Compact Muon Solenoid
LHC, CERN
Visit us: CMS Public Website, CMS Physics ;
Contact us: CMS Publications Committee
| CMS-PAS-BTV-20-001 | ||
| Calibration of charm jet identification algorithms using proton-proton collision events at $\sqrt{s}= $ 13 TeV | ||
| CMS Collaboration | ||
| March 2021 | ||
| Abstract: Many measurements at the LHC experiments require an efficient identification of heavy-flavour jets, i.e. jets originating from bottom (b) or charm (c) quarks. An overview of the algorithms used for c jet identification in the CMS experiment is given and a novel method to calibrate them is presented. The new method corrects the entire distribution expected as output when the algorithms are applied to jets of different flavours. It is based on an iterative method exploiting three distinct control regions that are enriched with either b jets, c jets or light-flavour jets. Finally, a validation of the method is performed by checking closure of the measured correction factors on the same collision data as well as by testing the method on toy datasets which emulate different miscalibration conditions. The calibrated results improve over traditional efficiency measurements and are expected to increase the sensitivity of future physics analysis by facilitating the use of the full distributions of heavy-flavour identification algorithm outputs, for example, as inputs to machine learning algorithms. | ||
|
Links:
CDS record (PDF) ;
inSPIRE record ;
CADI line (restricted) ;
These preliminary results are superseded in this paper, Submitted to JINST. The superseded preliminary plots can be found here. |
||
| Figures | |
png pdf |
Figure 1:
Normalised distributions of the CvsL (left) and CvsB (right) discriminators for the DeepCSV (dashed) and DeepJet (full) algorithms using jets from simulated hadronic $\mathrm{t\bar{t}}$ events with ${p_{\mathrm {T}}} > $ 20 GeV and $ {| \eta |} < $ 2.5. The distribution is shown for b jets (red), c jets (green) and light-flavour jets (blue) separately. |
png pdf |
Figure 1-a:
Normalised distributions of the CvsL (left) and CvsB (right) discriminators for the DeepCSV (dashed) and DeepJet (full) algorithms using jets from simulated hadronic $\mathrm{t\bar{t}}$ events with ${p_{\mathrm {T}}} > $ 20 GeV and $ {| \eta |} < $ 2.5. The distribution is shown for b jets (red), c jets (green) and light-flavour jets (blue) separately. |
png pdf |
Figure 1-b:
Normalised distributions of the CvsL (left) and CvsB (right) discriminators for the DeepCSV (dashed) and DeepJet (full) algorithms using jets from simulated hadronic $\mathrm{t\bar{t}}$ events with ${p_{\mathrm {T}}} > $ 20 GeV and $ {| \eta |} < $ 2.5. The distribution is shown for b jets (red), c jets (green) and light-flavour jets (blue) separately. |
png pdf |
Figure 2:
ROC curves showing the individual performance of the CvsL (left) and CvsB (right) discriminators for the DeepCSV (blue) and DeepJet (red) algorithms using jets from simulated hadronic $\mathrm{t\bar{t}}$ events with ${p_{\mathrm {T}}} > $ 20 GeV and $ {| \eta |} < $ 2.5. |
png pdf |
Figure 2-a:
ROC curves showing the individual performance of the CvsL (left) and CvsB (right) discriminators for the DeepCSV (blue) and DeepJet (red) algorithms using jets from simulated hadronic $\mathrm{t\bar{t}}$ events with ${p_{\mathrm {T}}} > $ 20 GeV and $ {| \eta |} < $ 2.5. |
png pdf |
Figure 2-b:
ROC curves showing the individual performance of the CvsL (left) and CvsB (right) discriminators for the DeepCSV (blue) and DeepJet (red) algorithms using jets from simulated hadronic $\mathrm{t\bar{t}}$ events with ${p_{\mathrm {T}}} > $ 20 GeV and $ {| \eta |} < $ 2.5. |
png pdf |
Figure 3:
Two-dimensional ROC contours showing the c-tagging efficiency as a simultaneous function of b jet and light-flavour jet mistagging rates for DeepCSV (blue) and DeepJet (red) algorithms using jets from simulated hadronic $\mathrm{t\bar{t}}$ events with ${p_{\mathrm {T}}} > $ 20 GeV and $ {| \eta |} < $ 2.5. |
png pdf |
Figure 4:
Normalised two-dimensional distributions showing the CvsL and CvsB discriminators on the x-axis and y-axis respectively. Distributions are shown using c jets (upper), b jets (middle) and light-flavour jets (lower) from simulated hadronic $\mathrm{t\bar{t}}$ events with ${p_{\mathrm {T}}} > $ 20 GeV and $ {| \eta |} < $ 2.5. The left-hand column shows the discriminators of the DeepCSV algorithm, whereas the right-hand column shows those of the DeepJet algorithm. |
png pdf |
Figure 4-a:
Normalised two-dimensional distributions showing the CvsL and CvsB discriminators on the x-axis and y-axis respectively. Distributions are shown using c jets (upper), b jets (middle) and light-flavour jets (lower) from simulated hadronic $\mathrm{t\bar{t}}$ events with ${p_{\mathrm {T}}} > $ 20 GeV and $ {| \eta |} < $ 2.5. The left-hand column shows the discriminators of the DeepCSV algorithm, whereas the right-hand column shows those of the DeepJet algorithm. |
png pdf |
Figure 4-b:
Normalised two-dimensional distributions showing the CvsL and CvsB discriminators on the x-axis and y-axis respectively. Distributions are shown using c jets (upper), b jets (middle) and light-flavour jets (lower) from simulated hadronic $\mathrm{t\bar{t}}$ events with ${p_{\mathrm {T}}} > $ 20 GeV and $ {| \eta |} < $ 2.5. The left-hand column shows the discriminators of the DeepCSV algorithm, whereas the right-hand column shows those of the DeepJet algorithm. |
png pdf |
Figure 4-c:
Normalised two-dimensional distributions showing the CvsL and CvsB discriminators on the x-axis and y-axis respectively. Distributions are shown using c jets (upper), b jets (middle) and light-flavour jets (lower) from simulated hadronic $\mathrm{t\bar{t}}$ events with ${p_{\mathrm {T}}} > $ 20 GeV and $ {| \eta |} < $ 2.5. The left-hand column shows the discriminators of the DeepCSV algorithm, whereas the right-hand column shows those of the DeepJet algorithm. |
png pdf |
Figure 4-d:
Normalised two-dimensional distributions showing the CvsL and CvsB discriminators on the x-axis and y-axis respectively. Distributions are shown using c jets (upper), b jets (middle) and light-flavour jets (lower) from simulated hadronic $\mathrm{t\bar{t}}$ events with ${p_{\mathrm {T}}} > $ 20 GeV and $ {| \eta |} < $ 2.5. The left-hand column shows the discriminators of the DeepCSV algorithm, whereas the right-hand column shows those of the DeepJet algorithm. |
png pdf |
Figure 4-e:
Normalised two-dimensional distributions showing the CvsL and CvsB discriminators on the x-axis and y-axis respectively. Distributions are shown using c jets (upper), b jets (middle) and light-flavour jets (lower) from simulated hadronic $\mathrm{t\bar{t}}$ events with ${p_{\mathrm {T}}} > $ 20 GeV and $ {| \eta |} < $ 2.5. The left-hand column shows the discriminators of the DeepCSV algorithm, whereas the right-hand column shows those of the DeepJet algorithm. |
png pdf |
Figure 4-f:
Normalised two-dimensional distributions showing the CvsL and CvsB discriminators on the x-axis and y-axis respectively. Distributions are shown using c jets (upper), b jets (middle) and light-flavour jets (lower) from simulated hadronic $\mathrm{t\bar{t}}$ events with ${p_{\mathrm {T}}} > $ 20 GeV and $ {| \eta |} < $ 2.5. The left-hand column shows the discriminators of the DeepCSV algorithm, whereas the right-hand column shows those of the DeepJet algorithm. |
png pdf |
Figure 5:
Feynman diagrams showing production of charm quarks in association with W boson at the LHC (left and middle) along with the major background (right). |
png pdf |
Figure 6:
Pre-calibration distributions of CvsL (left) and CvsB (right) obtained from the DeepCSV (upper) and DeepJet (lower) taggers for jets selected in the W+c (OS-SS) selection. The bin corresponding to a tagger value of $-1$ is plotted at $-0.1$ for DeepCSV. |
png pdf |
Figure 6-a:
Pre-calibration distributions of CvsL (left) and CvsB (right) obtained from the DeepCSV (upper) and DeepJet (lower) taggers for jets selected in the W+c (OS-SS) selection. The bin corresponding to a tagger value of $-1$ is plotted at $-0.1$ for DeepCSV. |
png pdf |
Figure 6-b:
Pre-calibration distributions of CvsL (left) and CvsB (right) obtained from the DeepCSV (upper) and DeepJet (lower) taggers for jets selected in the W+c (OS-SS) selection. The bin corresponding to a tagger value of $-1$ is plotted at $-0.1$ for DeepCSV. |
png pdf |
Figure 6-c:
Pre-calibration distributions of CvsL (left) and CvsB (right) obtained from the DeepCSV (upper) and DeepJet (lower) taggers for jets selected in the W+c (OS-SS) selection. The bin corresponding to a tagger value of $-1$ is plotted at $-0.1$ for DeepCSV. |
png pdf |
Figure 6-d:
Pre-calibration distributions of CvsL (left) and CvsB (right) obtained from the DeepCSV (upper) and DeepJet (lower) taggers for jets selected in the W+c (OS-SS) selection. The bin corresponding to a tagger value of $-1$ is plotted at $-0.1$ for DeepCSV. |
png pdf |
Figure 7:
Pre-calibration distributions of CvsL (left) and CvsB (right) obtained from the DeepCSV (upper) and DeepJet (lower) taggers for jets selected in the $\mathrm{t\bar{t}}$ selection. The bin corresponding to a tagger value of $-1$ is plotted at $-0.1$. |
png pdf |
Figure 7-a:
Pre-calibration distributions of CvsL (left) and CvsB (right) obtained from the DeepCSV (upper) and DeepJet (lower) taggers for jets selected in the $\mathrm{t\bar{t}}$ selection. The bin corresponding to a tagger value of $-1$ is plotted at $-0.1$. |
png pdf |
Figure 7-b:
Pre-calibration distributions of CvsL (left) and CvsB (right) obtained from the DeepCSV (upper) and DeepJet (lower) taggers for jets selected in the $\mathrm{t\bar{t}}$ selection. The bin corresponding to a tagger value of $-1$ is plotted at $-0.1$. |
png pdf |
Figure 7-c:
Pre-calibration distributions of CvsL (left) and CvsB (right) obtained from the DeepCSV (upper) and DeepJet (lower) taggers for jets selected in the $\mathrm{t\bar{t}}$ selection. The bin corresponding to a tagger value of $-1$ is plotted at $-0.1$. |
png pdf |
Figure 7-d:
Pre-calibration distributions of CvsL (left) and CvsB (right) obtained from the DeepCSV (upper) and DeepJet (lower) taggers for jets selected in the $\mathrm{t\bar{t}}$ selection. The bin corresponding to a tagger value of $-1$ is plotted at $-0.1$. |
png pdf |
Figure 8:
Pre-calibration distributions of CvsL (left) and CvsB (right) obtained from the DeepCSV (upper) and DeepJet (lower) tagger for jets selected in the DY+jet selection. The bin corresponding to a tagger value of $-1$ is plotted at $-0.1$. |
png pdf |
Figure 8-a:
Pre-calibration distributions of CvsL (left) and CvsB (right) obtained from the DeepCSV (upper) and DeepJet (lower) tagger for jets selected in the DY+jet selection. The bin corresponding to a tagger value of $-1$ is plotted at $-0.1$. |
png pdf |
Figure 8-b:
Pre-calibration distributions of CvsL (left) and CvsB (right) obtained from the DeepCSV (upper) and DeepJet (lower) tagger for jets selected in the DY+jet selection. The bin corresponding to a tagger value of $-1$ is plotted at $-0.1$. |
png pdf |
Figure 8-c:
Pre-calibration distributions of CvsL (left) and CvsB (right) obtained from the DeepCSV (upper) and DeepJet (lower) tagger for jets selected in the DY+jet selection. The bin corresponding to a tagger value of $-1$ is plotted at $-0.1$. |
png pdf |
Figure 8-d:
Pre-calibration distributions of CvsL (left) and CvsB (right) obtained from the DeepCSV (upper) and DeepJet (lower) tagger for jets selected in the DY+jet selection. The bin corresponding to a tagger value of $-1$ is plotted at $-0.1$. |
png pdf |
Figure 9:
The shape calibration scale factor maps for DeepCSV- (left) and DeepJet- (right) based c-taggers for c jets are shown. SF$_c^{(-1)}$ denotes the SF for c jets with defaulted discriminator values, along with statistical (first term) and systematic (second term) uncertainties. The total uncertainty is denoted by red envelopes around the central values, while statistical uncertainties alone are denoted by black lines. Grey datapoints with hatched uncertainties denote bins with statistics insufficient for SF evaluation. |
png pdf |
Figure 10:
The shape calibration scale factor maps for DeepCSV- (left) and DeepJet- (right) based c-taggers for b jets are shown. SF$_b^{(-1)}$ denotes the SF for b jets with defaulted discriminator values, along with statistical (first term) and systematic (second term) uncertainties. The total uncertainty is denoted by red envelopes around the central values, while statistical uncertainties alone are denoted by black lines. Grey datapoints with hatched uncertainties denote bins with statistics insufficient for SF evaluation. |
png pdf |
Figure 11:
The shape calibration scale factor maps for DeepCSV- (left) and DeepJet- (right) based c-taggers for light-flavour jets are shown. SF$_{\mathrm {udsg}}^{(-1)}$ denotes the SF for light-flavour jets with defaulted discriminator values, along with statistical (first term) and systematic (second term) uncertainties. The total uncertainty is denoted by red envelopes around the central values, while statistical uncertainties alone are denoted by black lines. Grey datapoints with hatched uncertainties denote bins with statistics insufficient for SF evaluation. |
png pdf |
Figure 12:
Contribution of each source of SF uncertainty, calculated as the square of the relative uncertainty in jet yield and expressed as the maximum of the up and down variations, at various values of the DeepCSV CvsL (left) and CvsB (right) discriminators for c (upper), b (middle) and light (lower) flavours. The effective total relative uncertainty values per bin are also shown in grey, for reference. The bin corresponding to a tagger value of $-1$ is plotted at $-0.1$. Statistical uncertainties are not shown. |
png pdf |
Figure 12-a:
Contribution of each source of SF uncertainty, calculated as the square of the relative uncertainty in jet yield and expressed as the maximum of the up and down variations, at various values of the DeepCSV CvsL (left) and CvsB (right) discriminators for c (upper), b (middle) and light (lower) flavours. The effective total relative uncertainty values per bin are also shown in grey, for reference. The bin corresponding to a tagger value of $-1$ is plotted at $-0.1$. Statistical uncertainties are not shown. |
png pdf |
Figure 12-b:
Contribution of each source of SF uncertainty, calculated as the square of the relative uncertainty in jet yield and expressed as the maximum of the up and down variations, at various values of the DeepCSV CvsL (left) and CvsB (right) discriminators for c (upper), b (middle) and light (lower) flavours. The effective total relative uncertainty values per bin are also shown in grey, for reference. The bin corresponding to a tagger value of $-1$ is plotted at $-0.1$. Statistical uncertainties are not shown. |
png pdf |
Figure 12-c:
Contribution of each source of SF uncertainty, calculated as the square of the relative uncertainty in jet yield and expressed as the maximum of the up and down variations, at various values of the DeepCSV CvsL (left) and CvsB (right) discriminators for c (upper), b (middle) and light (lower) flavours. The effective total relative uncertainty values per bin are also shown in grey, for reference. The bin corresponding to a tagger value of $-1$ is plotted at $-0.1$. Statistical uncertainties are not shown. |
png pdf |
Figure 12-d:
Contribution of each source of SF uncertainty, calculated as the square of the relative uncertainty in jet yield and expressed as the maximum of the up and down variations, at various values of the DeepCSV CvsL (left) and CvsB (right) discriminators for c (upper), b (middle) and light (lower) flavours. The effective total relative uncertainty values per bin are also shown in grey, for reference. The bin corresponding to a tagger value of $-1$ is plotted at $-0.1$. Statistical uncertainties are not shown. |
png pdf |
Figure 12-e:
Contribution of each source of SF uncertainty, calculated as the square of the relative uncertainty in jet yield and expressed as the maximum of the up and down variations, at various values of the DeepCSV CvsL (left) and CvsB (right) discriminators for c (upper), b (middle) and light (lower) flavours. The effective total relative uncertainty values per bin are also shown in grey, for reference. The bin corresponding to a tagger value of $-1$ is plotted at $-0.1$. Statistical uncertainties are not shown. |
png pdf |
Figure 12-f:
Contribution of each source of SF uncertainty, calculated as the square of the relative uncertainty in jet yield and expressed as the maximum of the up and down variations, at various values of the DeepCSV CvsL (left) and CvsB (right) discriminators for c (upper), b (middle) and light (lower) flavours. The effective total relative uncertainty values per bin are also shown in grey, for reference. The bin corresponding to a tagger value of $-1$ is plotted at $-0.1$. Statistical uncertainties are not shown. |
png pdf |
Figure 13:
Contribution of each source of SF uncertainty, calculated as the square of the relative uncertainty in jet yield and expressed as the maximum of the up and down variations, at various values of the DeepJet CvsL (left) and CvsB (right) discriminators for c (upper), b (middle) and light (lower) flavours. The effective total relative uncertainty values per bin are also shown in grey, for reference. The bin corresponding to a tagger value of $-$1 is plotted at $-$0.1. Statistical uncertainties are not shown. |
png pdf |
Figure 13-a:
Contribution of each source of SF uncertainty, calculated as the square of the relative uncertainty in jet yield and expressed as the maximum of the up and down variations, at various values of the DeepJet CvsL (left) and CvsB (right) discriminators for c (upper), b (middle) and light (lower) flavours. The effective total relative uncertainty values per bin are also shown in grey, for reference. The bin corresponding to a tagger value of $-$1 is plotted at $-$0.1. Statistical uncertainties are not shown. |
png pdf |
Figure 13-b:
Contribution of each source of SF uncertainty, calculated as the square of the relative uncertainty in jet yield and expressed as the maximum of the up and down variations, at various values of the DeepJet CvsL (left) and CvsB (right) discriminators for c (upper), b (middle) and light (lower) flavours. The effective total relative uncertainty values per bin are also shown in grey, for reference. The bin corresponding to a tagger value of $-$1 is plotted at $-$0.1. Statistical uncertainties are not shown. |
png pdf |
Figure 13-c:
Contribution of each source of SF uncertainty, calculated as the square of the relative uncertainty in jet yield and expressed as the maximum of the up and down variations, at various values of the DeepJet CvsL (left) and CvsB (right) discriminators for c (upper), b (middle) and light (lower) flavours. The effective total relative uncertainty values per bin are also shown in grey, for reference. The bin corresponding to a tagger value of $-$1 is plotted at $-$0.1. Statistical uncertainties are not shown. |
png pdf |
Figure 13-d:
Contribution of each source of SF uncertainty, calculated as the square of the relative uncertainty in jet yield and expressed as the maximum of the up and down variations, at various values of the DeepJet CvsL (left) and CvsB (right) discriminators for c (upper), b (middle) and light (lower) flavours. The effective total relative uncertainty values per bin are also shown in grey, for reference. The bin corresponding to a tagger value of $-$1 is plotted at $-$0.1. Statistical uncertainties are not shown. |
png pdf |
Figure 13-e:
Contribution of each source of SF uncertainty, calculated as the square of the relative uncertainty in jet yield and expressed as the maximum of the up and down variations, at various values of the DeepJet CvsL (left) and CvsB (right) discriminators for c (upper), b (middle) and light (lower) flavours. The effective total relative uncertainty values per bin are also shown in grey, for reference. The bin corresponding to a tagger value of $-$1 is plotted at $-$0.1. Statistical uncertainties are not shown. |
png pdf |
Figure 13-f:
Contribution of each source of SF uncertainty, calculated as the square of the relative uncertainty in jet yield and expressed as the maximum of the up and down variations, at various values of the DeepJet CvsL (left) and CvsB (right) discriminators for c (upper), b (middle) and light (lower) flavours. The effective total relative uncertainty values per bin are also shown in grey, for reference. The bin corresponding to a tagger value of $-$1 is plotted at $-$0.1. Statistical uncertainties are not shown. |
png pdf |
Figure 14:
ROC curves showing the individual performance of the CvsL (left) and CvsB (right) discriminators for the DeepCSV (blue) and DeepJet (red) algorithms for simulated jets (dashed lines) and the estimation of the same for jets in data (solid lines with uncertainty bands). |
png pdf |
Figure 14-a:
ROC curves showing the individual performance of the CvsL (left) and CvsB (right) discriminators for the DeepCSV (blue) and DeepJet (red) algorithms for simulated jets (dashed lines) and the estimation of the same for jets in data (solid lines with uncertainty bands). |
png pdf |
Figure 14-b:
ROC curves showing the individual performance of the CvsL (left) and CvsB (right) discriminators for the DeepCSV (blue) and DeepJet (red) algorithms for simulated jets (dashed lines) and the estimation of the same for jets in data (solid lines with uncertainty bands). |
png pdf |
Figure 15:
ROC contours showing c-tagging efficiencies as a simultaneous function of b and light-flavour jet misidentification rate, for the DeepCSV (left) and DeepJet (right) algorithms for simulated jets (dashed lines) and the estimation of the same for jets in data (solid lines). |
png pdf |
Figure 15-a:
ROC contours showing c-tagging efficiencies as a simultaneous function of b and light-flavour jet misidentification rate, for the DeepCSV (left) and DeepJet (right) algorithms for simulated jets (dashed lines) and the estimation of the same for jets in data (solid lines). |
png pdf |
Figure 15-b:
ROC contours showing c-tagging efficiencies as a simultaneous function of b and light-flavour jet misidentification rate, for the DeepCSV (left) and DeepJet (right) algorithms for simulated jets (dashed lines) and the estimation of the same for jets in data (solid lines). |
png pdf |
Figure 16:
Relative contributions of each source of uncertainty to the total uncertainty (statistical + systematic) for both CvsL and CvsB discrimination and for both DeepCSV and DeepJet taggers, quantified by the square of the change in area under ROC curves. |
png pdf |
Figure 17:
Post-calibration DeepCSV CvsL (left) and CvsB (right) distributions of jet samples selected from W+c (upper), $\mathrm{t\bar{t}}$ semi- and dileptonic (middle), and DY+jet (lower) events after application of DeepCSV c-tagger shape calibration scale factors. |
png pdf |
Figure 17-a:
Post-calibration DeepCSV CvsL (left) and CvsB (right) distributions of jet samples selected from W+c (upper), $\mathrm{t\bar{t}}$ semi- and dileptonic (middle), and DY+jet (lower) events after application of DeepCSV c-tagger shape calibration scale factors. |
png pdf |
Figure 17-b:
Post-calibration DeepCSV CvsL (left) and CvsB (right) distributions of jet samples selected from W+c (upper), $\mathrm{t\bar{t}}$ semi- and dileptonic (middle), and DY+jet (lower) events after application of DeepCSV c-tagger shape calibration scale factors. |
png pdf |
Figure 17-c:
Post-calibration DeepCSV CvsL (left) and CvsB (right) distributions of jet samples selected from W+c (upper), $\mathrm{t\bar{t}}$ semi- and dileptonic (middle), and DY+jet (lower) events after application of DeepCSV c-tagger shape calibration scale factors. |
png pdf |
Figure 17-d:
Post-calibration DeepCSV CvsL (left) and CvsB (right) distributions of jet samples selected from W+c (upper), $\mathrm{t\bar{t}}$ semi- and dileptonic (middle), and DY+jet (lower) events after application of DeepCSV c-tagger shape calibration scale factors. |
png pdf |
Figure 17-e:
Post-calibration DeepCSV CvsL (left) and CvsB (right) distributions of jet samples selected from W+c (upper), $\mathrm{t\bar{t}}$ semi- and dileptonic (middle), and DY+jet (lower) events after application of DeepCSV c-tagger shape calibration scale factors. |
png pdf |
Figure 17-f:
Post-calibration DeepCSV CvsL (left) and CvsB (right) distributions of jet samples selected from W+c (upper), $\mathrm{t\bar{t}}$ semi- and dileptonic (middle), and DY+jet (lower) events after application of DeepCSV c-tagger shape calibration scale factors. |
png pdf |
Figure 18:
Post-calibration DeepJet CvsL (left) and CvsB (right) distributions of jet samples selected from W+c (upper), $\mathrm{t\bar{t}}$ semi- and dileptonic (middle), and DY+jet (lower) events after application of DeepJet c-tagger shape calibration scale factors. |
png pdf |
Figure 18-a:
Post-calibration DeepJet CvsL (left) and CvsB (right) distributions of jet samples selected from W+c (upper), $\mathrm{t\bar{t}}$ semi- and dileptonic (middle), and DY+jet (lower) events after application of DeepJet c-tagger shape calibration scale factors. |
png pdf |
Figure 18-b:
Post-calibration DeepJet CvsL (left) and CvsB (right) distributions of jet samples selected from W+c (upper), $\mathrm{t\bar{t}}$ semi- and dileptonic (middle), and DY+jet (lower) events after application of DeepJet c-tagger shape calibration scale factors. |
png pdf |
Figure 18-c:
Post-calibration DeepJet CvsL (left) and CvsB (right) distributions of jet samples selected from W+c (upper), $\mathrm{t\bar{t}}$ semi- and dileptonic (middle), and DY+jet (lower) events after application of DeepJet c-tagger shape calibration scale factors. |
png pdf |
Figure 18-d:
Post-calibration DeepJet CvsL (left) and CvsB (right) distributions of jet samples selected from W+c (upper), $\mathrm{t\bar{t}}$ semi- and dileptonic (middle), and DY+jet (lower) events after application of DeepJet c-tagger shape calibration scale factors. |
png pdf |
Figure 18-e:
Post-calibration DeepJet CvsL (left) and CvsB (right) distributions of jet samples selected from W+c (upper), $\mathrm{t\bar{t}}$ semi- and dileptonic (middle), and DY+jet (lower) events after application of DeepJet c-tagger shape calibration scale factors. |
png pdf |
Figure 18-f:
Post-calibration DeepJet CvsL (left) and CvsB (right) distributions of jet samples selected from W+c (upper), $\mathrm{t\bar{t}}$ semi- and dileptonic (middle), and DY+jet (lower) events after application of DeepJet c-tagger shape calibration scale factors. |
png pdf |
Figure 19:
DeepCSV CvsB (first row), DeepCSV CvsL (second row), DeepJet CvsB (third row) and DeepJet CvsL (last row) discriminators of soft-muon-bias-free semileptonic $\mathrm{t\bar{t}}$ jets, before (left) and after (right) application of scale factors. |
png pdf |
Figure 19-a:
DeepCSV CvsB (first row), DeepCSV CvsL (second row), DeepJet CvsB (third row) and DeepJet CvsL (last row) discriminators of soft-muon-bias-free semileptonic $\mathrm{t\bar{t}}$ jets, before (left) and after (right) application of scale factors. |
png pdf |
Figure 19-b:
DeepCSV CvsB (first row), DeepCSV CvsL (second row), DeepJet CvsB (third row) and DeepJet CvsL (last row) discriminators of soft-muon-bias-free semileptonic $\mathrm{t\bar{t}}$ jets, before (left) and after (right) application of scale factors. |
png pdf |
Figure 19-c:
DeepCSV CvsB (first row), DeepCSV CvsL (second row), DeepJet CvsB (third row) and DeepJet CvsL (last row) discriminators of soft-muon-bias-free semileptonic $\mathrm{t\bar{t}}$ jets, before (left) and after (right) application of scale factors. |
png pdf |
Figure 19-d:
DeepCSV CvsB (first row), DeepCSV CvsL (second row), DeepJet CvsB (third row) and DeepJet CvsL (last row) discriminators of soft-muon-bias-free semileptonic $\mathrm{t\bar{t}}$ jets, before (left) and after (right) application of scale factors. |
png pdf |
Figure 19-e:
DeepCSV CvsB (first row), DeepCSV CvsL (second row), DeepJet CvsB (third row) and DeepJet CvsL (last row) discriminators of soft-muon-bias-free semileptonic $\mathrm{t\bar{t}}$ jets, before (left) and after (right) application of scale factors. |
png pdf |
Figure 19-f:
DeepCSV CvsB (first row), DeepCSV CvsL (second row), DeepJet CvsB (third row) and DeepJet CvsL (last row) discriminators of soft-muon-bias-free semileptonic $\mathrm{t\bar{t}}$ jets, before (left) and after (right) application of scale factors. |
png pdf |
Figure 19-g:
DeepCSV CvsB (first row), DeepCSV CvsL (second row), DeepJet CvsB (third row) and DeepJet CvsL (last row) discriminators of soft-muon-bias-free semileptonic $\mathrm{t\bar{t}}$ jets, before (left) and after (right) application of scale factors. |
png pdf |
Figure 19-h:
DeepCSV CvsB (first row), DeepCSV CvsL (second row), DeepJet CvsB (third row) and DeepJet CvsL (last row) discriminators of soft-muon-bias-free semileptonic $\mathrm{t\bar{t}}$ jets, before (left) and after (right) application of scale factors. |
png pdf |
Figure 20:
DeepCSV CvsB (first row), DeepCSV CvsL (second row), DeepJet CvsB (third row) and DeepJet CvsL (fourth row) discriminators of soft-muon-bias-free dileptonic $\mathrm{t\bar{t}}$ jets, before (left) and after (right) application of scale factors. |
png pdf |
Figure 20-a:
DeepCSV CvsB (first row), DeepCSV CvsL (second row), DeepJet CvsB (third row) and DeepJet CvsL (fourth row) discriminators of soft-muon-bias-free dileptonic $\mathrm{t\bar{t}}$ jets, before (left) and after (right) application of scale factors. |
png pdf |
Figure 20-b:
DeepCSV CvsB (first row), DeepCSV CvsL (second row), DeepJet CvsB (third row) and DeepJet CvsL (fourth row) discriminators of soft-muon-bias-free dileptonic $\mathrm{t\bar{t}}$ jets, before (left) and after (right) application of scale factors. |
png pdf |
Figure 20-c:
DeepCSV CvsB (first row), DeepCSV CvsL (second row), DeepJet CvsB (third row) and DeepJet CvsL (fourth row) discriminators of soft-muon-bias-free dileptonic $\mathrm{t\bar{t}}$ jets, before (left) and after (right) application of scale factors. |
png pdf |
Figure 20-d:
DeepCSV CvsB (first row), DeepCSV CvsL (second row), DeepJet CvsB (third row) and DeepJet CvsL (fourth row) discriminators of soft-muon-bias-free dileptonic $\mathrm{t\bar{t}}$ jets, before (left) and after (right) application of scale factors. |
png pdf |
Figure 20-e:
DeepCSV CvsB (first row), DeepCSV CvsL (second row), DeepJet CvsB (third row) and DeepJet CvsL (fourth row) discriminators of soft-muon-bias-free dileptonic $\mathrm{t\bar{t}}$ jets, before (left) and after (right) application of scale factors. |
png pdf |
Figure 20-f:
DeepCSV CvsB (first row), DeepCSV CvsL (second row), DeepJet CvsB (third row) and DeepJet CvsL (fourth row) discriminators of soft-muon-bias-free dileptonic $\mathrm{t\bar{t}}$ jets, before (left) and after (right) application of scale factors. |
png pdf |
Figure 20-g:
DeepCSV CvsB (first row), DeepCSV CvsL (second row), DeepJet CvsB (third row) and DeepJet CvsL (fourth row) discriminators of soft-muon-bias-free dileptonic $\mathrm{t\bar{t}}$ jets, before (left) and after (right) application of scale factors. |
png pdf |
Figure 20-h:
DeepCSV CvsB (first row), DeepCSV CvsL (second row), DeepJet CvsB (third row) and DeepJet CvsL (fourth row) discriminators of soft-muon-bias-free dileptonic $\mathrm{t\bar{t}}$ jets, before (left) and after (right) application of scale factors. |
png pdf |
Figure 21:
Distribution of the SF pulls, quantified as the differences between the injected SFs and the SFs retrieved by the fits in units of the statistical uncertainties in the latter ($\frac {\mathrm {SF}_{\mathrm {extracted}}-\mathrm {SF}_{\mathrm {injected}}}{\sigma _{\mathrm {extracted}}}$), across all bins in the CvsB-CvsL plane, for the SF map with "mild" (left) and "strong" (right) SFs. |
png pdf |
Figure 21-a:
Distribution of the SF pulls, quantified as the differences between the injected SFs and the SFs retrieved by the fits in units of the statistical uncertainties in the latter ($\frac {\mathrm {SF}_{\mathrm {extracted}}-\mathrm {SF}_{\mathrm {injected}}}{\sigma _{\mathrm {extracted}}}$), across all bins in the CvsB-CvsL plane, for the SF map with "mild" (left) and "strong" (right) SFs. |
png pdf |
Figure 21-b:
Distribution of the SF pulls, quantified as the differences between the injected SFs and the SFs retrieved by the fits in units of the statistical uncertainties in the latter ($\frac {\mathrm {SF}_{\mathrm {extracted}}-\mathrm {SF}_{\mathrm {injected}}}{\sigma _{\mathrm {extracted}}}$), across all bins in the CvsB-CvsL plane, for the SF map with "mild" (left) and "strong" (right) SFs. |
| Tables | |
png pdf |
Table 1:
Summary of the heavy-flavour tagging definitions for both b- and c-tagging using the DeepCSV and DeepJet taggers. |
png pdf |
Table 2:
The combined jet yield and contribution of each flavour of jet to each selection is shown. The number of events is reported from data, while the per-flavour contribution is determined from simulation. The purity of each selection (row) is highlighted in bold text. |
| Summary |
|
This note presents a novel method to calibrate the full differential shape of the discriminator distributions used for c jet identification at CMS. The method uses three different sets of event selection criteria, targeting topologies enriched in W+c, top quark pairs and DY+jet events. These topologies are highly enriched in c, b and light-flavour jets respectively, resulting in purities of a given jet-flavour that range between 81% and 93%. By employing an iterative fitting approach in each of these three regions, scale factors are derived to match the simulated discriminator distributions to those observed in data. Given that the c-tagging algorithm is composed of two discriminators, one to discriminate c from b jets (CvsB) and another to discriminate c from light-flavour and gluon jets (CvsL), the scale factors are derived as a function of the two-dimensional CvsL and CvsB discriminator values. An adaptive binning is adopted to optimise the granularity of the provided calibration with respect to the statistical uncertainty in each bin. We present validation and closure tests that validate the robustness of the method. Although not reported here, the method has also been demonstrated to work in the context of a search for associated production of a Higgs boson with a vector boson, where the Higgs boson decays into a pair of charm quarks [9]. The calibration of the full differential discriminator shape allows to use the c-tagging discriminators as an input to multivariate techniques (based on machine learning) or by fitting the discriminator shapes to data to extract observables which are sensitive to the jet-flavour. The shape-calibration extends the use of the c-tagging algorithms beyond the application of discrete working points, and will result in more advanced use-cases for c jet identification in physics analyses. |
| References | ||||
| 1 | CMS Collaboration | Identification of b-quark jets with the CMS experiment | JINST 8 (2013) P04013 | CMS-BTV-12-001 1211.4462 |
| 2 | CMS Collaboration | Identification of heavy-flavour jets with the CMS detector in pp collisions at 13 TeV | JINST 13 (2018) P05011 | CMS-BTV-16-002 1712.07158 |
| 3 | CMS Collaboration | Identification of c-quark jets at the CMS experiment | CMS-PAS-BTV-16-001 | CMS-PAS-BTV-16-001 |
| 4 | E. Bols et al. | Jet flavour classification using DeepJet | JINST 15 (2020) P12012 | 2008.10519 |
| 5 | CMS Collaboration | Performance of the DeepJet b tagging algorithm using 41.9/fb of data from proton-proton collisions at 13 TeV with Phase 1 CMS detector | CDS | |
| 6 | D. Guest et al. | Jet flavor classification in high-energy physics with deep neural networks | Physical Review D 94 (2016) | 1607.08633 |
| 7 | CMS Collaboration | Observation of Higgs boson decay to bottom quarks | PRL 121 (2018) 121801 | CMS-HIG-18-016 1808.08242 |
| 8 | S. Moortgat | When charm and beauty adjoin the top. First measurement of the cross section of top quark pair production with additional charm jets with the CMS experiment | PhD thesis, Vrije U., Brussels, May, 2019 CERN-THESIS-2019-051 | |
| 9 | CMS Collaboration | A search for the standard model Higgs boson decaying to charm quarks | JHEP 03 (2020) 131 | CMS-HIG-18-031 1912.01662 |
| 10 | CMS Collaboration | CMS technical design report for the pixel detector upgrade | CDS | |
| 11 | CMS Collaboration | The CMS experiment at the CERN LHC | JINST 3 (2008) S08004 | CMS-00-001 |
| 12 | CMS Collaboration | The CMS trigger system | JINST 12 (2017) P01020 | CMS-TRG-12-001 1609.02366 |
| 13 | CMS Collaboration | CMS luminosity measurement for the 2017 data-taking period at $ \sqrt{s} = $ 13 TeV | ||
| 14 | T. Sjostrand et al. | An introduction to PYTHIA 8.2 | Computer Physics Communications 191 (2015) 159 | |
| 15 | P. Skands, S. Carrazza, and J. Rojo | Tuning PYTHIA 8.1: the Monash 2013 Tune | EPJC 74 (2014) 3024 | 1404.5630 |
| 16 | NNPDF Collaboration | Parton distributions for the LHC Run II | JHEP 04 (2015) 040 | 1410.8849 |
| 17 | P. Nason | A new method for combining NLO QCD with shower Monte Carlo algorithms | JHEP 11 (2004) 040 | hep-ph/0409146 |
| 18 | S. Frixione, P. Nason, and C. Oleari | Matching NLO QCD computations with parton shower simulations: the POWHEG method | JHEP 11 (2007) 070 | 0709.2092 |
| 19 | S. Alioli, P. Nason, C. Oleari, and E. Re | A general framework for implementing NLO calculations in shower Monte Carlo programs: the POWHEG BOX | JHEP 06 (2010) 043 | 1002.2581 |
| 20 | J. M. Campbell, R. K. Ellis, P. Nason, and E. Re | Top-pair production and decay at NLO matched with parton showers | JHEP 04 (2015) 114 | 1412.1828 |
| 21 | J. Alwall et al. | The automated computation of tree-level and next-to-leading order differential cross sections, and their matching to parton shower simulations | JHEP 07 (2014) 079 | 1405.0301 |
| 22 | J. Alwall et al. | Comparative study of various algorithms for the merging of parton showers and matrix elements in hadronic collisions | EPJC 53 (2008) 473 | 0706.2569 |
| 23 | Y. Li and F. Petriello | Combining QCD and electroweak corrections to dilepton production in FEWZ | PRD 86 (2012) 094034 | 1208.5967 |
| 24 | S. Frixione et al. | Single-top hadroproduction in association with a W boson | JHEP 07 (2008) 029 | 0805.3067 |
| 25 | E. Re | Single-top Wt-channel production matched with parton showers using the POWHEG method | EPJC 71 (2011) 1547 | 1009.2450 |
| 26 | N. Kidonakis | NNLL threshold resummation for top-pair and single-top production | Phys. Part. Nucl. 45 (2014) 714 | 1210.7813 |
| 27 | J. M. Campbell and R. K. Ellis | MCFM for the Tevatron and the LHC | NPPS 205-206 (2010) 10 | 1007.3492 |
| 28 | T. Gehrmann et al. | W$ ^+ $W$ ^- $ production at hadron colliders in next to next to leading order QCD | PRL 113 (2014) 212001 | 1408.5243 |
| 29 | S. Agostinelli et al. | GEANT4---a simulation toolkit | NIM A 506 (2003) 250 -- 303 | |
| 30 | CMS Collaboration | Particle-flow reconstruction and global event description with the CMS detector | JINST 12 (2017) P10003 | CMS-PRF-14-001 1706.04965 |
| 31 | CMS Collaboration | Performance of missing transverse momentum reconstruction in proton-proton collisions at $ \sqrt{s} = $ 13 TeV using the CMS detector | JINST 14 (2019) P07004 | CMS-JME-17-001 1903.06078 |
| 32 | M. Cacciari, G. P. Salam, and G. Soyez | The anti-$ k_t $ jet clustering algorithm | JHEP 04 (2008) 063 | 0802.1189 |
| 33 | M. Cacciari, G. P. Salam, and G. Soyez | FastJet user manual | EPJC 72 (2012) 1896 | 1111.6097 |
| 34 | CMS Collaboration | Jet energy scale and resolution in the CMS experiment in pp collisions at 8 TeV | JINST 12 (2017) P02014 | CMS-JME-13-004 1607.03663 |
| 35 | CMS Collaboration | Pileup mitigation at CMS in 13 TeV data | JINST 15 (2020), no. arXiv:2003.00503. CMS-JME-18-001-003. 09, P09018. 57 p, . Submitted to JINST | |
| 36 | CMS Collaboration | Measurement of $ B\bar{B} $ angular correlations based on secondary vertex reconstruction at $ \sqrt{s}= $ 7 TeV | JHEP 03 (2011) 136 | CMS-BPH-10-010 1102.3194 |
| 37 | CMS Collaboration | Measurement of associated production of a W boson and a charm quark in proton-proton collisions at $ \sqrt{s} = $ 13 TeV | The European Physical Journal C 79 (2019) 269 | CMS-SMP-17-014 1811.10021 |
| 38 | CMS Collaboration | Electron and photon reconstruction and identification with the CMS experiment at the CERN LHC | Accepted by JINST | CMS-EGM-17-001 2012.06888 |
| 39 | CMS Collaboration | Performance of the CMS muon detector and muon reconstruction with proton-proton collisions at $ \sqrt{s}= $ 13 TeV | JINST 13 (2018) P06015 | CMS-MUO-16-001 1804.04528 |
| 40 | CMS Collaboration | Performance of the reconstruction and identification of high-momentum muons in proton-proton collisions at $ \sqrt{s} = $ 13 TeV | JINST 15 (2020) P02027 | CMS-MUO-17-001 1912.03516 |
| 41 | CMS Collaboration | Measurement of the inelastic proton-proton cross section at $\sqrt{s} = $ 13 TeV | JHEP 07 (2018) 161 | CMS-FSQ-15-005 1802.02613 |
| 42 | M. G. Bowler | e+ e- production of heavy quarks in the string model | Z. Phys. C 11 (1981) 169 | |
| 43 | B. Andersson, G. Gustafson, G. Ingelman, and T. Sjostrand | Parton fragmentation and string dynamics | PR 97 (1983) 31 | |
| 44 | T. Sjostrand | Jet fragmentation of nearby partons | NPB 248 (1984) 469 | |
| 45 | ALEPH Collaboration | Study of the fragmentation of b quarks into B mesons at the Z peak | PLB 512 (2001) 30 | hep-ex/0106051 |
| 46 | DELPHI Collaboration | A study of the b quark fragmentation function with the DELPHI detector at LEP I and an averaged distribution obtained at the Z pole | EPJC 71 (2011) 1557 | 1102.4748 |
| 47 | OPAL Collaboration | Inclusive analysis of the b quark fragmentation function in Z decays at LEP | EPJC 29 (2003) 463 | hep-ex/0210031 |
| 48 | SLD Collaboration | Measurement of the b quark fragmentation function in $ {\mathrm{Z^0}} $ decays | PRD 65 (2002) 092006 | hep-ex/0202031 |
| 49 | CMS Collaboration | Measurement of the associated production of a Z boson with charm or bottom quark jets in proton-proton collisions at $ \sqrt{s} = $ 13 TeV | Physical Review D 102 (2020) | CMS-SMP-19-004 2001.06899 |
|
|
Compact Muon Solenoid LHC, CERN |
|
|
|
|
|
|