2023 USRA Projects
1. Machine learning with SuperCDMS SNOLAB
Contacts: Prof. Scott Oser & Dr. Yan Liu | Email: oser@phas.ubc.ca and yanliusp@gmail.com
After decades’ search of what accounts for about 75% of the total mass in the universe, dark matter still remains to be discovered. SuperCDMS SNOLAB aims to detect tiny scattering or absorption signals from dark matter particles. The student will participate in the software and computing development of the experiment.
1.) In searching for new discoveries, it is important for experimenters to avoid (sometimes subconsciously) biases towards false positives and this is commonly done by blinding potential signals during the initial analysis stage. The student will help develop the blinding scheme for the upcoming SuperCDMS data analyses using generative adversarial network (GAN).
2.) We are currently developing an advanced signal extraction technique based on the pulse shape differences of the signal. The student will be involved with the effort of categorizing different groups of signals using machine learning techniques such as auto encoder.
Experience with python is required.
2. Statistics of CMB Polarization
Contact: Dr. Douglas Scott | Email: dscott@phas.ubc.ca | Web: https://www.astro.ubc.ca/people/scott/basic.html
The cosmic microwave background allows us to probe the Universe on the largest length scales possible. There are several hints or "anomalies" that may suggest modifications to physics on large scales or at very early times in the history of the Cosmos. In order to assess if such anomalies are real or just mild statistical excursions in the data, it is necessary to find new ways to probe the large-scale Universe. One such new probe is provided by sensitive measurements of CMB polarization, which comes from new modes in the early Universe. The latest maps of large-angle polarization have been provided by the Planck satellite. In this project we will study aspects of sky polarization, and investigate statistical techniques that can be used to distinguish the cosmological signals and to test for deviations from statistical anisotropy. Additionally, it will be useful to assess the power of future (more sensitive) polarization measurement using simulations.
3. Deep Learning in Astronomy
Contact: Dr. Douglas Scott | Email: dscott@phas.ubc.ca | Web: https://www.astro.ubc.ca/people/scott/basic.html
There are many data analysis problems in astronomy that are best approached using simple likelihood function methods. However, there are other questions (involving non-linear selection tasks, or pattern-matching in huge databases) that are more efficiently performed with "machine-learning" (ML) methods, such as neural networks. One downside to the use of ML approaches is that it is often difficult to determine robust uncertainties on derived parameters. Another unresolved issue is how to combine traditional and ML methods in tasks that use both approaches for different parts. We will investigate these topics by looking at the use of ML in astronomy, combining data at multiple wavelengths to identify and categorize distant galaxies and assess their statistical properties.
4. Multi-Task Deep Learning for Segmentation and Outcome Prediction in PET/CT Images of Cancer Patients
Contact: Arman Rahmim | Email: arman.rahmim@ubc.ca | Web: http://qurit.ca
Outline of Research Project:
We aim, in our multidisciplinary lab, to perform advanced image analysis on PET and PET/CT images to improve and simplify diagnosis, cancer staging, therapy response assessment and prediction of outcome for cancer patients, in close collaboration with nuclear medicine physicians. Thorough analyses on medical images, leading to the field of radiomics, often rely on accurate segmentations of tumors; meanwhile, a fully automated segmentation step is a bottleneck for applying radiomics studies on large datasets.
Deep learning especially via Convolutional Neural Networks (CNNs) has enabled improved performance in different medical imaging tasks. In particular, improved outcome prediction for cancer patients can enable better personalization of therapies, though it has been less frequently considered via CNNs since they were not as efficient compared to alternative non-deep-learning frameworks. The deep features learned by (convolutional) layers of CNNs for tumor segmentation (i.e., U-net) have the potential to guide the outcome prediction network by exploiting the scarce and precious radiomics features. We aim to evaluate this potential by suggesting and evaluating the models for simultaneous tumor segmentation and outcome prediction from PET/CT images of patients with different cancers.
Our proposed plan at the Quantitative Radiomolecular Imaging & Therapy (Qurit) lab (Qurit.ca) is to develop multi-task learning approaches for segmentation and outcome prediction in parallel. This innovative idea has only once been applied to PET/CT radiomics and we aim to propose different multi-task frameworks. Studies have shown that multi-task learning, when the tasks are related, could improve the efficiency of the individual tasks. In other words, the deep features that are learned for segmenting tumor regions and the relevant features from tumor regions in outcome prediction can co-facilitate learning. We aim to develop and validate such deep learning frameworks for the analysis of PET/CT images of cancer patients.
The Student's Role:
After joining Qurit lab, the student will gain familiarity with AI-based techniques including deep learning approaches especially for segmentation and outcome prediction by studying reading material and test data provided by Qurit lab, and with the help of the post-doctoral researchers and PhD students in the lab. In the next step, sub-projects that the student is supposed to work on (in collaboration with other lab members) will be refined adaptively based on student capabilities and interests. The main part of the student's project is to provide help to develop and improve the deep models for simultaneous segmentation and outcome prediction. The student will contribute to one or more clusters (research sub-groups within our lab for lymphoid, cervical, prostate and/or lung cancers) to evaluate the capability of the developed multi-task models in different cancers. The student will experience an exciting research program and may have the opportunity to participate in AI competitions and/or conference/journal works by team members.
The trainee will be (i) mentored on a daily/weekly basis, (ii) immersed in a multi-disciplinary environment, (iii) provided with in-depth understanding of recent developments in imaging & AI, and (iv) the opportunity to interact with our clinical and technical collaborators. The trainee will present and discuss research ideas to the team and respond to questions in a safe yet intellectually inspiring environment. The program will also emphasize the trainee's communication skills, a strong area of mentorship in our lab.
5. Looking for New Physics with ATLAS Precision Measurements
Contact: Alison Lister | Email: alister@phas.ubc.ca | Web: https://atlas.cern/
Q: How do we learn something about new physics beyond the Standard Model (BSM) without measuring it directly? A: We look for its impact on things we can measure! The UBC ATLAS group is working to constrain new physics using precision measurements of Standard Model particles. Different hypothetical BSM particles can cause subtle changes to what we see in the detector. By putting together these measurements we can look for any anomalies that could hint at new physics.
The student will work on translating individual measurements into a combined framework, and optimizing variables to maximize sensitivity. See more information about ATLAS and new physics here.
6. Constructing Silicon Inner Tracker (ITk) for ATLAS Detector Upgrade
Contact: Alison Lister | Email: alister@phas.ubc.ca | Web: https://atlas.cern/
The UBC ATLAS group is among several institutions around the world participating in the construction of the new, all silicon, Inner Tracker (ITk) for the upgrade of the ATLAS detector for the High Luminosity Large Hadron Collider (HL-LHC) at CERN. Each module of the silicon strip tracker must undergo a series of thermal and electrical quality control measurements before they are installed in the ATLAS detector. Here at UBC, we are performing these critical tests in our newly commissioned cleanroom.
The student will work on building and improving the test setup, optimizing and automating the testing procedure, and analyzing and presenting the results from electrical tests to the wider ATLAS community. See more information about the ATLAS silicon Inner Tracker here.
7. Deep Learning with ATLAS
Contact: Alison Lister | Email: alister@phas.ubc.ca | Web: https://atlas.cern/
The ATLAS UBC Group is developing new deep learning techniques for both signal vs. background classification problems as well as inference problems (given what we see in our detector, what are the most likely properties of the particles that produce that signature). The students will work on further improvements of the method as well as develop techniques for mitigation of the impact of the systematic uncertainties on the deep learning model.
Experience and familiarity with Python is required.
8. Quantum Coherent Control
Contact: V. Milner | Email: vmilner@phas.ubc.ca | Web: http://coherentcontrol.sites.olt.ubc.ca/
Our research group on Quantum Coherent Control uses ultrafast lasers to control and study the behaviour of molecular "super-rotors" and their interaction with quantum media, such as helium nanodroplets or ultracold plasmas. Super-rotors are extremely fast rotating molecules produced in our laboratory (and not available anywhere else!) using a unique laser system known as an "optical centrifuge". Many fascinating properties of molecular super-rotors have been theoretically predicted. A few of them have been already shown by our group in the last five years, but many more await discovery.
In the summer of 2023, the USRA student will help a senior PhD student with an ongoing experiment on the laser centrifugation of molecules captured by the beam of helium nanodroplets. For specific tasks and projects, please contact Dr. Milner at vmilner@phas.ubc.ca.
9. Magnetic Resonance with Non-Aqueous and Water Protons in the Brain
Contacts: Alex Mackay & Carl Michal | Email: mackay@phas.ubc.ca & michal@phas.ubc.ca Web: https://phas.ubc.ca/users/alex-mackay & https://phas.ubc.ca/~michal/
Magnetic resonance imaging (MRI) is heavily used in medicine because it produces images with high contrast between different soft tissue types and between healthy and pathological tissue.
The physical mechanisms which determine this exquisite tissue contrast are still not clearly understood. MR images from the brain are generated from signals coming from hydrogen nuclei in water in the brain; however, the signal from the water protons are influenced by interactions between water and non-aqueous protons attached to lipids and proteins. The proposed project will involve in vitro and ex vivo NMR experiments designed to enable us to better understand the interactions between non-aqueous protons and water protons in the brain. Having a clearer understanding of these interactions may enable us to extract more specific and quantitative information about brain microstructure using MRI.
10. Moire effects in 2D quantum materials
Contacts: Ziliang Ye | Email: zlye@phas.ubc.ca | Web: http://ye.physics.ubc.ca
MoS2 is a 2D quantum material exhibiting a series of fascinating optical properties such as direct bandgap at Brillouin zone corners, optically accessible valley degree of freedom, and strong excitonic effects. Recently, these new properties are found to be modifiable by making heterostructures with other 2D quantum materials, where electronic and optical properties can be altered on the nanometer scale by forming so-called Moire superlattices. As a result, a range of exotic phases including Mott insulator and unconventional superconductor arise out of such an emergent order. This summer, we would like to invite students interested in experiencing experimental condensed matter research to join us to study the emerging property of the Moire superlattice formed by twisted 2D materials. The intern will have the opportunity to learn hands-on skills of making high-quality heterostructures as well as advanced optical spectroscopy and data analysis techniques.
11. Improving the Performance of the Advanced LIGO Gravitational Wave Detectors
Contacts: Jess McIver & Raymond Ng | Email: mciver@phas.ubc.ca and rng@cs.ubc.ca | Web: https://gravitational-waves/phas.ubc.ca
Gravitational-wave detector data, including the LIGO detectors, contains a high rate of instrumental artifacts that mask or mimic true astrophysical gravitational wave (GW) signals. This project will characterize noise sources in the Advanced LIGO detectors with the goal of reducing the number of 'false alarm' GW candidates and improving the reach of GW searches for the next observing run. Students will work with a team of physicists and data scientists, and gain transferable skills in data visualization, Python programming, gravitational-wave astrophysics, large-scale physics experimental instrumentation, and potentially machine learning (if desired). Familiarity with Python is preferred.
Find out more about the UBC LIGO Group and the LIGO Scientific Collaboration.
12. Satellite Observations for the Protection of the Dark Sky
Contact: Aaron Boley | Email: aaron.boley@ubc.ca | Web: acboley | UBC Physics & Astronomy
Large constellations of satellites are poised to cause major interference with astronomy. Astronomers are engaged in discussions with governments and industry to identify mitigations for existing interference and to establish paths toward minimizing interference by future satellite constellations. As part of this effort, observations of so-called megaconstellation satellites and large debris are needed in astronomical observing bands to establish the brightness distribution of satellites and their variability, as well as to assess the effectiveness of mitigations. Such observations are further needed to emphasize the different ways that satellites will impact astronomy.
This project will observe megaconstellation satellites and large debris in different observing bands and assess their brightness range. The project will further compare the results with satellite brightness modelling and assess intrinsic variability.
13. Orbital Evolution of Hot Jupiters
Contact: Aaron Boley | Email: aaron.boley@ubc.ca | Web: acboley | UBC Physics & Astronomy
Planet-planet and planet-star interactions create variations in the transit times of planets. Such variations are observable signatures of planetary orbital evolution and offer constraints on planet formation and planetary system models. So-called Hot Jupiters, which are giant planets that orbit their host star with periods ranging from days to about two weeks, may be particularly prone to tidal orbital decay. The decay leads to a decrease in the planet’s semi-major axis, and increase in its period, and quite possibly, the eventual destruction of the planet.
This project will analyze citizen science data and archival data from national facilities to search for timing signatures of planetary orbital decay.
14. A 2D Material Workstation to Stack Membranes for the Quantum Materials and Device Foundry.
Contact: Ke Zou | Email: kzou@phas.ubc.ca | Web: kzou | UBC Physics & Astronomy
The student would work with an integrated nanofabrication workstation, based on a precision inert-atmosphere glovebox with a computer-controlled microscope interfaced to our existing molecular-beam epitaxy (MBE) systems in the Quantum Materials and Devices Foundry (QMDF), part of the Quantum Materials Institute (QMI) at UBC. MBE, used widely both in research and industry, is the most advanced technique for thin-film growth because it can provide atomic-level control of new crystalline phases of materials. Because MBE can produce layered phases of quantum materials not possible by any other method, this workstation promises the fabrication of previously unavailable freestanding two-dimensional (2D) membranes that possess new and unexpected electronic properties, providing highly valuable samples to a large number of groups at QMI and other institutes.
15. Charting the Growth of Galaxies
Contact: Allison Man | Email: aman@phas.ubc.ca | Web: https://phas.ubc.ca/users/allison-man
Galaxies evolve on astronomical timescales of millions or even billions of years. The study of galaxy evolution is therefore based on inferring connections between various galaxy populations across cosmic time. This requires knowledge of galaxy properties, such as distances, sizes, masses, ages, and star formation rates. The student will learn now to extract such information from galaxy images and spectra. Driven by the student's interest, the project will tackle these important scientific questions: What triggers or shuts down star formation in galaxies? How do active supermassive black holes influence star formation of their host galaxies? What happens to galaxies when they collide with each other?
The student will apply their Python computing skills to handle large datasets and images, to visualize and to present findings. These skills are relevant for a variety of projects in astronomy, other research disciplines and beyond academia.
Experience with Python programming is required. Knowledge of physics, astronomy, statistics, data analysis, LaTeX and Git will be considered a plus. The ideal candidate will have taken at least one ASTR course at the 200-level and above.
16. Laboratory for Atomic Imaging Research
Contact: Sarah Burke & Doug Bonn | Email: saburke@phas.ubc.ca; bonn@phas.ubc.ca | Web: https://lair.phas.ubc.ca
The "Laboratory for Atomic Imaging Research" run by profs Burke and Bonn offers projects focused on design, analysis and experiment in atomic scale characterization of the structure and electronic states, of surfaces, molecules, and quantum materials by scanning probe microscopy. Interested students may wish to reach out to discuss particular interests.
17. Single-molecule investigations of competition among DNA secondary structures and consequences for biomolecular interactions
Contact: Sabrina Leslie | Email: leslielab@msl.ubc.ca | Web: https://leslielab.msl.ubc.ca/
DNA, as a long, double-stranded polymer, can take on a variety of secondary structures. Several of these structures can be induced by DNA supercoiling – how twisted the DNA is compared to its relaxed state. The total amount of supercoiling in a DNA molecule is fixed, so when there are multiple sites within a single molecule susceptible to secondary structures, the sites will compete among each other to form.
In this project, the student will study the competition among secondary structures in a DNA molecule and compare their results to predictions from a statistical mechanics model. The student will learn how to genetically modify DNA in order to introduce new structures, and study the presence of structures using molecular biology and single-molecule microscopy techniques. Specifically, she will study the competition between two ‘unwinding sites’, regions of DNA that become single-stranded when subject to negative supercoiling.
The student will receive hands-on training and guidance from the Leslie Biophysics Research Group, including day-to-day guidance from a senior graduate student in the lab, regular meetings with the research team and principal investigator, and surrounding interdisciplinary environment at the Michael Smith Labs. Through the course of the summer, the student will hone her oral communication skills through opportunities to present at group meetings and interact at local events such as the SBME and Michael Smith Labs Summer Poster Fairs.
18. Single-molecule microscopy of oligo-oligo and oligo-enzyme kinetics
Contact: Sabrina Leslie | Email: leslielab@msl.ubc.ca | Website: https://leslielab.msl.ubc.ca/
Antisense oligonucleotide (ASO) therapeutics is an emerging technology capable of altering mRNA expression by targeted cleaving via RNase H1. The molecular mechanisms and interactions by which this process works are not well understood, and secondary structures of the ASO and/or RNA target can have a strong impact on the interaction rates. In this research project, the student will use fluorescence lifetime correlation spectroscopy (FLCS) to detect and measure oligo-oligo and oligo-enzyme reactions, and write data analysis code in MATLAB to quantify kinetic rates from microscopy data.
The student will receive training in single-molecule fluorescence microscopy and quantitative FLCS analysis and will work closely with a research fellow in the lab. Weekly meetings with the Leslie biophysics team, and daily interactions with members of our interdisciplinary colleagues in the Michael Smith Labs, will support and guide the researcher’s training and development. In addition to gaining hands-on research experience, anticipated outcomes of this summer research project include oral presentations for lab members and poster presentations at the Michael Smith Labs Summer Poster Fairs and SBME Summer Poster Fairs, providing key training in written and oral communication.
This project is ideally suited to an undergraduate student in the UBC Biophysics program who is interested in continuing towards a senior thesis with the Leslie Lab the following year.
19. Single-particle microscopy of mRNA lipid nanoparticle complexes
Contact: Sabrina Leslie | Email: leslielab@msl.ubc.ca | Web: https://leslielab.msl.ubc.ca/
Nanoparticles are increasingly used in pharmaceutical applications. This research project will use single-particle confinement microscopy to investigate the biophysical properties, stoichiometry and kinetics of nanoparticle assemblies and their interactions. This technique entraps particles in femto-liter reaction wells and allows prolonged monitoring of reactions. It also enables direct visualization of interactions between chemical species and nanoparticles.
For this project, the student will perform single-particle experiments to investigate and quantify probe-nanoparticle interactions. For example, particle tracking algorithms can be used to extract valuable information such as diffusion coefficients and fusion kinetics. This project is best suited for engineering physics students with an interest in computer science and biology since we will be applying physical tools to understand out data.
The student will receive training in quantitative image analysis as well as hands-on microscopy and will work closely with a research fellow and graduate student. Weekly meetings with the supervisor and collaborators, and daily interactions with members of our interdisciplinary research group including the nanomedicine network, will support and guide the project. In addition to gaining hands-on research experience, anticipated outcomes of this summer research project include virtual presentations with lab members, providing key training in writing and oral communication.
20. Particle Physics research with the DarkLight Collaboration at TRIUMF
Contact: Prof. Michael Hasinoff | Email: hasinoff@physics.ubc.ca | Web: https://phas.ubc.ca/users/michael-hasinoff
Dr. Katherine Pachal | Email: kpachal@triumf.ca | Web: http://darklightariel.mit.edu
We are preparing a new experiment at the TRIUMF/ARIEL e-linac accelerator to search for a possible new particle that could be the mediator of a new short range 5th force in Particle Physics. The student will participate in measurements of the timing and energy response of a prototype scintillation counter using a solid state PMT. She/He will learn to use the sophisticated CERN data analysis program ROOT as well as hone their skills in both C++ and Python based analysis. All the work will take place at TRIUMF and the student will be able to participate in all the activities planned for the other ~30 Co-op/USRA students working at TRIUMF.
21. High Throughput Solid State Synthesis for Machine Learning Applications
Contact: Alannah Hallas | Email: hallas@phas.ubc.ca | Web: https://qmi.ubc.ca/
The ultimate goal of this research project will be to generate a robust data set that will ultimately be used for machine learning.
High entropy oxides (HEOs) are a topic that have generated significant excitement in the fields of chemistry, physics, and materials science following their first report in 2015. The reason for this excitement is that HEOs, with their deliberately high configurational disorder due to the combination of many elements sharing a single crystal lattice, are a fundamentally new platform for engineering materials with tailored properties. The high configurational entropy in these materials allows new phases to be stabilized and imbues them with enhanced structural stability, a necessary characteristic for a wide-range of applications.
Recent work from the Hallas group, has led to the discovery of a highly tunable family of magnetic HEOs with the spinel crystal structure. The materials under investigation in this work are composed of six elements (Cr,Mn,Fe,Co,Ni,Ga)3O4 and by tuning the concentration of only one element over a narrow range we were able to show remarkable tunability in the magnetic properties. However, due to the site selectivity of the spinel structure, these properties vary in a non-systematic manner, making the prediction of a compound with targeted properties intractable by conventional theoretical methods. Furthermore, thoroughly surveying the phase space of different ratios of six elements is also experimentally intractable due to the millions of possible combinations. This is therefore an ideal problem for a machine learning approach.
This project will unfold in two stages. The summer research position offered here is related to the first stage, which is the development of a robust training data set to be used as an input for machine learning. Two summer students will be hired to work on this task. They will randomly survey 200+ compositions to experimentally assess their magnetic properties. This will involve developing an experimental protocol and methodology using a microwave synthesis method. The protocol will be iteratively optimized for high throughput sample preparation without compromising sample quality. Sample quality will then be assessed using x-ray diffraction. A protocol for measuring the magnetic properties will also be developed. The output from this project will be a training data set that will be used to perform the first proof-of-principle machine learning study to predict the composition of a material with targeted magnetic properties, an experimentally intractable problem. Summer students will have the opportunity to participate in the machine learning component of the project towards the end of the work term.
22. Atomistic Calculations of High Entropy Oxides
Contact: Prof. Joerg Rottler, Prof. Alannah Hallas, Dr. Solveig Aamlid | Email: solveig.aamlid@ubc.ca
Web: https://qmi.ubc.ca/research/atomistic-approach-disordered-materials/
High entropy oxides are a novel class of materials that find promising applications in energy conversion and storage, thermoelectric devices, and catalysis. In these crystals, up to five elements share one crystallographic site, which leads to properties that are vastly different from their simpler building blocks. However, predicting the phase stability and crystal structure of such materials is challenging. One approach to solve this problem is to use density functional theory (DFT) to calculate pairwise interaction parameters and further use those parameters in Monte Carlo simulations.
In this USRA project based at the Quantum Matter Institute, the student will help identify relevant elements and crystal structures for high entropy materials, run and optimize calculations of the chosen combinations, and evaluate the results to guide the synthesis of new materials. The student will receive training in computational methods that are commonly used in solid state physics. Good knowledge of computational methods, ideally some experience with Python/Matlab/Unix OS would be very helpful.
23. Quantitative magnetic resonance imaging (MRI)
Contact: Shannon Kolind | Email: shannon.kolind@ubc.ca | Web: kolind | UBC Physics & Astronomy
Quantitative magnetic resonance imaging (MRI) is far more reproducible and interpretable than conventional qualitative MRI. Our lab works to develop advanced MRI techniques specific to various biological components of the brain and spinal cord, such as myelin, axons, or inflammatory cells. These techniques are extremely important in the study of neurological diseases such as multiple sclerosis (MS).
The goal of this summer project is to help develop and validate novel markers of disease progression based on conventional and quantitative MRI data, collected as part of a nation-wide study that aims to improve our understanding of disease progression for Canadians living with MS. This huge dataset provides a unique opportunity for creative application of image normalization, de-noising, machine learning, template creation, and other data analysis techniques. The student will also help with a related project, mapping characteristics of the healthy human brain based on quantitative MRI data. We are looking for a highly driven and motivated student who is keen to learn about and develop these widely applicable skills.