Back to all members...
Tiarnan Doherty
Postdoc, started 2022
Tiarnan is a Schmidt Science Fellow at the Electron Microscopy Group, University of Cambridge with Paul Midgely, and is an academic visitor at OATML as part of his fellowship. Tiarnan completed his PhD in Physics at the University of Cambridge.His PhD involved understanding the fundamental properties of metal halide perovskites, a revolutionary new technology for inexpensive, scalable, and highly efficient solar panels. As a Schmidt Science Fellow, he will pivot from materials science to computer science, with an ambition to develop new AI approaches that will greatly accelerate the discovery and development of materials for applications in energy, healthcare, and computing. Tiarnan was an [Oppenheimer Research Fellow] and an [Early Career Research Fellow at Murray Edwards College].
Publications while at OATML • News items mentioning Tiarnan Doherty • Reproducibility and Code • Blog Posts
Publications while at OATML:
Hybrid Physics-Machine Learning Models for Quantitative Electron Diffraction Refinements
High-fidelity electron microscopy simulations required for quantitative crystal structure refinements face a fundamental challenge: while physical interactions are well-described theoretically, real-world experimental effects are challenging to model analytically. To address this gap, we present a novel hybrid physics-machine learning framework that integrates differentiable physical simulations with neural networks. By leveraging automatic differentiation throughout the simulation pipeline, our method enables gradient-based joint optimization of physical parameters and neural network components representing experimental variables, offering superior scalability compared to traditional second-order methods. We demonstrate this framework through application to three-dimensional electron diffraction (3D-ED) structure refinement, where our approach learns complex thickness distributions directly from diffraction data rather than relying on simplified geometric models. This method achie... [full abstract]
Shreshth Malik, Tiarnan Doherty, Benjamin Colmey, Stephen J. Roberts, Yarin Gal, Paul A. Midgley
arXiv
[paper]