Finding new open clusters with Gaia
Galaxy redshifts with machine learning
Ever-larger photometric galaxy catalogues require sophisticated techniques to derive redshifts efficiently and robustly. In my Master’s project, I used a mixture density network (a type of neural network) to derive accurate redshifts for galaxies in the CANDELS GOODS-South field. I showed that machine learning could be viable for galaxy redshift determination, although more training data would be necessary to improve the results of the method.
De-reddening Cepheid variables with Gaia
Cepheid variable stars are the bottom rung on the cosmic distance ladder. Accurately measuring their characteristics is essential to derive robust cosmological parameters, but this is greatly complicated by stellar extinction. I used Gaia parallaxes and a Bayesian inference framework in Python to independently and precisely derive Leavitt law (period-luminosity relation) parameters, along with extinctions to each individual star.