1

Realistic top Quark Reconstruction for Vertex

Detector Optimisation

Talini Pinto Jayawardena (RAL)Kristian Harder (RAL)

LCFI Collaboration Meeting23/09/08

2

Overview• Purpose of the analysis• Choice of detector and simulation

software• Algorithms used for the

reconstruction• Reconstruction of the top mass –

for 6 jet channel• Vertex detector studies• Conclusion and discussion

3

TTBar Analysis – Purpose and Aim

• Reconstruction of heavy particle decay to multi jet final state events crucial for ILC physics

• Event reconstruction optimisation for ILC energies using only realistic algorithms

• Performance studies of detector options considered for the ILC– Enables the optimisation of vertex detector,

calorimeter currently undergoing R&D

4

Top Reconstruction – 6 Jet Channel

• Three main top pair event topologies:•tt bbqqqq – 6 jets•tt bbqqlν – 4 jets•tt bblνlν – 2 jets

• 44% of the sample decaying to 6 jets give good statistics for reconstruction

• Large fraction of the energies in the final state visible

–Small loss of energy in neutrinos

Leptonic decays result in missing momentum

5

Simulation of Detector•Use of the LDCPrime detector concept

•GEANT4 simulation software MOKKA used to describe the detector model

•A good description of geometry, although with limited detail on support structure, cabling etc.

Ladder structure in the vertex detector

6

Algorithms for Reconstruction

• Uses the Marlin framework for the reconstruction of events from raw (simulated) data of the detector

• Applies PandoraPFA for particle flow –Clustering and particle flow done in a single stage

• Track reconstruction includes full LDC tracking

• Secondary (+tertiary) vertexing and heavy flavour tagging performed by LCFIVertex

Tracking+vertexing and calorimeter readout reconstructed without the use of MC information

7

Top Reconstruction – Selecting the events

•Durham jet finder used to force 6 jets for all events – Ensures that the actual 6 jet events are reconstructed well

•All hadronic jet channel selected by – Applying a veto on events with a large missing momentum– Requiring 2 b tags for each event (neural net jet likelihood)

Momentum Cut of 25GeV

8

Top Reconstruction – Jet Combinations

• Need to separate the b jets from those of W

– Jets having the two highest probability selected as the b jets

• Remaining 4 jets need to be separated to their correct W decay jet pairs

– Combinations where the 2 masses are the least different are chosen as the best jet combinations

• Corresponding W-b jet combinations used to obtain the top mass

– Jet combinations where the 2 masses are most similar regarded as the best combinations

9

Top Reconstruction – The top Mass

•Clear top mass peak observed

•Broader peak seen below the signal at higher energies

– Jet finder fails to group the correct particles belonging to a given jet

– Incorrect reconstructed mass due to wrong jet combinations

Mean of main mass distribution: 167.7GeVWidth: 8.2 GeV

NB: Purity/Efficiency values obtained for sample with background from the other top pair event topologies only

90.4% Purity42.57% Efficiency

10

The Vertex Detector•Reference digitiser uses a simple gaussian smearing of the hits reconstructed•Two realistic vertex detector options considered:

– CCD based sensor technology

– Active pixel technology (DEPFET)

•The designs described through digitisers during particle reconstruction

– Considers external effects on the charge during collection and transfer between pixels

•Performance analysed with the aid of top mass resolution

CCD

DEPFET

S. Uebelacker

11

Reconstruction in the Vertex Detector – CCD Digitiser

• Create ionisation points along a track with a given distance

• Charge distribution calculated for two different areas:– Depleted zone – Based on electron mobility and Lorentz

shift– Undepleted zone – Diffusion of charge subjected to

magnetic effects• Signal and electronic noise simulated with Poisson

and Gaussian distributions, respectively• Reconstruction achieved by using an algorithm to

obtain the centre of gravity • Each hit is reconstructed separately and hit position

corrected for magnetic effects• Correction applied for the shift in the interaction

plane due to the insensitive bulk

12

• Ionisation points created as in the CCD digitiser• Diffusion of charge from each ionisation point

obtained through a Gaussian distribution– Width of the distribution proportional to the root of

distance between collection plane and ionisation point

• Shift due to magnetic field effect taken into account• Signal and electronic noise simulated as in the CCD

Digitiser• Reconstruction obtained by finding the centre of

gravity where the columns/rows are weighted by the mean amount of deposited electrons

Reconstruction in the Vertex Detector – DEPFET Digitiser

13

Performance Analysis

Mean of narrow Gaussian: 168.3GeV Width: 7.96GeV Efficiency: 47% Pixel Size: 20x20microns

Mean of narrow Gaussian: 167.7GeV Width: 8.06GeV Efficiency: 46.7% Pixel Size: 25x25micronsNB: Reference Gaussian hit resolution applied:

2.8x2.8microns

14

Summary• Good top mass reconstruction with

simple selection criteria enables performance of different detectors to be analysed

• Performance of the two realistic digitisers has a mass resolution very similar to that of the reference (dummy) digitiser, as expected

• LCNote to be written up summarizing the performance study