LHCb PatVeloTT Performance

34
LHCb PatVeloTT LHCb PatVeloTT Performance Performance Adam Webber Adam Webber

description

LHCb PatVeloTT Performance. Adam Webber. Why Upgrade?. Currently we de-focus the beams LHCb Luminosity ~ 2x10 32 cm -2 s -1 ~ 1 interaction per bunch crossing Design Luminosity ~ 10 34 cm -2 s -1 Upgrade Luminosity ~ 2x10 33 cm -2 s -1 Cleverer trigger More sensitive detector. - PowerPoint PPT Presentation

Transcript of LHCb PatVeloTT Performance

Page 1: LHCb PatVeloTT Performance

LHCb PatVeloTT LHCb PatVeloTT PerformancePerformance

Adam WebberAdam Webber

Page 2: LHCb PatVeloTT Performance

Why Upgrade?Why Upgrade?

Currently we de-focus the beamsCurrently we de-focus the beamso LHCb Luminosity ~ 2x10LHCb Luminosity ~ 2x103232 cm cm-2-2ss-1-1

o ~ 1 interaction per bunch crossing~ 1 interaction per bunch crossing

o Design Luminosity ~ 10Design Luminosity ~ 1034 34 cmcm-2-2ss-1-1

Upgrade Luminosity ~ 2x10Upgrade Luminosity ~ 2x1033 33 cmcm-2-2ss-1-1

o Cleverer triggerCleverer trigger

o More sensitive detectorMore sensitive detector

22

Page 3: LHCb PatVeloTT Performance

The DetectorThe Detector

Single arm forward spectrometerSingle arm forward spectrometer

33

Page 4: LHCb PatVeloTT Performance

Sector: a set of sensors Sector: a set of sensors

connected to the same connected to the same

readout chipreadout chip

Split into 1, 2, 3 and 4 Split into 1, 2, 3 and 4

sensor long sectorssensor long sectors

Trigger Tracker (TT)Trigger Tracker (TT)

44

Page 5: LHCb PatVeloTT Performance

Upgrade Project 1: Upgrade Project 1: TT GranularityTT Granularity

Sector Y-Granularity ~ tens cmSector Y-Granularity ~ tens cm

4 layers – 2 at 5º rotation4 layers – 2 at 5º rotation

PatVeloTT:PatVeloTT:

VELO

55

Page 6: LHCb PatVeloTT Performance

First ResultsFirst Results Two luminosities have been investigated, the initial LHCb target of Two luminosities have been investigated, the initial LHCb target of

2x102x103232 cm cm-2-2ss-1-1 and the upgrade target of and the upgrade target of 2x102x103333 cm cm-2-2ss-1-1. Initial studies . Initial studies have looked at the following questions:have looked at the following questions:

How many candidate VeloTT tracks are there?How many candidate VeloTT tracks are there? How often do we get it right?How often do we get it right? What is the (1/Pt) resolution of the ‘best’ track?What is the (1/Pt) resolution of the ‘best’ track? How do the ΧHow do the Χ2 2 of the ‘best’ tracks compare to the rejected ones?of the ‘best’ tracks compare to the rejected ones?

The following graphs are made from simulated Bs->The following graphs are made from simulated Bs->ϕϕϕϕ events using events using the Minimal Upgrade Layout. Two luminosity data sets:the Minimal Upgrade Layout. Two luminosity data sets:

2x102x1032 32 cmcm-2-2ss-1-1 – 3,430 events: 140,258 VELO tracks – 3,430 events: 140,258 VELO tracks 2x102x1033 33 cmcm-2-2ss-1-1 – 1,808 events: 178,620 VELO tracks – 1,808 events: 178,620 VELO tracks

66

Page 7: LHCb PatVeloTT Performance

Track CandidatesTrack Candidates

2x1032 cm-2s-1 2x1033 cm-2s-1

77

In PatVeloTT, each VELO track will have a number of candidate In PatVeloTT, each VELO track will have a number of candidate tracks associated with it.tracks associated with it.

Each candidate track will be reconstructed from a set of clusters in Each candidate track will be reconstructed from a set of clusters in the TT. the TT.

Page 8: LHCb PatVeloTT Performance

Track Candidates - SectorsTrack Candidates - Sectors

(the error bars are smaller than the coloured markers)(the error bars are smaller than the coloured markers)

Here is the mean number of candidates from the tracks which went through Here is the mean number of candidates from the tracks which went through sectors of a particular length. The average number of candidate tracks is sectors of a particular length. The average number of candidate tracks is considerably larger nearer the beam pipe (where the sectors are smaller).considerably larger nearer the beam pipe (where the sectors are smaller).

88

Page 9: LHCb PatVeloTT Performance

Track Candidates - PtTrack Candidates - Pt Here is the average number of candidate tracks for a range of transverse Here is the average number of candidate tracks for a range of transverse

momentum (of the MC particle associated with the VELO track. The points momentum (of the MC particle associated with the VELO track. The points represent the centre of the Pt bins (i.e. 0-0.5 GeV is at 0.25GeV, etc).represent the centre of the Pt bins (i.e. 0-0.5 GeV is at 0.25GeV, etc).

99

Page 10: LHCb PatVeloTT Performance

Number of ClustersNumber of Clusters

2x1032 cm-2s-1 2x1033 cm-2s-1

1010

In PatVeloTT, each VeloTT track will have a number of TT clusters In PatVeloTT, each VeloTT track will have a number of TT clusters associated with it.associated with it.

Both luminosities peak at 4 clusters. The higher luminosity Both luminosities peak at 4 clusters. The higher luminosity distribution has a larger fraction of tracks with 5-7 hits.distribution has a larger fraction of tracks with 5-7 hits.

Page 11: LHCb PatVeloTT Performance

Correct Clusters? - PtCorrect Clusters? - Pt It is of interest how often we correctly pick the right clusters when It is of interest how often we correctly pick the right clusters when

reconstructing a track. Using MC information we can see how many of the reconstructing a track. Using MC information we can see how many of the clusters associated to the track are correct. The below plot shows how the clusters associated to the track are correct. The below plot shows how the match percentage varies with Pt.match percentage varies with Pt.

1111

Page 12: LHCb PatVeloTT Performance

Correct Clusters? - PtCorrect Clusters? - Pt What happens when we make a cut on the reconstructed Pt?What happens when we make a cut on the reconstructed Pt? Reconstructed Pt > 1 GeV:Reconstructed Pt > 1 GeV:

1212

Page 13: LHCb PatVeloTT Performance

Correct Clusters? - PtCorrect Clusters? - Pt Also for reconstructed Pt > 1.5 Gev:Also for reconstructed Pt > 1.5 Gev:

1313

Page 14: LHCb PatVeloTT Performance

Correct Clusters? - SectorsCorrect Clusters? - Sectors We can also look at how the match percentage varies with sector length.We can also look at how the match percentage varies with sector length.

1414

Page 15: LHCb PatVeloTT Performance

Correct Clusters? - SectorsCorrect Clusters? - Sectors Reconstructed Pt > 1 GeV:Reconstructed Pt > 1 GeV:

1515

Page 16: LHCb PatVeloTT Performance

Correct Clusters? - SectorsCorrect Clusters? - Sectors Reconstructed Pt > 1.5 GeVReconstructed Pt > 1.5 GeV

1616

Page 17: LHCb PatVeloTT Performance

Success Rate – 2x10Success Rate – 2x103232

A successful match is defined as when either:A successful match is defined as when either:• At least 70% of the TT cluster hits are matched to MC truth.At least 70% of the TT cluster hits are matched to MC truth.• All but one of the TT hits is matched to MC truth (i.e. 2/3).All but one of the TT hits is matched to MC truth (i.e. 2/3).

1717

Page 18: LHCb PatVeloTT Performance

Success Rate – 2x10Success Rate – 2x103333

1818

The fraction of unsuccessful cluster matches is much higher at 2x10The fraction of unsuccessful cluster matches is much higher at 2x103333. .

Page 19: LHCb PatVeloTT Performance

Success RateSuccess Rate The mean success rates for the various cuts on the reconstructed Pt:The mean success rates for the various cuts on the reconstructed Pt:

1919

Page 20: LHCb PatVeloTT Performance

1/Pt Resolution1/Pt Resolution Resolution of (1/Pt) = [(1/Pt)Resolution of (1/Pt) = [(1/Pt)MCMC – (1/Pt) – (1/Pt)measuredmeasured] / (1/Pt)] / (1/Pt)MCMC

This was plotted for multiple bins of Pt ranging from 0-4GeV (see below This was plotted for multiple bins of Pt ranging from 0-4GeV (see below examples).examples).

2020

Page 21: LHCb PatVeloTT Performance

1/Pt Resolution1/Pt Resolution The widths of the Gaussians (from each of the Pt bins) was measured and The widths of the Gaussians (from each of the Pt bins) was measured and

is shown below as a function of Pt.is shown below as a function of Pt. Statistics were low for the higher Pt bins, hence the large errors. Statistics were low for the higher Pt bins, hence the large errors.

2121

Page 22: LHCb PatVeloTT Performance

ΧΧ22 of ‘Best’ and Rejected Tracks of ‘Best’ and Rejected Tracks

2x1032 cm-2s-1 2x1033 cm-2s-1

Below are plots of the pseudo χBelow are plots of the pseudo χ2 2 for the ‘best’ track against the rejected for the ‘best’ track against the rejected tracks. This pseudo χtracks. This pseudo χ2 2 should be treated as a quality parameter rather than should be treated as a quality parameter rather than a regular statistical χa regular statistical χ22. .

There is a cut on the pseudo χThere is a cut on the pseudo χ2 2 of the best track at 10of the best track at 1044. The majority of the . The majority of the points on these graphs are located very close to zero. Over the next few points on these graphs are located very close to zero. Over the next few slides we will look at particular regions of these graphs. slides we will look at particular regions of these graphs.

2222

Page 23: LHCb PatVeloTT Performance

ΧΧ22 of ‘Best’ and Rejected Tracks of ‘Best’ and Rejected Tracks

2323

Page 24: LHCb PatVeloTT Performance

ΧΧ22 of ‘Best’ and Rejected Tracks of ‘Best’ and Rejected Tracks

2x1032 cm-2s-1 2x1033 cm-2s-1

Conditions:Conditions: Discarded Tracks pseudo χDiscarded Tracks pseudo χ22 < 10 < 1044.. Both best and discarded pseudo χBoth best and discarded pseudo χ22 > 400. > 400.

2424

Page 25: LHCb PatVeloTT Performance

ΧΧ22 of ‘Best’ and Rejected Tracks of ‘Best’ and Rejected Tracks

2x1032 cm-2s-1 2x1033 cm-2s-1

Conditions:Conditions: Both best and discarded pseudo χBoth best and discarded pseudo χ22 < 50. < 50.

2525

Page 26: LHCb PatVeloTT Performance

ΧΧ22 Distributions Distributions Mean pseudo χMean pseudo χ2 2 for all tracks and for tracks with a successful MC cluster for all tracks and for tracks with a successful MC cluster

match: match:

2626

Page 27: LHCb PatVeloTT Performance

ΧΧ22 Distributions Distributions The pseudo χThe pseudo χ2 2 for tracks with an unsuccessful MC cluster match is for tracks with an unsuccessful MC cluster match is

considerable larger: considerable larger:

2727

Page 28: LHCb PatVeloTT Performance

Extras: Hi SteveExtras: Hi Steve08/02/10:08/02/10: In PatVeloTT the candidate tracks are cut first on how many layers of the In PatVeloTT the candidate tracks are cut first on how many layers of the

TT have cluster hits in them, and then on pseudo χTT have cluster hits in them, and then on pseudo χ22. I thought it might be . I thought it might be interesting to see how the pseudo χinteresting to see how the pseudo χ2 2 compare when:compare when:

• # of layers with hits is equal for the ‘best’ and discarded candidates.# of layers with hits is equal for the ‘best’ and discarded candidates.• # of layers with hits is larger for the ‘best’ track.# of layers with hits is larger for the ‘best’ track.

This is shown on the following slides…This is shown on the following slides… Also, I seem to have accumulated millions of graphs… if there’s anything Also, I seem to have accumulated millions of graphs… if there’s anything

else that you can think of which you’d like to see then there’s a good else that you can think of which you’d like to see then there’s a good chance I’ve already made it. Otherwise I’m sure I can make it for you, let chance I’ve already made it. Otherwise I’m sure I can make it for you, let me know.me know.

12/02/10:12/02/10: I’ve tried making the 1/Pt resolution graphs for different cluster MC match I’ve tried making the 1/Pt resolution graphs for different cluster MC match

fractions (as you suggested), but the statistics aren’t there for an analysis. fractions (as you suggested), but the statistics aren’t there for an analysis. I’ve included a couple of graphs (slides 34-35) for 100% cluster match and I’ve included a couple of graphs (slides 34-35) for 100% cluster match and for 75%-87.5% match. These were the only two for which there were for 75%-87.5% match. These were the only two for which there were enough tracks to make sensible graphs (there was also enough 0% enough tracks to make sensible graphs (there was also enough 0% matches but these were nonsensical).matches but these were nonsensical). 2828

Page 29: LHCb PatVeloTT Performance

ΧΧ22 of ‘Best’ and Rejected Tracks of ‘Best’ and Rejected Tracks

2x1032 cm-2s-1 2x1033 cm-2s-1

Conditions:Conditions: Number of layers with cluster hits is equal.Number of layers with cluster hits is equal. Discarded Tracks pseudo χDiscarded Tracks pseudo χ22 < 10 < 1044..

2929

Page 30: LHCb PatVeloTT Performance

ΧΧ22 of ‘Best’ and Rejected Tracks of ‘Best’ and Rejected Tracks

Conditions:Conditions: Number of layers with cluster hits is greater for the ‘best’ tracks.Number of layers with cluster hits is greater for the ‘best’ tracks. Discarded Tracks pseudo χDiscarded Tracks pseudo χ22 < 10 < 1044..

2x1032 cm-2s-1 2x1033 cm-2s-1

3030

Page 31: LHCb PatVeloTT Performance

1D χ1D χ22 distribution examples distribution examples(see slide 28)(see slide 28)

3131

Page 32: LHCb PatVeloTT Performance

1/Pt Resolution1/Pt ResolutionCluster Match % = 100%Cluster Match % = 100%

3232

Page 33: LHCb PatVeloTT Performance

1/Pt Resolution1/Pt Resolution75% < Cluster Match % < 87.5%75% < Cluster Match % < 87.5%

3333

Page 34: LHCb PatVeloTT Performance

1/Pt Resolution1/Pt Resolution 1/Pt resolution plots were also made by ‘binning’ the data in equal chunks 1/Pt resolution plots were also made by ‘binning’ the data in equal chunks

of 1/Pt. This covers the same range of Pt as the previous plot (0 - 4 GeV).of 1/Pt. This covers the same range of Pt as the previous plot (0 - 4 GeV).

(please note that the furthest two points on the right actually contain data from (please note that the furthest two points on the right actually contain data from 0 – 100 MeV. But how do you plot a point midway between 0.01 and ∞ ?!)0 – 100 MeV. But how do you plot a point midway between 0.01 and ∞ ?!) 3434