The LiC Detector Toy (LDT) Tracking detector optimization with fast simulation VERTEX 2011, Rust M. Valentan, R. Frühwirth, M. Regler, M. Mitaroff

The LiC Detector Toy (LDT)Tracking detector optimization

with fast simulation

VERTEX 2011, Rust M. Valentan, R. Frühwirth, M. Regler, M. Mitaroff

LiC Detector Toy (LDT)

In a nutshell• The “LiC Detector Toy” (short “LDT”) allows

investigation of the track parameter resolution via Monte Carlo, for optimizing a detector setup

• Simulation of particle tracks through the track sensitive part of a collider detector with a solenoid magnetic field, including material effects

• Support of measurements by semiconductor pixel and strip detectors, and a TPC

• Track reconstruction by a Kalman filter, including tests of goodness of the fits

• Written in MatLab® (a language and IDE by MathWorks)

• Version 2.0 available for GNU Octave

June 22, 2011 2M. Valentan (HEPHY Vienna)


Motivation

• Compare track parameter resolutions of various detector setups, for both barrel and forward/backward regions

• Optimize size and position of the track sensitive devices, and of the detector material budgets

• A simple tool – easy to understand, handle and modify• Can easily be adapted to meet individual needs• Can be installed on a desktop or laptop PC• Quick results by “shorter than a coffee break”• Ideal tool for investigating the effect of local variations



• Single or double Si strip layers, pixel layers, TPC• Passive material possible• Homogeneous magnetic field (by a solenoid),

rotational symmetry w.r.t. the z-axis of the detector setup

• Start parameters for simulated tracks are user-defined:– Vertex position range,transverse momentum range, range of polar

angle θ, number of tracks from the vertex.– Latest custom version reads start parameter from file

• Output:– Track parameter resolutions– Impact parameter resolutions– Test quantities


General features


Graphic User Interface (GUI)

Choose and display detector geometry

View output

Simulation parameters

Save and loadresults



Detector model

• Human readable text file• Coaxial cylinder layers of

arbitrary length and position• Circular plane layers,

perpendicular to z-axis• Passive layers• 2D measurement– Barrel: RΦ and z– Forward: u and v

• Resolution in TPC may depend on z.

x

v

δ1

y

δ2



2D and 3D Detector Display



• Exact helix track model• Data corruption by measurement errors, multiple

scattering and detector inefficiencies only• Gaussian (e.g. for TPC) or uniformly distributed

measurement errors• Material budget assumed to be concentrated in “thin”

layers, no special treatment of electrons:– Multiple scattering: |p| conserved, correct path length traversed;– Thickness of scatterers given in radiation lengths;– Scattering angles Gaussian distributed (Rossi-Greisen-Highland’s

formula);– No energy loss (by ionisation or bremsstrahlung) simulated.


Track simulation


Complex intermediate region

• Arbitrary sequence of cylindric and plane detector layers



• No pattern recognition• Single track fit by an exact Kalman filter• Linear track model: Expansion point is a “reference track”• Fitted parameters defined at the inside of the innermost

layer:– { Φ, z, θ, β = φ-Φ, κ = ±1/rH } with sign(κ) = sign(dφ/ds),

and corresponding 5x5 covariance matrix

• Goodness of the fit monitored by pull quantities and χ²


Track reconstruction


Automatic loops

Several pairs of barrel and forward

detector setups

Several different start values for e.g. the transverse momentum

Compare detector setupsas function of – momentum

– polar angle



Automatic loopsrms of Δpt/pt

2 σ of projected impact

These plots are directoutputs of the program!

• blue:• green:• red:

ILDILD without SITILD without SET



Fine grained simulation mode

• Normal: – Simulate many tracks

– track parameter resolution from distribution

• Quick & dirty:– simulate one track

– track parameter resolutions from covariance matrix

– enables fine scanning of detector setup

13M. Valentan (HEPHY Vienna)June 22, 2011

quite clean


LDT 2.0 for GNU Octave

• Octave is free• Command line based

(no GUI)• All features except for

3D detector display• Factor 10 slower



LDT and other software

• Successful validation against– Fullsim: Mokka/Marlin (ILD)– Fastsim: SGV (ILD)

• External list of start parameters– Instead of simulating internally– Connection with event generators

• Output as csv file for further analysis– Successfully tested with RAVE vertex

fitting toolkit



Wishlist

• Conical detector surfaces– > e.g. slanted part of Belle II SVD

• Segmentation in RΦ– > Simulation of ladders

• Simulation of e-/e+energy loss by bremsstrahlung;

• Continuous multiple scattering (e.g. in front of muon detectors).





Updated TPC resolution

• TPC resolution as commented by Ron Settles:

• Includes:– Constant point resolution of endplates– Term dependent on angle β between track and normal

vector of detector layer– Term dependent on polar angle θ and charge

spreading


Documents

The LiC Detector Toy (LDT) Tracking detector optimization with fast simulation VERTEX 2011, Rust M. Valentan, R. Frühwirth, M. Regler, M. Mitaroff