Evolving initial conditions far forward in time requires specialized techniques. For the first few hundred thousand years, up until the epoch of recombination (when the first atoms formed), the universe was relatively simple, with fluctuations gentle enough that analytical techniques suffice. But later, the situation becomes too complex. If we decompose the density fluctuations into Fourier modes, they evolve independently at first, but as their amplitudes increase (as structure grows), they begin to interact. To continue, we must turn to numerical methods. We represent the matter in the universe as a system of discrete particles, calculating their interactions numerically. State-of-the-art simulations use billions of particles. At Carnegie Mellon, I run models of the universe on a parallel computer built by Professor Bob Nichol and myself in the Physics Department.

We would like to include all possible physical processes in our computer simulations. The most important is gravity, already impossible to calculate analytically for systems of more than two particles. Gas dynamics is also included; it manifests itself as gentle pressure forces felt by diffuse intergalactic gas clouds as well as shock heating when supersonic streams of gas fall into the outer layers of galaxies. The universe also contains photons, which, unlike matter, are able to easily propagate through space and have an influence on the remote regions, heating and ionizing material. Including intergalactic radiative transfer is one of the latest improvements and is one of the things I currently work on.

Confronting our model predictions for the remote voids are observations of the diffuse gas between galaxies. These observations are among the most exciting results from the new generation of 10 meter class telescopes, led by the Keck in Hawaii (which I was fortunate to use a few years ago). The extremely sparse gas cannot be seen in emission since it produces too few photons. But with distant bright quasars providing the ideal background light source, the neutral hydrogen and helium content of these regions (which was produced primordially in the Big Bang) shows up readily as absorption lines. These lines are themselves spatially clustered, revealing structure even in the most desolate parts of space.

For me, the most exciting aspect of this work lies in the fact that in the very early universe such a remote void region was located closer than the radius of a nucleon from material now forming the densest and largest galaxy clusters. As the universe expanded vastly, information about this early structure was preserved in the distribution of matter. At present these same regions are separated by hundreds of millions of light years. The initial quantum fluctuations can be probed by looking at the structure in the quasar absorption lines. This is a valuable complement to measurements of the cosmic microwave background, which probe much larger scales.

Testing our ideas for the initial conditions and the contents of the universe requires that we compare against as many different aspects of the observed universe as possible. The sparse intergalactic medium extends the range of our comparisons. A theory that correctly predicts the numbers of dense massive galaxy clusters must also describe the most dissimilar environments, these intergalactic voids. We find that the measurements are consistent with our standard model, and can also help us constrain some free parameters. This fine-tuning in turn helps us produce better simulations, providing a better understanding of structure formation. This interplay between structure formation and cosmology will continue. Our next steps will involve looking at the first sources of light in the universe and how they were able to ionize atoms in intergalactic space. The early formation of these sources (quasars and stars) is a very stringent test of our understanding of how structure forms on the smallest scales yet investigated. Our ultimate goal is to be able to predict the nature of all structure, from intergalactic voids down to galaxies, planets and people, directly from the Big Bang.