Jon Gans – STAR Physics Analysis Meeting – Oct 22, 2002

Vertex Finding and Pileup in the p+p data

Jon GansYale University

STAR Collaboration

1) Detector Setup

2) Vertexing Bias – p+p and Au+Au

3) Influence of Bias on the Data

4) New Methodology to Remove Bias

(Works for me, but probably not for you)

5) Remarks on pileup rejection

6) Contamination

7) Outlook

Jon Gans – STAR Physics Analysis Meeting – Oct 22, 2002

420 CM

Time Projection Chamber:

2.1 meter drift over 40 μs

|η| < 1.5 , Full Azimuth

Central Trigger Barrel (Scintillator ):

Trigger For Au+Au (not for p+p)

6° x 1 m – 240 Cover Entire Barrel

Sample 11 (Δt 110ns) times per trigger

Beam Beam Counters:

+/- 3meters , 3.5 < | η | < 5.5

Use for p+p Trigger (Coincidence)

70cm Vertex Resolution

Must Use TPC Tracks For Vertexing

Jon Gans – STAR Physics Analysis Meeting – Oct 22, 2002

60 Total Bunches in Ring ( I only show 5)55 Are Filled With Protons5 are empty213 ns between each bunch

Since the empty bunches are only in the detector for short period of time, upstream, beam gas and pileup events are sampled.

Vertex Efficiency with Embedding

RECO

VertexEfficiency

Empty BunchEvent

Jon Gans – STAR Physics Analysis Meeting – Oct 22, 2002

Raw Multiplicity

Vertex Finding Efficiency

Events with abs(ppLMVz – mcz) < 1cm

Total Number of EventsEfficiency =

Why don’t we use these numbers to

correct data?

I Haven’t Told you the Whole Truth

Loses More Low Multiplicity Eventsthan High Multiplicity Events

Vertex Finder Correct ~ 74%

Jon Gans – STAR Physics Analysis Meeting – Oct 22, 2002

Vertex Finding Resolution

Monte Carlo Z Vertex – ppLMV Z Vertex (cm)

Co

unt

s

What Do We Do With Misidentified Vertices

Jon Gans – STAR Physics Analysis Meeting – Oct 22, 2002

Vertex Finding Contamination

Events with abs(ppLMVz – mcz) > 1cm

Total Number of EventsContamination =

0

.1

.2

Co

nta

min

atio

n

Raw Multiplicity

Contamination ~ 11%

Raw Multiplicity

0

.2

.4Lost Vertices ~ 14%

Jon Gans – STAR Physics Analysis Meeting – Oct 22, 2002

Does Embedding and Real Data Agree?

Events with Reconstructed Vertex(no residual cut)

Total Number of EventsEfficiency =

DATA

EMBEDDING

Jon Gans – STAR Physics Analysis Meeting – Oct 22, 2002

Spectra Using Primary Vertex Analysis MethodS

TA

R

/ U

A1

Correction For η Acceptance:(STAR abs(η) < .5, UA1 abs(η) < 2.5

pt GeV/c

STAR / UA1RAWVertex Corrected

No Efficiency or Acceptance Corrections (which would move spectra up anyways!)

Jon Gans – STAR Physics Analysis Meeting – Oct 22, 2002

Consequences of Multiplicity Bias

Mean pt scales with multiplicityUA1 – Phys Lett B 366 (1996) 434

Momentum bias tied to vertex finding bias.

Vertex Finder looses low multiplicity events

=> Spectra is shifted to higher pt

0 25

MultiplicityM

ean

pt G

eV/c

0.2

0

.6

Jon Gans – STAR Physics Analysis Meeting – Oct 22, 2002

HIJING + TRS

Data

fitPoints < 15

What about Au+Au data?Thanks to P. Fachini

HIJING + TRS

Data

No fitpoints cut

Events with Reconstructed Vertex(no residual cut)

Total Number of EventsEfficiency =

Jon Gans – STAR Physics Analysis Meeting – Oct 22, 2002

Vertex Found

HIJING + TRS

Wrong Vertex

HIJING + TRS

No Vertex

HIJING + TRS

# good global tracks:– # fit points

25– flag 0

# good global tracks:– # fit points

25– flag 0

# good global tracks:– # fit points

25– flag 0

More on Au+Au HIJING StudiesThanks to P. Fachini

Shapes Similar to proton+proton Embedding

Jon Gans – STAR Physics Analysis Meeting – Oct 22, 2002

Determining Spectra Without Vertexing

x(z) = -0.304233 cm + -0.00065732 * zy(z) = 0.407226 cm + 0.00012396 * z

1) Project Track To Beam Line Check for 1cm DCA or less

Beamline position is well known in STAR.For events with a found Vertex: Parameterize X and Y position of Vertex versus Z

Z Vertex Position (cm)-200 0 200Y

Ver

tex

Pos

ition

(cm

)

-1.6

1.

2

STAR Preliminary

Jon Gans – STAR Physics Analysis Meeting – Oct 22, 2002

-2

2

Match Track To CTB To Reject Pileup

Lower Luminosity (Early in Run) Higher Luminosity (Late in Run)

Track Matches CTB Hit

Track Projects To CTBCTB Match Eff. =

Good Measure Of PileupLower Matching Efficiency => Higher Pileup

Azimuth

1

0.5

00 180 360

Z P

osi

tio

n

Azimuth0 180 360

STAR Preliminary STAR Preliminary

Note: 5-10% Matching Rate When CTBs Rotated by 90 degrees

Jon Gans – STAR Physics Analysis Meeting – Oct 22, 2002

-2

2

Same Studies with SVT Matched Tracks

Lower Luminosity (Early in Run) Higher Luminosity (Late in Run)

Lower Matching Efficiency => Higher Pileup

Z P

osi

tio

n

Azimuth0 180 360

STAR Preliminary STAR Preliminary

0 180 360Azimuth

Appears that SVT Matched Tracks Suffer Less Pileup!!Can we use the SVT for pilup rejection?

Very Preliminary Result, needs more attention.

1

.5

0

Note: 5-10% Matching Rate When CTBs Rotated by 90 degrees

Jon Gans – STAR Physics Analysis Meeting – Oct 22, 2002

ΔZ

Contamination

STAR Preliminary

Drawbacks to No Vertex Method

Z position of vertex unknownFrom Higher Multiplicity Events Get Z position

Due To Added Material outside +/- 30 cm(Inner Vertex Detector Electronics and Cooling)Acceptance and Efficiency Corrections Difficult

-100cm -30 cm 30cm 100 cm

STAR Preliminary

~ 8%

Jon Gans – STAR Physics Analysis Meeting – Oct 22, 2002

ST

AR

/

UA

1

Correction For η Acceptance:(STAR abs(η) < .5, UA1 abs(η) < 2.5

pt GeV/c

STAR / UA1

UNCORRECTED Spectra1/

nEve

nts

1/p t

d((

h+ +

h- )/

2)/d

ηdp t

BLACK – UA1 C. Albajar et al., Nucl. Phys. B335, 261 (1990)

RED – STAR Uncorrected Spectra

NEVER Trust A Log Plot

STAR Preliminary

Jon Gans – STAR Physics Analysis Meeting – Oct 22, 2002

Future and Outlook

Investigating methods to constrain Z Vertex Position.JG and P. Fachini investigating methods to accomplish this.

Efficiency and Acceptance Corrections currently being studied in pp

ITTF Vertex Finder

-60 -40 -20 0 20 40 60

Mean z Projection - mcVertz