Scientific

The COVID-19 global pandemic has led to unprecedented public interest in mathematical modelling as a tool to understand the dynamics of disease spread and predict the impact of public health interventions. In this pair of talks, we will describe how mathematical models are being used, with particular reference to the British Columbia epidemic.

In the first talk, Prof. Caroline Colijn (Dept. of Mathematics, Simon Fraser University) will outline the key features of the British Columbia data and focus on how modelling has allowed us to estimate the effectiveness of the provincial response. In the second talk, Prof. Daniel Coombs (Dept. of Mathematics and Inst. of Applied Mathematics, University of British Columbia) will describe forward-looking modelling approaches that can provide some guidance as the province moves towards partial de-escalation of measures. Each talk will be 30 mins in length and followed by a question and discussion period.

For more details on the group's work and to contact the team, please visit https://bccovid-19group.ca/

Until very recently most cancer biologists operated with the assumption that the most common route to metastasis involved cells of the primary tumor transforming to a motile single-cell phenotype via complete EMT (the epithelial-mesenchymal transition). This change allowed them to migrate individually to distant organs, eventually leading to clonal growths in other locations. But, a new more nuanced picture has been emerging, based on advanced measurements and on computational systems biology approaches. It has now been realized that cells can readily adopt states with hybrid properties, use these properties to move collectively and cooperatively, and reach distant niches as highly metastatic clusters. This talk will focus on the accumulating evidence for this revised perspective, the role of biological physics theory in instigating this whole line of investigation, and on open questions currently under investigation.

Classical results of Dobrushin and Lanford-Ruelle establish, in rough terms, that for a local energy function on a subshift without too much long-range order, the translation-invariant Gibbs measures are precisely the equilibrium measures. There are multiple definitions of a Gibbs measure in the literature, which do not always coincide. We will discuss two of these definitions, one introduced by Capocaccia and the other used by Dobrushin-Lanford-Ruelle, and outline a proof (available at [arxiv.org/abs/2003.05532]) that they are equivalent.

We will also discuss forthcoming work, in which we show that Gibbsianness is preserved by pushforward through a certain kind of almost invertible factor map. As an application in one dimension, we show that for a sufficiently regular potential, any equilibrium measure on an irreducible sofic shift is Gibbs. As far as we know, this is the first reasonably general result of the Lanford-Ruelle type for a class of subshifts without the topological Markov property.

Joint work with Luísa Borsato, with extensive advice from Brian Marcus and Tom Meyerovitch.

This lecture was given in two parts. The video on this page was distributed as a pre-recorded session ahead of a second live lecture.

Caroline Series' The modular surface and continued fractions

In this talk, we will use compartmental models to examine the power of age-targeted mitigation strategies for COVID-19. We will present evidence that, in the context of strategies which end with herd immunity, age-heterogeneous strategies are better for reducing direct mortalities across a wide parameter regime. And using a model which integrates empirical data on age-contact patterns in the United States and recent estimates of COVID-19 mortality and hospitalization rates, we will present evidence that age-targeted approaches have the potential to greatly reduce mortalities and ICU utilization for COVID-19, among strategies which ultimately end the epidemic by reaching herd immunity. This is joint work with Maria Chikina.