Reconstructing EVolutionary histories for organisms on EArth and stars in gaLaxies (REVEAL)

Stars in galaxies undergo compositional evolution, i.e. a star’s chemical make-up changes over generations. Stars are born from a gas reservoir, into which they dump new elements when they die, forever changing the chemical composition of future generations of stars. When stars have evolved on their own, their changes are well modelled by tree-like relationships. In reality, stars evolve in distinct environments and then mix, which produces a network-like structure, specifically directed networks, or “phylogenetic networks” as they are called in biology. Challenges in fitting such networks to physical systems to reveal evolutionary histories include 1.) difficulties in fitting a large number of parameters (~ N² parameters, where N is the number of nodes); 2.) modelling migration events that mix material between nodes; 3.) incomplete datasets; 4.) propagating uncertainty in a high-dimensional parameter space. The abundance of data for chemical elements in stars in our Milky Way invites us to tackle these challenges now. 

In this PhD project, the student will develop generalizable auto-differentiable phylogenetic networks for evolving stars experiencing mass and continual migration events. They will additionally explore methods for efficiently fitting networks to massive datasets through reducing the number of network nodes and edges, and accelerating existing Bayesian methods designed for exploring high-dimensional parameter spaces. They will then apply the new network for extracting evolutionary histories to an astrophysical case study.

The new network tools will equip astrophysicists with a novel way to transform the new massive datasets arriving for millions of stars in our Milky Way into discoveries regarding stellar interactions and stellar evolution.

Open to any UK or international candidates. Up to 30% of our UKRI funded studentships can be awarded to candidates paying international rate fees.

You will need to meet the minimum entry requirements for our PhD programme.

Programming experience in Python is desirable, particularly in PyTorch.