Jobid=639502614625256224 (0.0503)
You will be part of a larger project at the Uncertainty in Complex Systems lab (PI: Dr Max Hinne) aimed at learning which statistical and/or computational model is most suited for a provided data set and scientific question. In practice, statistical models contain many arbitrary choices that may (a) have a detrimental effect on what we can learn from it and (b) greatly reduces the reproducibility of scientific research, as different researchers make different decisions. A Bayesian multiverse is a systematic way of reducing these adverse effects and aims to make statistical analyses more robust.
We are looking for a research assistant to contribute to the development of bamojax (Bayesian modelling in JAX), our open-source Python toolbox for Bayesian modelling and model comparison. Throughout your employment, you will have the opportunity to work with others within our group to see what modelling choices they make, and how our new Bayesian multiverse tools can help with this. There will be weekly meetings with the PI to discuss the project's progress and next steps, as well as bi-weekly lab meetings.
You will have two core tasks. The first is to implement advanced methods for Bayesian model comparison in JAX. These techniques will then be used to systematically explore alternative models for user-defined statistical problems. The second task is to construct an efficient search procedure for variants of a given model.
Deel deze vacature: