The main problem I've worked on is how stars and star clusters form in giant molecular clouds. Gravity, MHD, atomic/molecular physics, and crucially stellar feedback all play a role, and we can use numerical simulations to disentangle these physics. These processes are a key uncertain piece of microphysics in galaxy formation, and give rise to the population of globular clusters, which carry important clues about our Galaxy's ancient history.
I like to come up with ways to make astrophysical simulations more realistic, more efficient, and more adaptive.
The initial mass function of stars is a fundamentally important quantity in practically all fields of astrophysics. In my own work on GMCs and star clusters, the details of where and how individual stars form are now the leading-order uncertainty. And yet what physics are responsible for it, and how it is expected to vary, remain deeply uncertain. My collaboration is working on the next generation of self-consistent star formation simulations, which will serve as a virtual laboratory for mapping out the physical origins of the IMF in realistic GMCs.
Giant molecular clouds are the progenitors of stars and star clusters in our Galaxy. I would like to understand their lifecycle: how they form, how their properties influence star formation, and how they disperse.
Most baryonic matter in the Universe is in a state of magnetohydrodynamic turbulence. I want to understand how the properties of astrophysical systems emerge from the behaviour of turbulence. Particularly important for star formation is understanding exactly how gravity and feedback alter and drive turbulence.
I am a regular contributor to the FIRE collaboration, working on high-resolution cosmological zoom-in simulations that resolve the multi-phase ISM and stellar feedback processes. I am using my high-resolution GMC simulations to calibrate the next generation of star formation prescriptions, extending galaxy simulations' predictive power into dense gas.
pytreegrav is a fast, OpenMP-parallel Barnes-Hut style tree-code for computing gravitational potentials and fields from particle data implemented entirely in python. With numba as its backend, it can crunch forces/potentials at a rate that is nearly competitive with state-of-the-art N-body codes.
meshoid (MESHless Operations such as Integrals and Derivatives) is a multi-purpose tool for performing a variety of useful operations for analyzing or visualizing meshless or unstructured mesh simulation data.
MakeCloud is an initial conditions generator for GMC simulations, which can include initial turbulent velocity and magnetic fields with an arbitrary mixture of compressive and solenoidal modes. Compressive magnetic fields not recommended.
CloudPhinder uses an algorithm related to the Subfind algorithm to identify self-gravitating gas structures (ie. the progenitors of star clusters) in galaxy simulation data.
SinkVis is my collaboration's in-house visualization tool for making images and movies of star formation simulations, with meshoid's projection/deposition routines as its backend.