Undergraduate Research Positions

Fix a field F of characteristic zero, and let the symmetric group $S_n$ act on the polynomial ring $F[x_1,\ldots,x_n]$ by permuting the variables. It is well-known that the ring of symmetric polynomials over F (the subring of $S_n$-invariants) is itself a polynoimal ring, either in the \emph{elementary symmetric functions} $s_k = \sum_{A\in\binom{[n]}{k}} \prod_{i\in A} A_i$ ($1\leq k\leq n$), or alternatively in the \emph{power sum polynoimals} $t_k = \sum_{i=1}^{n} x_i^k$ (same range of $k$).

A similar result holds when we replace the polynomial ring with the set of Laurent polynomials $F[\{x_i\}^\pm\}]$ (for technical reasons this is not a ring, though that can be fixed). The Laurent polynomials admit an action by rescaling the variables, and in particular have a "subring" consisting of the elmeents of degree zero (e.g. the symmetric polynomial $\sum_{i\neq j} x_i/x_j$). We will investigate generating sets for Laurent polynomials of homogenous degree zero and for higher-dimensional generalizations.

We will investigate the p-adic character of recent physics papers on "p-adic string theory", specifically whether harmonic analysis on trees rather than harmonic analysis on p-adic groups is sufficient for the computations.

This project is concerned with two-component competing system on lattice having both local (perimeter) and nonlocal interactions ($H^{-s}$). First one derives the most general form of $\Gamma$-convergence energy functional which involves fractional $s$-perimeter and Riesz potential interaction $\frac{1}{|x-y|^{n-2t}}$ where $s$ and $t$ are both in the range of $(0,1)$. Then we want to find (numerically and variationally) the most stable lattice shape for this system on a two-dimensional or three-dimensional lattice. Basic knowledge of multi-variable calculus, functional analysis, PDE and computer programming are needed.

Arrays of boxes filled with positive integers subject to certain rules are called Young tableaux. They generate functions known as Schur functions, which date from Cauchy in 1815, and arise today in many areas including combinatorics, algebra, quantum physics and algebraic geometry. In this project we will investigate a natural generalization of skew Schur functions with the aim of discovering their algebraic and combinatorial properties. We may also investigate connections to other areas. A strong background (A+) in abstract algebra such as Math 322 or Math 323 is required, and an aptitude for combinatorics and programming skills will be an asset.

A problem in extremal combinatorics asks how many (discrete) objects are possible subject to some condition. We consider simple matrices which are (0,1)-matrices with no repeated columns. We say a simple matrix A has F as a configuration if a submatrix B of A is equal to a row and column permutation of F. Much has been determined about the maximum number of columns an m-rowed simple (0,1)-matrix can have subject to condition it has no configuration F for some given F. There is an attractive conjecture (A. and Sali 2005) to pursue with some challenging open problems. We might pursue some 4-rowed choices for F that when forbidden in an m-rowed simple matrix A should yield a quadratic bound (quadratic in m) on the number of columns of A. The proofs will no doubt use a multiset approach (A. and Lu). There are many variations of this problem including considering specially structured families of forbidden configurations. We will have plenty of problems to work on. An interested student could contact undergrads Zachary Pellegrin or Kim Dinh who worked with me summers 2019 and 2020 respectively.

In this project, the student will implement, analyze, and validate a computational model for the development of dentition in reptiles. The model will be a quantification of a hypthesis first proposed in the mid 70s by J.W Osborn. We will collaborate with the Richman Lab in the Faculty of Dentistry at UBC who have been quantifying the patterning of tooth eruption across the jaw in leopard geckos. The student will carry out simulations of both deterministic and stochastic versions of the model and use nonlinear regression and maximum likelihood estimation to fit parameters using the Richman Lab’s data.

The group of Khanh Dao Duc at the Math department of the University of British Columbia (Vancouver, Canada) is seeking to recruit a USRA student, to participate in the development of mathematical and computational methods for investigating biological structures from cryogenic-electron microscopy (cryo-EM) data. For the following project, Dr. David Dynerman (Chan Zuckerberg Biohub, San Francisco, USA) will be co-supervisor. The group is also working in close collaboration with other research groups and scientists, including Professor Nina Miolane (ECE Department, UC Santa Barbara) and Dr. Frédéric Poitevin (LCLS, Stanford Linear Accelerator Center) so this position will also give the opportunity to learn more about the strong coupling between instrumentation in electron microscopy, advanced mathematical concepts and machine learning algorithms.

As the level of detail in 3D models produced by cryogenic-electron microscope (cryo-EM) continues to increase, one barrier faced by all commonly used software is assuming that experimental images are linear projections of the protein’s 3D density. This linearity assumption starts to break down at high resolution due to an electron optical effect called Ewald Sphere Curvature. We recently developed some basic theory on how to account for this curvature and directly recover 3D structures from 2D cryo-EM images without relying on the standard linear assumption. Humorously, our approach to correct this non-linear optical effect involves constructing and solving a large, but tractable, linear system. Our goal is to now run numerical simulations to prove that the method improves 3D model resolution. This project would involve inventing an efficient algorithm to construct this linear system as well as applying standard numerical analysis techniques to solve it. Once this prototype implementation is complete, we plan to analyze its robustness against noise and other perturbations as a next step to bringing these ideas into common use in the cryo-EM field.

Biologists have discovered remarkable patterns of cell migration in many essential biological processes, including development of embryos and wound healing. For example, the so-called neural crest cells migrate in streams along the spine of the embryos of chicks, frogs and zebrafish. Recently, our group built a model to explain cell polarization and migration in terms of the chemical signaling inside the cells. This has been used to study the cell-cell coordination in collective migration of neural crest cells, with interesting results. However, our simulations so far have been in two dimensions only, on a planar surface. The USRA project for summer 2021 extends the existing model to 3D to simulate how a cell migrates on a textured substrate, and on a 3D scaffold of collagen fibers. The project will require programming in Python and Rust, with training and help from a graduate student.

Minimization problems arise naturally in many branches of mathematics and science. For example, problems in navigation involve finding paths of least length (geodesics) on the earth's surface. Minimal surfaces, which are two-dimensional analogs of geodesics, are minimizers (or simply critical points) of the area function, and arise naturally in material science; for example in fluid interface problems and elasticity problems. A simple physical example of a minimal surface is the soap film that forms after dipping a wire frame into a soap solution. By the laws of surface tension this soap film has the property that it is stable, that is it becomes larger under slight deformations. The theory of minimal surfaces has had striking applications, for example to general relativity and low dimensional topology. This project will study connections between an eigenvalue problem on surfaces with boundary and minimal surfaces in the unit ball.

Canada has the third-largest crude oil reserve in the world behind Venezuela and Saudi Arabia. Canadian oil reserve is believed to include approximately 170 billion barrels of oil or 11% of total global oil reserves. The oil sands industry is both a significant contributor to the Canadian economy and is widely regarded as a cause of adverse environmental effects, e.g. it has been estimated to account for 10% of GHG emission in Canada.

Recent studies show that anaerobic microorganisms contribute to the degradation of Naphtha hydrocarbons and naphthenic acids in the FFT and MFT layers of oil sands tailings ponds, producing methane and CO2, both potential causes of GHG emissions. The FFT and MFT layers are colloidal suspensions, which behave like viscoplastic fluids with time-dependent rheology: changing both with age and depth in the pond. The key feature of a viscoplastic fluid is its yield stress: the material flows only if the imposed stress exceeds the yield stress. This raises questions regarding the stability of bubbles, which are trapped in a yield stress fluid, that we try to answer in this project through a series of targeted experiments.

We are seeking 2 students to work on two subprojects: (i) the growth and stability of bubble clouds in a viscoplastic fluid; (ii) the growth and stability of a single bubble or multiple bubbles introduced in a perfectly degassed fluid. This might lead to fundamental understanding of how the rheology of the fluid and interaction between the stress field around the bubbles affect the onset of motion. The students will assist in all operations related with the experiment: fluid preparation, rheometry measurements of the fluids, running experiments and image processing. The students will learn the physical background to the experiments and may help in design of new components and implement changes to the current apparatus.

To model the shape, motility, and behaviour of biological cells, a variety of computational platforms are commonly used. This project will be based on the cellular Potts model approach, whose origin is based on the physics of foams and soap bubbles. The student will be taught and then expected to use the open-source software (Morpheus) to simulate single and multiple interacting cells. The research will consist of the following (1) Deriving conditions for simple cell behaviour based on the underlying mathematics of the CPM. (Briefly: to use calculus to find minima of an energy function called the Hamiltonian for simple geometries.) (2) To test the analytic predictions using simulations and (3) To assist in developing simulations for interactions of (real) cells in several biological case-studies.

Many systems that evolve in time, for instance the firing of neurons, appear to be governed by a combination of dynamical (deterministic) and stochastic (random) causes. This project will involve the study of neuron models (see ref. [1]), for example the Morris-Lecar neuron model, using mathematical analysis together with numerical simulation (with Matlab, or Python, etc.). The models we consider will include ODEs, SDEs (stochastic differential equations: ODEs influenced by random noise), and possibly PDEs and SPDEs, depending on background and interest.

References:

[1]. P.E. Greenwood, L.M. Ward, Stochastic Neuron Models, Springer 2016 (free on-line)

Inference problems involving 3-dimensional point data (i.e., data where each observation is comprised of a collection of points in 3d-euclidian space) are abundant in data science with many application both in science and engineering. For example, the learning of force models from atomic configurations and the classification of 3-dimensional objects based on their geometric shapes can be formulated as a regression and classification task on 3D point data, respectively.
In order to perform inference on such 3D point data efficiently certain symmetries --induced by the Euclidian symmetry group and the type of predicted quantity-- should be directly built into the respective inference model. For example, an inference model used for the prediction of inter-atomic forces should be invariant under translation of the input configurations and covariant with respect to rotation of input configurations.
A wealth of techniques has been developed in recent years within the applied mathematics and material science community for the learning of force fields. The aim of this project is to use and extend some of these techniques in the form of new algorithmic implementations for use in a broader scope of applications (e.g., the above mentioned classification of geometrical shape data.)
The student should be interested both in theoretical aspects of the project as well as computational aspects pertaining to the implementation of the developed algorithmic approaches. Implementation of algorithms will likely require coding in PyTorch (a python API for deep learning) and/or Julia.

Geometry optimisation in molecular modelling typically involves the computation of stable equilibria, i.e. minima of a potential energy surface, or more generally transition states (saddle points) or transition paths. Minimizing computational cost is of course a general goal of numerical optimisation but the extremely high computational cost of accurate molecular potential energy models makes this even more critical. This project would initially focus on a mutually agreed upon mathematical aspect of geometry optimisation, for example the design of accelerated dynamical systems to compute saddle points, or the development of a molecular multi-grid algorithm. In a next step, the mathematical theory would be implemented in a prototype code. Finally, in collaboration with a modelling group, the new algorithms could be integrated into a general purpose molecular modelling software and tested on real-world modelling scenarios. The project would likely involve aspects of numerical optimisation, linear algebra, and ODEs. For further details and references see http://www.math.ubc.ca/~ortner/research/optimisation/

In compressed sensing and it's generalizations, one captures linear measurements of a signal (i.e., an n-tuple of real numbers) which has some known structure, i.e., the signal is known to belong to a particular subset of n-tuples of real numbers. The ability to recover the signal from these measurements depends on the invertibility of the linear measurements restricted to this subset, moreover stable recovery is implied if the linear measurements act as a near isometry restricted to this subset. In a recent paper, we gave a way of determining a number of random linear measurements sufficient to achieve this near isometry depending upon the signal structure and the tail bounds of the random measurements. In this USRA, we wish to show that the results of that paper are essentially unimprovable. This project involves an in-depth understanding of sub-Gaussian random variables and high-dimensional geometry; ideally it results in a publication. We believe that the load is most appropriate for two USRA students.

The Nullstellensatz is a foundational theorem in algebraic geometry proved by David Hilbert around 1890. The classical version (often covered in Math 423) is for polynomials with coefficients in an algebraically closed field. There are also variants for polynomials with real coefficients and more generally, with coefficients in a p-closed field. This project is inspired by the recent preprint
https://arxiv.org/abs/1911.10595
which proves a version of the Nullstellensatz for polynomials with quaternion coefficients. Here are the specific activities I have in mind:
1. learning existing versions of the Nullstellensatz for polynomials with commutative coefficients,
2. learning background material on p-closed fields and central simple algebras,
3. trying to generalize the above preprint.
Prerequisites: Strong background and interest in abstract algebra, preferably Math 423.

Tensor networks are diagrams (or graphs) that allow one to describe a complicated decomposition of a given tensor (multidimensional array). They are used in quantum physics since they represent good approximations of steady states of different quantum systems.
Since the edges in tensor networks generally signify physical proximity between quantum particles, it makes sense to consider tensor networks given by hypergraphs in dimensions higher than 1. In this project we are going to try constructing a meaningful version of the tensor network called MERA (Multiscaled Entanglement Renormalization Ansatz) in dimensions 2 and up, which uses hypergraphs and satisfies all the properties that 1 dimensional MERA does.

With the emergence of big data, information more and more often comes in the shape of a multi-dimensional array (or tensor). The importance of finding a decomposition of such an object is at least two-fold. First, it often uncovers additional information about the incoming data. For example, in the famous Netflix prize problem, finding the decomposition allows us to find different user types, and predict movie preferences. Second, and sometimes more important, the decomposition allows for efficient storage of the tensor.
A nonnegative tensor decomposition is a decomposition in which each of the factors is nonnegative. Such decompositions appear in many applications, including in hidden variable models, such as the model in the Netflix problem. Despite its usefulness, nonnegative tensor decomposition is an NP-hard problem.
In this project we are going to study the space of tensors of given nonnegative rank. It has been shown that tensors of nonnegative rank at most 2 correspond to probability distributions that are totally positive, a type of distribution that signifies strong dependence between random variables. A similar result has been shown for 2 X 2 X 2 tensors of nonnengative rank at most 3. Here, we will study larger tensors of nonnegative rank at most 3, as well as higher nonnegative ranks.

In this project we are going to study decompositions of tensors according to an orthogonal tensor network.

We will design algorithms for finding such decompositions, and study the properties of tensors that can decompose according to an orthogonal tensor network..

New measurement technologies like single-cell RNA sequencing are bringing ‘big data’ to biology. My group develops mathematical tools for analyzing time-courses of high-dimensional gene expression data, leveraging tools from probability and optimal transport. We aim to develop a mathematical theory to answer questions like: How does a stem cell transform into a muscle cell, a skin cell, or a neuron? How can we reprogram a skin cell into a neuron?
We model a developing population of cells with a curve in the space of probability distributions on a high-dimensional gene expression space. We design algorithms to recover these curves from samples at various time- points and we collaborate closely with experimentalists to test these ideas on real data. We have recently applied these ideas to shed light on cellular reprogramming (to learn more, see here: https://broadinstitute.github.io/wot )
We are looking for students with some background in optimization, probability, and mathematical programming. We are looking to hire several students to work on separate aspects of this large-scale project.

In biology, quorum sensing is an intercellular form of communication that bacteria use to coordinate group behavior such as biofilm production and the production of antibiotics and virulence factors. It is also responsibe for the mechanism underlying the onset of luminescence production in cultures of the marine bacterium Vibrio fischeri. Luminescence and, more generally, quorum sensing are important for V. fischeri to form a mutualistic symbiosis with a small Hawaiian squid, Euprymna scolopes. During the day the bacterial cells grow to a high density in the light organ of the squid, which then triggers by sunset a luminescence behavior that camouflages the squid at night from its predators by eliminating its shadow in thin pools of ocean water.
In this USRA project, the candidate will work together with members of my research group to analyze and perform computer simulations of various nonlinear ODE systems characterizing quorum sensing behavior in 3-D domains. These model ODE systems have been recently derived from an asymptotic reduction of more elaborate PDE/ODE systems of quorum sensing. The ideal candidate will have strong calculus skills, a good background in ODEs, some exposure to PDEs, strong computational skills in a scientific programming language such as MATLAB, as well as a willingness to learn some of the biological issues from reading some journal articles with members of my research group. Mathematically, the student will learn techniques in nonlinear ODE dynamics, some bifurcation theory, and be exposed to an interesting interdisciplinary application.
For two recent references from my group on this general topic see:
https://www.math.ubc.ca/~ward/papers/lux_new.pdf
https://www.math.ubc.ca/~ward/papers/qpde_revise.pdf

An elliptic curve over the rationals can be expressed via an equation in variables x and y that is cubic in x and quadratic in y. They play an important role in many areas of number theory and have been the subject of research for over a century. To each elliptic curve is associated a prime-indexed complex sequence that has interesting statistical properties, and a supersingular prime for an elliptic curve is then a prime at which the corresponding value in the sequence vanishes.
We can ask - how often do such primes arise? Their occurrence for a large family of elliptic curves has been conjectured by Lang and Trotter. We will investigate variations of this question further through computation and the study of heuristic models.
This project involves studying the background of elliptic curves, understanding the history of the problem, and working with some analytical machinery as well as the program SAGE (no prior experience needed) to gain further insight into the distributions of supersingular primes. A student would benefit from a background in algebra and complex analysis.

We want to investigate reaction-diffusion systems with nonlocal diffusions. Nonlocal diffusions are ubiquitous in nature. In this project we will study several prototype fractional Gierer-Meinhardt, Schnakenberg, Gray-Scott, Brusselator systems. One difficulty is that these are quite nonlocal problems and even the definition requires singular integrals. We will study the effects of different fractional indices, different boundary conditions, combinations of nonlinearities on the existence and stability of localized patterns.

The phenomenon of Anderson localization for random Schrödinger operators is well established in low dimensions or strong disorder. One of its facets is the fact that eigenvectors for eigenvalues at the bottom of the spectrum are well localized in space. However, the determination of the localization centers of these eigenvectors is a difficult problem. The use of an effective potential has recently been advertised and shown to be numerically very effective. This project will explore analytical aspects of the relation between localization and the effective potential with an eye towards a new proof of localization via the effective potential.

The classical gambler's ruin problem can be thought of as a random walk stopped upon exiting an interval. Since simple random walk on Euclidean lattice approximates Brownian motion on the Euclidean space, one might naturally expect that simple random walk killed upon exiting a set approximates Brownian motion killed upon exiting a continuous domain. In this project, we consider simple random walks killed (or stopped) upon exiting a domain. The goal of this project is to obtain quantitative estimates that compare transition probabilities of random walk and Brownian motions killed upon exiting (a  large class of) domains. The student will learn tools from analysis (Whitney decomposition, singular integrals) and probability (Doob transform, heat kernel bounds).

The translation of mRNA into protein is a fundamental, yet complex biological process, mediated by ribosomes. To explain what can affect its efficiency, it is crucial to unravel the interactions between the ribosomes and other molecular complexes, but also to take into account other factors at a larger scale. The main goal of this proposal is to draw a global picture of the role played by ribosomes in translational systems across different scales, encompassing molecular, cellular and evolutionary aspects. More specifically, we shall investigate 1) the biophysical properties and evolution of the ribosome exit tunnel, 2) different modes of translation, transport and remodeling of the ribosome in response to specific spatial cellular organizations and 3) the limiting factors that drive protein translation at the system level. These completing approaches will both elucidate the functional impact of the ribosome structure, and conversely, determine how spatial or resources management impose evolutionary and design constraints at the molecular level.

We will analyze the biophysical properties of the tunnel from cryo EM data, and elucidate the interplay between the electrostatics and geometry of the exit tunnel. These biophysical properties will also be studied in the context of evolution, with cryo EM structures of the ribosomes available for many species. Studying the evolution of the ribosome and its tunnel will require inferring ancestral shapes that can potentially explain different modes of translation.

At the mesoscopic scale, we will focus on two important modes of translation: First, the translation of membrane protein genes, which involves transport to the endoplasmic reticulum (ER). Interestingly, this gives rise to geometric patterns of polyribosomes on the ER membrane. To explain these patterns, we will study a new biophysical model of translation, and compare the patterns with imaging data. The second local mode of translation to investigate occurs in dendritic regions, located far from the cell nucleus. By combining theoretical analysis of modes of transport, with differential expression data, we will build statistical tools to distinguish, for different genes, their mode of transport, and infer the associated local translation dynamics. At the system scale, we are interested in the metabolic cost of translation. In vitro timeseries measurements of protein levels plateau, with multiple potential limiting factors. Upon fitting a mathematical model to a various set of experimental conditions, we will disentangle the contribution of these factors. This modeling approach will serve as a first step for high throughput measurement of translation rates. In vivo, living systems also need to manage the ribosomal population. In particular, "ribophagy" pathway can not only decrease the ribosomal pool, but also allows the recycling of ribosomal components. To assess the robustness and optimality of this pathway, we will use optimal control.

The study of homological invariants of knots has benefited from the existence of knot tables (see). This gives a collection from which to draw examples, compute, and gain intuition about the invariants in question. This has been the source of new conjectures and, ultimately, structure related to these novel invariants. Knots are sometimes studied by decomposing them into simpler pieces called Conway tangles. In the case of knot Fleor homology and Khovanov homology, we now have a good idea of the form these invariants take when evaluated on a Conway tangle. However, Conway tangles are more difficult to tabulate than knots, and as a result there is not a table of Conway tangles that researchers can work with. This project will take a step in this direction: we’ll restrict to Conway tangles of a particular form, and systematically build a table of these special examples. With this in hand, appealing to existing software, we will compute the aforementioned invariants. This project will require some algebraic background and an interest in topology. Familiarity with Python and or C++ will be very useful. 

A problem in extremal combinatorics asks how many (discrete) objects are possible subject to some condition. We consider simple matrices which are (0,1)-matrices with no repeated columns. We say a simple matrix A has F as a configuration if a submatrix B of A is equal to a row and column permutation of F. Much has been determined about the maximum number of columns an m-rowed simple (0,1)-matrix can have subject to condition it has no configuration F for some given F configuration. There is an attractive conjecture (A. and Sali) to pursue with some challenging open problems. The bounds we obtain are sometimes exact, sometimes asymptotic (Theta(f(m))) and sometimes the coefficient of the leading term in the bound is exact. A variation is to consider a specially structured family of forbidden configurations. Under the supervision of Richard Anstee, we will have plenty of problems to work on.

Experimental biologists have acquired lots of data on the distribution and mobility of B cell receptors (BCR) on the surface of B cells of the immune system. The goal of this project will be to build and parameterize a suite of stochastic mathematical models that describe the mobility and organization of BCR. We will design the models based on reasonable biophysical assumptions and with careful reference to existing models in the literature. The models will be implemented as stochastic computer simulations in Matlab or Python. The parameters will then be fit to the available data using Approximate Bayesian Computation approaches (ABC). The parameterized models from this project will lead to improved quantitative understanding of B cell signaling, which may ultimately impact our understanding of a range of diseases. Students working on this project with Dan Coombs will learn about B cell biology, cell biophysics, image analysis, modeling and scientific computation. There may be an opportunity to assist with hands-on experimental work as well.

B cell receptors (BCR) are mobile on the surface of B cells. In this primarily experimental project, a student will use two-colour fluorescence microscopy to study the motion of individual receptors either in the presence of a bulk-labeled background of receptors or other surface molecules, or in relation to the motion of other nearby receptors. In the first case, the goal will be to study correlations between the mobility parameters of the focal receptor and the density of labelled background. In the second case, correlations will be sought between the mobility parameters of nearby receptors. The experiments will require culture and labeling of cells, imaging, image analysis and then fitting to mathematical models of receptor motion. Students working on this project with Dan Coombs will learn about B cell biology, microscopic imaging and image analysis, modeling and scientific computation.

The development of multicellular organisms is coordinated by both mechanical and biochemical regulation. A recent model of tissue development that coupled mechanical and biochemical regulation demonstrated a surprisingly large variety of dynamical behaviours. Preliminary analysis indicates that the model has a rarely seen co-dimension two bifurcation that is still in need of elucidation. This project will require a student to carry out numerical bifurcation analysis of the model using both existing bifurcation software (like XPPAUT) and some self-written code (in matlab / python). Some background and an interest in learning more about ODEs (MATH 215), bifurcation theory (MATH 345), and numerical methods would be useful.

By 2013, more than 550,000 oil and gas wells had been drilled across the country with the majority located in the Western Canada Sedimentary Basin with 5,000-20,000 new wells drilled each year. After a certain operating period, all wells will need to be plugged and abandoned (P&A). As of December 2016, there were 24,802 wells in British Columbia. Of these, 11% were abandoned and 27.4% were inactive.

In Alberta, 151,000 wells have been abandoned which represents 35% of all wells in the province. The consequences of poor P&A are wellbore leakage which can contaminate groundwater, near-surface ecology and damage to marine ecosystems offshore. Wellbore leakage also contributes to greenhouse gas emissions, may present explosive and/or health risks. P&A is a potential massive societal and environmental problem for Canada.

Wellbore leakage occurs through leakage pathways located within the cement sheath which surrounds the casings. Many operational factors could contribute to the formation of these leakage pathways such as poor primary cementing or repeat pressurization of the casing for the purpose of hydraulic fracturing, just to name a few. Primary cementing can be affected by things such as well orientation, number of cementing stages, well depth, etc.

We are seeking 2 students to work with Ian Frigaard on two streams defined to study this process.

First stream will investigate this process from a fluid mechanics properties. And the second stream aims to get a better understanding of well leakage through the modeling of leakage pathways based on statistical data.

A) Experimental investigation of annular displacement

In the field, after the well is drilled, the casing is lowered into the open well. In this point, the space inside and outside the casing is occupied by the drilling mud that keeps the hydrostatic balance between the hole and the formation. Then, cement slurry is pumped downwards inside the casing, reaches the casing's bottom, and flows up into the annular section displacing the drilling mud upwards. A good seal will not leave residual mud anywhere.
In the lab, we use two flow loops to simulate the field process. We have carefully designed and built the loops to achieve dynamic similarity. We can control the key parameters of the process, such as flow rate, eccentricity, rheology, and fluid's densities. The data acquisition is through imaging with high sensitivity cameras and automated instrumentation. The objective is to capture experimental data relevant to theoretical predictions of the fluid-fluid displacement flows under a wide variety of scenarios.

1. What You Will Do:
The student will perform some combination of experimental work. The student will assist in all operations related with the experiment: fluid preparation, running experiment, image processing of the data, rheometry measurements of the fluids and data analysis. The student will learn the physical background to the experiments and may help in design of new components, undertake bits of machining/manufacturing, and implement changes to the current apparatus

2. Supervision Received:
The graduate student mentor will support the student on a daily basis, as will another PhD student involved in the project. Professor Ian Frigaard will facilitate a number of group meetings.

3. Skills for Success:
Active listening, communication, creative thinking, critical thinking, problem solving. Basic programming and machining skills. Interest in fluid mechanics.

B) Data analysis - Oil & Gas Well Data

The research project aims to get a better understanding of well leakage through the modeling of leakage pathways based on statistical data. Provincial regulator such as the British Columbia Oil & Gas Commission (BCOGC) and the Alberta Energy Regulator (AER) keep large databases concerning various well aspects. The student would work on extracting relevant information from these databases and analyzing the data in ways which could contribute to leakage pathways statistical modeling or modeling of well geometry for primary cementing purposes.

1. What You Will Do:
The student will perform some data analysis using either MATLAB or R or other useful software. The student should be able to handle large amount of data. The student should have a general interest in fluid mechanics and be creative and critical thinking, self-taught individual. Programming abilities, i.e.; ability to write code to extract text from PDF is really essential.

2. Supervision Received:
The graduate student mentor will support the student on a daily basis, as will another PhD student involved in the project. Professor Ian Frigaard will facilitate a number of group meetings.

At least one USRA position is offered for talented undergraduate(s) in the group of Leah Keshet to combine mathematical modeling and computational biology. Interested students will be working with other group members on several projects related to cell motility and the collective behaviour of cells in a tissue. Model outcomes will be compared to both controlled single cell experiments (e.g. cells in microfluidic devices), and experiments in full tissues (e.g. developing chick). Image processing is used to extract data from experimental images.

Students will use computational software (such as the open-source package Morpheus), contribute to writing code (guided by a postdoctoral fellow) and to running and testing codes for a variety of in-silico experiments.

Students with a computational background and familiarity with Python, MatLab, C++, and similar programming languages, as well as an interest in applied mathematics / mathematical biology are encouraged to apply.

A Sidon set is a set whose pairwise sums a+b are all distinct (other than the trivial a+b = b+a). If we let R(n) denote the size of the largest Sidon subset of {1, 2, ..., n}, then it is known that R(n) is roughly νn for large enough n. Erdös asked whether it is possible that | R(n) - νn | is bounded; such an unlikely result would indicate some extreme regularity to Sidon sets. The student will work with Greg Martin, where we propose to settle this question in the negative. Most constructions of large Sidon sets proceed by constucting a Sidon set modulo some integer q (for example, the construction might take place in a finite field) and then noting that any Sidon set (mod q) remains a Sidon set when considered as a set of integers. Moreover, the (mod q) construction is usually preserved under affine transformations x --> ax+b (mod q), a property that can be detected using exponential sums. If one of these affine transformations ends up moving the whole Sidon set so that it does not intersect the interval [m, q], then the resulting Sidon set is still large but is now a subset of {1, 2, ... , m}, resulting in a large enough value for R(m) to answer Erdös's.

With the emergence of big data, information more and more often comes in the shape of a multi-dimensional array (or tensor). The importance of finding a decomposition of such an object is at least two-fold. First, it often uncovers additional information about the incoming data. For example, in the famous Netfix prize problem, finding the decomposition allows us to find different user types. Second, and sometimes more important, the decomposition allows for efficient storage of the tensor.
Despite its usefulness, tensor decomposition remains a hard problem, both computation-ally and statistically. We will focus on its computational aspects. It has been shown that decomposing a general tensor or finding any of its eigenvectors is NP-hard. Although these are hard problems for general tensors, solving them efficiently for specific families is possible.
Elina Robeva aim to hire two students focusing on two different families of tensors: tensors decomposing into an orthogonal tensor network and incoherently decomposable tensors. For each family we will aim to design algorithms for exact decomposition, give an efficiently checkable criterion for when a tensor lies in the family, and describe the eigenvalues and eigenvectors of a tensor in the family.

New measurement technologies like single-cell RNA sequencing are bringing 'big data' to biology. My group develops mathematical tools for analyzing time-courses of high-dimensional gene expression data, leveraging tools from probability and optimal transport. We aim to develop a mathematical theory to answer questions like How does a stem cell transform into a muscle cell, a skin cell, or a neuron? How can we reprogram a skin cell into a neuron?
We model a developing population of cells with a curve in the space of probability distributions on a high-dimensional gene expression space. We design algorithms to recover these curves from samples at various time-points and we collaborate closely with experimentalists to test these ideas on real data. We have recently applied these ideas to shed light on cellular reprogramming (to learn more, see here)
We are looking for students with some background in optimization, probability, and mathematical programming. We are looking to hire several students to work on separate aspects of this large-scale project.

The time dependent solution of a large class of Fokker-Planck equations for the distribution functions of electrons and/or reactive species can be obtained numerically with an efficient pseudospectral method defined with non-classical basis polynomials. The solutions are expressed in terms of the eigenfunctions and eigenvalues of the linear Fokker-Planck equation. The Fokker-Planck eigenvalue problem is isospectral with the Schroedinger equation so that the pseudospactral methods developed can be applied to both eigenvalue problems. There are large number of quantum problems of current interest that involve the Yukawa, Wood-Saxon and Hulthen potentials.

There are well defined projects for 2-4 undergraduate summer students working with Bernie Shizgal. A publication on the Fokker-Planck equation with a previous undergraduate student was recently published and can be found at the link;

A publication on the Schroedinger equation with a previous undergraduate student was recently published and can be found at the link;

The pseudospectral methods have been described in the book "Spectral Methods in Chemistry and Physics", Springer 2015

Description: A knot is a non-intersecting closed curve in a three-dimensional space, considered up to deforming the curve. Using computer calculations we will investigate properties of modular knots, a family of knots (more precisely) highlighted by É. Ghys.

Interested students should have programming experience. Background in combinatorics or topology is a plus.

In biology, quorum sensing is an intercellular form of communication that bacteria use to coordinate group behavior such as biofilm production and the production of antibiotics and virulence factors. It is also responsible for the mechanism underlying the onset of luminescence production in cultures of the marine bacterium Vibrio fischeri. Luminescence and, more generally, quorum sensing are important for V. fischeri to form a mutualistic symbiosis with a small Hawaiian squid, Euprymna scolopes. During the day the bacterial cells grow to a high density in the light organ of the squid, which then triggers by sunset a luminescence behavior that camouflages the squid at night from its predators by eliminating its shadow in thin pools of ocean water.

In this USRA project, the candidate will work together with my two members of my research group to analyze and perform computer simulations of various nonlinear ODE systems characterizing quorum sensing behavior in 3-D domains. These model ODE systems have been recently derived from an asymptotic reduction of more elaborate PDE/ODE systems of quorum sensing. The ideal candidate will have strong calculus skills, a good background in ODEs, some exposure to PDEs, strong computational skills in a scientific programming language such as MATLAB, as well as a willingness to learn some of the biological issues from reading some journal articles with members of my research group. Mathematically, the student will learn techniques in nonlinear ODE dynamics, some bifurcation theory, and be exposed to an interesting interdisciplinary application.

Arrays of boxes filled with positive integers subject to certain rules are called Young tableaux. They generate functions known as Schur functions, which date from Cauchy in 1815, and arise today in many areas including combinatorics, algebra, quantum physics and algebraic geometry. In this project we will investigate a natural generalization of Schur functions and their associated Young tableaux, with the aim of discovering combinatorial rules for them, such as product rules. We may also investigate connections to other areas.
A strong background in abstract algebra such as Math 322 or Math 323 is required, and an aptitude for combinatorics and programming skills will be an asset.

At least one USRA position is offered for talented undergraduate(s) in the group of Leah Keshet to combine mathematical modeling and computational biology. Interested students will be working with other group members on several projects related to cell motility and the collective behaviour of cells in a tissue. Students will use computational software (for example: Cellular Potts Model), contribute to writing code (guided by a postdoctoral fellow) and to running and testing codes for a variety of conditions. Students with a computational background and familiarity with Python, MATLAB, C++, and similar programming languages, as well as an interest in applied mathematics will be seriously considered.

Required skills: Computation using OpenFOAM

By 2013, more than 550,000 oil & gas well had been drilled in Canada. Before hydrocarbons can be produced every well undergoes primary cementing. This operation consists on sealing the annular section between the steel pipe that stabilizes the well (named casing), and the rock formation. The seal should increase production and prevent subsurface fluids from percolating to surface. Nevertheless, gas leakage to surface is common. A public perception is that 10-20% of wellbores leak, which has both health & safety consequences and environmental/ecological impact. This project will study this process from a fluid mechanics perspective.

In the field, after the well is drilled, the casing is lowered into the open well. In this point, the space inside and outside the casing is occupied by the drilling mud that keeps the hydrostatic balance between the hole and the formation. Then, cement slurry is pumped downwards inside the casing, reaches the casing’s bottom, and flows up into the annular section displacing the drilling mud upwards. A good seal will not leave residual mud anywhere.

In the lab, we use two flow loops to simulate the field process. We have carefully designed and built the loops to achieve dynamic similarity. We can control the key parameters of the process, such as flow rate, eccentricity, rheology, and fluid’s densities. The data acquisition is through imaging with high sensitivity cameras and automated instrumentation. The objective is to capture experimental data relevant to theoretical predictions of the fluid-fluid displacement flows under a wide variety of scenarios.

We are seeking 2 students, who will perform some combination of experimental work and associated computations with the group of Ian Frigaard. Experimentally, the students will assist in all operations associated with the experiment: fluid preparation, running experiment, image processing of the data, rheometry measurements of the fluids and data analysis. The students will need to understand the physical background to the experiments and may need to help in design of new components, undertake bits of machining/manufacturing, and implement changes to the current apparatus. Computationally, the students will run and analyze selected simulations using OpenFOAM.

Interested students should contact Alondra Renteria alondra.renteria.ruiz@gmail.com and Ali Etrati etrati@alumni.ubc.ca with resume and transcript. The intention is that the successful students be funded under the NSERC USRA program, so grades should be good!

Khovanov homology, in its original form devised by Khovanov about 20 years ago, is a combinatorially defined homology theory for knots and links. Bar-Natan extended this theory to tangles using a certain cobordism category, more precisely a category whose objects are crossingless tangles and whose morphisms are embedded surfaces that interpolate between such tangles. This made calculations of Khovanov homology much more efficient. Very recently, Kotelskiy-Watson-Zibrowius observed that if one specializes to tangles with four ends, a certain version of Bar-Natan's theory can be interpreted in terms of immersed curves on the 3-punctured disc. This is particularly interesting in the light of similar interpretations of Heegaard Floer and instanton homology.

This project has one main goal: compute the immersed curves associated with as many 4-ended tangles as possible. A computer program to find the immersed curves from Bar-Natan's invariant has already been written by Zibrowius. What is missing is a program that computes Bar-Natan's invariant in this setting. One option would be to adapt one of the already existing implementations of Khovanov homology. Alternatively, it might be easier to just write an implementation of the algorithm from scratch.

Prerequisites for this project with Liam Watson are an excellent grade in a first course in algebraic topology (covering classification of surfaces, simplicial or singular homology and the Euler characteristic) as well as some experience in programming, ideally python and/or C++. Java would be required if we decide to adapt existing implementations of Bar-Natan's theory.

The matrix completion problem is as follows: Given a subset of entries of a low-rank matrix, the goal is to fill in the missing entries by leveraging the low-rank structure. There are competing convex optimization programs for this, but while one has the strongest theoretical backing (nuclear-norm minimization), another seems to work better in practice (max-norm minimization). This project with Yaniv Plan studies the difference both theoretically and through numerical simulations. A key will be to understand the underlying geometry of the corresponding convex sets.

The transpose operation makes the ring of n x n complex matrices into an algebra-with-involution. That is, the ring is an algebra over the complex numbers, and is equipped with an additive self map A -> A* that reverses the order of multiplication, and so that A**=A. We will say that an r-tuple (A1, A2, ... , Ar) of n x n matrices _generates_ this algebra if no strict subalgebra-with-involution contains every element of the r-tuple. Let X(n,r) denote the space of all such generating r-tuples. This project will study the topology of the space X(n,r) and of a related space Y(n,r) where the r-tuples are considered only up to an involution-preserving change of basis. It will study both particularly in the case of small values of n (and r) where it is possible to understand the spaces completely.

The project with Ben Williams will require a strong understanding of linear algebra, knowledge of homology will be highly desirable. Some commutative algebra or algebraic geometry will also be helpful.

In some imaging applications (X-ray crystallography, for example) the resulting image essentially captures the magnitude (amplitude) of the Fourier Transform of a function of interest. However, the Fourier transform is complex-valued, so information (the phase) is lost.

The problem can be rephrased as an optimization problem -- of selecting a candidate function whose Fourier transform best matches the observed signal. This project with Lior Silberman will investigate whether techniques of regularized optimal transport can be brought to bear on this problem by quantifying distances and gradients in the space of candidate functions.

Background in real analysis and some programming experience are required.

A problem in extremal combinatorics asks how many (discrete) objects are possible subject to some condition. We consider simple matrices which are (0,1)-matrices with no repeated columns. We say a simple matrix A has F as a configuration if a submatrix B of A is equal to a row and column permutation of F. Much has been determined about the maximum number of columns an m-rowed simple (0,1)-matrix can have subject to condition it has no configuration F for some given F. There is an attractive conjecture (A. and Sali) to pursue with some challenging open problems. We will pursue some 4-rowed choices for F that when forbidden in A should yield a quadratic bound on the number of columns of A. There are many variations of this problem including considering specially structured families of forbidden configurations. We will have plenty of problems to work on.

Interested students could contact my previous USRA students Santiago Salazar and Jeffrey Dawson and Cindy Tan.

A student might work on a problem from the following small book which is available on line. The background needed is some knowledge of simple systems of ode's and some probability. There are problems at many levels.

[BOOK] Stochastic neuron models PE Greenwood, LM Ward - 2016 - Springer In this book we describe a large number of open problems in the theory of stochastic neural systems, with the aim of enticing probabilists to work on them. These include problems arising from stochastic models of individual neurons as well as those arising from stochastic models of the activities of small and large networks of interconnected neurons. We sketch the necessary neuroscience background to these problems so that probabilists can grasp the context in which they arise. This project with Priscilla Greenwood focuses on the mathematics of the models and theories.

Elliptic curves are a fundamental object in number theory. The problem of cataloguing elliptic curves (up to various symmetries) is a difficult computational problem and continues to attract significant attention. Recently Bennett developed a connection between elliptic curves over Q and solutions of a family of Diophantine equations called Thue equations. These equations are very difficult to solve rigorously, however by exploiting properties of continued fractions, one can solve them heuristically. In this project we will investigate the link between solutions of Thue equations and continued fractions over other fields - such as the Gaussian integers.

The group of Leah Keshet offers placement for USRA students with computational experience to work on problems in mathematical biology. Our group is interested in modeling cell shape, cell motility, tissue migration and other phenomena associated with developmental and cell biology. We currently work on understanding the behavior of normal and malignant cells, as well as intracellular chemicals that influence cell shape, adhesion, contractility, and motility. .

The student will be in charge of refining and running publicly availble software (Compucell3D, Morpheus) or helping with development and testing of new software, to simulate collective cell behaviour corresponding to mathematical models.

Students with engineering-math or CS-math double major are especially encouraged to apply. Particulaly promising students who succeed at the USRA research will have an opportunity of being strongly considered for future MSc graduate studies with the Keshet group.

A problem in extremal combinatorics asks how many (discrete) objects are possible subject to some condition. We consider simple matrices which are (0,1)-matrices with no repeated columns. We say a simple matrix A has F as a configuration if a submatrix B of A is equal to a row and column permutation of F. Much has been determined about the maximum number of columns an m-rowed simple (0,1)-matrix can have subject to condition it has no configuration F for some given F. There is an attractive conjecture (A. and Sali) to pursue with some challenging open problems. A variation is to consider a specially structured family of forbidden configurations. We will have plenty of problems to work on.

Interested students could contact my previous USRA students Santiago Salazar and Jeffrey Dawson.

The time dependent solution of a large class of Fokker-Planck equations for the translational distribution functions of electrons and/or reactive species can be obtained numerically with an efficient pseudospectral method defined with non-classical basis polynomials. The solutions are expressed in terms of the eigenfunctions and eigenvalues of the linear Fokker-Planck equation. The Fokker-Planck eigenvalue problem is isospectral with the Schroedinger equation so that the pseudospactral methods developed can be applied to both eigenvalue problems. There are large number of quantum problems of current interest that involve the Yukawa, Krazner, Pseudo-harmonic, Wood-Saxon and Hulthen potentials.

There are well defined projects for up to four undergraduate summer students working with Bernie Shizgal. A publication with two undergraduate students was recently published; Shizgal, Ho and Zang, “The computation of radial integrals with nonclassical quadratures for quantum chemistry and other applications”, J. Math. Chem. 55, 413 (2017). The pseudospectral methods have been described in “Spectral Methods in Chemistry and Physics”, Springer 2015

Biologists have discovered remarkable patterns of collective cell migration during early development of animal embryos. For example, the so-called neural crest cells (NCC) migrate in streams along the spine of the embryos of chicks, frogs and zebrafish. Recently, our group built a model to explain NCC collective migration in terms of the chemical signaling inside the cells and between the neighbors.

The project for summer 2018 extends the existing model to explore an intriguing phenomenon: how NCC cells negotiate obstacle courses, as they are shown to be able to do in experiments. The student will generalize our current ODE-based model to account for chemoattraction and filopodial sensing. The project will require MATLAB and Python programming.

We seek a summer student interested in theoretical molecular biophysics. The goal of the project is to understand how baculovirus can break the protein gel filling the pores of the nuclear pore complex and hence enter the cell nucleus. This will be done using molecular simulations.

A background in biophysics and experience in programming/scientific computing (python, C, Linux OS etc) are required. This is a joint project between Prof. Joerg Rottler (Physics) and Prof. James Feng (Mathematics)

The proposed research centres on the problem of tabulating elliptic curves over quadratic fields, to test various conjectures (Birch, Swinnerton-Dyer, Modularity, etc.) and to complement work done in the LMFDB (the London Database of L-function calculations). The work would have a definite computational flavour, but could otherwise be tailored to fit the skill-set of the student, involving classical invariant theory (as used to great effect recently by Bhargava), algebraic number theory and complex analysis. It is likely that code would be produced, in one or more of the standard computational algebra packages (Sage, Magma, Pari, etc). Mike Bennett

Oil & gas wells are sealed by a process called primary cementing, in which sequences of fluids are pumped and displace one another along an annular flow path. The fluids are non-Newtonian, have different densities and other rheologies. At UBC we are studying these flows by both experimental means and by modelling/simulation. For this project we seek a motivated individual to help in running fluid flow experiments within our custom-built flow loops, participating in some construction, design and calibration tasks, possibly running some numerical simulations etc..

The applicant needs to have practical skills, a good physical sense of fluid flows and proficiency in a lab setting. This could be suitable for an Engineering or Science student with some experience and interest in fluid mechanics, plus physical intuition. Ian Frigaard

We will investigate numerically the spectral behaviour of groups defined through topology. In a recent breakthrough Kaluba--Nowak--Ozawa recently showed that computational techniques can show an eigenvalue gap in the actions of a discrete group, the outer automorphism group of the free group F_5 (more precisely, the group Out(F_5) has "Kazhdan Property (T)").

A related family of groups, "mapping class groups", are connected to the geometry and topology of surfaces, and it is not known whether they have Property (T) or not. We will investigate this question numerically.

Background in group theory and linear algebra (at the level of Math 223,322) is required. Lior Silberman

In some imaging applications (X-ray crystallography, for example) the resulting image essentially captures the magnitude (amplitude) of the Fourier Transform of a function of interest. However, the Fourier transform is complex-valued, so information (the phase) is lost.

The problem can be rephrased as an optimization problem -- of selecting a candidate function whose Fourier transform best matches the observed signal. We will investigate whether techniques of regularized optimal transport can be brought to bear on this problem by quantifying distances and gradients in the space of candidate functions.

Background in real analysis and some programming expeirence are required. Lior Silberman

A graph, G, is a set of dots, known as vertices, connected to each other by a set of lines, known as edges. A proper colouring of G is a colouring of the vertices such that no two vertices joined by an edge are the same colour. If we are given k colours then the number of ways of colouring G, as a polynomial in k, is known as the chromatic polynomial. If we have infinitely many colours then the chromatic polynomial generalizes to the chromatic symmetric function. This project will aim to generalize the chromatic symmetric function yet further, with the goal of establishing its basic properties, and trying to resolve well-known conjectures regarding the chromatic symmetric function. One such example is that it is known that if G is a tree, that is a graph with no set of edges forming a cycle, then the chromatic polynomial is the same for every tree with the same number of vertices. However, it is conjectured that two trees are non-isomorphic if and only if they have distinct chromatic symmetric functions. This project will be under the supervision of Stephanie van Willigenburg and Samantha Dahlberg.

A strong background in abstract algebra such as Math 322 or Math 323, an aptitude for combinatorics, and programming skills will be an asset.

This project, which will be jointly supervised by one or more professors in a biology department, will involve computational analysis of multicolour microscopic images of cell surface receptors, likely on B cells. This project will require good programming skills (using MATLAB or Python) and willingness to work closely with the experimental team.

Students could be from Math, CompSci, (Bio)Physics or Engineering.

Minimization problems arise naturally in many branches of mathematics and science. For example, problems in navigation involve finding paths of least length (`geodesics') on the earth's surface. Minimal surfaces, which are two-dimensional analogs of geodesics, are minimizers (or simply critical points) of the area function, and arise naturally in material science; for example in fluid interface problems and elasticity problems. A simple physical example of a minimal surface is the soap film that forms after dipping a wire frame into a soap solution. By the laws of surface tension this soap film has the property that it is stable, that is it becomes larger under slight deformations. The theory of minimal surfaces (and submanifolds) has had striking applications, for example to general relativity and low dimensional topology.

This project will involve studying existence of minimal surfaces, properties of minimal surfaces, and applications. Ailana Fraser

A problem in extremal combinatorics asks how many (discrete) objects are possible subject to some condition. We consider simple matrices which are matrices of integer entries with no repeated columns. We say a simple matrix A has F as a configuration if a submatrix B of A is equal to a row and column permutation of F. We typically restrict our attention to (0,1)-matrices and much has been determined about the maximum number of columns an m-rowed simple (0,1)-matrix can have subject to condition it has no configuration F for some given F. We will consider variations where the simple matrices are restricted to have entries in {0,1,2} and have some conjecture to explore arising from joint work with Attila Sali.

Interested students could contact my previous USRA students Foster Tom, Maxwell Allman, Farzad Fallahi, and Santiago Salazar.

Biologists have discovered remarkable patterns of collective cell migration during early development of animal embryos. For example, the so-called neural crest cells (NCC) migrate in streams along the spine of the embryos of chicks, frogs and zebrafish. Moreover, NCCs from different sources manage to stay unmixed while migrating side by side. Later, they seem to be directed to different destinations along the spine, and then toward the front of the body, where they form various tissues and organs.

There are several mysteries about the collective migration. How do cells interact with each other to maintain cohesion among those from the same source, while keeping a boundary between cell clusters from difference sources? How do the cells decide where to stop or turn into a different route? The intensive efforts by biologists have produced some hypotheses. But as these questions involve the intimate coupling between biochemical signaling and cell mechanics, answering them requires the help of quantitative analysis.

We have developed a mathematical model and numerical techniques for modeling the interaction and collective migration of neural crest cells. For this summer project, the student will generalize our existing tool to study two scenarios of interest: (i) Given an externally imposed gradient of chemoattractant, how does a cluster of NCCs behave? (ii) What chemical and geometric cues may guide different streams of NCCs into different migration paths? Our model involves ODEs describing the dynamics of the signaling molecules and molecular motors, as well as ODEs or PDEs governing the mechanical behaviour of the cells and tissues. The computer program includes MATLAB and Python programming.

Displacement of one fluid by another of different properties is a common process in many industrial applications. Our focus is in displacements of miscible fluids in inclined ducts, where density and viscosity difference exists. Depending on the fluid properties, flow rates and inclinations, different flow regimes and behaviours are observed. For this project we seek a motivated individual to help in running a large number of numerical simulations, covering a wide range of parameters in displacement flows. The simulations are in a 2D channel with high aspect ratio and are written using PELICANS, an open-source C++ package. The simulations will be sent to WestGrid and the results will be post-processed using existing MATLAB codes.

The applicant will need to set the correct parameters for each case, submit the jobs to WestGrid, collect and organize the data, and help with post-processing the results. Other data management tasks may also be included. Experience with Linux, MATLAB and knowledge of C++ is an advantage.

This could be suitable for an Engineering or Science student with some experience and interest in fluid mechanics, plus physical intuition. Ian Frigaard

Oil and gas wells are cemented to reduce environmental leakage, to strengthen/support them structurally and to zonally isolate production zones in the reservoir. A model of this process has been developed at UBC consisting of 2D hydraulic calculations of viscoplastic fluids in a narrow channel coupled with a vector concentration equation. The model is implemented and solved in MATLAB using finite difference method. For this project we aim to validate this model with experimental/field data that we have available and the main focus will be to aid in this validation. The work will include running different cementing scenarios, post-processing the outcomes and comparing with the experimental data in an intelligent way.

An ideal candidate: is proficient in MATLAB (reading and compiling MATLAB .m files, troubleshooting MATLAB programs), has strong mathematical background (multi-variable calculus, ordinary and partial differential equations), is familiar with numerical algorithms (Numerical derivatives, numerical integration, root finding methods), can demonstrate strong verbal communication, problem solving and critical thinking.

This could be suitable for an Engineering or Science student with some experience and interest in fluid mechanics, plus physical intuition. Ian Frigaard

We are looking for a student to perform computations in linear algebra and finite group actions which will contribute to a project in the representation theory of p-adic groups. The only required background is linear algebra and group theory (say at the level of Math 223 and Math 322). Familiarity with computer algebra systems such as SAGE is desirable but not required. No knowledge of of more advanced mathematics is needed for the USRA project.

The student will have the opportunity to learn some of the advanced mathematics underlying the project as a whole. Lior Silberman

The behaviour of cells in a tissue depends on their chemical and mechanical interactions, and on stimuli they receive from their environment. Our group studies mathematical models for intracellular signaling and its effect on cell shape, motility, and tissue dynamics. Publicly available software exists for simulating such cellular systems. This project will consist of adapting such software (CompuCell3D, Morpheus, CHASTE, and others) to the specific models for signaling studied by members of our group.

This project is suited for CPSC, MATH or PHYS majors who can demonstrate ability to work on large open-source projects as part of an interdisciplinary team. Specifically, we are looking for motivated students with experience in test-driven development, object-oriented programming (C++, Python or MATLAB) and Linux/Unix server administration. Applicants with scientific computing experience in C++ using template class libraries for numerical methods, linear algebra, mesh manipulation and multi-core processing (e.g. Boost, PETSc, MPI, OpenMP, OpenMesh, etc.) are strongly encouraged to apply. Successful accomplishment in this summer work could lead to future research opportunities, including a graduate (MSc.) position in the Mathematical Biology group of Leah Keshet.

The Brownian motion of particles is a basic fundamental physical process. Consider the problem of particle moving randomly on a curved surface, such as a cell wall or material substrate. Starting from a point, what is the average time for such a particle to reach a certain "trapping region"? This is known as the "mean first passage time".

We can approach such problems using partial differential equations (PDEs). In this project, we have several goals: (1) to compute solutions to diffusion PDEs on surfaces using the Closest Point Method; (2) to implement particle simulations based on closest point representations of surfaces; and (3) to investigate the role of curvature in first passage time problems.

Useful skills include proficiency in MATLAB/Octave or Python, a background in numerical algorithms (such as finite differences and interpolation), and some knowledge of differential equations. Knowledge/interest in collaborative software development with Git would be helpful. Colin MacDonald

An etale algebra E over a field F is a finite-dimensional algebra of the form E = E1 × · · · × Er, where each Ei is a finite separable field extension of F . If r = 1, i.e., E is a field, the primitive element theorem asserts that E can be generated over F by a single element. If F is an infinite field, the primitive element theorem re-mains valid for etale algebras. However, it breaks down if F is a finite field. The project is to find the minimal number of generators in this case, under various assumptions on E (e.g., assuming that |F | = q and dimF (E) = n is fixed or assuming that q, n and r are fixed).

This problem came up in my joint work with UBC postdoc Uriya First. Uriya plans to be involved in this USRA project in person while he is still at UBC (his appointment ends on June 10). He may also be available remotely after he leaves. We plan to begin with background reading in Galois theory, the classification and basic properties of finite fields, the primitive element theorem for fields and etale algebras, and some related linear algebra (in particular, the Vandermonde determinant and its variants).

Prerequisites are Math 223, 322 and 323 or equivalent. This general topic and the specific question should be accessible to an undergraduate; we are hoping to get a definitive answer to our main question (or at least some variants of it) by the end of the summer. We also hope that working on this topic will prepare a student for a more in-depth study of abstract algebra and related areas of pure mathematics. Zinovy Reichstein