Wes Armour MPhys PhD

Professor Wes Armour is a Professor of Scientific Computing and Associate Head (Digital and Information) of the Mathematical, Physical and Life Sciences (MPLS) Division at the University of Oxford. He previously served as Associate Head of Department for Research in the Department of Engineering Science, having prior to that been the Director of the Oxford e-Research Centre (OeRC)—an interdisciplinary institute within the department. To date, he has been awarded over £31M in funding as Principal Investigator or Co-Investigator across a diverse research portfolio spanning high-performance computing, machine learning, signal processing, time-domain radio astronomy, and protein crystallography.

He was also the Director and Principal Investigator of JADE and JADE2, a national tier-2 High Performance Computing facility that served as the first national infrastructure in the UK dedicated entirely to advancing Artificial Intelligence, deep learning, and data-intensive science.

He completed an MPhys degree in Fundamental Particle Physics and Cosmology in 2001, followed by a PhD in 2004 focusing on Lattice Gauge Theory and Chiral Perturbation Theory (the computational and effective field theories of the strong nuclear force).

His research focuses on understanding and improving ways to extract science from data by addressing core bottlenecks in modelling, simulation, and data processing. Specifically, his work tackles fundamental challenges in numerical analysis, hardware-software co-design, and high-performance machine learning to extract insights from extreme-scale datasets. At the direct intersection of AI architecture and hardware capability, his research group specializes in designing next-generation algorithms optimized for modern computing systems.

An early contributor to the UK e-Science programme, Professor Armour was part of the group that developed GridPP in the early 2000s and was a pioneer in the adoption of GPU technology for scientific computing. In 2008, he utilised early NVIDIA GPGPUs (sm_10 architecture) to accelerate conjugate gradient routines used to simulate electron transport in graphene. In 2012, he was a member of a team based at Diamond Light Source that implemented machine learning methods to enable the automated merging of multiple crystal datasets in protein crystallography. Since then, he has introduced the use of GPUs across a range of fields, notably pioneering their application to Square Kilometre Array (SKA) data processing.

His recent work focuses heavily on high-performance AI optimization, scalability, and algorithmic sustainability. This includes leading comprehensive benchmarking studies on the real-time energy and carbon footprints of deep learning models and vision backbones during inference, evaluating built-in GPU power sensors, and maximizing the energy efficiency of Fast Fourier Transforms (FFTs) for real-time edge computing. In optimization, his group designs advanced second-order and natural gradient descent methods—such as Fisher-Orthogonal Projection and runtime-orchestrated frameworks—to drastically improve convergence speeds during large-batch training and scalable LLM training on modern data-centre hardware. This algorithmic research also extends to resolving variance and modality competition within multimodal deep learning models.

In the domains of radio astronomy and computer vision, Professor Armour’s group develops core data pipelines and intensity-sensitive image quality assessment indexes for next-generation instruments, strategically aligned with key deployment milestones for the international SKA project. Through the SPOTLIGHT project and partnerships with the Giant Metrewave Radio Telescope (GMRT), his team deploys real-time, high-throughput filtering pipelines. Additionally, his recent work encompasses reduced-precision signal processing, and the development of open-source, GPU-accelerated software packages like AnACor2.0 for high-performance analytical absorption corrections in X-ray crystallography.

• Data Science
• Scientific Computing
• GPU Computing
• Signal Processing
• Modelling and Simulation
• High Performance Computing

• Astro Accelerate - GPU accelerate real-time signal processing for time-domain radio-astronomy
• SKA SDP - the science data processor for the Square Kilometre Array (a next-generation radio telescope);
• ENErGize - Energy Efficient GPU Computing, focusing on tuning and optimisation of GPU libraries and algorithms for energy efficiency
• FUN - Fast GPU Convolutions, focusing on accelerating Fourier Domain convolutions on GPU architectures
• AFAR - Acceleration of AI learning, focusing on the acceleration of CNNs
• APT - AI HPC Intersection. This project investigates the use of AI to either predict best values for characterisations in HPC codes or to increase accuracy in simulation results.

Publications: (see Google Scholar)

2017

Jayanth Chennamangalam, David MacMahon, Jeff Cobb, Aris Karastergiou, Andrew PV Siemion, Kaustubh Rajwade, Wes Armour, Vishal Gajjar, Duncan R Lorimer, Maura A McLaughlin, Dan Werthimer, Christopher Williams. SETIBURST: A Robotic, Commensal, Realtime Multi-Science Backend for the Arecibo Telescope The Astrophysical Journal Supplement Series 228 (Number 2), (2017)

2016

K Adámek, W Armour. A Real-time Single Pulse Detection Algorithm for GPUs Astronomical Data Analysis Software and Systems, ADASS XXVI, (2016)

K Adámek, J Novotný, W Armour. A polyphase filter for many-core architectures Astronomy and Computing, Volume 15, (2016)

Kaustubh Rajwade, Jayanth Chennamangalam, Duncan Lorimer, Aris Karastergiou, Dan Werthimer, Andrew Siemion, David MacMahon, Jeff Cobb, Christopher Williams, Wes Armour. Commissioning of ALFABURST: initial tests and results. arXiv preprint arXiv:1602.04389, (2016)

2015

W Armour, S Hands, C Strouthos. Strong Interaction Effects at a Fermi Surface in a Model for Voltage-Biased Bilayer Graphene. Phys. Rev. B 92 (Issue 23), (2015)

S Dimoudi, W Armour. Pulsar Acceleration Searches on the GPU for the Square Kilometre Array. Proceedings for Astronomical Data Analysis Software and Systems XXV, (2015)

Jayanth Chennamangalam, Aris Karastergiou, David MacMahon, Wes Armour, Jeff Cobb, Duncan Lorimer, Kaustubh Rajwade, Andrew Siemion, Dan Werthimer, Christopher Williams. ALFABURST: A realtime fast radio burst monitor for the Arecibo telescope. Proceedings of the Fourteenth Marcel Grossmann Meeting on General Relativity, (2015)

Jayanth Chennamangalam, Aris Karastergiou, David MacMahon, Wes Armour, Jeff Cobb, Duncan Lorimer, Kaustubh Rajwade, Andrew Siemion, Dan Werthimer, Christopher Williams. ALFABURST.  WSPC Proceedings, July 2015

A Karastergiou, J Chennamangalam, W Armour, C Williams, B Mort, F Dulwich, S Salvini, A Magro, S Roberts, M Serylak, A Doo, AV Bilous, RP Breton, H Falcke, J-M Grießmeier, JWT Hessels, EF Keane, VI Kondratiev, M Kramer, J Van Leeuwen, A Noutsos, S Osłowski, C Sobey, BW Stappers, P Weltevrede. Limits on Fast Radio Bursts at 145 MHz with ARTEMIS, a real-time software backend. Monthly Notices of the Royal Astronomical Society Volume 452 (Issue 2), (2015)

J Chennamangalam, A Karastergiou, W Armour, C Williams, M Giles. Artemis: A real-time data processing pipeline for the detection of fast transients. Radio Science Conference (URSI AT-RASC), 2015 1st URSI Atlantic, (2015)

S Hands, W Armour, C Strouthos. Graphene as a Lattice Field Theory. 9th International Workshop on Critical Point and Onset of Deconfinement (CPOD 2014), (2015)

2011-2014

J. Foadi, P. Aller, Y. Alguel, A. Cameron, D. Axford, R. L. Owen, W. Armour, D. G. Waterman, S. Iwata and G. Evans. Clustering procedures for the optimal selection of data sets from multiple crystals in macromolecular crystallography. Acta Cryst. D69, 1617-1632 (2013).

Anna J. Warren, Wes Armour, Danny Axford, Mark Basham, Thomas Connolley, David R. Hall, Sam Horrell, Katherine E. McAuley, Vitaliy Mykhaylyk, Armin Wagner and Gwyndaf Evans. Visualization of membrane protein crystals in lipid cubic phase using X-ray imaging. Acta Cryst. D69, 1252–1259 (2013).

Wesley Armour, Simon Hands and Costas Strouthos. Monte Carlo study of strongly interacting degenerate fermions: A model for voltage-biased bilayer graphene. Physical Review D 87, 065010 (2013)

M. Serylak, A. Karastergiou, C. Williams, W. Armour, M. Giles and LOFAR Pulsar Working Group.Observations of transients and pulsars with LOFAR international stations and the ARTEMIS system. Proceedings of IAUS 291 (Neutron Stars and Pulsars: Challenges and Opportunities after 80 years, 2012).

M. Serylak, A. Karastergiou, C. Williams, W. Armour and LOFAR Pulsar Working Group. Observations of transients and pulsars with LOFAR international stations. ASP Conference Series, 466, (Electromagnetic Radiation from Pulsars and Magnetars, 2012).

W. Armour, A. Karastergiou, M. Giles, C. Williams, A. Magro, K. Zagkouris, S. Roberts, S. Salvini, F. Dulwich and B. Mort. A GPU-based survey for millisecond radio transients using ARTEMIS. ASP Conference Series, 461, 33 (ADASS XXI, 2012).

David R Merrifield, Vasuki Ramachandran, Kevin J Roberts, Wesley Armour, Danny Axford, Mark Basham, Thomas Connolley, Gwyndaf Evans, Katherine E McAuley, Robin L Owen and James Sandy. A novel technique combining high-resolution synchrotron x-ray microtomography and x-ray diffraction for characterization of micro particulates Measurement Science and Technology Vol 22, 11 (2011)

Wesley Armour, Simon Hands and Costas Strouthos. Monte Carlo simulation of monolayer graphene at nonzero temperature Physical Review B 84, 075123 (2011)

Wesley Armour, Simon Hands, John B. Kogut, Biagio Lucini, Costas Strouthos and Pavlos Vranas. Magnetic monopole plasma phase in (2+1)d compact quantum electrodynamics with fermionic matter Physical Review D 84, 014502 (2011)

Radio Astronomy, Theoretical Physics, Condensed Matter Physics and Grid Computing publications can be found here.

I currently have places available for research students.