We require completed CVs, along with a Letter of Recommendation from your university tutor, by 21st February 2018.
Stop-press: We are extending the deadline to end of Sun 25 Feb. Please get CVs to us by then; don't panic if Letters are one or two days later.
Possible projects include the following.
Feel free to tell us in your CV in which projects you would be particular interested (can be one or more). If you have no strong preference, that is fine. We will be looking for the best students and will try to match them up to suitable projects.
Tracking Detector Design for ATLAS Upgrade
In ATLAS preparations are ongoing to upgrade the detectors for data-taking around 2025 when the Large Hadron Collider will collide protons with a much higher rate than today. Already in the ongoing data-taking, up to 60 low-energy collisions coincide with each interesting physics one. In 2025 the situation will be even worse and up to 200 additional interactions will occursimultaneously. This will be a challenge for the detectors and the physics analyses. For this upgrade, a new tracking detector will be build in which the group here at RAL is involved.
The summer student will help with the validation of the design for this new tracking detector whose final design details still needs to be fixed. He/she will study the tracking performance for the new detector. To do so he/she will use simulations to study the reconstruction of electrons or muons which traverse the detector and which are produced by particles coming from possible new physics.
Proposed dates of placement: 8 week in the period between June to end of August.
Student specification: For this task he/she will use C++ and/or Python code to analyse the output of the ATLAS reconstruction software suite. Therefore, the student should have interest in computing, but no deep prior knowledge in C++ or Python is required.
High-Level Tracking Triggers for ATLAS
The ATLAS Trigger system makes fast, real-time, decisions on whether to keep
data from interesting proton-proton collision events to be studied later, or
discard them. We can only keep about 1 in 100,000 collisions. The High-Level
Trigger (HLT) includes fast software algorithms that process information from
the Inner Detector to find charged particle tracks. Because of the huge number of
particles produced in LHC collisions, the Inner Detector tracking software uses
a lot of computing power - almost half of the HLT computing resources are used
to reconstruct tracks in real time. ATLAS has developed custom-built
electronics (called the Fast TracKer, FTK) to find tracks before the start of
the HLT; this hardware will be deployed and operated for the first time this
year. In this project, you will use the ROOT analysis package to perform
validation of the tracks from FTK and commission new FTK-based triggers ready
to be used online later in the year.
Proposed dates of placement: 8 weeks in the period June to August.
Student specification: You should have an interest in computing with some
experience of programming in C++ or a similar language. Some knowledge of ROOT
would be helpful but is not essential.
Pixel Module Mounting for the ATLAS Tracker
The ATLAS Pixel detector will be replaced for the High-Luminosity Upgrade of the ATLAS Tracker. At RAL, we are responsible for mounting Pixel Modules on the carbon-fibre C-sections.
The student will be involved in:
test assemblies/targets and comparing accuracy using the RAL module placement system and Smartscope.
Qualification of different adhesives for module mounting as necessary.
Measuring pot-life of adhesives to be used and programming the new glue dispenser to compensate.
DAQ and test setup for electrical testing of assembled half-rings.
Utilising Machine Learning to optimise Data Placement for the ATLAS experiment
the start of LHC data taking, the ATLAS experiments has produced hundreds of
petabytes of data. The data produced by ATLAS during LHC running is only
a small fraction of the total as there are many derived data formats for
specific types of analysis as well as a huge amount of Monte Carlo simulations.
ATLAS data management system distributes this data across 120 sites around the
world. While the data movement is handle automatically, decisions about
what data to replicate, archive or delete are still made by humans. The
aim of the project is to utilise machine learning techniques to analyse the
data access patterns of ATLAS jobs in order to predict future usage. This
can then be used to optimise data placement to improve the rate at which data
is processed as well as saving resources.
Particle Reconstruction Algorithms for the CMS Level-1
In 2026, after a series of upgrades, the Large Hadron
Collider will be colliding protons at a significantly higher rate, in order to
increase the sensitivity of the ATLAS and CMS detectors to evidence for new
physics. The CMS detector's level-1 trigger system identifies the most
interesting collisions within a few microseconds, by reconstructing the
numerous particles produced in each collision using high-speed programmable
electronics (FPGAs). From 2026, identifying the most interesting collisions
will become significantly more difficult due to the increased rate of proton
collisions, and so the level-1 trigger system will be upgraded to use
state-of-the-art technologies. This is a very challenging project, and research
is underway here to optimise our proposed solution.
The summer student will use C++ software running on
simulated LHC collision events to develop algorithms that identify the
particles produced from each collision - either reconstructing their
trajectories within the tracking systems, or identifying particular types of
particles by combining these reconstructed tracks with information from other
detector components (such as calorimeters).
Since these algorithms must ultimately be run in FPGAs,
the student will need to keep them as simple as possible, and understand the
limitations and strengths of the electronics. Depending on how quickly the
project progresses, the student may also be able to write firmware that implements
the algorithms in FPGAs using a high-level synthesis language.
Proposed dates of placement: 8 weeks during a period to
be agreed from June to August.
Student specification: You need to have a logical mind,
and should be familiar with computer programming, and ideally with C++. An
interest in particle physics or electronics would be a bonus.
Deep Neural Networks in CMS
The CMS experiment at the Large Hadron Collider is
collecting large volumes of complex data. The data is being examined for evidence of physics beyond
the standard model, such as new types of Higgs Boson. Traditional data analysis techniques are
evolving to include using sophisticated machine learning algorithms. The increase in computing power is making recent
developments such as Deep Neural Networks viable in the large and complex data environment at the LHC. The
student will use neural networks to study their capability for identifying signatures of interesting events in the
CMS detector. This will involve the use of GPUs and possibly FPGAs.
Student specification: The student should have familiarity with computer
programming, and ideally with C++ and an interest in Particle Physics.
studies of the LUX-ZEPLIN Dark Matter experiment
The nature of dark matter is one of the open and fundamental questions in
physics, and the LZ experiment is at the forefront of technology designed to
pursue this question. The experiment will be using a two-phase Time
Projection Chamber with the ability to measure both the scintillation light
from the liquid and the electroluminescence light from the gas above the
liquid. The success of the LZ experiment depends on a careful understanding of
the response of the detector.
The summer student will use simulation to study the impact of a non-uniformity
of the gas gap above the liquid on the electroluminescence light generation.
Proposed dates of placement: 8 weeks during June-August
should have an interest in particle physics and computing, with some experience
of programming in C . Some knowledge of ROOT and Finite Element Methods would
be helpful but is not essential.