My research was been primarily focused on extragalactic astronomy and cosmology. My thesis work explored a major outstanding problem in astrophysics - dark matter. In order to study dark matter I run simulations on supercomputers; a visualization of such a simulation is in the background of this page.

The dominant form of mass in the standard cosmological model is collisionless cold dark matter (CDM). The distribution of dark matter forms something of a cosmic scaffolding on which luminous matter aggregates. The original distribution of dark matter is due to primordial random density perturbations (via the power spectrum) and then over the 13.9 billions years of the Universe the dark mark has collapsed into virialized halos due to the relentless long range force of gravity. Baryonic matter (all regular matter, such as stars, or hydrogen gas) aggregates in these dark matter over densities and forms luminous galaxies, like our own Milky Way. Simulations and observations on cosmological scales are in strong agreement with the CDM theory. However the distribution of dark matter is poorly understood on smaller galactic scales as indicated by a lack of concordance between CDM predictions and observations: 1) The inner density profile of galaxies inside the scale radius is much shallower than expected. Additionally the central density of dark matter halos is observed to be constant (with intrinsic scatter) independent of halo mass. This is the so called core/cusp problem. 2) The number of observed galaxies in the Local Group is an order of magnitude lower than that predicted by CDM simulations. This is the so called missing satellite problem. There are many other interesting apparent peculiarities of dark matter on small scales and as observations improve our understanding of these phenomena will improve, in fact, even the problems above may merely be artifacts of our limited observational ability. All of these issues would be ameliorated by mechanisms which lower the central density of galaxies (or raise the phase space density of dark matter).

Mechanisms which are relevant include baryonic feedback and (mildly) collisional dark matter (contrary to what some astronomers presume particle physicts have demonstrated that dark matter can have a relatively large interaction cross section that would not yet be detected astrophiycally). Evidence is mounting that including the effects of baryons in high resolution numerical simulations such as gas cooling, star formation, and gas outflows driven by supernovae can better match the observations such as flatter cores in dwarf galaxies. In fact, the reality is that galaxies most certainly do experience feedback effects from stars, but simulations of such physics is difficult. The possibility of cold, nondissipative, but collisional and self-interacting dark matter is also being raised. A dark matter interaction with a mean free path between 1 kpc and 1 Mpc would only alter the evolution of cold dark matter in high density galaxy environments and large scale power spectrum measurements would be insensitive to the this modification of dark matter. My thesis work tested the galactic scale distribution of dark matter by using detailed numerical simulations (including a full range of baryonic physics including gas cooling, star formation, and supernovae driven gas outflows) to explore the effects of cosmologically consistent self-interacting dark matter.

The simulations were run on massively parallel computing systems using the N-body+SPH code ChANga. The simulations were run in a fully cosmological context beginning from high redshift with a power spectrum modified by a transfer function. We evolve our simulations for billions of simulated years, but only millions of wall-clock CPU hours. Many of our most scrutinized galaxies were high-resolution zoom ins of interesting halos, galaxies that look like canonical dwarf galaxies of the Milky Way.

If you are interested in the awesome visualizations of these simulations I encourage use of these videos for academic or educational purposes with attribution ("Source: Alexander B. Fry, University of Washington, ChANga numerical code"). You can access a bunch of simulations here here. You can find embedable videos of simulations here (dwarf galaxy showing dark matter, CDM vs SIDM) and here (dwarf galaxy showing gas, CDM vs SIDM) or even here.