Research Fellows Directory
David van Dyk
Professor David van Dyk
Imperial College London
Recent technological advances have dramatically increased the quality and quantity of astronomical data. New instruments provide massive new surveys, high resolution spectrography and imaging, and incredibly detailed movies of explosive processes in the solar atmosphere. These new instruments are helping scientists make impressive strides in our understanding of the universe, but at the same time are generating massive data analysis challenges. This proposal aims to develop statistical methods to bridge the gap between newly available data and the sophisticated scientific questions astronomers are asking of it.
Astrophysics-based computer simulation models can describe complex phenomena that defy mathematical formulation. These models aim to quantify astrophysical hypotheses and predict data observed from astrophysical sources. This proposal’s primary goal is to develop principled statistical and computational techniques for complex physical systems, components of which may require computer models. In particular, we aim to develop new methods for comparing model predictions with data from high-tech telescopes with the goals of fitting, assessing, and ultimately improving models and thus bettering our understanding of the underlying astrophysics. This may require combining complex models for the multiple sources composing an observation along with models for the instruments making the observations. The results are highly-structured statistical models that require sophisticated and computationally intensive algorithms. Developing the necessary computational methods along with techniques for model building, checking, and improvement forms the core of this project.
The new methods that we develop for astronomy will be used as spring boards for new general statistical methods for the complex data and models that are ubiquitous in the engineering, physical, and biological sciences. The researcher is also taking a lead role in a new Astrostatistics Centre at Imperial.
Interests and expertise (Subject groups)