Dust obscures the light from stars, which is one of the primary was we detect its presence. Absorbing starlight heats the dust, which causes it to shine by itself in the infrared. Dust also scatters starlight, deflecting it onto new paths which may add or remove light from our observations. If we want to understand both dust and the stars it obscures, we therefore must understand how light and dust interact along the path to our telescopes.
To do this, we simulate the entire process of light propagating through space and interacting with dust; this is known as radiative-transfer modelling. Since the beginning of my Ph.D. I have made extensive use of this, in particular for interpreting three-dimensional environments. These cases are particularly difficult to work with, because the radiative-transfer problem does not (in general) have an analytical solution in these cases. We therefore turn to a statistical technique called Monte Carlo radiative transfer (MCRT), which samples from the probability distributions of the properties of light, dust and their interactions to produce an approximate solution to the problem. Although it is approximate, we know that in the limit of infinite computing time it would converge to the exact solution, meaning we can choose the runtime to give us a solution that is good enough. This statistical approach is able to give an approximate solution for almost any radiative-transfer problem imaginable, although it will always be slower than methods which solve specific problems with exact solutions that work in those limited cases. This allows us to simulate images (or other kinds of data) which can be compared directly to the data taken by telescopes.
My Ph.D. thesis was based extensively on using MCRT and developing codes to solve this problem, and several papers on which I am a (co-)author have exploited this technique. This has included work on evolved stars, interstellar dust and star formation. Unfortunately, none of the codes that I contributed to are open-source, at present, which is something I plan to change in future. As a result, my more recent papers make use of open-source codes such as HYPERION.
Peter Scicluna 