layer to the modules with infinite resolution (10 times bigger).
5) Are (3) and (4) consistent ?
7) * Can we implement tilted pixel modules - perhaps in the FPIX region ? There are examples on how to do this in Mersi’s private area.
Specific tklayout designs to compare (Camilla+Jennifer):
0) Compile tklayout resolutions we have already done into a spreadsheet. Each design included should only make one change relative to previous designs so we know the effect of each change.
1) Compile results for Roland straw man 5.
2) Compile results for Phase 1 design (default in phase 2 configuration).
3) Compile results for Phase 1 design (default in phase 2 configuration), but doubling the material. tripling the material.
4) Compile results for Phase 1 design, but using better resolution in layer 1.
5) Compile results for Phase 1 design, but using better resolution in layer 1, and increasing the material in layer 1 proportionately.
6) compile results for Phase 1 design, but improving resolution to new values (but keeping the same bpix, fpix module locations)
7) compile results for Phase 1 design, but improving resolution to new values (but keeping the same bpix, fpix module locations), and adding in extra material
Analysis work (afternoon):
1) Build up H→bb analysis finder that can be used in all HH→bb+X analysis
2) Each person takes a different HH channel and inputs the object ID found above, and determines the total number of events selected.
3) Add b-tagging efficiencies to HH channel selection.
4) Next run analysis selection over irreducible backgrounds. Ie, those that have the same real objects as the signal. (tt→bbWW is background for HH→bbWW).
5) Next run analysis selection over reducible backgrounds using estimates of b-jet mis-identification on non-b-jets. For instance, for signal HH→bbbb, run over ttbar background and apply mis-identification to charm jets and light jets and b-jet id to b-jets. For HH→bbbb, run over QCD sample, and apply b-jet, charm-jet, light jet b-tag parametrizations.
