# Introduction

Public collision plots approved by the tau group which have not been superseded by notes/papers yet.

Superseded plots can be found here.

All tau public results can be found here.

## Performance plots

### Performance Plot for HL-LHC Workshop 2017 (October 2017)

 Tau identification efficiency for the three working points (Loose, Medium and Tight) using the algorithm optimized for HL-LHC detector and conditions (HL-LHC tuning”) as a function of pT for reconstructed $\tau$ candidates, shown for one-prong taus. Tau identification efficiency for the three working points (Loose, Medium and Tight) using the algorithm optimized for HL-LHC detector and conditions (HL-LHC tuning”) as a function of pT for reconstructed $\tau$ candidates, shown for three-prong taus. Jet rejection as a function of $\tau$ efficiency for the algorithm optimized for HL-LHC detector and conditions (HL-LHC tuning”) for $\tau$ candidates within $|\eta|<4.0$ (black) as well as for tau candidates restricted to $|\eta|<2.5$ (blue), compared to the Run 2 performance optimized for the Run-2 detector and conditions (Run-2 performance”) for $\tau$ candidates within $|\eta|<2.5$ (green), shown for one-prong taus. Jet rejection as a function of $\tau$ efficiency for the algorithm optimized for HL-LHC detector and conditions (HL-LHC tuning”) for $\tau$ candidates within $|\eta|<4.0$ (black) as well as for tau candidates restricted to $|\eta|<2.5$ (blue), compared to the Run 2 performance optimized for the Run-2 detector and conditions (Run-2 performance”) for $\tau$ candidates within $|\eta|<2.5$ (green), shown for three-prong taus.

### Performance Plot for ICHEP 2016 (July 2016)

 The visible mass reconstructed using isolated muons and offline tau candidates passing the offline loose identification requirement. The Z mass peak is observed in an enriched sample of $Z\to\tau\tau\to\mu\tau$(had) events from the 2016 dataset in 13 TeV collisions, corresponding to an integrated luminosity of 7.1fb-1. These events are collected using a single muon trigger. Event selections and background estimations are described in ATL-PHYS-PUB-2015-025 and in Eur. Phys. J. C75 (2015) 303. Only statistical uncertainties are shown.

### Performance plots for LHCC 2015 (December 2015)

 (Plot 1:) The BDT tau identification score for offline tau candidates passing the offline medium identification requirement. The tau candidates are observed in an enriched sample of $Z\to\tau\tau\to\mu\tau$(had) events from the 2015 dataset in 13 TeV collisions, corresponding to an integrated luminosity of 3.3fb-1. These events are collected using a single muon trigger. Event selections and background estimations are described in ATL-PHYS-PUB-2015-025 and in Eur. Phys. J. C75 (2015) 303. Only statistical uncertainties are shown. (Plot 2:) The visible mass reconstructed using isolated muons and offline tau candidates passing the offline medium identification requirement. The Z mass peak is observed with high purity in an enriched sample of $Z\to\tau\tau\to\mu\tau$(had) events from the 2015 dataset in 13 TeV collisions, corresponding to an integrated luminosity of 3.3fb-1. These events are collected using a single muon trigger. Event selections and background estimations are described in ATL-PHYS-PUB-2015-025 and in Eur. Phys. J. C75 (2015) 303. Only statistical uncertainties are shown.

### Performance plots for tau identification, tau e veto and tau energy scale (Moriond2013 ID and data set) (February 2013)

 (Plot 1:) Inverse background efficiency as a function of the signal efficiency with a Boosted Decision Tree (BDT) algorithm for 1-prong $\tau_\mathrm{had-vis}$ candidates with a $p_\mathrm{T} > 15$ GeV and $|\eta| < 2.5$. The signal efficiencies are obtained using $Z\to\tau \tau$, $Z'\to\tau \tau$ and $W\to\tau \nu$ simulated events. The background efficiencies are derived using 2012 collision data after applying a multi-jet selection and are calculated with respect to all candidates with exactly one reconstructed track. The Winter 2013 BDT uses $\pi^{0}$-related variables that increase its performance. (Plot 2:) Inverse background efficiency as a function of the signal efficiency with a BDT algorithm for multi-prong $\tau_\mathrm{had-vis}$ candidates with a $p_\mathrm{T} > 15$ GeV and $|\eta| < 2.5$. Multi-prong $\tau$ candidates are defined as reconstructed $\tau$ candidates with 2 or 3 tracks and for the signal efficiency only true $\tau$ leptons decaying into three charged particles are considered. The signal efficiencies are obtained using $Z\to\tau\tau$, $Z'\to\tau\tau$ and $W\to\tau\nu$ simulated events. The background efficiencies are derived using 2012 collision data after applying a multi-jet selection and are calculated with respect to all candidates with two or three reconstructed tracks. (Plot 3:) Signal efficiencies of the Winter 2013 TauBDT for 1-prong $\tau_\mathrm{had-vis}$ candidates as a function of the number of reconstructed vertices for $p_\mathrm{T} > 15$ GeV and $|\eta| < 2.5$. The identification is performed using a BDT algorithm at a loose, medium or tight working point. The efficiencies are obtained using $Z\to\tau\tau$, $Z'\to\tau\tau$ and $W\to\tau\nu$ simulated events with one reconstructed track with respect to all true taus decaying into one charged particle. (Plot 4:) Signal efficiencies of the Winter 2013 TauBDT for multi-prong $\tau_\mathrm{had-vis}$ candidates as a function of the number of reconstructed vertices for $p_\mathrm{T} > 15$ GeV and $|\eta| < 2.5$. The identification is performed using a BDT algorithm at a loose, medium or tight working point. The efficiencies are obtained using $Z \to \tau\tau$, $Z' \to \tau\tau$ and $W \to \tau\nu$ simulated events with two or three reconstructed tracks with respect to all true taus decaying into three charged particles. (Plot 5:) Background efficiencies of the Winter 2013 TauBDT for 1-prong $\tau_\mathrm{had-vis}$ candidates as a function of the number of reconstructed vertices for $p_\mathrm{T} > 15$ GeV and $|\eta| < 2.5$. The identification is performed using a BDT algorithm at a loose, medium or tight working point. The efficiencies are derived using 2012 collision data after applying a multi-jet selection and are calculated with respect to all candidates with exactly one reconstructed track. (Plot 6:) Background efficiencies of the Winter 2013 TauBDT for multi-prong $\tau_\mathrm{had-vis}$ candidates as a function of the number of reconstructed vertices for $p_\mathrm{T} > 15$ GeV and $|\eta| < 2.5$. The identification is performed using a BDT algorithm at a loose, medium or tight working point. The efficiencies are derived using 2012 collision data after applying a multi-jet selection and are calculated with respect to all candidates with two or three reconstructed tracks. (Plot 7:) Distribution of the number of tracks associated with hadronically-decaying tau candidates from the Z to tau tau tag & probe selection in 2012 data, before any tau identification requirement is applied. The signal contribution from simulated Z events is outlined in red. The jet background is taken from data control regions, while the electron background is taken from simulated samples. A fit to the distribution of the number of tracks provides the number of hadronically-decaying tau leptons before any identification criteria are applied, for the purpose of the identification efficiency measurement [using the same method as described in ATLAS-CONF-2012-142]. (Plot 8:) Distribution of the number of tracks associated with hadronically-decaying tau candidates from the Z to tau tau tag & probe selection in 2012 data, after the medium tau identification requirement is applied. The signal contribution from simulated Z events is outlined in red. The jet background is taken from data control regions, while the electron background is taken from simulated samples. The 1-prong (multi- prong) template is based on tau candidates with exactly one (two or more) reconstructed tracks in a cone of deltaR<0.2 around the tau axis. A fit to the distribution of the number of tracks provides the number of hadronically-decaying tau leptons after the identification criteria are applied, for the purpose of the identification efficiency measurement [using the same method as described in ATLAS-CONF-2012-142]. (Plot 9:) Distribution of the number of tracks associated with hadronically-decaying tau candidates from the W to tau nu tag & probe selection in 2012 data, before any tau identification requirement is applied. The signal contribution from simulated W events is outlined in red. The jet background is taken from data control regions, while the electron background is taken from simulated samples. A fit to the distribution of the number of tracks provides the number of hadronically-decaying tau leptons before any identification criteria are applied, for the purpose of the identification efficiency measurement [using the same method as described in ATLAS-CONF-2012-142]. (Plot 10:) Distribution of the number of tracks associated with hadronically-decaying tau candidates from the W to tau nu tag & probe selection in 2012 data, after the medium tau identification requirement is applied. The signal contribution from simulated W events is outlined in red. The jet background is taken from data control regions, while the electron background is taken from simulated samples. A fit to the distribution of the number of tracks provides the number of hadronically-decaying tau leptons after the identification criteria are applied, for the purpose of the identification efficiency measurement [using the same method as described in ATLAS-CONF-2012-142]. (Plot 11:) Response curves as a function of the reconstructed visible tau momentum at LC scale for one-prong tau decays in bins of |η|. The momentum range display corresponds to transverse momenta greater than 15 GeV. The tau response is defined as the ratio of the reconstructed visible momentum at LC scale divided by the true visible momentum and binned in the true visible momentum and |η|. Uncertainties are statistical only. The method is equivalent to that described in ATLAS-CONF-2012-064 for 2011 data. (Plot 12:) Response curves as a function of the reconstructed visible tau momentum at LC scale for multi-prong tau decays in bins of |η|. The momentum range display corresponds to transverse momenta greater than 15 GeV. The multi-prong tau decays refer to the hadronic decay modes with at least two reconstructed tracks. The tau response is defined as the ratio of the reconstructed visible momentum at LC scale divided by the true visible momentum and binned in the true visible momentum and |η|. Uncertainties are statistical only. The method is equivalent to that described in ATLAS-CONF-2012-064 for 2011 data. (Plot 13:) Systematic uncertainty on the tau energy scale (TES) for 1-prong tau decays in the central region (|η| < 0.3), as a function of PT. Each different marker represents a separate source of uncertainty as indicated in the legend. The violet band shows the combined uncertainty from all sources. The method is similar to that described in ATLAS-CONF-2012-064 for 2011 data. (Plot 14:) Systematic uncertainty on the tau energy scale (TES) for multi-prong tau decays in the forward region ( 1.6< |η| < 2.5), as a function of PT. The multi-prong tau decays refer to the hadronic decay modes with at least two reconstructed tracks. Each different marker represents a separate source of uncertainty as indicated in the legend. The violet band shows the combined uncertainty from all sources. The method is similar to that described in ATLAS-CONF-2012-064 for 2011 data. (Plot 15:) Momentum resolution for 1-prong tau decays and multi-prong tau decays as function of PT . The multi-prong tau decays refer to the hadronic decay modes with at least two reconstructed tracks. The resolution is calculated as the difference between the reconstructed and generated PT . The resolution is obtained from a Gaussian fit by dividing the σ of the Gaussian by the mean value of generated P. The method is similar to that described in ATLAS-CONF-2012-064 for 2011 data. (Plot 16:) Background efficiency as a function of signal efficiency with a Boosted Decision Tree (BDT) algorithm for truth-matched 1-prong tau candidates with pT > 15 GeV and eta < 2.47. The signal efficiency is obtained using Z to tau tau simulated events. The background efficiency is obtained using Z to ee simulated events. Candidates are required to pass loose tau identification and not overlap within a cone R=0.2 with a reconstructed electron candidate which passes tight electron identification. (Plot 17:) Signal efficiency of the Winter 2013 electron veto for truth-matched 1-prong tau candidates as a function of reconstructed transverse momentum for pT > 15 GeV and eta < 2.47. The identification is performed using a Boosted Decision Tree (BDT) algorithm at a loose, medium, or tight working point. The signal efficiency is obtained using Z to tau tau simulated events. Candidates are required to pass loose tau identification and not overlap within a cone R=0.2 with a reconstructed electron candidate which passes tight electron identification. (Plot 18:) Signal efficiency of the Winter 2013 electron veto for truth-matched 1-prong tau candidates as a function of reconstructed transverse momentum for pT > 15 GeV and eta < 2.47. The identification is performed using a Boosted Decision Tree (BDT) algorithm at a loose, medium, or tight working point. The signal efficiency is obtained using Z to tau tau simulated events. Candidates are required to pass loose tau identification and not overlap within a cone R=0.2 with a reconstructed electron candidate which passes tight electron identification. (Plot 19:) Signal efficiency of the Winter 2013 electron veto for truth-matched 1-prong tau candidates as a function of the number of reconstructed vertices for pT > 15 GeV and eta < 2.47. The identification is performed using a Boosted Decision Tree (BDT) algorithm at a loose, medium, or tight working point. The signal efficiency is obtained using Z to tau tau simulated events. Candidates are required to pass loose tau identification and not overlap within a cone R=0.2 with a reconstructed electron candidate which passes tight electron identification. (Plot 20:) Background efficiency of the Winter 2013 electron veto for 1-prong tau candidates as a function of transverse momentum for pT > 15 GeV and eta < 2.47. The identification is performed using a Boosted Decision Tree (BDT) algorithm at a loose, medium, or tight working point. The background efficiency is obtained using Z to ee simulated events. Candidates are required to pass loose tau identification and not overlap within a cone R=0.2 with a reconstructed electron candidate which passes tight electron identification. (Plot 21:) Background efficiency of the Winter 2013 electron veto for 1-prong tau candidates as a function of transverse momentum for pT > 15 GeV and eta < 2.47. The identification is performed using a Boosted Decision Tree (BDT) algorithm at a loose, medium, or tight working point. The background efficiency is obtained using Z to ee simulated events. Candidates are required to pass loose tau identification and not overlap within a cone R=0.2 with a reconstructed electron candidate which passes tight electron identification. (Plot 22:) Background efficiency of the Winter 2013 electron veto for 1-prong tau candidates as a function of the number of reconstructed vertices for pT > 15 GeV and eta < 2.47. The identification is performed using a Boosted Decision Tree (BDT) algorithm at a loose, medium, or tight working point. The background efficiency is obtained using Z to ee simulated events. Candidates are required to pass loose tau identification and not overlap within a cone R=0.2 with a reconstructed electron candidate which passes tight electron identification.

### Performance plots for Tau Identification and Tau Energy Scale in 2012 (14 August 2012)

 (Plot 1:) Final systematic uncertainty on the tau energy scale (TES) for 1-prong decays in the central region (abs(eta)< 0.3), as a function of PT. Each different marker represents a separate source of uncertainty as indicated in the legend. The violet band shows the combined uncertainty from all sources. The method is equivalent to that described in ATLAS-CONF-2012-064 for 2011 data. (Plot 2:) Final systematic uncertainty on the tau energy scale (TES) for multi-prong decays in the endcap region, as a function of PT. Each different marker represents a separate source of uncertainty as indicated in the legend. The violet band shows the combined uncertainty from all sources. The method is equivalent to that described in ATLAS-CONF-2012-064 for 2011 data. (Plot 3:) Transverse momentum resolution for one prong decays in different eta regions as function of PT . The transverse resolution is calculated as the difference between the reconstructed and generated PT . The resolution is obtained from a Gaussian fit by dividing the σ of the Gaussian by the mean value of generated PT. (Plot 4:) Signal eﬃciencies for 1-prong τ candidates as a function of the number of reconstructed vertices for loose (green), medium (blue) and tight (red) selection on all jets discriminants with a pT>20 GeV and |η| <2.3. The identiﬁcation was performed using a projective Log-LikeliHood score. The eﬃciencies were obtained using Z->ττ, Z'->ττ and W->τν Pythia8 samples. (Plot 5:) Background eﬃciencies for 1-prong τ candidates as a function of the number of reconstructed vertices for loose (green), medium (blue) and tight (red) selection on all jets discriminants with a pT>20 GeV and |η| <2.3. The identiﬁcation was performed using a projective Log-LikeliHood score. The eﬃciencies were obtained using 2012 dijet data samples with an integrated luminosity of 740 pb-1. (Plot 6:) Signal eﬃciencies for multi-prong τ candidates as a function of the number of reconstructed vertices for loose (green), medium (blue) and tight (red) selection on all jets discriminants with a pT>20 GeV and |η| <2.3. The identiﬁcation was performed using a projective Log-LikeliHood score. The eﬃciencies were obtained using Z->ττ, Z'->ττ and W->τν Pythia8 samples. (Plot 7:) Background eﬃciencies for multi-prong τ candidates as a function of the number of reconstructed vertices for loose (green), medium (blue) and tight (red) selection on all jets discriminants with a pT>20 GeV and |η| <2.3. The identiﬁcation was performed using a projective Log-LikeliHood score. The eﬃciencies were obtained using 2012 dijet data samples with an integrated luminosity of 740 pb-1. (Plot 8:) Signal eﬃciencies for multi-prong τ candidates as a function of the number of reconstructed vertices for loose (green), medium (blue) and tight (red) selection on all jets discriminants with a pT>20 GeV and |η| <2.3. The identiﬁcation was performed using a Boosted Decision Tree algorithm. The eﬃciencies were obtained using Z->ττ, Z'->ττ and W->τν Pythia8 samples. (Plot 9:) Background eﬃciencies for multi-prong τ candidates as a function of the number of reconstructed vertices for loose (green), medium (blue) and tight (red) selection on all jets discriminants with a pT>20 GeV and |η| <2.3. The identiﬁcation was performed using a Boosted Decision Tree algorithm. The eﬃciencies were obtained using 2012 dijet data samples with an integrated luminosity of 740 pb-1.

### Performance Plots for Tau Identification in 2012 (12 Jun 2012)

 (Plot 1:) Signal eﬃciencies for 1-prong τ candidates as a function of the number of reconstructed vertices for loose (green), medium (blue) and tight (red) selection for pT>20 GeV and |η| <2.3. The identiﬁcation was performed using a Boosted Decision Tree algorithm. The eﬃciencies were obtained using Z->ττ, Z'->ττ and W->τν Pythia8 samples. (Plot 2:) Background eﬃciencies for 1-prong τ candidates as a function of the number of reconstructed vertices for loose (green), medium (blue) and tight (red) selection for pT>20 GeV and |η| <2.3. The identiﬁcation was performed using a Boosted Decision Tree algorithm. The eﬃciencies were obtained using 2012 dijet data samples with an integrated luminosity of 740 pb-1. (Plot 3:) Inverse background eﬃciency as a function of the signal eﬃciency for 1-prong τ candidates with a pT > 20 GeV and |η| < 2.3. The signal eﬃciencies were obtained using Z->ττ,Z'->ττ and W->τν Pythia 8 samples, while the background efficiencies were derived using 2012 dijet data samples. (Plot 4:) Inverse background eﬃciency as a function of the signal eﬃciency for multi-prong τ candidates with a pT > 20 GeV and |η| < 2.3. Multi-prong τ candidates are deﬁned as reconstructed τ candidates with 2 or 3 tracks and for the signal efficiency only true τ leptons decaying into three charged particles are considered. The signal eﬃciencies were obtained using Z->ττ,Z'->ττ and W->τν Pythia 8 samples, while the background efficiencies were derived using 2012 dijet data samples.

## Event displays

### Z->ττ candidate event display (07 October 2010)

 This is a display of an event with a candidate Z->τ+τ-->μ+νντ-hν decay in the ATLAS detector where τh denotes a hadronic tau decay.Event properties: pT(μ) = 18 GeV pTvis(τh) = 26 GeV mvis(μ , τh) = 47 GeV mT(μ , ETmiss) = 8 GeV ETmiss = 7 GeV The hadronic tau candidate has three well identified tracks. The muon and tau candidate have opposite sign reconstructed charges. No additional object (electron, muon or jet) was reconstructed in this event.

### W->τν candidate event display (20 July 2010)

 A candidate for a W->τν decay, with a hadronically decaying tau, collected on 24 May 2010. Event properties: pT(τ) = 29 GeV ETmiss = 39 GeV Δφ(τ, ETmiss ) = 3.1 mT = 68 GeV No additional object (electron, muon or jet) was found in the event.