Robert Baldock
Dr Robert Baldock
Member of Queens' College
Postdoc in Prof Payne's group
Office: 507 Mott Bld
Email: rjnb2 @ cam.ac.uk
TCM Group, Cavendish Laboratory
19 JJ Thomson Avenue,
Cambridge, CB3 0HE UK.
Research
I am interested in the statistical mechanics of classical and quantum systems, and also in inverse statistical mechanics. My Ph.D. at Cambridge was supervised by Dr. Gábor Csányi and Prof. Mike Payne in the enginnering and physics departments respectively.
In my PhD I adapted an algorithm called nested sampling, originally developed by the astrophysicist Dr. John Skilling for Bayesian data analysis. My adaptation allows one to calculate the partition function of a material as an explicit function of temperature, and therefore to do statistical mechanics from first principles. The nested sampling algorithm unbiasedly explores the entirity of configuration space, and therefore discovers expected and unexpected phases, together with their phase transitions and structures. In this way, my implementation of nested sampling automatically uncovers phase diagrams, equations of state, and all other quantities that characterise the equilibrium statistical mechanics of a material.
I have applied my algorithm to study the phase diagrams of binary Lennard-Jonesium "alloys", and collaborate with Dr. Livia Bartók-Pártay who has been applying my algorithm to study the phase diagrams of aluminium and the shape-memory alloy NiTi. Indeed, Livia and Gábor are now leading a research effort to apply this algorithm to many interesting materials questions. I am also collaborating with Dr. Robert Horton in the group of Prof. Mike Finnis at Imperial College, who is studying the behaviour of hydrogen in steel. Previously I have also worked with Prof. David Wild and his group at the University of Warwick, exploring the application of nested sampling to biomolecules.
In Plain English
Scientists and engineers would like to be able to predict the behaviour of materials from the atoms that make them up. I study how it is possible to do this. This will allow us to design powerful new materials with radically different properties to the materials we have today, and these new properties will enable new technologies.
To go into slightly more detail, scientists use particlar language to describe materials: we use ideas such as pressure, temperature, and so forth. Of course, materials are made up from atoms, and the language we use to describe atoms is very different: atoms are described by quantum mechanics. To predict the behaviour of materials we must make a link between pressure, temperature etc., and quantum mechanics. The link between these two languages is the theory of statistical mechanics. Happily, statistical mechanics has been well understood for decades. Unhappily, statistical mechanics is very difficult to apply in all but the simplest imagined scenarios: the mathematical calulations are just too difficult for humans. I work on new ways of looking at the same mathematics that allow computers to guess the answer. The clever part is that we can understand the uncertainty in their guesses, and that it is perfectly feasible to make the uncertainty very small, so that we get the true result.
Featured Publications
- Determining pressure-temperature phase diagrams of materials Phys. Rev. B 93 174108 (2016)
- Exploiting molecular dynamics in Nested Sampling simulations of small peptides Comput. Phys. Commun. 201 8 (2016)