-- ElisabettaGallo - 2017-05-04

Performance of online b tagging in proton-proton collisions at 13 TeV at the CMS experiment

CMS DP_2017-011

Online vs offline b-tagging performance (uds-quark)
Performance of the online (red) and offline (blue) b jet identification efficiency demonstrating the probability for uds jets to be misidentified as b jet as a function of the efficiency to correctly identify b jets. Both online and offline b-tagging are performed using CSVv2 algorithm. The curves are obtained for the eight leading online and offline jets with pT> 30 GeV and abs(eta)<2.4 in simulated tt events, b jets from gluon splitting to a pair of b quarks are considered as b jets. The plot is obtained using the 2016 detector conditions with 35 pile-up collisions. Only events having a sum of online jet pT > 250 GeV are considered. The performance in this figure serves as an illustration since the b jet identification efficiency depends on the pT and eta distribution of the jets in the topology as well as the amount of b jets from gluon splitting in the sample. [Get pdf version]
Contact: Silvio Donato
img5192a6d8997a3ebabb040fc6ea87937b.png

Online vs offline b-tagging performance (c-quark)
Performance of the online (red) and offline (blue) b jet identification efficiency demonstrating the probability for c jets to be misidentified as b jet as a function of the efficiency to correctly identify b jets. Both online and offline b-tagging are performed using CSVv2 algorithm. The curves are obtained for the eight leading online and offline jets with pT> 30 GeV and abs(eta)<2.4 in simulated tt events, b jets from gluon splitting to a pair of b quarks are considered as b jets. The plot is obtained using the 2016 detector conditions with 35 pile-up collisions. Only events having a sum of online jet pT > 250 GeV are considered. The performance in this figure serves as an illustration since the b jet identification efficiency depends on the pT and eta distribution of the jets in the topology as well as the amount of b jets from gluon splitting in the sample. [Get pdf version]
Contact: Silvio Donato
img314b7cbfe0bf513c8bb97a72965d1bd1.png

Online vertex resolution
Online vertex resolution using the full tracker (red) and only the pixel tracker (blue). The plots are obtained using simulated multijet events generated with a flat pT parton distribution between 15 GeV and 3000 GeV. The plot is obtained using the 2016 detector conditions with 35 pile-up collisions. Only events having a sum of online jet pT > 250 GeV are considered. [Get pdf version]
Contact: Silvio Donato
imgbb1abc520145617995f2607597ef9b90.png

Fast primary vertex resolution
Resolution of the fast primary vertex algorithm that finds the primary vertex projecting the pixel hits back to the beam line using the jets direction. The fast primary vertex algorithm is used as a first step of the online b tagging, in order to reconstruct only the tracks close to the primary vertex. In 2017, a z-depended bias correction will be applied in order to remove the two-peak feature of the plot. The plot is obtained using simulated multijet events generated with a leading jet pT distribution flat between 15 GeV and 3000 GeV. The plot is obtained using the 2016 detector conditions with 35 pile-up collisions. Only events having a sum of online jet pT > 250 GeV are considered. [Get pdf version]
Contact: Silvio Donato
imga1156ef79331e6927f474574805b8a96.png

Online b-tagging computing time
Processing time of the online regional tracking to b tag the leading eight jets with pT > 30 GeV (red) and of the online tracking for particle flow (blue). For reference, the limit on the average global time of the HLT farm is about 200 ms. The plot is obtained using 2016 data with an average pile-up of 49 and having a sum of online jet pT > 250 GeV. The online b tagging was run in ~6% of events accepted by L1 trigger. The number in the parentheses is the average processing time per event. Overflow events are shown in the last bin and are taken into account in the average processing time computation. [Get pdf version]
Contact: Silvio Donato
img3bbaed07d61348352bc14b1e0fdc62a1.png
Topic attachments
I Attachment History Action Size Date Who Comment
PDFpdf btagPaper.pdf r1 manage 21.0 K 2017-05-04 - 09:21 ElisabettaGallo  
PNGpng btagPaper.png r1 manage 31.8 K 2017-05-04 - 09:21 ElisabettaGallo  
PDFpdf btagPaperFastPV.pdf r1 manage 15.0 K 2017-05-04 - 09:21 ElisabettaGallo  
PNGpng btagPaperFastPV.png r1 manage 27.0 K 2017-05-04 - 09:21 ElisabettaGallo  
PDFpdf btaggingVsC.pdf r1 manage 18.2 K 2017-05-04 - 09:21 ElisabettaGallo  
PNGpng btaggingVsC.png r1 manage 22.3 K 2017-05-04 - 09:21 ElisabettaGallo  
PDFpdf btaggingVsUds.pdf r1 manage 18.3 K 2017-05-04 - 09:21 ElisabettaGallo  
PNGpng btaggingVsUds.png r1 manage 23.4 K 2017-05-04 - 09:21 ElisabettaGallo  
PDFpdf timing.pdf r1 manage 16.8 K 2017-05-04 - 09:21 ElisabettaGallo  
PNGpng timing.png r1 manage 28.3 K 2017-05-04 - 09:21 ElisabettaGallo  
Edit | Attach | Watch | Print version | History: r2 < r1 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r2 - 2017-05-04 - ElisabettaGallo
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    CMSPublic All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback