Tom Nye
I am a Senior Lecturer in statistics in the School of Mathematics, Statistics and Physics at Newcastle University.
These pages describe my research interests, my research group and software projects. For details about my teaching, Newcastle students should visit canvas.
Research interests
- Bayesian inference of phylogenetic trees
- Non-Euclidean geometric statistics for data sets of trees and graphs
- Tree-valued stochastic processes
- Statistical modelling and manifold-valued data in computational neuroscience
Projects
Bayesian inference of phylogenies: I have supervised a number of PhD students on projects involving inference of trees from genetic sequence data. Projects have included modelling compositional heterogeneity across branches, investigating non-reversible Markov substitution models and developing branch-heterogeneous Lie Markov models. I currently collaborate on a Leverhulme grant with Martin Smith at Durham University inferring phylogenies for fossils using morphological data.
Principal component analysis in BHV tree space: Billera-Holmes-Vogtmann (BHV) tree space is a beautiful space which encapsulates all possible evolutionary histories for a given set of species. Since BHV tree space is not a Euclidean vector space, PCA must be adapted in order to work in the new setting. I have developed methods for fitting principal geodesics and minimal principal surfaces.
Stochastic processes in BHV tree space: I have established the fundamental theory of Brownian motion in BHV tree space, and I am developing inference methods for fitting these to data with a current PhD student.
Information geometry for phylogenetic trees: In collaboration with researchers at Gottingen University, Germany, I have recently developed a new space of phylogenetic trees in which the metric measures the distance between trees in terms of similarity of the distributions on genetic sequences induced by the trees. Called wald space, there are still many unanswered mathematical questions, and there’s great potential for developing statistical methods in this space.
Non-Euclidean statistics in computational neuroscience: I collaborate with the computational neuroscience lab at Newcastle and co-supervise a number of students in this area.
Publications
Details of my publications are on Google scholar.
Background
My PhD was in pure mathematics (specifically gauge theory and global analysis); a significant output was the proof of the “Nye-Singer index theorem” for Dirac operators on $S^1\times \mathbb{R}^3$. After that I worked in industry as a mathematical consultant for a couple of years. I returned to academia with a postdoc position at the MRC Biostatistics Unit working on problems in statistical bioinformatics. I started my lectureship at Newcastle in January 2006.
As a statistician I tend to be Bayesian. Geometrical ideas often influence the way I think and solve problems.
Contact
Email: tom.nye@ncl.ac.uk
Phone: +44 (0)191 2085369
School of Mathematics, Statistics and Physics
Herschel Building
Newcastle University
Newcastle upon Tyne
NE1 7RU