FTKPublicResults
Introduction
This page collects public results for the performance of the FTK system. The technical design report can be found here
. Figures from the TDR are collected on this page.
April 2019 Run 2 results
| Distribution of track transverse momentum (pT) from an FTK slice. A single Data Formatter (DF) board provides partial coverage of Tower22, which spans -1.5 < η < 0 and 1.6 < φ < 2.0. The plot covers a subset of 380,000 events from Run 364485, a special high pile-up run with average interactions per crossing μ = 82, collected in October 2018. FTK tracks with Insertable B-layer (IBL) hits were excluded, due to a FTK module ordering problem that caused incorrect hit positions in the run. The cause is understood and the fix is being implemented. FTK tracks are matched to offline tracks within Δ R < 0.02. The last bin contains the overflow. |
 png pdf |
| The distribution of FTK track pseudo-rapidity (η) vs. azimuthal angle ( φ) from an FTK slice. A single Data Formatter (DF) board provides partial coverage of Tower22, which spans -1.5 < η < 0 and 1.6 < φ < 2.0. The plot covers a subset of 380,000 events from Run 364485, a special high pile-up run with average interactions per crossing μ = 82, collected in October 2018. FTK tracks with Insertable B-layer (IBL) hits were excluded, due to a FTK module ordering problem that caused incorrect hit positions in the run. The cause is understood and the fix is being implemented. FTK tracks are matched to offline tracks within Δ R < 0.02. |
 png pdf |
| Number of tracks produced per event from an FTK slice. A single Data Formatter (DF) board provides partial coverage of Tower22, which spans -1.5 < η < 0 and 1.6 < φ < 2.0. The plot covers a subset of 380,000 events from Run 364485, a special high pile-up run with average interactions per crossing μ = 82, collected in October 2018. FTK tracks with Insertable B-layer (IBL) hits were excluded, due to a FTK module ordering problem that caused incorrect hit positions in the run. The cause is understood and the fix is being implemented. FTK tracks are matched to offline tracks within Δ R < 0.02. The last bin contains the overflow. |
 png pdf |
| The expected FTK tracking efficiency with respect to MC truth from functional emulation of the full FTK system using FTKSim, vs. the x position of the luminous centroid of the beamspot. A fully simulated sample of muons with longitudinal impact parameter |z0| < 110 cm and pseudo-rapidity |η| &\lt; 2.5 is used. The efficiency when treating FTK sectors as patterns (black points) is compared to the efficiency for the nominal patterns (red points) and the efficiencies after the first (green points) and second (blue points) tracking stages. The nominal beamspot used for training the sectors and constants and patterns has x = -0.5 mm and y = -0.9 mm. |
 png pdf |
| The expected FTK tracking efficiency with respect to MC truth from functional emulation of the full FTK system using FTKSim, vs. the x position of the luminous centroid of the beamspot. A fully simulated sample of muons with longitudinal impact parameter |z0| < 110 cm and pseudo-rapidity |η| < 2.5 is used. The colored dots represent patterns trained with the quoted beamspot x position. |
 png pdf |
| The expected FTK tracking efficiency with respect to MC truth from functional emulation of the full FTK system using FTKSim, vs. track transverse momentum pT and the transverse impact parameter d0. A specialized pattern bank is used, in which 30% of the patterns are dedicated to high momentum tracks with large d0. A fully simulated sample of muons with longitudinal impact parameter |z0| < 110 cm and pseudo-rapidity |η| < 2.5 is used. |
 png pdf |
| The resolution of the FTK track curvature, from functional emulation of FTK using FTKSim (FTK Full Simulation) and a parameterized uncertainty model (FTK Fast Simulation). The difference between the FTK reconstructed and true values of track charge divided by twice the track transverse momentum (&Deltaa; Q/2pT) is plotted. A fully simulated sample of muons with longitudinal impact parameter |z0| < 110 cm and pseudo-rapidity |η| < 2.5 is used. The resolution for the fast simulation is modeled as a double gaussian derived from full simulation. |
 png pdf |
| FTK vs. offline track transverse momentum (pT) residuals from an FTK slice. The difference between the FTK and matched offline track pT are shown for (black dots) the FTK slice, (shaded red histogram) functional emulation of the FTK slice using FTKSim, and (dashed line) FTK refit. A single Data Formatter (DF) board provides partial coverage of Tower22, which spans -1.5 < η < 0 and 1.6 < φ < 2.0. The plot covers a subset of 380,000 events from Run 364485, a special high pile-up run with average interactions per crossing μ = 82, collected in October 2018. FTK tracks with Insertable B-layer (IBL) hits were excluded, due to a FTK module ordering problem that caused incorrect hit positions in the run. The cause is understood and the fix is being implemented. FTK tracks are matched to offline tracks within Δ R < 0.02. |
 png pdf |
| FTK vs. offline track pseudo-rapidity (η) residuals from an FTK slice. The difference between the FTK and matched offline track η are shown for (black dots) the FTK slice, (shaded red histogram) functional emulation of the FTK slice using FTKSim, and (dashed line) FTK refit. A single Data Formatter (DF) board provides partial coverage of Tower22, which spans -1.5 < η < 0 and 1.6 < φ < 2.0. The plot covers a subset of 380,000 events from Run 364485, a special high pile-up run with average interactions per crossing μ = 82, collected in October 2018. FTK tracks with Insertable B-layer (IBL) hits were excluded, due to a FTK module ordering problem that caused incorrect hit positions in the run. The cause is understood and the fix is being implemented. FTK tracks are matched to offline tracks within Δ R < 0.02. |
 png pdf |
| FTK vs. offline track azimuthal angle (φ) residuals from an FTK slice. The difference between the FTK and matched offline track ϕ are shown for (black dots) the FTK slice, (shaded red histogram) functional emulation of the FTK slice using FTKSim, and (dashed line) FTK refit. A single Data Formatter (DF) board provides partial coverage of Tower22, which spans -1.5 < η < 0 and 1.6 < φ < 2.0. The plot covers a subset of 380,000 events from Run 364485, a special high pile-up run with average interactions per crossing μ = 82, collected in October 2018. FTK tracks with Insertable B-layer (IBL) hits were excluded, due to a FTK module ordering problem that caused incorrect hit positions in the run. The cause is understood and the fix is being implemented. FTK tracks are matched to offline tracks within Δ R < 0.02. |
 png pdf |
August 2018 12-layer event displays
| Event display for two proton-proton collision events including FTK information recorded on 20th August 2018 showing the ATLAS inner detector in x-y (transverse) projection, top, and r-z (lateral) projection, bottom. One FTK tower was enabled covering 1/64 of the Inner Detector geometrical coverage. The red lines show trajectories corresponding to track parameters obtained from an offline fit to the cluster positions determined by FTK. The FTK cluster positions are shown in white points and, for SCT strips in the lateral projection, lines. The primary vertices determined from tracks reconstructed offline are shown in purple. |
 png |
| Event display for two proton-proton collision events including FTK information recorded on 20th August 2018 showing the ATLAS inner detector in x-y (transverse) projection, top, and r-z (lateral) projection, bottom. One FTK tower was enabled covering 1/64 of the Inner Detector geometrical coverage. The red lines show trajectories corresponding to track parameters obtained from an offline fit to the cluster positions determined by FTK. The FTK cluster positions are shown in white points and, for SCT strips in the lateral projection, lines. The primary vertices determined from tracks reconstructed offline are shown in purple. |
 png |
August 2018 8-layer tracks AUX-ROS plots
| Number of 8-layer tracks produced per event from an FTK slice. The slice contains four Input Mezzanines, one Data Formatter board, one Auxiliary Card, and one Associative Memory Board. In this slice, instead of sending 8-layer track information to the second- stage board, the auxiliary card directly outputs to a ROS. The 8-layer track information consists of hit coordinates, pattern identification information, and which layers are included in the 8-layer track fit. The slice spans -0.2 < η < 1.2 and 2.4 < ɸ < 2.8. This plot covers a subset of Run 358395, a 13 TeV pp run which began on August 15, 2018 in which 859196 events were processed by FTK at a rate of 63 kHz. |
 png pdf |
| The Auxiliary Card fits 8-layer tracks with a minimum of 7 hits. This plot shows the fraction of tracks output by the FTK slice with missing hits in a Pixel layer (blue), in an SCT layer (red), and those with no missing hits. Fractions are plotted as a function of recorded event number. The slice contains four Input Mezzanines, one Data Formatter board, one Auxiliary Card, and one Associative Memory Board. In this slice, instead of sending 8-layer track information to the second- stage board, the Auxiliary Card directly outputs to a ROS. The 8-layer track information consists of hit coordinates, pattern identification information, and which layers are included in the 8-layer track fit. The slice spans − 0.2 < η < 1.2 and 2.4 < φ < 2.8. This plot covers a subset of Run 358395, a 13 TeV pp run which began on August 15, 2018 in which 859196 events were processed by FTK at a rate of 63 kHz. |
 png pdf |
| Fraction of FTK slice output tracks that are matched to tracks produced by running FTK functional simulation (FTKSim) [ATL-DAQ-PROC-2014-030] on RAW data. The RAW data corresponds to a subset of the events processed by the FTK slice. The slice contains four Input Mezzanines, one Data Formatter board, one Auxiliary Card, and one Associative Memory Board. In this slice, instead of sending 8-layer track information to the second-stage board, the Auxiliary Card directly outputs to a ROS. The 8-layer track information consists of hit coordinates, pattern identification information, and which layers are included in the 8-layer track fit. The slice spans − 0.2 < η < 1.2 and 2.4 < φ < 2.8. This plot cov- ers a subset of Run 358395, a 13 TeV pp run which began on August 15, 2018 in which 859196 events were processed by FTK at a rate of 63 kHz. |
 png pdf |
| Fraction of tracks produced by running FTK functional simulation (FTKSim) [ATL-DAQ-PROC-2014-030] on RAW data that are matched to FTK slice output tracks. The RAW data corresponds to a subset of the events processed by the FTK slice. The slice contains four Input Mezzanines, one Data Formatter board, one Auxiliary Card, and one Associative Memory Board. In this slice, instead of sending 8-layer track information to the second-stage board, the Auxiliary Card directly outputs to a ROS. The 8-layer track information consists of hit coordinates, pattern identification information, and which layers are included in the 8-layer track fit. The slice spans − 0.2 < η < 1.2 and 2.4 < φ < 2.8. This plot covers a subset of Run 358395, a 13 TeV pp run which began on August 15, 2018 in which 859196 events were processed by FTK at a rate of 63 kHz. |
 png pdf |
| The p_T distribution of FTK functional simulation (FTKSim) [ATL-DAQ-PROC- 2014-030] 8-layer tracks that were matched to tracks output by the FTK slice. RAW data corresponding to a subset of the events processed by the FTK slice is used as input to the simulation. Because the FTK slice output tracks do not contain track parameter information, the value of p_T is taken from FTKSim. The distribution of matched simulated tracks (grey) is compared to all simulated tracks for the same events (purple). The slice contains four Input Mezzanines, one Data Formatter board, one Auxiliary Card, and one Associative Memory Board. In this slice, instead of sending 8-layer track information to the second- stage board, the Auxiliary Card directly outputs to a ROS. The 8-layer track information consists of hit coordinates, pattern identification information, and which layers are included in the 8-layer track fit. The slice spans − 0.2 < η < 1.2 and 2.4 < φ < 2.8. Were processed by FTK at a rate of 63 kHz. This plot covers 500 events from Run 358395, a 13 TeV pp run which began on August 15, 2018. These 500 events represent a subset of recorded events that were also included in the ATLAS bytestream. |
 png pdf |
| The fraction of tracks output by the FTK slice that were matched to FTK functional simulation (FTKSim) [ATL-DAQ-PROC-2014-030] 8-layer tracks as a function of p_T . RAW data corresponding to a subset of the events processed by the FTK slice is used as input to the simulation. Because the FTK slice output tracks do not contain track parameter information, the value of p_T is taken from FTKSim. The slice contains four Input Mezzanines, one Data Formatter board, one Auxiliary Card, and one Associative Memory Board. In this slice, instead of sending 8-layer track information to the second-stage board, the Auxiliary Card directly outputs to a ROS. The 8-layer track information consists of hit coordinates, pattern identification information, and which layers are included in the 8-layer track fit. The slice spans − 0.2 < η < 1.2 and 2.4 < φ < 2.8. Were processed by FTK at a rate of 63 kHz. This plot covers 500 events from Run 358395, a 13 TeV pp run which began on August 15, 2018. These 500 events represent a subset of recorded events that were also included in the ATLAS bytestream. |
 png pdf |
Wildcard Performance
| Figure 1 shows the FTK track finding efficiency from FTK functional simulation [ATL-DAQ-PROC-2014-030] of Monte Carlo single muon events as a function of the muon pseudorapidity η . The black line shows the simulation output assuming perfect detector operation. The green line shows the simulation output when hits from the list of disabled pixel and SCT modules from 2017 running are excluded from the pattern recognition and track finding, emulating typical detector operation. A drop in the efficiency can be seen when disabled modules are taken into account. The red line shows the efficiency using the WildCards algorithm, in which disabled modules are assumed to always have hits in the pattern recognition stage. The track fitting stage is unchanged by the algorithm. Using this algorithm, the efficiency improves significantly. |
 png pdf |
2017 Pattern Bank Production Plots
| Coverage distribution of the generated patterns. One billion patterns per tower are generated using the FTK fit constants. Each pattern corresponds to 15-bit hit coordinates measured in eight detector planes (three pixel detector planes and five strip detector planes). Each pixel coordinate encodes information along two directions in a single 15-bit word. There are 32 barrel (|η|<1.6) and 32 endcap (|η|>1.6) towers. Within the 109 patterns generated, there may be duplicates. The coverage is defined as the multiplicity at which a particular pattern is generated. The coverage distribution is shown separately for barrel and endcap towers. Patterns with large coverage are the most important for achieving good pattern recognition efficiencies. |
 png pdf |
| Distributions of the number NX of active ternary bits per AM address (top left), the number of generated patterns per AM address Ngen (top right), and the fraction of generated to encoded patterns per AM address Ngen/2NX (lower row). The distributions are shown separately for barrel (|η|<1.6) and endcap (|η|>1.6) towers. For each of the 64 towers 16.8 million AM addresses are available, whereas 109 patterns have been generated. Each AM address holds eight 15-bit words of information about hits in the eight detector planes. Each 15-bit word has up to three ternary bits, which encode the information 0, 1 or X, where X means the ternary bit is "active" and the values 0 or 1 are both valid. The ternary bits allow for large numbers (2NX) of patterns to be encoded at a single AM address. For this figure, ternary bit patterns are generated for the configuration (1+1)pix(1)sct with 1 ternary bit for each of the two directions in each of the three pixel (pix) layers and with 1 ternary bit for each of the five strip (sct) layers. The generated patterns are ordered by decreasing coverage and are assigned to AM addresses. The resulting NX distribution exhibits tails to large NX, corresponding to AM addresses which encode a large number of possible patterns and thus are susceptible to random coincidences. Such patterns are also reflected by low values of Ngen/2NX. |
 png pdf (NX png pdf) (Ngen png pdf) (Ngen/2NX png pdf) |
| Single muon track reconstruction efficiency in the FTK system as a function of the pseudorapidity η. The efficiencies are shown for the configuration (1+1)pix(1)sct of FTK, using 1 ternary bit for each of the two directions in each of the three pixel (pix) layers, 1 ternary bit for each of the five strip (sct) layers and no limit on the number NX of active ternary bits per AM address. The resulting efficiency (red) shows distinct drops near η=±1.2. These are related to a change of geometry of detector modules from a barrel-like to a disk-like orientation. The efficiency is levelled by iterating the assignment of patterns to AM addresses in bins of η. For each iteration, the number of available AM addresses is increased for low-efficiency bins and decreased for high-efficiency bins, such that the total number of AM addresses per tower is kept constant. The result is satisfactory after one iteration (blue) and does not improve much when adding another iteration (dark yellow). |
 png pdf |
| Distributions of the number NX of active ternary bits per AM address (top left), the number of generated patterns per AM address Ngen (top right), and the fraction of generated to encoded patterns per AM address Ngen/2NX (lower row). The distributions are shown separately for barrel (|η|<1.6) and endcap (|η>1.6) towers. For each of the 64 towers 16.8 million AM addresses are available, whereas 109 patterns have been generated. Each AM address holds eight 15-bit words of information of the eight detector planes. Each 15-bit word has up to three ternary bits, which encode the information 0, 1 or X, where X means the ternary bit is active and the values 0 or 1 are both valid. The ternary bits allow for large numbers 2NX of patterns to be encoded at a single AM address, where NX is the total number of ternary bits in status X at a given AM address. For this figure, ternary bit patterns are generated for the configuration (1+1)pix(3)sct with 1 ternary bit for each of the two directions in each of the three pixel (pix) layers and with 3 ternary bit for each of the five strip (sct) layers. The generated 109 patterns are ordered by decreasing coverage and are assigned to AM addresses. When assigning patterns to AM addresses, limits on NX are introduced, NX<8 for barrel towers and NX<5 for endcap towers, respectively. The resulting NX distribution shows that many patterns have NX values near the cut-off. The limit in NX is also reflected in the distribution Ngen, which falls off steeply. |
 png pdf (NX png pdf) (Ngen png pdf) (Ngen/2NX png pdf) |
| Single muon track reconstruction efficiency in the FTK system as a function of the pseudorapidity η. The efficiencies are shown for ternary bits calculated using the configuration (1+1)pix(3)sct, using 1 ternary bit for each of the two directions in each of the three pixel (pix) layers, 3 ternary bit for each of the five strip (sct) layers and a limit on the number of active ternary bits NX<8 (NX<5) for the barrel (endcap) towers. The resulting efficiency (red) shows efficiency drops drops near η±1.2. These are related to a change of geometry of detector modules from a barrel-like to a disk-like orientation. The efficiency is levelled by iterating the assignment of patterns to AM addresses in bins of η. For each iteration, the number of available AM addresses is increased for low-efficiency bins and decreased for high-efficiency bins, such that the total number of AM addresses per tower is kept constant. The result is satisfactory after one iteration (blue) and does not improve much when adding another iteration (dark yellow). |
 png pdf |
| Comparison of the single-muon reconstruction efficiency achieved with the FTK system for two different algorithms to determine the ternary bits in the pattern. The algorithm (1+1)pix(1)sct (blue) is using 1 ternary bit for each of the two directions in each of the three pixel (pix) layers, 1 ternary bit for each of the five strip (sct) layers and no limit on the number NX of active ternary bits per AM address. The preferred algorithm (1+1)pix(3)sct (red) is using 1 ternary bit for each of the two directions in each of the three pixel layers, 3 ternary bit for each of the five strip layers and limits NX<8 (NX<5) for the barrel (endcap) towers. |
 png pdf |
|
Average number of FTK roads (i.e. patterns with hits on 7 or 8
planes) per tower as a function of the number of collisions per
bunch crossing μ, determined from simulated
pp→t |
 png pdf ((1+1)pix(1)sct png pdf) ((1+1)pix(3)sct,NXbarr<8,NXec<5 png pdf) |
|
Average number of
FTK eight-layer fits per tower as
a function of the number of collisions per bunch crossing μ,
determined from simulated pp→t |
 png pdf ((1+1)pix(1)sct png pdf) ((1+1)pix(3)sct,NXbarr<8,NXec<5 png pdf) |
| Distributions of the encoded pattern multiplicity M for barrel (left) and endcap (right) towers, using two different algorithms to determine the ternary pattern bits. The (1+1)pix(1)sct algorithm (red) is using 1 ternary bit for each of the two directions in each of the three pixel (pix) layers, 1 ternary bit for each of the five strip (sct) layers and no limit on the number NX of active ternary bits per AM address ; the (1+1)pix(3)sct algorithm (blue) is using 1 ternary bit for each of the two directions in each of the three pixel layers, 3 ternary bit for each of the five strip layers and limits N<>sub>X<8 (NX<5) for the barrel (endcap) towers. For the algorithm without cut on NX, patterns with identical non-ternary bits are always assigned to a single AM address, possibly at the cost of generating large NX. Each encoded pattern, after expanding all ternary bits, is stored only once (M=1). In contrast, for the algorithm with limit on the number NX, patterns with identical non-ternary bits may be assigned to several distinct AM addresses, if the overlap in the ternary bits is small and the resulting NX thus would exceed the limit. This procedure avoids large NX but does create duplicate patterns. As visible from the blue curve, a moderate level of duplicate patterns (M>1) is created. |
 png pdf (barrel png pdf) (endcap png pdf) |
| Comparison of the bank volume and the fraction of generated coverage loaded into the AM chip for different algorithms to set ternary bits. The barrel (endcap) results are shown as filled (open) circles. The bank volume is defined as the average number of patterns encoded at a given AM address, 〈2NX〉. The fraction of generated coverage corresponds to the fraction of the originally generated 109 tracks per tower which are processed in sequence before all available AM addresses are filled. Two classes of algorithms are compared: algorithms without limit on the number NX of ternary bits (blue squares) and algorithms with a limit on NX (red circles). For the algorithms shown with blue squares, the number of available ternary bits per detector plane is indicated for each single point. For example, the configuration (1+1)pix(1)4(2)1 is using 1 ternary bit for each of the two directions of the 3 pixel (pix) planes, 1 ternary bit for the innermost 4 strip planes and 2 ternary bits for the outermost strip plane. The configuration (1+1)pix(2)sct is using 1 ternary bit for each of the two directions of the 3 pixel planes and 2 ternary bits of each of the five strip (sct) planes. For each of the red points the configuration of ternary bits per detector plane is fixed to (1+1)pix(3)sct, but the corresponding limit on NX varies and is indicated. In the limit NX<1, the ternary bits behave as normal bits, because the value X is never assigned to a ternary bit. The corresponding bank volume thus is equal to unity. Increasing fractions of generated coverage correspond to better track reconstruction efficiencies. Increasing bank volumes correspond to an increased level of random coincidences, and challenge the dataflow. The endcap towers generally require a much smaller bank volume as compared to barrel towers, when looking at the same fraction of coverage. This is related to the geometrical tower definition and is intentional, as the endcap patterns correspond to tracks passing regions of high hit densities. When comparing algorithms with and without limits on NX at a similar fraction of coverage, the algorithms with limits on NX result in a smaller bank volume and thus perform better. |
 png pdf |
2014 Performance Plots
| The ATLAS HLT beam spot finding and primary vertex reconstruction algorithm adapted for FTK tracks are used to find all primary vertices in the event. The resolution along the beam axis (z) of the FTK primary vertices is dependent on the number of tracks associated with the vertex. The plot shows the z resolution as a function of the number of FTK tracks per primary vertex for all primary vertices in the event. For a 20-track vertex the z-resolution is 40m. The sample used is tt MC with IBL for 14 TeV pp collisions with an average of 60 overlapping pp interaction. This plot is an update of figure 28 of the FTK TDR [CERN-LHCC-2013-007] which showed a z-resolution of 70m for a 20-track vertex. |
 png pdf |
|
Track multiplicities of candidates computed with FTK tracks for a signal cone with d?R < 0.1
(top) and for an isolation ring with 0.1 < ?dR < 0.4 (bottom). A signal sample of from gluon fusion
H -> tautau -> had had events is shown in blue, and a multijet QCD background sample is shown in red. Both
signal and background samples are produced with |
 png pdf  png pdf |
|
The plot shows the identification e ciency measured in a signal sample of gluon fusion
H ->tautau-> had had (red triangles, blue circles) and a multijet background sample (black triangles) for a
cut-based FTK HLT selection. The HLT-FTK selection is defined as: N_tracks(?dR < 0.1) < = 3 and
N_tracks(0.1 <d ?R < 0.4) <= 2. The signal e ciency with respect to truth (red triangles) is defined as
the fraction of Level-1 tau s matched to a true hadronic tau decay and to an o
ffline- tau that pass the HLT-FTK
selection; the e fficiency with respect to the o
ffline selection (blue circles) is defined as the fraction of
Level-1 taus matched to a true hadronic tau decay and to an offl
ine-tau identified with the o
ffline loose Boosted
Decision Tree (BDT) [ATLAS-CONF-2013-064] that pass the HLT-FTK selection. The background
e fficiency (black triangles) is obtained from a multijet QCD sample and is defined as the fraction of
Level-1 taus matched to an o
ffline-tau that pass the HLT-FTK selection. The matching of the Level-1
tau with the true hadronic tau or with the o
ffline- tau is satisfied if it is within ?dR < 0.2 from the Level-1
direction. In both the signal and background samples, an offl
ine tau match is found for > 97% of the
Level-1 candidates. The true taus s are required to have pT > 8 GeV and |eta| < 2.4. The o
ffline taus are o
ffline
reconstructed objects with |eta|< 2.5. Both signal and background samples are produced with |
 png pdf |
|
The plot shows the tau identification efficiency as a function of Level-1 tau candidate pT for taus from a signal sample of gluon fusion H->tau tau-> had ha events (blue triangles) and from a multijet background sample (black triangles) when a selection based on a multivariate discriminant that uses FTK tracking is applied at the HLT (FTK-BDT).
The efficiency is defined as the fraction of Level-1 taus matched to a true hadronic tau decay and to an offline-tau identified with the offline loose Boosted Decision Tree (BDT) [ATLAS-CONF-2013-064] that pass the FTK-BDT selection at HLT for the signal and as the fraction of Level-1 taus matched to an offline-tau that pass the FTK-BDT selection for the multijet background.
The FTK-BDT selection uses only variables based on FTK tracking. The Level-1 tau candidates are divided in 1-prong (also containing 0-prong) and multi-prong candidates based on the number of FTK associated in the signal cone. The variables that show the most separation power are the invariant mass of the tracks, and the sum of the pT of all tracks in the isolation ring divided by the sum of the pT of all the tracks in the signal region.
The two BDTs are then combined to obtain an overall efficiency per Level-1 tau candidate for signal and background.
This plot is an update and a summary with respect to Figures 39 and 42 of the FTK TDR [CERN-LHCC-2013-007] that now includes the use of IBL and a BDT tau selection based on FTK Tracks. The Level-1 seed has been updated to the expected 2015 trigger configuration and the background sample changed from WH->lnuqq to the multijet sample (with |
 png pdf |
| The plot shows the tau identification efficiency as a function of the offline-tau pT when applying the FTK selection (blue) and calorimeter clusters selection (red) at HLT. The efficiency is defined as the fraction of Level-1 tau matched to a true hadronic tau decay and to an offline tau identified with the offline loose Boosted Decision Tree (BDT) [ATLAS-CONF-2013-064]. The FTK selection at HLT requires N_tracks (deltaR < 0.1) <= 3 and N_tracks (0.1< deltaR < 0.4) <= 2 while the calorimeter clusters selection requires clusters with E_T>25 GeV, f_core0.079 where f_core is ratio of the transverse energy in a cone of radius 0.2 to the transverse energy in a cone of radius 0.4 around the tau direction and R_cal = (Sigma E_T * dR)/ (Sigma E_T). In both cases, the selections have be tuned to give the same background rejection factor of 2 per Level-1 tau candidate (with p_T > 12 GeV and calorimetric isolation of 4 GeV). This plot is an update and a summary with respect to figure 39 of the FTK TDR [CERN-LHCC-2013-007] that now includes the use of IBL. |
 png pdf |
|
The plot shows the efficiency per event for a sample of boosted gluon fusion H->tautau->hadhad events, defined by requiring the Higgs pT > 60 GeV, versus the multijet QCD background efficiency for several HLT selections.
The Level-1 selection used requires two Level-1 tau candidates with 12 and 20 GeV and a calorimetric isolation of 4 GeV separated in delta R as well as a Level-1 jet candidate with pT> 25 GeV. The red downwards pointing triangle shows the per event efficiency of the HLT selection which requires that the jet and at least one of the two tau to be consistent with the primary vertex (PV).
The red circle shows the per event efficiency for applying the FTK selection at HLT (N_tracks (delta R < 0.1) <= 3 and N_tracks (0.1< delta R < 0.4) <= 2 ).
The red upwards pointing triangle show the per event efficiency for applying the FTK selection at HLT and the primary vertex consistency requirement.
The black star shows the efficiency of applying the calorimeter cluster selection (f_core < 0.59 and R_cal>0.079 where f_core is ratio of the transverse energy in a cone of radius 0.2 to the transverse energy in a cone of radius 0.4 around the tau direction and R_cal = ( Sigma E_T * dR )/ (Sigma E_T) ) and adding the requirement that both the Level-1 tau candidates have E_T > 32 GeV. The black triangle shows the efficiency for the same selection described above but with Level-1 tau candidates satisfying E_T >40 GeV for the first and E_T > 25 GeV for the second.
The plot summarizes tau studies in the FTK TDR [CERN-LHCC-2013-007] that have been updated and now include the use of the IBL. The Level-1 seed has been updated to the one expected in the 2015 run and the background sample changed from WH-> lnuqq to the multijet sample (with |
 png pdf |
| The transverse impact parameter is shown for tracks associated to light-flavor (black) and heavy-flavor (red) jets. The solid lines show the distribution for the offline tracks, whereas the points show the FTK tracks. The impact parameter is signed such that track displacements in the direction of the jet have positive values, while tracks with displacements opposite of the jet direction are negative. This plot is an update with respect to figure 19 of the FTK TDR [CERN-LHCC-2013-007] that now includes the use of IBL. |
 png pdf |
| The transverse impact parameter and its significance are shown for tracks associated to light-flavor (black) and heavy-flavor (red) jets. The solid lines show the distribution for the offline tracks, whereas the points show the re-fitted FTK tracks. The first and third plots show the distributions in the barrel and the second and fourth plots show the distributions in the end-caps. The transverse impact parameter significance is defined as the transverse impact parameter divided by its associated uncertainty. The impact parameter is signed such that track displacements in the direction of the jet have positive values, while tracks with displacements opposite of the jet direction are negative. This plot is an update with respect to figure 19 of the FTK TDR [CERN-LHCC-2013-007] that now includes the use of IBL and the refit of FTK clusters associated to the FTK tracks using the offline algorithm. The clusters associated to the FTK tracks are fit with the global Chi^2 algorithm. A Chi^2 < 2 cut is applied to the FTK tracks after refit to increase track purity. The narrower d_0 significance core for FTK refitted tracks is a result of overestimated impact parameter uncertainties for the refitted FTK tracks. |
 png pdf  png pdf  png pdf png pdf |
| In run 1, b-jet tagging in the HLT began with a calorimeter-based jet pre-selection that tightened the jet pT thresholds. With the FTK, track finding can be run with looser HLT jet thresholds without putting additional load on the HLT processors. The plot shows two such working points from a draft run-2 menu along with some examples of re-optimized working points in which the b-tagging is run with lower jet thresholds. Each of the working points shown includes a pre-selection requiring four L1 jets above 20 GeV. %The FTK enables this approach without adding additional load to the HLT processors. The output rates of the various trigger items are shown as a function of the event-level tth (h->bb) efficiency. All operating points assume the re-fitted FTK performance. The black points show b-jet items that will fit into the HLT constraints without FTK, red points show options with significantly larger input rates possible with FTK. The efficiencies are quoted with respect to the inclusive signal and include the L1 efficiency. The trigger names specify the jet multiplicities, pT thresholds and b-tagging operating points. For example, the ``2b55_4j55_Medium'' trigger requires at least two medium b-tagged jets above 55 GeV and four or more jets above 55 GeV. The plot summarizes the impact of the FTK b-tagging described in the FTK TDR [CERN-LHCC-2013-007], updated to include the use of the IBL and the refit of FTK tracks. |
 png pdf |
| In run 1, b-jet tagging in the HLT began with a calorimeter-based jet pre-selection that tightened the jet \pt thresholds. With the FTK, track finding can be run with looser HLT jet thresholds without putting additional load on the HLT processors. The plot shows two such working points from a draft run-2 menu along with some examples of re-optimized working points in which the b-tagging is run with lower jet thresholds. Each of the working points shown includes a pre-selection requiring four L1 jets above 20 GeV. The output rates of the various trigger items are shown as a function of the event-level signal efficiency for an exotic 4b signature (G(m=300 GeV)->hh->4b). All operating points assume the re-fitted FTK performance. The black points show proposed b-jet items that will fit into the HLT constraints without FTK, red points show options with significantly larger input rates possible with FTK. The efficiencies are quoted with respect to an offline requirement of four identified b-jets and include the L1 efficiency. The trigger names specify the jet multiplicities, pT thresholds and b-tagging operating points. For example, the ``2b55_4j55_Medium'' trigger requires at least two medium b-tagged jets above 55 GeV and four or more jets above 55 GeV. The plot summarizes the impact of the FTK b-tagging described in the FTK TDR [CERN-LHCC-2013-007], updated to include the use of the IBL and the refit of FTK tracks. |
 png pdf |
| Et distribution for offline b-tagged jets in simulated ttbar events. The offline jets, shown in yellow, are reconstructed using the anti-kt(R=0.4) algorithm and are required to have |eta| < 2.5 and satisfy the 80% working point of the IP2D b-tagging algorithm [ATLAS-CONF-2010-091]. The black points show the sub-set of offline jets matching a 20 GeV L1 jet (left) or an FTK track-jet (right). FTK track jets are reconstructed from FTK tracks [CERN-LHCC-2013-007] with |z0 x sin(theta)| < 2 mm using the anti-kt(R=0.4) algorithm. The z0 is calculated with respect to the primary vertex found using the FTK tracks. The FTK track jets are required to consist of at least two tracks and have transverse momentum greater than 5 GeV. L1 jets are reconstructed from trigger towers using a sliding window algorithm. The bottom panels show the ratio of the matched jets to the full unbiased distribution. The high CPU consumption required by the tracking in the High Level Trigger means for b-tagging, the tracking must be seeded by jets found at L1 with a reasonably high (~20 GeV) Et threshold in order to reduce the rate to a level where the tracking can be performed. With the FTK, full scan b-tagging can be performed at the full Level 1 output rate, independently of jets found at L1. This can increase the trigger acceptance for b-jets at low transverse energies in events triggered at Level 1 by additional signatures besides the b-jets themselves. In addition, where the jet Et is low, the FTK tracks can help to distinguish jets from the primary interaction from additional low Et jets arising from pileup interactions. |
 eps pdf  eps pdf |
| Pixel cluster size distribution for a tt̄ sample with pile-up 60 at 14 TeV. The clustering algorithm used is the grid clustering with the grid dimensions being 8 columns and 21 rows. |  |
| Figure 2: Comparison of the resolution of track parameter η for a single muon sample with no pile-up using the three different clustering algorithms. The ratio indicated by the blue markers is the ratio of the fully realistic implementation over the ideal, while the the red markers show the ratio of the partially realistic implementation over the ideal. The ideal clustering algorithm places no limits on the size of the cluster, while in the ideal centroid calculation the centroid is calculated as the ToT -weighted average of the cluster. On the realistic clustering implementation, the grid clustering is used, with a grid of 8 × 21 pixels, limiting the maximum cluster size, while in realistic centroid calculation the centroid is the center of a bounding box containing the cluster. |  |
| Comparison of the resolution of track parameter η for a tt̄ sample with pile-up 60 at √s = 14 TeV using the three different clustering algorithms. The ratio indicated by the blue markers is the ratio of the fully realistic implementation over the ideal, while the the red markers show the ratio of the partially realistic implementation over the ideal. The ideal clustering algorithm places no limits on the size of the cluster, while in the ideal centroid calculation the centroid is calculated as the ToT -weighted average of the cluster. On the real istic clustering implementation, the grid clustering is used, with a grid of 8 × 21 pixels, limiting the maximum cluster size, while in realistic centroid calculation the centroid is the center of a bounding box containing the cluster. |  |
| Comparison of the χ 2 distribution for a single muon sample with no pile-up using the three different clustering algorithms. The ratio indicated by the blue markers is the ratio of the fully realistic implementation over the ideal, while the the red markers show the ratio of the partially realistic implementation over the ideal. The ideal clustering algorithm places no limits on the size of the cluster, while in the ideal centroid calculation the centroid is calculated as the ToT -weighted average of the cluster. On the realistic clustering implementation, the grid clustering is used, with a grid of 8 × 21 pixels, limiting the maximum cluster size, while in realistic centroid calculation the centroid is the center of a bounding box containing the cluster. |  |
|
Comparison of the χ 2 distribution for a tt̄ sample with pile-up 60 at √s = 14 TeV using the three different clustering algorithms. The ratio indicated by the blue markers is the ratio of the fully realistic implementation over the ideal, while the the red markers show the ratio of the partially realistic implementation over the ideal. The ideal clustering algorithm places no limits on the size of the cluster, while in the ideal centroid calculation the centroid is calculated as the ToT -weighted average of the cluster. On the realistic clustering implementation, the grid clustering is used, with a grid of 8 × 21 pixels, limiting the maximum cluster size, while in realistic centroid calculation the centroid is the center of a bounding box containing the cluster. |  |
2017 8-layer tracks AUX-ROS plots -- Superceeded by 2018 plots
| Number of 8-layer tracks produced per event from an FTK slice. The slice contains four Input Mezzanines, one Data Formatter board, one Auxiliary Card, and one Associative Memory Board. In this slice, instead of sending 8-layer track information to the second- stage board, the auxiliary card directly outputs to a ROS. The 8-layer track information consists of hit coordinates, pattern identification information, and which layers are included in the 8-layer track fit. The slice spans -0.2 < η < 1.2 and 2.4 < ɸ < 2.8. This plot covers a subset of Run 340453, a 13 TeV pp run which began on November 9, 2017, in which 6247 events were processed by FTK at a rate of 35 kHz. |
 png pdf |
| Fraction of FTK slice output tracks that are matched to tracks produced by running FTK functional simulation (FTKSim) [ATL-DAQ-PROC-2014-030] on RAW data. The RAW data corresponds to a subset of the events processed by the FTK slice. The slice contains four Input Mezzanines, one Data Formatter board, one Auxiliary Card, and one Associative Memory Board. In this slice, instead of sending 8-layer track information to the second-stage board, the auxiliary card directly outputs to a ROS. The 8-layer track information consists of hit coordinates, pattern identification information, and which layers are included in the 8-layer track fit. The slice spans -0.2 < η < 1.2 and 2.4 < ɸ < 2.8. This plot covers 100 events from Run 340453, a 13 TeV pp run which began on November 9, 2017. These 100 events represent a subset of recorded events that were also included in the ATLAS bytestream. |
 png pdf |
| The p T distribution of FTK functional simulation (FTKSim) [ATL-DAQ-PROC-2014-030] 8-layer tracks that were matched to tracks output by the FTK slice. RAW data corresponding to a subset of the events processed by the FTK slice is used as input to the simulation. Because the FTK slice output tracks do not contain track parameter information, the value of p T is taken from FTKSim. The distribution of matched simulated tracks (grey) is compared to all simulated tracks for the same events (purple). The slice contains four Input Mezzanines, one Data Formatter board, one Auxiliary Card, and one Associative Memory Board. In this slice, instead of sending 8- layer track information to the second-stage board, the auxiliary card directly outputs to a ROS. The 8-layer track information consists of hit coordinates, pattern identification information, and which layers are included in the 8-layer track fit. The slice spans -0.2 < η < 1.2 and 2.4 < ɸ < 2.8. Were processed by FTK at a rate of 35 kHz. This plot covers 100 events from Run 340453, a 13 TeV pp run which began on November 9, 2017. These 100 events represent a subset of recorded events that were also included in the ATLAS bytestream. |
 png pdf |
| The ɸ distribution of FTK functional simulation (FTKSim) [ATL-DAQ-PROC-2014-030] 8-layer tracks that were matched to tracks output by the FTK slice. RAW data corresponding to a subset of the events processed by the FTK slice is used as input to the simulation. Because the FTK slice output tracks do not contain track parameter information, the value of ɸ is taken from FTKSim. The distribution of matched simulated tracks (grey) is compared to all simulated tracks for the same events (purple). The slice contains four Input Mezzanines, one Data Formatter board, one Auxiliary Card, and one Associative Memory Board. In this slice, instead of sending 8- layer track information to the second-stage board, the auxiliary card directly outputs to a ROS. The 8-layer track information consists of hit coordinates, pattern identification information, and which layers are included in the 8-layer track fit. The slice spans -0.2 < η < 1.2 and 2.4 < ɸ < 2.8. This plot covers 100 events from Run 340453, a 13 TeV pp run which began on November 9, 2017. These 100 events represent a subset of recorded events that were also included in the ATLAS bytestream. |
 png pdf |
| The η distribution of FTK functional simulation (FTKSim) [ATL-DAQ-PROC-2014-030] 8-layer tracks that were matched to tracks output by the FTK slice. RAW data corresponding to a subset of the events processed by the FTK slice is used as input to the simulation. Because the FTK slice output tracks do not contain track parameter information, the value of η is taken from FTKSim. The distribution of matched simulated tracks (grey) is compared to all simulated tracks for the same events (purple). The slice contains four Input Mezzanines, one Data Formatter board, one Auxiliary Card, and one Associative Memory Board. In this slice, instead of sending 8- layer track information to the second-stage board, the auxiliary card directly outputs to a ROS. The 8-layer track information consists of hit coordinates, pattern identification information, and which layers are included in the 8-layer track fit. The slice spans -0.2 < η < 1.2 and 2.4 < ɸ < 2.8. This plot covers 100 events from Run 340453, a 13 TeV pp run which began on November 9, 2017. These 100 events represent a subset of recorded events that were also included in the ATLAS bytestream. |
 png pdf |
| The auxiliary card fits 8-layer tracks with a minimum of 7 hits. This plot shows the fraction of tracks output by the FTK slice with missing hits in a Pixel layer (blue), in an SCT layer (red), and those with no missing hits. Fractions are plotted as a function of recorded event number. The slice contains four Input Mezzanines, one Data Formatter board, one Auxiliary Card, and one Associative Memory Board. In this slice, instead of sending 8-layer track information to the second-stage board, the auxiliary card directly outputs to a ROS. The 8-layer track information consists of hit coordinates, pattern identification information, and which layers are included in the 8-layer track fit. The slice spans -0.2 < η < 1.2 and 2.4 < ɸ < 2.8. This plot covers a subset of Run 340453, a 13 TeV pp run which began on November 9, 2017, in which 6247 events were processed by FTK at a rate of 35 kHz |
 png pdf |
Major updates:
-- JohnAlison - 20 Aug 2014 -- LaurenTompkins - 13 May 2014 Responsible: LaurenTompkins
Subject: public
| I | Attachment | History | Action | Size | Date | Who | Comment |
|---|---|---|---|---|---|---|---|
 png |
8x10.png | r1 | manage | 15.2 K | 2016-05-30 - 12:23 | StamatiosGkaitatzis | |
 eps |
FTKTrkJetTurnOnWrtBtag.eps | r1 | manage | 21.1 K | 2014-08-20 - 18:05 | JohnAlison | |
 pdf |
FTKTrkJetTurnOnWrtBtag.pdf | r1 | manage | 18.1 K | 2014-08-20 - 16:13 | JohnAlison | |
| png |
FTKTrkJetTurnOnWrtBtag.png | r1 | manage | 18.9 K | 2014-08-20 - 16:16 | JohnAlison | |
| png |
JiveXML_358656_913511153_Hashed-EventInfo-YX-RZ-2018-09-09-09-50-44.png | r1 | manage | 376.0 K | 2018-09-14 - 14:05 | LaurenTompkins | 2018 AUX->ROS plots |
| png |
JiveXML_358656_962058206_Hashed-EventInfo-YX-RZ-2018-09-09-09-54-02.png | r1 | manage | 379.5 K | 2018-09-14 - 14:05 | LaurenTompkins | 2018 AUX->ROS plots |
| eps |
L1TurnOnWrtBtag.eps | r1 | manage | 20.2 K | 2014-08-20 - 16:13 | JohnAlison | |
| pdf |
L1TurnOnWrtBtag.pdf | r1 | manage | 17.8 K | 2014-08-20 - 16:12 | JohnAlison | |
| png |
L1TurnOnWrtBtag.png | r1 | manage | 18.8 K | 2014-08-20 - 16:17 | JohnAlison | |
| pdf |
ROI_efficiency_sig_bkg.pdf | r1 | manage | 56.8 K | 2014-05-26 - 13:16 | LaurenTompkins | more tau plots |
| png |
ROI_efficiency_sig_bkg.png | r1 | manage | 10.4 K | 2014-05-26 - 13:07 | LaurenTompkins | |
| pdf |
Sig_L1_TAU12I_eff_BDT_tau_pt.pdf | r1 | manage | 60.3 K | 2014-05-26 - 13:16 | LaurenTompkins | more tau plots |
| png |
Sig_L1_TAU12I_eff_BDT_tau_pt.png | r1 | manage | 10.6 K | 2014-05-26 - 13:16 | LaurenTompkins | more tau plots |
| pdf |
VertexNTrks_vs_VertexZerr_profile.pdf | r1 | manage | 15.3 K | 2014-05-26 - 11:33 | LaurenTompkins | Primary vertex pdf |
| png |
VertexNTrks_vs_VertexZerr_profile.png | r2 r1 | manage | 12.4 K | 2014-05-26 - 11:50 | LaurenTompkins | Vertex png |
| pdf |
Xhh300RefitClean.pdf | r1 | manage | 47.9 K | 2014-06-27 - 12:03 | LaurenTompkins | |
| png |
Xhh300RefitClean.png | r1 | manage | 9.0 K | 2014-06-27 - 12:03 | LaurenTompkins | |
| png |
can_OffAll_b_60Refit_d0Sig_signed_b.png | r1 | manage | 29.6 K | 2014-05-26 - 13:50 | LaurenTompkins | btagging pngs |
| pdf |
can_OffAll_b_60Refit_d0_signed_b.pdf | r1 | manage | 60.7 K | 2014-05-26 - 13:53 | LaurenTompkins | |
| png |
can_OffAll_b_60Refit_d0_signed_b.png | r1 | manage | 12.6 K | 2014-05-26 - 13:50 | LaurenTompkins | btagging pngs |
| png |
can_OffAll_b_60_d0_signed_b.png | r1 | manage | 12.5 K | 2014-05-26 - 13:50 | LaurenTompkins | btagging pngs |
| pdf |
can_OffAll_e_60Refit_d0Sig_signed_e.pdf | r1 | manage | 59.9 K | 2014-05-26 - 13:52 | LaurenTompkins | btagging |
| png |
can_OffAll_e_60Refit_d0Sig_signed_e.png | r1 | manage | 29.8 K | 2014-05-26 - 13:50 | LaurenTompkins | btagging pngs |
| pdf |
can_OffAll_e_60Refit_d0_signed_e.pdf | r1 | manage | 61.1 K | 2014-05-26 - 13:52 | LaurenTompkins | btagging |
| png |
can_OffAll_e_60Refit_d0_signed_e.png | r1 | manage | 28.1 K | 2014-05-26 - 13:50 | LaurenTompkins | btagging pngs |
| pdf |
compare3Resolutions_eta_noIBL_deltaRplot.pdf | r1 | manage | 17.2 K | 2019-04-25 - 15:40 | DavidMStrom | |
| png |
compare3Resolutions_eta_noIBL_deltaRplot.png | r1 | manage | 25.4 K | 2019-04-25 - 15:40 | DavidMStrom | |
| pdf |
compare3Resolutions_phi_noIBL_deltaRplot.pdf | r1 | manage | 16.8 K | 2019-04-25 - 15:40 | DavidMStrom | |
| png |
compare3Resolutions_phi_noIBL_deltaRplot.png | r1 | manage | 22.5 K | 2019-04-25 - 15:40 | DavidMStrom | |
| pdf |
compare3Resolutions_pt_noIBL_deltaRplot.pdf | r1 | manage | 17.2 K | 2019-04-25 - 15:40 | DavidMStrom | |
| png |
compare3Resolutions_pt_noIBL_deltaRplot.png | r1 | manage | 22.5 K | 2019-04-25 - 15:40 | DavidMStrom | |
| pdf |
eff_events_ggF.pdf | r1 | manage | 51.4 K | 2014-05-26 - 13:16 | LaurenTompkins | tau sig eff plots |
| png |
eff_events_ggF.png | r1 | manage | 11.2 K | 2014-05-26 - 13:16 | LaurenTompkins | tau sig eff plots |
| pdf |
eff_vs_pt_BDT.pdf | r1 | manage | 15.4 K | 2014-05-26 - 13:37 | LaurenTompkins | tau eff plots |
| png |
eff_vs_pt_BDT.png | r1 | manage | 11.2 K | 2014-05-26 - 13:37 | LaurenTompkins | tau eff plots |
| pdf |
efficency_2D.pdf | r1 | manage | 68.7 K | 2019-04-25 - 15:26 | DavidMStrom | |
| png |
efficency_2D.png | r1 | manage | 87.5 K | 2019-04-25 - 15:26 | DavidMStrom | |
| eps |
efficiency.eps | r1 | manage | 33.8 K | 2018-06-01 - 06:14 | LaurenTompkins | Wildcard Efficiencies |
| pdf |
efficiency.pdf | r1 | manage | 22.0 K | 2018-06-01 - 06:14 | LaurenTompkins | Wildcard Efficiencies |
| png |
efficiency.png | r1 | manage | 26.7 K | 2018-06-01 - 06:14 | LaurenTompkins | Wildcard Efficiencies |
| pdf |
efficiency_2D.pdf | r1 | manage | 68.7 K | 2019-04-25 - 15:28 | DavidMStrom | |
| png |
efficiency_2D.png | r1 | manage | 87.5 K | 2019-04-25 - 15:28 | DavidMStrom | |
| pdf |
efficiency_beamspotx.pdf | r1 | manage | 36.6 K | 2019-04-25 - 15:22 | DavidMStrom | |
| png |
efficiency_beamspotx.png | r1 | manage | 93.4 K | 2019-04-25 - 15:22 | DavidMStrom | |
| pdf |
efficiency_stepbystep.pdf | r1 | manage | 43.7 K | 2019-04-25 - 15:19 | DavidMStrom | |
| png |
efficiency_stepbystep.png | r1 | manage | 86.2 K | 2019-04-25 - 15:19 | DavidMStrom | |
| pdf |
fastsim.pdf | r1 | manage | 65.0 K | 2019-04-25 - 15:32 | DavidMStrom | |
| png |
fastsim.png | r1 | manage | 46.7 K | 2019-04-25 - 15:32 | DavidMStrom | |
| pdf |
ftkPlots170303_fig1.pdf | r1 | manage | 6.9 K | 2017-03-04 - 12:49 | StefanSchmitt | Coverage distribution |
| png |
ftkPlots170303_fig1.png | r1 | manage | 55.3 K | 2017-03-04 - 12:49 | StefanSchmitt | Coverage distribution |
| pdf |
ftkPlots170303_fig10.pdf | r1 | manage | 6.7 K | 2017-03-04 - 13:01 | StefanSchmitt | Survey of parameter space with and with out cut on NX |
| png |
ftkPlots170303_fig10.png | r1 | manage | 82.1 K | 2017-03-04 - 13:01 | StefanSchmitt | Survey of parameter space with and with out cut on NX |
| pdf |
ftkPlots170303_fig2.pdf | r1 | manage | 9.9 K | 2017-03-04 - 12:50 | StefanSchmitt | Contro plots (1+1)pix(1)sct, no cut on NX |
| png |
ftkPlots170303_fig2.png | r1 | manage | 86.0 K | 2017-03-04 - 12:50 | StefanSchmitt | Contro plots (1+1)pix(1)sct, no cut on NX |
| pdf |
ftkPlots170303_fig2a.pdf | r1 | manage | 3.9 K | 2017-03-04 - 12:50 | StefanSchmitt | Contro plots (1+1)pix(1)sct, no cut on NX |
| png |
ftkPlots170303_fig2a.png | r1 | manage | 40.2 K | 2017-03-04 - 12:50 | StefanSchmitt | Contro plots (1+1)pix(1)sct, no cut on NX |
| pdf |
ftkPlots170303_fig2b.pdf | r1 | manage | 7.1 K | 2017-03-04 - 12:50 | StefanSchmitt | Contro plots (1+1)pix(1)sct, no cut on NX |
| png |
ftkPlots170303_fig2b.png | r1 | manage | 53.8 K | 2017-03-04 - 12:50 | StefanSchmitt | Contro plots (1+1)pix(1)sct, no cut on NX |
| pdf |
ftkPlots170303_fig2c.pdf | r1 | manage | 4.2 K | 2017-03-04 - 12:50 | StefanSchmitt | Contro plots (1+1)pix(1)sct, no cut on NX |
| png |
ftkPlots170303_fig2c.png | r1 | manage | 43.0 K | 2017-03-04 - 12:50 | StefanSchmitt | Contro plots (1+1)pix(1)sct, no cut on NX |
| pdf |
ftkPlots170303_fig3.pdf | r1 | manage | 10.1 K | 2017-03-04 - 12:52 | StefanSchmitt | Efficiency for: (1+1)pix(1sct), no cut on NX |
| png |
ftkPlots170303_fig3.png | r1 | manage | 81.1 K | 2017-03-04 - 12:52 | StefanSchmitt | Efficiency for: (1+1)pix(1sct), no cut on NX |
| pdf |
ftkPlots170303_fig4.pdf | r1 | manage | 7.8 K | 2017-03-04 - 12:53 | StefanSchmitt | Control plots: (1+1)pix(3)sct) with cuts NX<8 (barrel), NX<5 (endcap) |
| png |
ftkPlots170303_fig4.png | r1 | manage | 84.0 K | 2017-03-04 - 12:53 | StefanSchmitt | Control plots: (1+1)pix(3)sct) with cuts NX<8 (barrel), NX<5 (endcap) |
| pdf |
ftkPlots170303_fig4a.pdf | r1 | manage | 4.1 K | 2017-03-04 - 12:53 | StefanSchmitt | Control plots: (1+1)pix(3)sct) with cuts NX<8 (barrel), NX<5 (endcap) |
| png |
ftkPlots170303_fig4a.png | r1 | manage | 44.3 K | 2017-03-04 - 12:53 | StefanSchmitt | Control plots: (1+1)pix(3)sct) with cuts NX<8 (barrel), NX<5 (endcap) |
| pdf |
ftkPlots170303_fig4b.pdf | r1 | manage | 5.2 K | 2017-03-04 - 12:53 | StefanSchmitt | Control plots: (1+1)pix(3)sct) with cuts NX<8 (barrel), NX<5 (endcap) |
| png |
ftkPlots170303_fig4b.png | r1 | manage | 49.9 K | 2017-03-04 - 12:53 | StefanSchmitt | Control plots: (1+1)pix(3)sct) with cuts NX<8 (barrel), NX<5 (endcap) |
| pdf |
ftkPlots170303_fig4c.pdf | r1 | manage | 4.1 K | 2017-03-04 - 12:53 | StefanSchmitt | Control plots: (1+1)pix(3)sct) with cuts NX<8 (barrel), NX<5 (endcap) |
| png |
ftkPlots170303_fig4c.png | r1 | manage | 43.3 K | 2017-03-04 - 12:53 | StefanSchmitt | Control plots: (1+1)pix(3)sct) with cuts NX<8 (barrel), NX<5 (endcap) |
| pdf |
ftkPlots170303_fig5.pdf | r1 | manage | 9.8 K | 2017-03-04 - 12:54 | StefanSchmitt | Efficiency: (1+1)pix(3)sct with cut NX<8 (barrel) and NX<5 (endcap) |
| png |
ftkPlots170303_fig5.png | r1 | manage | 73.4 K | 2017-03-04 - 12:54 | StefanSchmitt | Efficiency: (1+1)pix(3)sct with cut NX<8 (barrel) and NX<5 (endcap) |
| pdf |
ftkPlots170303_fig6.pdf | r1 | manage | 8.4 K | 2017-03-04 - 12:56 | StefanSchmitt | Efficiency comparison: algorithms with and without cuts on NX |
| png |
ftkPlots170303_fig6.png | r1 | manage | 64.2 K | 2017-03-04 - 12:56 | StefanSchmitt | Efficiency comparison: algorithms with and without cuts on NX |
| pdf |
ftkPlots170303_fig7.pdf | r1 | manage | 6.0 K | 2017-03-04 - 12:57 | StefanSchmitt | Number of roads: algorithms with and without cuts on NX |
| png |
ftkPlots170303_fig7.png | r1 | manage | 55.0 K | 2017-03-04 - 12:57 | StefanSchmitt | Number of roads: algorithms with and without cuts on NX |
| pdf |
ftkPlots170303_fig7a.pdf | r1 | manage | 4.6 K | 2017-03-04 - 12:57 | StefanSchmitt | Number of roads: algorithms with and without cuts on NX |
| png |
ftkPlots170303_fig7a.png | r1 | manage | 60.9 K | 2017-03-04 - 12:57 | StefanSchmitt | Number of roads: algorithms with and without cuts on NX |
| pdf |
ftkPlots170303_fig7b.pdf | r1 | manage | 4.6 K | 2017-03-04 - 12:57 | StefanSchmitt | Number of roads: algorithms with and without cuts on NX |
| png |
ftkPlots170303_fig7b.png | r1 | manage | 62.2 K | 2017-03-04 - 12:57 | StefanSchmitt | Number of roads: algorithms with and without cuts on NX |
| pdf |
ftkPlots170303_fig8.pdf | r1 | manage | 6.1 K | 2017-03-04 - 12:58 | StefanSchmitt | Number of 8-layer fits: algorithms with and without cut on NX |
| png |
ftkPlots170303_fig8.png | r1 | manage | 54.3 K | 2017-03-04 - 12:58 | StefanSchmitt | Number of 8-layer fits: algorithms with and without cut on NX |
| pdf |
ftkPlots170303_fig8a.pdf | r1 | manage | 4.7 K | 2017-03-04 - 12:58 | StefanSchmitt | Number of 8-layer fits: algorithms with and without cut on NX |
| png |
ftkPlots170303_fig8a.png | r1 | manage | 62.2 K | 2017-03-04 - 12:58 | StefanSchmitt | Number of 8-layer fits: algorithms with and without cut on NX |
| pdf |
ftkPlots170303_fig8b.pdf | r1 | manage | 4.7 K | 2017-03-04 - 12:58 | StefanSchmitt | Number of 8-layer fits: algorithms with and without cut on NX |
| png |
ftkPlots170303_fig8b.png | r1 | manage | 60.7 K | 2017-03-04 - 12:58 | StefanSchmitt | Number of 8-layer fits: algorithms with and without cut on NX |
| pdf |
ftkPlots170303_fig9.pdf | r1 | manage | 5.8 K | 2017-03-04 - 13:00 | StefanSchmitt | Duplicate patterns multiplicity |
| png |
ftkPlots170303_fig9.png | r1 | manage | 38.6 K | 2017-03-04 - 13:00 | StefanSchmitt | Duplicate patterns multiplicity |
| pdf |
ftkPlots170303_fig9a.pdf | r1 | manage | 4.5 K | 2017-03-04 - 13:00 | StefanSchmitt | Duplicate patterns multiplicity |
| png |
ftkPlots170303_fig9a.png | r1 | manage | 40.4 K | 2017-03-04 - 13:00 | StefanSchmitt | Duplicate patterns multiplicity |
| pdf |
ftkPlots170303_fig9b.pdf | r1 | manage | 4.6 K | 2017-03-04 - 13:00 | StefanSchmitt | Duplicate patterns multiplicity |
| png |
ftkPlots170303_fig9b.png | r1 | manage | 40.6 K | 2017-03-04 - 13:00 | StefanSchmitt | Duplicate patterns multiplicity |
| pdf |
ggHtthh_mu60_Ftk_CompJZ_numTrackCore.pdf | r1 | manage | 49.4 K | 2014-05-26 - 13:15 | LaurenTompkins | tau plots |
| png |
ggHtthh_mu60_Ftk_CompJZ_numTrackCore.png | r1 | manage | 8.5 K | 2014-05-26 - 13:07 | LaurenTompkins | |
| pdf |
ggHtthh_mu60_Ftk_CompJZ_numTrackIso.pdf | r1 | manage | 48.9 K | 2014-05-26 - 13:15 | LaurenTompkins | tau plots |
| png |
ggHtthh_mu60_Ftk_CompJZ_numTrackIso.png | r1 | manage | 8.9 K | 2014-05-26 - 13:07 | LaurenTompkins | |
| png |
muons_chi2.png | r1 | manage | 24.2 K | 2016-05-30 - 12:23 | StamatiosGkaitatzis | |
| png |
muons_eta.png | r1 | manage | 24.4 K | 2016-05-30 - 12:23 | StamatiosGkaitatzis | |
| pdf |
plotKinematics_etaphi_noIBL_histobugfix.pdf | r1 | manage | 15.6 K | 2019-04-25 - 15:04 | DavidMStrom | eta phi |
| png |
plotKinematics_etaphi_noIBL_histobugfix.png | r1 | manage | 18.3 K | 2019-04-25 - 15:05 | DavidMStrom | eta phi |
| pdf |
plotKinematics_ntrks_noIBL_histobugfix.pdf | r1 | manage | 14.2 K | 2019-04-25 - 15:14 | DavidMStrom | |
| png |
plotKinematics_ntrks_noIBL_histobugfix.png | r1 | manage | 14.3 K | 2019-04-25 - 15:14 | DavidMStrom | |
| pdf |
plotKinematics_pt_noIBL_histobugfix.pdf | r1 | manage | 14.2 K | 2019-04-25 - 15:02 | DavidMStrom | |
| png |
plotKinematics_pt_noIBL_histobugfix.png | r1 | manage | 14.1 K | 2019-04-25 - 11:50 | DavidMStrom | pT no IBL |
| eps |
r00340465_Bitmask_vs_Time_NoneMiss.eps | r1 | manage | 1972.3 K | 2018-05-31 - 06:50 | LaurenTompkins | AUX-ROS plots |
| pdf |
r00340465_Bitmask_vs_Time_NoneMiss.pdf | r1 | manage | 34.6 K | 2018-06-01 - 06:15 | LaurenTompkins | AUX ROS plots |
| png |
r00340465_Bitmask_vs_Time_NoneMiss.png | r1 | manage | 154.0 K | 2018-06-01 - 06:15 | LaurenTompkins | AUX ROS plots |
| eps |
r00340465_Fraction_of_Matched_Events_vs_eta_ignoreGuessed_ignoreMSBs_separately.eps | r1 | manage | 1972.7 K | 2018-05-31 - 06:50 | LaurenTompkins | AUX-ROS plots |
| pdf |
r00340465_Fraction_of_Matched_Events_vs_eta_ignoreGuessed_ignoreMSBs_separately.pdf | r1 | manage | 55.0 K | 2018-06-01 - 06:15 | LaurenTompkins | AUX ROS plots |
| png |
r00340465_Fraction_of_Matched_Events_vs_eta_ignoreGuessed_ignoreMSBs_separately.png | r1 | manage | 108.1 K | 2018-06-01 - 06:15 | LaurenTompkins | AUX ROS plots |
| eps |
r00340465_Fraction_of_Matched_Events_vs_phi_ignoreGuessed_ignoreMSBs_separately.eps | r1 | manage | 1972.1 K | 2018-05-31 - 06:50 | LaurenTompkins | AUX-ROS plots |
| pdf |
r00340465_Fraction_of_Matched_Events_vs_phi_ignoreGuessed_ignoreMSBs_separately.pdf | r1 | manage | 55.4 K | 2018-06-01 - 06:15 | LaurenTompkins | AUX ROS plots |
| png |
r00340465_Fraction_of_Matched_Events_vs_phi_ignoreGuessed_ignoreMSBs_separately.png | r1 | manage | 116.6 K | 2018-06-01 - 06:15 | LaurenTompkins | AUX ROS plots |
| pdf |
r00340465_Fraction_of_Matched_Events_vs_pt_ignoreGuessed_ignoreMSBs_separately.pdf | r1 | manage | 34.6 K | 2018-06-01 - 06:15 | LaurenTompkins | AUX ROS plots |
| png |
r00340465_Fraction_of_Matched_Events_vs_pt_ignoreGuessed_ignoreMSBs_separately.png | r1 | manage | 107.0 K | 2018-06-01 - 06:15 | LaurenTompkins | AUX ROS plots |
| eps |
r00340465_Fraction_of_Matched_Events_vs_pt_ignoreGuessed_ignoreMSBs_separately1.eps | r1 | manage | 1972.7 K | 2018-05-31 - 06:50 | LaurenTompkins | AUX-ROS plots |
| eps |
r00340465_Fraction_of_Real_Events_ignoreGuessed_ignoreMSBs_8.eps | r1 | manage | 1971.2 K | 2018-05-31 - 06:50 | LaurenTompkins | AUX-ROS plots |
| pdf |
r00340465_Fraction_of_Real_Events_ignoreGuessed_ignoreMSBs_8.pdf | r1 | manage | 28.4 K | 2018-06-01 - 06:15 | LaurenTompkins | AUX ROS plots |
| png |
r00340465_Fraction_of_Real_Events_ignoreGuessed_ignoreMSBs_8.png | r1 | manage | 78.1 K | 2018-06-01 - 06:15 | LaurenTompkins | AUX ROS plots |
| pdf |
r00340465_Tracks_per_Event.pdf | r1 | manage | 26.7 K | 2018-06-01 - 06:15 | LaurenTompkins | AUX ROS plots |
| png |
r00340465_Tracks_per_Event.png | r1 | manage | 69.8 K | 2018-06-01 - 06:15 | LaurenTompkins | AUX ROS plots |
| pdf |
r00358407_Bitmask_vs_Time_NoneMiss-eps-converted-to.pdf | r1 | manage | 89.9 K | 2018-09-14 - 14:19 | LaurenTompkins | 2018 AUX->ROS plots |
| png |
r00358407_Bitmask_vs_Time_NoneMiss.png | r1 | manage | 16.4 K | 2018-09-14 - 14:19 | LaurenTompkins | 2018 AUX->ROS plots |
| pdf |
r00358407_Fraction_of_Matched_Events_ignoreGuessed_ignoreMSBs_8-eps-converted-to.pdf | r1 | manage | 89.4 K | 2018-09-14 - 14:19 | LaurenTompkins | 2018 AUX->ROS plots |
| png |
r00358407_Fraction_of_Matched_Events_ignoreGuessed_ignoreMSBs_8.png | r1 | manage | 13.5 K | 2018-09-14 - 14:19 | LaurenTompkins | 2018 AUX->ROS plots |
| pdf |
r00358407_Fraction_of_Matched_Events_vs_pt_ignoreGuessed_ignoreMSBs-eps-converted-to.pdf | r1 | manage | 88.8 K | 2018-09-14 - 14:19 | LaurenTompkins | 2018 AUX->ROS plots |
| png |
r00358407_Fraction_of_Matched_Events_vs_pt_ignoreGuessed_ignoreMSBs.png | r1 | manage | 13.1 K | 2018-09-14 - 14:19 | LaurenTompkins | 2018 AUX->ROS plots |
| pdf |
r00358407_Fraction_of_Matched_Events_vs_pt_ignoreGuessed_ignoreMSBs_separately-eps-converted-to.pdf | r1 | manage | 88.6 K | 2018-09-14 - 14:19 | LaurenTompkins | 2018 AUX->ROS plots |
| pdf |
r00358407_Fraction_of_Matched_Events_vs_pt_ignoreGuessed_ignoreMSBs_separately-eps-converted-to1.pdf | r1 | manage | 88.6 K | 2018-09-14 - 14:05 | LaurenTompkins | 2018 AUX->ROS plots |
| eps |
r00358407_Fraction_of_Matched_Events_vs_pt_ignoreGuessed_ignoreMSBs_separately.eps | r1 | manage | 1974.1 K | 2018-09-14 - 14:05 | LaurenTompkins | 2018 AUX->ROS plots |
| png |
r00358407_Fraction_of_Matched_Events_vs_pt_ignoreGuessed_ignoreMSBs_separately.png | r1 | manage | 17.0 K | 2018-09-14 - 14:19 | LaurenTompkins | 2018 AUX->ROS plots |
| pdf |
r00358407_Fraction_of_Real_Events_ignoreGuessed_ignoreMSBs_8-eps-converted-to.pdf | r1 | manage | 89.3 K | 2018-09-14 - 14:05 | LaurenTompkins | 2018 AUX->ROS plots |
| png |
r00358407_Fraction_of_Real_Events_ignoreGuessed_ignoreMSBs_8.png | r1 | manage | 13.9 K | 2018-09-14 - 14:05 | LaurenTompkins | 2018 AUX->ROS plots |
| pdf |
r00358407_Tracks_per_Event-eps-converted-to.pdf | r1 | manage | 88.6 K | 2018-09-14 - 14:05 | LaurenTompkins | 2018 AUX->ROS plots |
| pdf |
r00358407_Tracks_per_Event-eps-converted-to1.pdf | r1 | manage | 88.6 K | 2018-09-14 - 14:05 | LaurenTompkins | 2018 AUX->ROS plots |
| eps |
r00358407_Tracks_per_Event.eps | r1 | manage | 1973.9 K | 2018-09-14 - 14:05 | LaurenTompkins | 2018 AUX->ROS plots |
| png |
r00358407_Tracks_per_Event.png | r1 | manage | 13.2 K | 2018-09-14 - 14:05 | LaurenTompkins | 2018 AUX->ROS plots |
| pdf |
ttHHLTRefitClean.pdf | r1 | manage | 47.9 K | 2014-06-27 - 12:03 | LaurenTompkins | |
| png |
ttHHLTRefitClean.png | r1 | manage | 9.0 K | 2014-06-27 - 12:03 | LaurenTompkins | |
| png |
ttbar_chi2.png | r1 | manage | 22.3 K | 2016-05-30 - 12:23 | StamatiosGkaitatzis | |
| png |
ttbar_eta.png | r1 | manage | 25.0 K | 2016-05-30 - 12:23 | StamatiosGkaitatzis |
Topic revision: r24 - 2019-05-09 - DavidMStrom
ATLAS 
Internal TWiki
Changes
Notifications
RSS Feed
Account 
Copyright &© 2008-2023 by the contributing authors. All material on this collaboration platform is the property of the contributing authors. or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback