D 0 reconstruction: 15 AGeV – 25 AGeV – 35 AGeV

D0 reconstruction:15 AGeV – 25 AGeV – 35 AGeV

M.Deveaux, C.Dritsa, F.Rami

IPHC Strasbourg / GSI Darmstadt

•Outline•Motivation•Simulation Tools•Results for 25AGeV•Results for 15AGeV•Results for 35AGeV•Intermediate Conclusions•Proton-Proton collisions: first attempt•Summary and Conclusions

Motivation

Feasibility study of D0 reconstruction for beam energy of 25AGeV is ongoing:

• Simulations with relatively high statistics are needed to improve precision of results.

• What are the S/B and the tagging efficiency results for this beam energy and for a specific geometry?

• How are the above results affected for different beam energy (35AGeV, 15 AGeV) but same geometry?

• Can we measure open charm at 15AGeV?

Questions to address for studying D0 reconstruction at 15 and 35 AGeV:

How to generate D0 with correct parameters (γ, T and σY) What is the signal acceptance for those energies? How is the pt-Y distribution affected once cuts are applied?

Select Candidate TracksSelect Candidate Tracks

Select Candidate PairsSelect Candidate Pairs

Apply Final CutsApply Final Cuts

Calculate S/B, signal efficiency…Calculate S/B, signal efficiency…

Apply “soft” pre-selection criteria

D0 π+K-

Tools of the simulation

Optimised for each geometry and energy using specific algorithm

Calculation of S/B: how is it done?

1. Generate signal and background*

2. Apply the final cuts.

3. Fit the background distribution with an exponential function.

4. Fit the signal distribution with a Gaussian function.

5. The background fit function is normalised with respect to detector’s lifetime (~1011 centr coll).

6. The signal fit function is normalised with respect to detector’s lifetime (~1011 centr coll) taking into account the cross section.

7. Integrate the functions in a region of 2σ around the mean value of signal.

*Part of the background is generated with the Super Event method: Mixing all particles of all events together.

Optimisation of selection criteria (cuts)

The procedure

The cut optimisation procedure is based on an iterative algorithm searching for a maximum on a multidimensional surface (developed by M.Deveaux).

Advantages:• It takes into account correlations between different cuts.• It is fast (not more than few hours)

Disadvantages:• May converge at local maxima.• Most cuts are implemented but not all yet.

(Ex. impact parameter not yet implemented)

The most important cuts

• Rejection of particles intersecting the primary vertex (χ2 primary)• Reject vertices with low fit quality (χ2 secondary)• Select vertices within a distance from the initial collision point

Au-Au @ 25 AGeVGeometry used:

3 MAPS 200 μm, 5μm spatial resolution ( 10-15-20cm)

1 HYBRID 750 μm, 50μm pixel size ( 30cm)

5 STRIPS 400 μm ( 50, 62.5, 75, 87.5, 100cm)

Statistics generated:

225 Millions equivalent central events using Super Event method

The two last stations were included in the hits but not in the tracks

mπK (MeV/c2)

Total thickness : 3.35mm

σ = 15.4 ± 0.4 MeV/c2

ZMC-ZRECO (cm)

σ = 84.0 ± 2.8 (μm)

Au-Au @ 25 AGeV; Input: Bg=225 Millions, Signal=9000 D0

0.9

S/B

1.2*10 -42.6%

D0 multiplicityEff

Number of D0 expected after one run (1,2*1011centr coll)

within the inv. mass range of mean +/- 2σ:

13000

En

trie

s 5

Me

V

mπK (GeV/c2)

mπK (GeV/c2)

En

trie

s /

10

0 M

eV

mπK (GeV/c2)

Geometrical Acceptance for 25AGeV, 9000 D0

4πGeometrical Acceptance in the full

rapidity range: 34%

Pt (

GeV

/c)

Pt (

GeV

/c)

Geometrical Acceptance + Cuts

Pt (

GeV

/c)

Y

YY

In the 2<Y<3 rapidity range:

Reconstruction Efficiency: ~ 5%

>> The rapidity region of interest is populated after applying final cuts

Au-Au @ 15 AGeV

Statistics generated:

249 Millions equivalent central events using Event Mixing method

• Same Geometry

σ = 89.8 ± 3.3 μm σ = 15.8 ± 0.5 MeV

mπK (MeV/c2) ZMC-ZRECO (cm)

Au-Au @ 15 AGeV: Signal Generation-Multiplicity-Normalisation

Generate Signal Pairs : The choice for the parametres follows the choice of parametres for generation of D0 @ 25AGeV:

Because of lack of information for determining a Temperaturethe value of T is not changed.

Finally, the normalisation is done with respect to the

detector’s lifetime which was estimated to be

1.4·1011 centr colisions(For 25AGeV the lifetime is

1.2∙1011)

pBeam = 25 AGeVpBeam = 15 AGeV

Gaussian rapidity width = 1

T = 300MeV (Inverse Slope Parameter)

25AGeV15 AGeV

The multiplicity was assumed to be 10-5

1000

Numb D0 exp

2.4

Eff %

10-50.2

D0 multiplicity

S/B

15 AGeV, Input: Bg=249 Millions, Signal=8000

Background and signal distributions after cuts – before normalisation.The fits are shown.

En

trie

s /

5 M

eV

En

trie

s /

50

Me

V

mπK (GeV/c2)

mπK (GeV/c2) mπK (GeV/c2)

4π Geometrical Acceptance: 27%

Pt (

GeV

/c)

Y

Efficiency: Geometrical acceptance for 15AGeV, 8000 D0P

t (G

eV/c

)

Y

Pt (

GeV

/c)

Y

In the 2<Y<3 rapidity range:

Reconstructed/Generated : ~ 5.6%

>> The rapidity region of interest is populated after applying final cuts

Au-Au @ 35 AGeV

Statistics generated:

121 Millions equivalent central events using Event Mixing method

• Same Geometry

ZMC-ZRECO (cm)mπK (MeV/c2)

σInvMass = 14.3 ± 0.4 MeV σ = 86.2 ± 3.3 μm

Au-Au @ 35 AGeV: Signal Generation-Multiplicity-Normalisation

Generate Signal Pairs : The choice for the parametres follows the choice of parametres for generation of D0 @ 25AGeV:

Because of lack of information for determining a Temperaturethe value of T is not changed.

Finally, the normalisation is done with respect to the

detector’s lifetime which was estimated to be

1011 centr colisions(For 25AGeV the lifetime is

1.2∙1011)

pBeam = 25 AGeVpBeam = 35 AGeV

Gaussian rapidity width = 1

T = 300MeV (Inverse Slope Parameter)

25AGeV35 AGeV

The multiplicity was assumed to be 10-3

En

trie

s /

5 M

eV

2 different selection criteria

113000

77000

Numb D0 exp

3.0

2.1

Efficiency %

10 -32.0

10 -38

D0 multiplicity

S/B

35 AGeV, Input: Bg=121 Millions, Signal=7000E

ntr

ies

/ 5

0 M

eV

S/B=8

Det. Eff = 2.1%

mπK

(GeV/c2) mπK

(GeV/c2)

mπK

(GeV/c2)

4π Geometrical Acceptance: 37%

Pt (

GeV

/c)

Y

Efficiency: Geometrical acceptance for 35AGeV, 7000 D0P

t (G

eV/c

)

Y

Pt (

GeV

/c)

Y

In the 2<Y<3 rapidity range:

Reconstruction Efficiency: 4.5%

>> The rapidity region of interest is populated after applying final cuts

Intermediate Summary & Conclusion

Next steps and open questions:- Explore other setups that allow D0 measurements with better results.- What is the physics we can do with the above results?- Make an error estimation on S/B- Update cut finding procedure (expect improved results)- How to produce signal pairs with more realistic parameters?

A comparison study between 25 , 15 and 35 AGeV was done:

• The IM resolution and secondary vertex resolution remain almost unchanged.

• The over-all reconstruction efficiency was not significantly different: 2%

• The S/B as much as the number of reconstructed D0 scale (roughly) with the multiplicity.

• S/B15 = 0.2 ; ~ 1000 D0

• S/B25 = 0.9 ; ~ 13.000 D0

• S/B35 = 8; ~ 77.000 D0

Preliminary results of proton-proton collisions

Motivation :

> Nucleon-nucleon reaction data provide a reference for the interpretation of nucleus-nucleus collisions.

> The measurement of open charm in proton-proton collisions is itself interesting as there are no data available at threshold energies.

Outline:•Motivation•Event generation•Input of the simulation•First preliminary results

Preliminary results of proton-proton collisions:PYTHIA vs UrQMD @ 25AGeV

Models already tried for event generation:

> PYTHIA> UrQMD

Both models were checked in terms of charged particle multiplicity and only UrQMD in terms of average transverse momentum for charged particles.

PYTHIA is not adapted for such low energies;

Preliminary results of proton-proton collisions:PYTHIA @ 25AGeV

Models for event generation:>PYTHIA @ 25AGeV

But UrQMD gives rather satisfactory results as they are closer to experimental data...

2

3.2

4

<multiplicity>/event

PYTHIA

0.1K+ and K-

1.5protons

3Pi+ and Pi-

<multiplicity>/event

Experimental data*Particle

*Rossi et al. , 1975, Nucl Physics B, page:267

PYTHIA gives a factor of 2 more protons and a factor of 20 more kaons

Preliminary results of proton-proton collisions: UrQMD @ 25 AGeV

-

480

409

317

350

UrQMD:

<pt> (MeV/c)

-

401

424

310

322

Experimental

data*

<pt> (MeV/c)

0.040.02K-

1.10.7Pi-

1.71.2Pi+

1.51.5protons

0.090.06K+

Experimental

data*

<multipl>/evt

UrQMD Model:

<multipl>/evtParticle

* Reference: Rossi et al. , 1975, Nucl Physics B, page:267

For 100.000 evts:

It seems that UrQMD reproduces better than PYTHIA the experimental data.

Preliminary results of proton-proton collisions:Input of the simulation

•CBMROOT FEB07

•STS geometry •2 MAPS ( 150 μm; 10,20cm)•6 STRIPS ( 400 μm; 30, 40, 50, 60, 75, 80, 100cm)

•NO signal

•NO TARGET material used for a first approach

•100.000 collisions

Preliminary results of proton-proton collisions:What is the acceptance?

4

7

10

15

22

37

%of evts

with N tracks in

acceptance

4308

6729

9934

15165

22495

37447

Num of evts

with N tracks in

acceptance

2

1

0

5

4

3

Number of tracks in

acceptance

0.005

0.03

0.1

0.4

1

2

% of evts

with N tracks in

acceptance

5

33

104

419

1047

2314

Num of evts

with N tracks in

acceptance

11

10

9

8

7

6

Nb of tracks in

acceptance

Summarizing:•75% of events have from 0 to 2 tracks in acceptance Primary vertex reconstruction either impossible or very difficult!•20% of events have from 3 to 5 tracks in acceptance•The rest 5% have more than 6 tracks inside acceptance

Preliminary results of proton-proton collisions:What is the primary vertex residual?

Only 4 or 5 tracks in acceptance; (10% events) Width of the distribution of the order of 80 um

ZRECO -ZMC

For Primary Vertex

Preliminary results of proton-proton collisions:Summary - Open questions

• The particle multiplicity for proton-proton is very low; for 75% of the events it is almost impossible to reconstruct the collision point.

• For 10% of the events (4-5 tracks in acceptance) the width of the distribution primary vertex residual is of the order of 80um

• Study other models for event generation (DPMJET, others?)

• More realistic simulation: Implement a target material•The target geometry from HADES is “waiting” to be implemented.• Is there a better candidate?

• Is there a modification in the tracking algorithm for primary vertex finding needed?

• Explore other setups?

• Study other systems: ex: p+C