After about one hundred thousand (10⁵) years of expansion following the Big Bang, the hot, ionized gas, or plasma, filling the universe had cooled sufficiently that it recombined. This recombination event is observable as a sea of microwaves pervading the universe. This background radiation shows us that the matter in the iniverse was almost completely smooth at that time. Yet, when we look at the nearby universe, some 10 billion (10¹⁰) years later, it is full of huge contrasts: very dense objects, galaxies, stars and planets, embedded in a universe with an average density far smaller that the best terrestrial vacuum.

Gravity plays a central role in growing today's structures from those small initial density fluctuations. We can follow how the ripples grow analytically until they become non-linear (analogous to a wave beginning to break), but beyond that, although the governing equations are basically Newtonian, the complex behaviour of the cosmic density field can only be modelled numerically.

Simulating the growth of cosmic structure, from the star clusters and gas within galaxies to the large superclusters containing tens of thousands of galaxies and beyond, is a huge computational challenge (indeed, it is has been called on of the "Grand Computational Challenges" along with global environmental modelling and protein folding for example) because of the vast range of mass (at least 10⁹) and timescale (at least 10⁵) involved. My research uses computers from desktops to massively parallel supercomputers to follow the forces that shape the the structure that we observe in the universe.