Tom Beardsley

PhD

Computational Physicist and Programmer with a sprinkle of Boxing Coach


I am a Computational Physicist and Programmer with experience in academic research (polymers), high-performance computing (HPC) and financial software.

Skilled in numerous programming languages and simulation techniques, including: C/C++, CUDA (GPUs), MPI, Monte Carlo, field-theoretic simulations & javascript, I also have experience in: Python, PHP, SQL, HTML, CSS and C#.

I am driven by my desire to produce creative and accurate solutions to difficult problems and a need for continual learning.

Never shy of a challenge, I am a qualified boxing coach, have trekked 150km in the Himalayas, fed flying vultures from a paraglider, skydived and bungee-jumped my way around the world.

Selected Projects

Langevin Field-Theoretic Simulations of Diblock Copolymers on GPUs


Languages: C/C++

Libraries/APIs: CUDA, Thrust


This code acted as the engine for a number of academic research projects that led to multiple publications. They are orders of magnitude faster than equivalent serial code, allowing for highly accurate statistics and the production of full phase diagrams from a field-based model.

Interactive Ising Model in the Browser


Languages: javascript, html, css

Libraries/APIs: chartjs


Parallel Monte Carlo simulations of the 2D Ising model in the user's browser. The Ising model is used to study ferromagnetic systems in statistical physics. Multiple temperatures are simulated concurrently, with results displayed on charts in real-time.

Lattice Monte Carlo Simulations of Diblock Copolymers with Parallel Tempering


Languages: C/C++

Libraries/APIs: MPI, OpenMP


A fast, particle-based model where monomers are constrained to the sites of a face-centred cubic lattice. Parallel tempering, where replicas of the system run in parallel and can swap their configurations, allowed defects to anneal out and the novel detection of order-disorder transitions via the heat capacity.

Well-Tempered Metadynamics (WTMD) Applied to Field-Theoretic Simulations of Diblock Copolymers


Languages: C/C++

Libraries/APIs: CUDA, Thrust


WTMD allows a system to overcome an energy barrier separating two competing phases by adding Gaussians to a bias potential, U(Ψ), where Ψ is a collective order parameter. This project showed how it can be used to detect the lamellar-disorder transition with very high accuracy.