A quantum mechanical treatment of electrons is essential for first principles prediction of material properties, and density functional theory (DFT) provides a framework for such computational experiments. Offering a tolerable balance of accuracy and computational cost, it enjoys widespread use and consumes a significant fraction of scientific supercomputing time worldwide.

Orbital-free density functional theory is less well developed than the standard approach, but it provides substantial computational savings—enabling larger and longer timescale simulations—because it bypasses wave functions entirely. However, while the orbital-free strategy has a rigorous foundation and is accurate for some cases (free-electron-like metals and some semiconductors), more theoretical work is required to achieve reliability across the periodic table. My research tackles this problem.

Cambridge researchers—Profs. Chris Pickard (MSM) and Richard Needs (Physics)—pioneered a simple but robust paradigm for computational materials discovery: ab initio random structure searching (AIRSS). I contribute to this ongoing effort, having showed recently that orbital-free density functional theory can accelerate structure searching.

Supported by a Schmidt Science Fellowship, I am exploring machine-learning-driven techniques for simulating polymers, aiming to address sustainability challenges associated with consumer plastics. This work includes collaboration with Profs. Chris Pickard (MSM), James Elliott (MSM), and Gábor Csányi (Engineering).