Research

Developing image analysis tools and pipelines and applying them to various data is my profession. In particular, I'm interested in inferring the morphology of galaxies in a precise and accurate way. With this, my collaborators and I are not only able to simulate observations of existing and prospective telescopes but also to measure small distortions in the shape of galaxies caused by gravitational lensing.

Gravitational lensing ...

... describes the bending of light rays coming from a distant star, quasar or galaxy (the source) induced by the gravitational pull of masses (the lens, in our case galaxies or galaxy clusters) along the light path.

It comes in a couple of flavors, which are separated by the strength of the effect. We differentiate between strong lensing, which shows prominent features like arcs and multiply imaged sources; weak lensing, where the effect of the lens can only be estimated by investigating trends in ensembles of sources; and microlensing, where the lightcurves of sources are monitored to find and characterize very small lenses.

Shapelets

An elegant method of shape description is the shapelet decomposition, which expands the images into a set of orthogonal functions, the Gauss-Hermite polynomials. This basis function system appears adequate to describe galaxies because of its compact centralized shape, with additional fluctuations being introduced by higher-order modes. Thus, it can describe the shape of a galaxy faithfully with only a limited number of coefficients.

Besides several advantages like an exact convolution formalism, the shapelet basis system also has drawbacks. Although it is in principle complete, in practice only a limited number of modes are considered (see above). Hence, galaxies, whose intrinsic shape does not resemble the general shape of the low-order shapelet modes, will have an insufficient description. This is true especially for elliptical galaxies, for which the weak-lensing estimates are systematically biased low.

KSB & DEIMOS

A less demanding approach of measuring the lensing-induced distortions is the traditional KSB method, which tries to relate the ellipticity of the observed galaxy to the gravitational shear its light experienced on the way to us. KSB was known to be troublesome, and we could recently show that this behaviour stems from several mathematical flaws.

We wanted to keep the excellent performance of KSB and its model-independent approach, but develop a mathematically consistent treatment of gravitational lensing and observations effects. And in the DEIMOS method, we united the advantages of KSB and shapelets, which allowed us to come up with a fast and accurate weak-lensing pipeline.

Scientific computing

Analyses in astronomy and physics often involve large-scale data processing. Thus, developing solutions for achieving results in an fast and reliable way is another object of interest for me. This involves the use of efficient algorithms, robust protocols and parallelization, but also the employment of Software Engineering methods.