Ruprecht Karls Universität Heidelberg

ATLAS Projects

The PI Heidelberg ATLAS group covers a variety of topics focusing on boosted top quark reconstruction, searches and measurements involving top quarks, the Fast Tracker (FTK) upgrade of the ATLAS detector and ATLAS upgrade projects for the high luminosity Large Hadron Collider.

Boosted Top Quarks

Reconstruction with the HEPTopTagger

Searches for New Physics

Top-quark-pair resonance search

Buckets of Top and Higgs

Single Vector Like Quark search

Work in Progress and Topics of Interest

FTK - Fast Track Trigger

ATLAS Upgrade Projects for High Luminosity LHC

Track Triggers

Technologies:

Monolithic Pixel Sensors for Inner Tracker (ITK)



Boosted Top Quarks


The top quark is the heaviest known particle ($m_{top} \simeq 173$ GeV). It is even heavier than the newly discovered Higgs boson ($m_{Higgs} \simeq 125$ GeV). Because of its high mass, the top quark plays an important role in both the Standard Model (in electroweak symmetry breaking, due to its large coupling to the Higgs boson) and in Physics beyond the Standard Model scenarios. At the LHC, due to the large center-of-mass energy of the proton-proton collisions, top quark pairs ($t\bar{t}$) are produced in large amounts. A significant fraction of the top quarks have momenta that are large compared to the top mass and are therefore called "boosted" top quarks.

Reconstruction with the HEPTopTagger

We use the HEPTopTagger (HTT) to reconstruct such boosted top quarks ($P_T>200$ GeV) in the fully hadronic final state, i.e.:
test

Top quark decaying to hadrons. $t\rightarrow b + W(\rightarrow q\bar{q}') = b+q+\bar{q}'$, with $q/q'=u,c,d,s,b$.

test

Sketch of a top quark decaying to hadrons, with the top quark at rest (left) or boosted (right).


The HTT uses large jets as inputs, assuming they contain all decay products of the top quark. The substructure of the large jets is then tested for compatibility with a hadronic top decay. We have tested the performance of the HTT with ATLAS data in the lepton+jets final state. Assuming one top quarks decays via $t\rightarrow b + W (\rightarrow \mu\nu)$ we reconstruct events with a muon $\mu$, missing transverse energy (to account for the neutrino) and apply the HTT to the hadronically decaying top quark:
test
Top quark candidate mass, reconstructed with the HTT in lepton+jets events. JHEP 06 (2016) 093

The reconstructed top mass (fully hadronic!) and various control distributions are well behaved and agree with the Monte Carlo simulation within uncertainties. Additionally the stability of the HTT with respect to additional proton-proton interactions, referred to as pile-up, has been tested and found to be excellent, by looking at the stability of the reconstructed mass versus the average number of interactions $<\mu>$:
test
HTT top quark candidate mass versus average number of interactions $<\mu>$ for events reconstructed with the HTT in the lepton+jets final state. ATLAS-CONF-2013-084

The PI ATLAS group has been leading in using the HTT in ATLAS, e.g. by
  • establishing the HTT algorithm in the ATLAS Collaboration, by showing its usability in ATLAS data and simulation,
  • providing calibrations and uncertainties for the HTT,
  • using the HTT in a published ATLAS search.

Searches for New Physics

Many theories beyond the Standard Model predict new heavy resonance that can decay to top quarks, e.g. $Z'$ bosons. A search for resonances decaying into top-quark pairs using fully hadronic decays has been performed at the PI (JHEP 1301 (2013) 116). By using the HTT to reconstruct two top quarks the di-top quark mass can be measured:
test
Distribution of the $t\bar{t}$ invariant mass $m_{t\bar{t}}$. The data, the Standard Model $t\bar{t}$ background prediction, the QCD multijet background prediction and a hypothetical $Z'$ signal with $m_{Z'} = 1$ TeV are shown. Data points show statistical uncertainties only. JHEP 1301 (2013) 116

The data are in agreement with the Standard Model prediction and limits on the mass and production cross-section of the hypothetical heavy particles have been set.

Top-quark-pair resonance search

The top-quark is the heaviest particle in the Standard Model, as a result of its large coupling to the Higgs boson. Due to such a fundamental role of the top-quark, many theories that could address existing questions within the Standard Model rely on new particles that also couple with the top quark. In many of such models, such as models for quantum gravitation, one would expect a new undiscovered particle to decay into a top-quark pair. Our group searches for a particle decaying into top-quark pair, by comparing the spectrum of the mass of the top-quark pair to a simulation, of what one would expect to see if a new particle exists.
test
Reconstructed di-top quark mass as predicted in simulation and measured in data. ATLAS-CONF-2016-014

Buckets of Top and Higgs

In addition to the HTT (which of course can also be employed in measurements of Standard Model processes like the production of a Higgs boson in association with a pair of top quarks ("$ttH$")) a second method to reconstruct pairs of hadronically decaying top quarks is used by the ATLAS PI group. The "Buckets of Top" algorithm (JHEP 1308 (2013) 086) aims to reconstruct top quarks with transverse momenta $100 \mathrm{ GeV} < p_T < 400$ GeV, by sorting jets into three "buckets". The first two buckets $B_1$/$B_2$ will ideally contain the decay products of the two top quarks, while the third one $B_{ISR}$ collects the extra radiation in the event. The performance and applicability of the method have been studied and method was validated in a Monte-Carlo to data comparison. For the $ttH$ channel the bucket algorithm could be used to solve the combinatorial problem of assigning four b-quarks to the two top decays and the $H\rightarrow b\bar{b}$ Higgs boson decay, as suggested in JHEP 1402 (2014) 130.
test
Sketch of the buckets of top algorithm, $j_1$ to $j_{10}$ correspond to jets recontructed in the event.

Single Vector Like Quark search

The Standard Model has several open questions waiting to be answered, such as why we have only six quarks and not more. The discovery of the Higgs boson excludes the existence of new quarks similar to the ones we already have observed. However, a new class of quarks, called Vector Like Quarks (VLQs) has not been excluded, as they do not receive their masses through a coupling to the Higgs boson.
test
Leading-order Feynman diagram of single Q (Q=T,Y) production in Wb fusion and subsequent decay into Wb. ATLAS-CONF-2016-072

Such Vector Like Quarks have not been observed, but they are a key component in several Beyond the Standard Model theories, which address existing problems of the Standard Model. Our group works on the search for a new Vector Like Quark particle, which would be produced in association with a b-quark and a light-quark, and further decays in a W-boson and a b-quark. This search is performed in the channel on which the W boson decays into an electron or muon.
test
Distributions of the reconstructed VLQ mass in simulation and data. ATLAS-CONF-2016-072

Work in Progress and Topics of Interest

  • Top quark and W boson tagging performance studies in 13 TeV data,
  • Searches for New Physics using boosted final states with top quarks or W bosons, e.g. heavy resonances, light stop partners in vector like quarks/supersymmetry, dark matter, etc.,
  • Standard Model Measurements involving top quarks, $ttH$, cross-sections, etc.,
  • Improvements for for tagging even more boosted objects,
  • Improvement of the "Buckets of Tops" algorithm.


FTK - Fast Track Trigger

The Fast TracKer (FTK) system is a dedicated pre-processor, that allows to do the global track reconstruction at 100 kHz rate. Track reconstruction at this rate and at instanteneous luminosity of up to $3\times 10^{34}\mathrm{cm}^{-2}\mathrm{s}^{-1}$ is very challenging. At the moment, usage of CPUs for such task is not possible, as it would be way too slow and expensive. Therefore, a dedicated hardware system was designed based on custom Associative Memory chips (see web page on Wikipedia) for pattern matching and FPGA's (see "FPGAs for dummies") for track reconstruction and data processing. Such design together with highly-paralell processing allows to do reconstruction of tracks in the full ATLAS semiconductor tracker down to 1 GeV.

test
A sketch of the ATLAS trigger dataflow system with FTK included. FTK runs for each event that has been accepted by the hardware-based Level-1 Trigger and provides inputs for the software-based High Level Trigger.

Our group has contributed into delivery of the first operational control and configuration tools for FTK. Group members are involved in ongoing commissioning of FTK and integration inside ATLAS. The group takes leadership role in development of monitroing applications, that are key ingredient in the commissioning. This development spans from tools that check basic status of the dataflow in the system to high-level frameworks that sample ATLAS events, run FTK simulation on inner-detector hits and compare output to the actual tracks reconstructed by FTK hardware in the same event. As of 2017, FTK is in the critical start-up phase, has an active community both in Heidelerg and in CERN and provides high visibility for active members.
In the past, we have also contributed in Associative Memory chip tests. Associative Memory allows a very fast search by content in contrast to the commonly-used RAM, which allows fast search by address. In FTK such devices are used to find combinations of hits, that roughly match pre-computed patterns corresponding to possible trajectories of charged particles, i.e. pattern recognition.

Possible topics for Master/Bachelor theses

There are plenty of possible studies one can do in context of master or bachelor thesis. Please, contact the group leader for more information.

ATLAS Upgrade Projects for High Luminosity LHC

Track Triggers

In the years 2023-2025 the LHC accelerator and the experiments will undergo major upgrades to increase the luminosity (Phase 2). Post-upgrade about 200 proton-proton interactions per collision are expected at a collision rate of 40 MHz. The higher event rates necessitate high selectivity in the event selection and in the trigger. A big challenge is the high pileup rate of 200 proton-proton interactions per collision which increases the confusion problem in the event reconstruction and reduces the resolution of event quantities. This problem arises already at trigger level where only the few interesting physics events are selected and all other events are rejected.
The pileup problem can be effectively reduced by using tracking information as tracks allow to distinguish the different event vertices. Therefore Several concepts are being investigated for reconstructing tracks already at the trigger level.
Our group is involved in several track trigger projects:
  • FTK++ is a Phase 2 upgrade of the Fast Tracker (FTK). FTK is a fast hardware track processor which reconstructs all tracks of a collision with an event rate of 100 kHz. These tracks are input for the Higher Level Trigger (HLT). FTK is currently in the commissioning state.
  • HWTT is a new hardware track trigger which will be able to reconstruct tracks only in regions of interest (ROI) with a rate of up to 1 (4) MHZ.
  • L0TT is a proposed track trigger capable of reconstructing all tracks at full collision frequency (40MHz). This proposal requires installation of pixel sensors at large radii in the ATLAS experiment.

Technologies:

For the hardware implementation of track triggers various track reconstruction techniques are used. One is the pattern lookup technique where measured hits are compared with pre-calculated patterns. Pattern lookup in hardware are extremely fast, for example if implemented in associated memories. The AMchip6 is such a memory chip which has been developed for the FTK project. Another method is fast track fitting which can be implemented in Data Signal Processors (DSP) or Graphical Processing Units (GPU).

test test
Sketch of tracks originating from different proton-proton collisions. Several bundles of tracks form jets (top). High momentum isolated track typical for electrons, muons and tau leptons

References:

Monolithic Pixel Sensors for Inner Tracker (ITK)

For High Luminosity-LHC the ATLAS tracking detector will be rebuilt. Because of the increased particle rates the tracker has to be more radiation tolerant than the current tracking detector and has to stand fluences of more than 10^16 protons per square centimeter. Therefore the new tracker will consist of silicon pixel and silicon strip detectors only.
test
Possible design of the ATLAS Inner Pixel Tracker at HL-LHC

Our group is developing in close collaboration with Prof. Peric (KIT) radiation hard monolithic active pixel sensors (MAPS) based on an industrial High Voltage CMOS process. This HV-CMOS process allows to fully deplete the sensing diodes leading to a fast collection of the charge carriers in strong electric fields. Recent measurements have demonstrated the radiation hardness of this technology. The monolithic design -- which combines the active sensor and the readout electronics in a single device -- reduces costs and significantly simplifies the construction of pixel tracking modules.
Monolithic active pixel sensors are discussed as an option for the outermost ATLAS pixel layer. They are also considered as baseline for future collider experiment, e.g. at future circular colliders (FCC project).
test
Picture of a monolithic HV-MAPS sensor which combines several functionalities in a single chip. The standard hybrid design for the ATLAS pixel detector is shown in the background.
References:
Webmaster:
EDV Abteilung