Approved Combined Inner Detector Plots

Introduction

This page shows early Run-1 combined performance plots, and has been superceded by InDetTrackingPerformanceApprovedPlots.

Figures

Detector alignment

Unbiased residual distribution in x, integrated over all hits-on-tracks in the pixel barrel for the nominal geometry and the preliminary aligned geometry. The residual is defined as the measured hit position minus the expected hit position from the track extrapolation. Shown is the projection onto the local x coordinate, which is the precision coordinate. Tracks are selected to have pT > 2 GeV, |d0|<50mm, |z0|<400mm (in other words they are required to go through the pixel L0). The distribution is shown for 5 cosmic runs: 91885, 91888, 91890, 91891, 91900, all have solenoid on - corresponding to ~25% of all the solenoid-on data taken in Sep/Oct 2008. New Tracking is used (in 14.5.2.1). A double Gaus fit is performed but only the mean and sigma of the narrower Gaussian are shown.	Approved_PixBarResX.eps
Unbiased residual distribution in y, integrated over all hits-on-tracks in the pixel barrel for the nominal geometry and the preliminary aligned geometry. The residual is defined as the measured hit position minus the expected hit position from the track extrapolation. Shown is the projection onto the local y coordinate, which is the non-precision coordinate. Tracks are selected to have pT > 2 GeV, |d0|<50mm, |z0|<400mm (in other words they are required to go through the pixel L0). The distribution is shown for 5 cosmic runs: 91885, 91888, 91890, 91891, 91900, all have solenoid on - corresponding to ~25% of all the solenoid-on data taken in Sep/Oct 2008. New Tracking is used (in 14.5.2.1). A single Gaussian fit is performed.	Approved_PixBarResY.eps
Unbiased residual distribution in x, integrated over all hits-on-tracks in the SCT barrel for the nominal geometry and the preliminary aligned geometry. The residual is defined as the measured hit position minus the expected hit position from the track extrapolation. Shown is the projection onto the local x coordinate, which is the precision coordinate. Tracks are selected to have pT > 2 GeV, |d0|<50mm, |z0|<400mm (in other words they are required to go through the pixel L0). The distribution is shown for 5 cosmic runs: 91885, 91888, 91890, 91891, 91900, all have solenoid on - corresponding to ~25% of all the solenoid-on data taken in Sep/Oct 2008. New Tracking is used (in 14.5.2.1). A single Gaussian fit is performed.	Approved_SCTBarResX.eps
TRT resolution in Fall 2008 cosmic data (Xenon) This plots is made with combined ID tracks, with the following requirements: - minimum pT>2GeV - min 2 Pixel hits - min 9 SCT hits - min 45 TRT hits - Event Phase in range 5 <= EP <= 30 - Reject tracks with endcap hits - Reject tracks that come from events with more than one reco track (< 5 % of events, avoids pattern reco mistakes ) The fit is a single Gaussian, where the fit was iterated until the range corresponded to +/- 1.5*sigma. The mean and sigma of the fit are reported. The tracks are for five golden runs with magnetic field from Fall 2008 (runs 91885, 91888, 91890, 91891, 91900). The reconstruction was done with Si geometry tag InDetCosmics_2008_03 The distribution is shown for both before and after (tag TRT_Comsics_2008_06) TRT L1 alignment. The after plot includes both a TRT L1 and L2 alignment.	See Atlas.ApprovedPlotsTRT for TRT position resolution in 2008 and 2009 data (with alignment, TRT standalone tracks).
Cosmic tracks crossing the entire ID leave hits in both the upper and lower halves of the ID. These tracks can be split near the interaction point and fit separately, resulting in two collision-like tracks that can then be compared. The plots shows the difference in the d0 track parameter between the two split tracks. Tracks are selected to have pT > 2 GeV, |d0|<50mm, |z0|<400mm (in other words they are required to go through the pixel L0). Tracks also are required to have a hit in the Pixel B layer, 3 Pixel hits and in total 7 Silicon hits. The distribution is shown for 5 cosmic runs: 91885, 91888, 91890, 91891, 91900, all have solenoid on - corresponding to ~25% of all the solenoid-on data taken in Sep/Oct 2008. New Tracking is used (in 14.5.2.1). A single Gaussian fit is performed.	Approved_DeltaD0.eps
Cosmic tracks crossing the entire ID leave hits in both the upper and lower halves of the ID. These tracks can be split near the interaction point and fit separately, resulting in two collision-like tracks that can then be compared. The plots shows the difference in the z0 track parameter between the two split tracks. Tracks are selected to have pT > 2 GeV, |d0|<50mm, |z0|<400mm (in other words they are required to go through the pixel L0). Tracks also are required to have a hit in the Pixel B layer, 3 Pixel hits and in total 7 Silicon hits. The distribution is shown for 5 cosmic runs: 91885, 91888, 91890, 91891, 91900, all have solenoid on - corresponding to ~25% of all the solenoid-on data taken in Sep/Oct 2008. New Tracking is used (in 14.5.2.1). A single Gaussian fit is performed.	Approved_DeltaZ0.eps
Cosmic tracks crossing the entire ID leave hits in both the upper and lower halves of the ID. These tracks can be split near the interaction point and fit separately, resulting in two collision-like tracks that can then be compared. The plots shows the difference in the Q/pT track parameter between the two split tracks. Tracks are selected to have pT > 2 GeV, |d0|<50mm, |z0|<400mm (in other words they are required to go through the pixel L0). Tracks also are required to have a hit in the Pixel B layer, 3 Pixel hits and in total 7 Silicon hits. The distribution is shown for 5 cosmic runs: 91885, 91888, 91890, 91891, 91900, all have solenoid on - corresponding to ~25% of all the solenoid-on data taken in Sep/Oct 2008. New Tracking is used (in 14.5.2.1). A single Gaus fit is performed.	Approved_DeltaQoPT.eps
Cosmic tracks crossing the entire ID leave hits in both the upper and lower halves of the ID. These tracks can be split near the interaction point and fit separately, resulting in two collision-like tracks that can then be compared. The plots shows the difference in the phi track parameter between the two split tracks. Tracks are selected to have pT > 2 GeV, |d0|<50mm, |z0|<400mm (in other words they are required to go through the pixel L0). Tracks also are required to have a hit in the Pixel B layer, 3 Pixel hits and in total 7 Silicon hits. The distribution is shown for 5 cosmic runs: 91885, 91888, 91890, 91891, 91900, all have solenoid on - corresponding to ~25% of all the solenoid-on data taken in Sep/Oct 2008. New Tracking is used (in 14.5.2.1). A double Gaussian fit is performed but only the mean and sigma of the narrower Gaus are shown.	Approved_DeltaPhi.eps
Unbiased residual distribution in x, integrated over all hits-on-tracks in the pixel barrel using the 2008 aligned geometry. The residual is defined as the measured hit position minus the expected hit position from the track extrapolation. Shown is the projection onto the local x coordinate, which is the precision coordinate. Tracks are selected to have pT > 1 GeV, |d0|<50mm, |z0|<400mm (in other words they are required to go through the pixel L0), and to have 1 B-layer, 3 Pixel, 8 SCT, and 25 TRT Hits. The distribution is shown using data taken in 2009, cosmic run: 121330, which had solenoid on. New Tracking is used (in AtlasTier0-15.2.0.7). A double Gaus fit is performed but only the mean and sigma of the narrower Gaussian are shown. Note: The distribution shown here is made with the new standard cuts (listed), we have verified that the results are not sensitive to the difference in track selection.	ApprovedPlots2009Data_PixelX.eps
Unbiased residual distribution in y, integrated over all hits-on-tracks in the pixel barrel using the 2008 aligned geometry. The residual is defined as the measured hit position minus the expected hit position from the track extrapolation. Shown is the projection onto the local y coordinate, which is the non-precision coordinate. Tracks are selected to have pT > 1 GeV, |d0|<50mm, |z0|<400mm (in other words they are required to go through the pixel L0), and to have 1 B-layer, 3 Pixel, 8 SCT, and 25 TRT Hits. The distribution is shown using data taken in 2009, cosmic run: 121330, which had solenoid on. New Tracking is used (in AtlasTier0-15.2.0.7). A single Gaus fit is performed and the mean and sigma are shown. Note: The distribution shown here is made with the new standard cuts (listed), we have verified that the results are not sensitive to the difference in track selection.	ApprovedPlots2009Data_PixelY.eps
Unbiased residual distribution in x, integrated over all hits-on-tracks in the SCT barrel using the 2008 aligned geometry. The residual is defined as the measured hit position minus the expected hit position from the track extrapolation. Shown is the projection onto the local x coordinate. Tracks are selected to have pT > 1 GeV, |d0|<50mm, |z0|<400mm (in other words they are required to go through the pixel L0), and to have 1 B-layer, 3 Pixel, 8 SCT, and 25 TRT Hits. The distribution is shown using data taken in 2009, cosmic run: 121330, which had solenoid on. New Tracking is used (in AtlasTier0-15.2.0.7). A single Gaus fit is performed and the mean and sigma are shown. Note: The distribution shown here is made with the new standard cuts (listed), we have verified that the results are not sensitive to the difference in track selection.	ApprovedPlots2009Data_SCTX.eps

ID track parameter resolutions

Cosmics muons traverse the whole Inner Detector and thus leave hits in the upper and lower parts of the detector. By dividing the track to its upper and lower half according to the value of the y coordinate of hits on track and refitting both hit collections, two collision-like tracks originating from the same cosmic muon are obtained. After alignment, track parameter resolutions are studied by comparing the difference (residual) of the track parameters at the perigee point. Since both tracks have an associated error, the quoted resolution is the RMS of the residual distribution of the particular track parameter divided by square root 2. The track parameter resolutions are studied dependant on variables like the pT or the d0 of the tracks. Shifted mean values of the residual distributions can be a sign of systematic detector deformations. The following distributions show data from runs 91885,91888,91890,91891 and 91900 taken in 2008. The tracks have been refitted using Athena release 15.0.0.7 (Tier0).

The following cuts have ben applied per track (if not stated otherwise):

• nPixelHits in the barrel >=2
• nSCTHits in the barrel >=6
• nTRTHits in the barrel >=25
• |d0| < 40mm
• pT >= 1 GeV

Additionally only events with an event phase between 5 and 30 ns are accepted to ensure proper timing of the sub-detectors. The plots show comparisons of tracks using the full Inner Detector (silicon and TRT detectors, closed triangles), only the silicon sub-detecors (open triangles) together with tracks from cosmic simulation using the full Inner Detector (stars). The requirement of at least 25 TRT hits is dropped for silicon only tracks. However, the cut on the event phase is retained to ensure proper timing and the comparability of the analyzed sets of tracks. There is no cut on the event phase for simulation events since the jitter in cosmic trigger timing is not simulated.

Transverse impact parameter resolution as a function of pT. In the low pT region, the resolution is dominated by multiple scattering effects. At higher values, the resolution is flat. Taking into account the TRT information improves the resolution. The difference to the MC curve indicates the remaining mislaignment.	eps figure
Transverse impact parameter resolution as a function of d0 itself. For this plot the d0 cut is released to 120 mm and the minum number of Pixel hits is set to one. In general the resolution for full ID tracks is better. The resolution is better in the central d0 region due to more Pixel layers crossed and less spread clusters in the Pixel detector. Dips are seen if the d0 of the tracks equal the radii of the pixel layers (indicated by dashed lines). Since the d0 is in these cases very close to a hit on a Pixel layer, the extrapolation to the perigee point is very small and the resolution improves. The MC distributions confirms the observed behaviour.	eps figure
Mean of the transverse impact parameter distribution as a function of pT. The expected value of the mean is 0 as confirmed by the MC distribution. In data a shift is seen for full ID and silicon only tracks. The shift increases with higher pT.	eps figure
Mean of the transverse impact parameter distribution as a function of d0 itself. For this plot the d0 cut is released to 120 mm and the minum number of Pixel hits is set to one. The expected value of the mean is 0 as confirmed by the MC distribution. In data a shift is seen for full ID and silicon only tracks. The shift is biggest in the central d0 region.	eps figure
Relative momentum resolution as a function of pT. The relative momentum resolution increases with higher pT due to stiffer tracks and a more difficult measurement of the sagitta. Including information from the TRT extends the lever arm and helps improving the resolution especially at high pT values. The difference to the MC curve indicates the remaining misalignment.	eps figure
Mean of the relative momentum distribution as a function of pT. The expected value of the mean is 0 as confirmed by the MC distribution. In data a shift is seen for full ID and silicon only tracks. The shift increases with higher pT.	eps figure
Resolution of the azimuthal angle as a function of pT. In the low pT region, the resolution is dominated by multiple scattering effects. At higher values, the resolution is flat. Taking into account the TRT information improves the resolution. The difference to the MC curve indicates the remaining mislaignment.	eps figure
Resolution of the polar angle as a function of eta. The resolution of the polar angle theta improves at larger eta due to broader pixel clusters that allow a more precise position measurement. Since the TRT effectively does not measure the z coordinate in the barrel region, the resolutions are equal for silicon only and full ID tracks. The difference to the MC curve indicates the remaining misalignment.	eps figure

Distributions from cosmic ray tracks

Warning: Cosmic ray spectra including comparisons with MC which appeared before on this page are not approved for external use at this stage. The MC does not model some details of the special cosmic trigger configuration and timing. It includes the two main access shafts, but not the two lift shafts. Until these effects are taken into account, the MC-data comparison is not useful.

Cosmic muons crossing the entire ATLAS detector leave hits in the Inner Detector (ID) and so tracks can be reconstructed. Two different tracking algorithms are in use for the reconstruction: the CTB (Cosmics and Test Beam) tracking and the New Tracking. The results below show the performance of the CTB.

Tracks are characterized by 5 parameters. These are defined in a reference point, the perigee, which is the point of closest approach to the z/beam axis. d0 is the signed distance to the z-axis, z0 is the z-coordinate of the perigee, phi0 is the angle in the x-y plane at the perigee, theta0 is the angle with the z-axis and q/p is the charge of the cosmic muon divided by its momentum.

In the next plots, the parameters of these cosmic muon tracks (reconstructed with CTB tracking and release 14.5.0.5, AtlasProduction) from the ATLAS combined cosmic run 91890 (Autumn 2008) are shown. Both toroid and solenoid were on during this run.

For the theta0 and z0 distributions, tracks are required to have Silicon hits (since these parameters are not measured by the TRT barrel).

Cosmic spectra (d0 distribution) as seen in run 91890.	eps figure
Cosmic spectra (z0 distribution) as seen in run 91890.	eps figure
Cosmic spectra (phi0 distribution) as seen in run 91890.	eps figure
Cosmic spectra (theta0 distribution) as seen in run 91890.	eps figure
Cosmic spectra (QoverP distribution) as seen in run 91890.	eps figure

Cosmic ray track statistics

Collected track statistics in autumn 2008:   field off field on total All tracks 4940000 2670000 7610000 SCT hit 1150000 880000 2030000 Pixel hit 230000 190000 420000

pdf figure

Event displays

Inner detector event display for event 757427 of the cosmic run 90270 with field ON.
Inner detector event display for event 62340 of the cosmic run 90731 (no Solenoid). Showing hits on track for both TRT endcaps, as well as SCT and pixel barrel and endcap hits.

ID performance, including charge assignment, with 2009 cosmic ray data

ID tracking performance studies with the sample of cosmic ray data taken in 2009 have been carried out, with particular attention to the efficiencies and resolutions for positive and negative tracks.

These plots were approved on 25 March 2010. We used the following 2009 Cosmic runs: 121198, 121238, 121275, 121330, 121366, 121368, 121416, 121457, 121513, 121569. We use the IDCosmic stream and require the L1_TRT trigger. The tracks were rereconstructed with the following alignment sets and configuration:

• InDetAlign Collision Robust 2009 05
• TRTAlign Collision Robust 2009 02
• IndetTrkErrorScaling-004-00
• CorrectionFactor = 0.93

For trigger, track selection efficiency and resolution studies, we split into non-pixel and pixel track categories with the following requirements:

• TRTEventphase >-30
• |pT | > 10 GeV
• |theta + 0.5pi| < .5 and |phi + 0.5pi| < .5
• NonPixel: |d0| < 299 mm Pixel: |d0|< 89 mm
• NonPixel: |z0| < 540 mm Pixel: |d0|< 350 mm
• NonPixel: NPixelBarrel = 0 Pixel: NPixelBarrel≥ 2
• NSCTBarrel ≥ 6
• NTRTBarrelHits > 25

For the track finding efficiency we require:

For TRT tracks:

• TRTEventphase >-30
• |pT | > 10 GeV
• |phi + 0.5pi| < .5
• NonPixel: |d0| < 299 mm
• NTRTHits top hemisphere > 20 and bottomhemisphere > 5

For Si tracks:

• TRTEventphase >-30
• |pT | > 10 GeV
• |theta + 0.5pi| < .5 and |phi + 0.5pi| < .5
• |z0| < 700 mm Pixel:
• NSCTEndCapHits = 0

Ratio of positive to negative TRT Trigger efficiency, measured with respect to the muon trigger in the different momentum bins for pixel tracks.	eps figure
Ratio of positive to negative TRT Trigger efficiency, measured with respect to the muon trigger in the different momentum bins for non-pixel tracks.	eps figure
Track finding efficiencies for the Silicon (blue solid) and TRT (red open) CTB track collection	eps figure
Ratio of positive to negative muon track finding efficiencies for the Silicon (blue solid) and TRT (red open) CTB track collections	eps figure
Track selection efficiencies for tracks with (black circles) and without (blue triangles) pixel hits as a function of the track momentum	eps figure
Ratio of positive to negative track selection efficiencies for tracks with pixel hits (black circles) and tracks with pixel hits (blue triangles).	eps figure
The fraction of events w = NOS/(NOS +NSS) where the two half tracks disagree on charge, versus measured momentum, p, for tracks without pixel hits. No charge disagreement has been found for tracks with pixel hits. For entries with 0 events, we use the convention of an uncertainty of +1 (which corresponds to roughly 68% credibility), converted into a rate. The probability of measuring a single (half)track with the wrong charge is approximately 0.5 times w, for w << 1.	eps figure
Curvature (1/pT ) resolution of upper/lower half tracks, versus momentum p of full track. The upward (downward) pointing triangular datapoints show positive (negative) tracks with pixel hits. The resolution appears independent of charge. For reference, the filled circular datapoints show tracks passing a tighter selection (most importantly |d0| < 40 mm), yielding tracks more similar to those found in collision events. The dotted (solid) lines shows the result of fitting the charge-integrated curvature resolution of pixel (collision-like) tracks versus p (see text for function). This yields an asymptotic resolution σ∞ = 0.00053±0.00001 GeV−1 (σ∞ = 0.00047±0.00001 GeV−1) for pixel (collision-like) tracks	eps figure
Curvature (1/pT ) bias of positive (upward pointing triangles) and negative (downward pointing triangles) half tracks with pixel hits, versus momentum p of full track. A fit of a zeroth-order polynomial (not shown) to these datapoints yields a bias of 17±2×10−5 GeV−1 (7±2×10−5 GeV−1) for positive (negative) halftracks.	eps figure
Curvature (1/pT ) resolution of upper/lower half tracks, versus momentum p of full track. The upward (downward) pointing triangular datapoints show positive (negative) tracks without pixel hits. The resolution appears independent of charge. For reference, the filled circular datapoints show tracks passing a tighter selection (most importantly |d0| < 40 mm), yielding tracks more similar to those found in collision events. The dotted (solid) lines shows the result of fitting the charge-integrated curvature resolution of non-pixel (collision-like) tracks versus p (see text for function). This yields an asymptotic resolution σ∞ = 0.00135±0.00001 GeV−1 (σ∞ = 0.00047±0.00001 GeV−1) for non-pixel (collision-like) tracks.	eps figure
Curvature (1/pT ) bias of positive (upward pointing triangles) and negative (downward pointing triangles) half tracks without pixel hits, versus momentum p of full track. A fit of a zeroth-order polynomial (not shown) to these datapoints yields a bias of −4±3×10−5 GeV−1 (−12±3×10−5 GeV−1) for positive (negative) halftracks.	eps figure
Resolution of transverse impact parameter, d0, of upper/lower half tracks, versus momentum p of full track. The upward (downward) pointing triangular datapoints show positive (negative) tracks with pixel hits. The resolution appears independent of charge. For reference, the filled circular datapoints show tracks passing a tighter selection (most importantly |d0| < 40 mm), yielding tracks more similar to those found in collision events. The dotted (solid) lines shows the result of fitting the chargeintegrated curvature resolution of pixel (collision-like) tracks versus p (see text for function). This yields an asymptotic resolution σ∞ = 29.3±0.4 μm (σ∞ = 19.4±0.3 μm) for pixel (collision-like) tracks.	eps figure
Bias of transverse impact parameter, d0, for positive (upward pointing triangles) and negative (downward pointing triangles) half tracks with pixel hits, versus momentum p of full track. A fit of a zeroth-order polynomial (not shown) to these datapoints yields a bias of −2±1μm (−0.8±1μm) for positive (negative) halftracks.	eps figure

Beam Conditions Monitor (BCM) timing distributions

During the combined Inner Detector running in November 2008, out of 50 million triggers, over 100 events with signals in the BCM were observed. The BCM reads out 31 consecutive bunch crossings for each trigger, and the initial timing calibration from measured cable and fibre lengths puts the signal around bin 19. The timing distribution is narrower for events triggered by the TRT than for events triggered by the RPC muon chambers.

BCM timing distribution for RPC triggered events	eps figure
BCM timing distribution for TRT triggered events	eps figure

Beam Conditions Monitor (BCM) beam aborts

Beam aborts from aperture scans. These plots were approved on 16 February 2010.

Total number of BCM high and low gain channel hits integrated over 40 us versus time during a Beam 2 IR1 horizontal aperture scan with LHC fill 923. The post mortem buffer covers 1177 LHC revolutions (105 ms). It is frozen 9 ms (~100 orbits) after an abort. An increase of the number of BCM hits is visible during a step of the aperture scan starting at -26 ms before the end of the buffer. A beam abort initiated by BCM occurred at -9.8 ms before the end of the buffer and the BCM hit activity returns to zero after the beam is dumped.

eps figure

Timelines of all 16 BCM readout channels in full 390 ps sampling resolution covering 3 25 ns bunch crossings. The are recorded with the BCM post mortem buffer during a Beam 2 IR1 horizontal aperture scan with LHC fill 923 in the moment of a BCM initiated beam abort. The left timelines are recorded by one, the right by another ROD. High gain (HG) channel timelines have a green, low gain (LG) ones a yellow background. The top four timelines are from A side detectors, the bottom four from C side detectors. The 3 HG + 3 LG beam abort condition was reached for both RODs and the HG channel show, as expected, longer pulse widths. The pulses of the C side detectors are about 12.5 ns before the A side detector pulses as expected for beam background from Beam 2 (C->A).

eps figure

Comparison of Atlas.NewTracking and CTBTracking

Comparison of NewTracking versus CTB Tracking for real data from the IDCosmic stream of run 91885 (field ON) reconstructed in the first reprocessing in December 2008. (These are technical plots, comparing the results of two different tracking algorithms).

The following cuts have ben applied:

• 2*nPixelHits+nSCTHits>=8
• number of TRT hits >=40
• pT >= 5 GeV

Distribution of pt for NewTracking and CTB Tracking.	eps figure
Distribution of q/p for NewTracking and CTB Tracking.	eps figure
Number of Hits (nPixel+nSCT+nTRT) on the track.	eps figure
Distribution of d0 for NewTracking and CTB Tracking.	eps figure
Distribution of phi0 for NewTracking and CTB Tracking.	eps figure
Distribution of z0 for NewTracking and CTB Tracking.	eps figure
Distribution of η for NewTracking and CTB Tracking. The double-peak structure at η ≈ -0.4 und η ≈ 0.3 is due to the construction shafts through which the ATLAS detector was lowered into the cavern.	eps figure

Impact of Random Misalignments

The figures below are taken from the ATLAS PUB note The Impact of Inner Detector Misalignments on Selected Physics Processes. Shown here is the impact of the Day-1 and Day-100 random ID misalignments on reconstructed Zmumu, Jpsi and Bd masses in Monte Carlo simulated samples of these processes.

The ID reconstructed Z mass distribution for a Zmumu Monte Carlo sample reconstructed using the Day-1 and Day-100 ID misalignments and the perfect ID alignment. The ID reconstructed Z mass is the invariant mass formed from the two highest pT Inner Detector tracks, with both tracks satisfying pT>15 GeV and having opposite charge. One can see that the random smearing of module positions in the Day-1 and Day-100 alignment constants clearly impact the Z mass resolution.	eps figure
The difference between the ID reconstructed Z mass and the true Z mass for a Zmumu Monte Carlo sample reconstructed using the Day-1 and Day-100 ID misalignments and the perfect ID alignment. The ID reconstructed Z mass is the invariant mass formed from the two highest pT Inner Detector tracks, with both tracks satisfying pT>15 GeV and having opposite charge. The true Z mass is the invariant mass of the two true particles can be associated to the ID tracks used in the Z reconstruction. A Gaussian is fitted in the range [mu-RMS,mu+RMS], and the mean and width of this Gaussian stated in the plot. The random module position smearing of the The Day-1 misalignments, at the level of 20microns, produces a Z mass resolution degraded by ~50% when compared to the perfect alignment case. The Day-100 geometry uses a reduced random smearing of the module positions, at the level of 10microns, and consequently the impact on the Z mass resolution is smaller, a ~13% degradation.	eps figure
The difference between the ID reconstructed Z mass and the true Z mass for a Zmumu Monte Carlo sample reconstructed using the Day-1 and Day-100 ID misalignments and the perfect ID alignment, where the tracks have been restricted to |eta|<1.0 (barrel region).	eps figure
The results of a Gaussian fit to the JPsi mass for events in a Bd Monte Carlo sample that have been reconstructed with Day-1 and Day-100 ID alignment constants and perfect ID alignment. The JPsi candidates are selected by looking for opposite charged track pairs which originate from a common vertex and have a mass within 3 sigma of the nominal JPsi mass. One can see that the impact of random misalignments on the JPsi mass resolution is not significant.	eps figure
The distribution of the event-by-event difference between the JPsi mass when reconstructed with the ideal ID alignment, to the mass when the same event is reconstructed using the Day-1 or Day-100 ID random misalignments. The mean and width of a Gaussian fit made to the core of the distribution is reported in the plot. The Day-1 and Day-100 misalignments degrade the JPsi mass resolution relative to the ideal alignment case by only ~23MeV and ~12MeV respectively. This effect is not significant when added in quadrature to the ideal alignment JPsi resolution of 48MeV (see plot above).	eps figure
The results of a Gaussian fit to the Bd mass for events in a Bd Monte Carlo sample that have been reconstructed with Day-1 and Day-100 ID alignment constants and perfect ID alignment. For details of the Bd mass reconstruction see the CSC book pages 1124-1126. One can see that the impact of random misalignments on the Bd mass resolution is not significant.	eps figure
The distribution of the event-by-event difference between the Bd mass when reconstructed with the ideal ID alignment, to the mass when the same event is reconstructed using the Day-1 or Day-100 ID random misalignments. The mean and width of a Gaussian fit made to the core of the distribution is reported in the plot. The Day-1 and Day-100 misalignments degrade the Bd mass resolution relative to the ideal alignment case by only ~32MeV and ~16MeV respectively. This effect is not significant when added in quadrature to the ideal alignment Bd resolution of 77MeV (see plot above).	eps figure

Impact of Global Systematic Misalignments

The figures below are taken from the ATLAS PUB note The Impact of Inner Detector Misalignments on Selected Physics Processes. Shown here is the impact of the Curl, Twist, Elliptical and Telescope global systematic ID misalignments on combined ID track reconstruction and on reconstructed Zmumu, Jpsi and Bd masses.

The \chisquareddof~distributions of Inner Detector tracks when reconstructed with each of the Curl, Twist, Elliptical and Telescope weak mode ID misalignments compared to the ideal ID alignment case. A Zmumu Monte Carlo sample is used, and all tracks with reconstructed pT>2GeV are shown.	Curl eps figure Twist eps figure Elliptical eps figure Telescope eps figure
The difference between the reconstructed Inner Detector track Q/pT and the associated truth particle Q/pT when tracks are reconstructed with the Curl-Large and Curl-Small ID misalignments compared to the ideal ID alignment case. A Zmumu Monte Carlo sample is used, and all tracks with reconstructed pT>2GeV are shown. The different plots cover different track eta regions.	Barrel eps figure ECA eps figure ECC eps figure
Ratio of the reconstructed Inner Detector track pT to the associated truth particle pT as a function of the truth particle Q*pT for track |eta|<1.0. Tracks from a Zmumu Monte Carlo sample are reconstructed with the Curl-Large and Curl-Small ID misalignments and the ideal alignment case. Each point represents the mean of the pT^{reco}/pT^{truth} distribution in that bin, with the error bars representing the error on the mean. The plot is made separately for positively and negatively charged tracks.	Positive eps figure Negative eps figure
The ID reconstructed Z mass distribution for a \zmumu~Monte Carlo sample reconstructed using the Curl misalignments, compared with the ideal ID alignment. The ID reconstructed Z mass is the invariant mass formed from the two highest pT Inner Detector tracks, with both tracks satisfying pT>15GeV and having opposite charge. An impact on the Z mass resolution is clearly produced by the Curl misalignment. The magnitude of this impact is better seen in the figure below.	eps figure
The difference between the ID reconstructed Z mass and the truth Z mass, for a Zmumu Monte Carlo sample reconstructed using the Curl-Large and Curl-Small ID misalignments and the ideal ID alignment. The two plots differ in the Eta restrictions applied to the tracks used to reconstruct the Z boson (either no restriction or requiring both legs |eta|<1.0). A Gaussian is fitted in the range [mu-RMS,mu+RMS], and the mean and width of this Gaussian stated in the plot.	eps figure Barrel eps figure
The distribution of the event-by-event difference between the JPsi mass when reconstructed with the ideal ID alignment, to the mass when the same event is reconstructed using the Curl-Large and Curl-Small ID misalignments. A Bd simulated Monte Carlo sample is used. The mean and width of a Gaussian fit made to the core of the distribution is reported in the plot.	eps figure
The distribution of the event-by-event difference between the Bd mass when reconstructed with the ideal ID alignment, to the mass when the same event is reconstructed using the Curl-Large and Curl-Small ID misalignments. A Bd simulated Monte Carlo sample is used. The mean and width of a Gaussian fit made to the core of the distribution is reported in the plot.	eps figure
Figure description ...	figure

Links

• Inner Detector tracking performance Wiki
• Inner Detector software Wiki
• Plots prepared and approved for the LHC Detector Alignment Workshop in June 2009. For presentations with more technical ID alignment information. Please contact the ID Alignment convener before using them.