Thomas Flöss
Postdoctoral researcher in theoretical and computational cosmology at the University of Vienna.
I’m a postdoc at the University of Vienna in the computational cosmology group of Oliver Hahn. I work primarily on the use of late time cosmological probes, such as galaxy surveys and intensity mapping (e.g. 21-cm) to study the primordial universe, in particular cosmic inflation and primordial non-Gaussianity. In my research I make use analytical, numerical and state-of-the-art machine learning methods. Currently, I am working on developing methods for field-level inference, as a means of optimal cosmological data analysis. I received my PhD from the University of Groningen, under the supervision of Daan Meerburg, Diederik Roest and Léon Koopmans.
I have (co-)written several codes for cosmological data analysis and forecasting. In particular, my JAX-based code BFast is able to rapidly evaluate the bispectrum of cosmological simulation boxes, using GPU acceleration, and was used to rapidly compute the bispectra of the 32,768 Quijote Big Sobol Sequence simulations. Together with Oliver Philcox, I co-wrote the PolyBin3D package, which allows for fast evaluation of polyspectra from surveys, including survey window deconvolution, also on GPU.