Björn Sprungk of Technical University Bergakademie Freiberg will receive the 2020 SIAM Activity Group on Uncertainty Quantification Early Career Prize. The prize will be awarded at the 2020 SIAM Conference on Uncertainty Quantification (UQ20), to be held March 24-27, 2020 at the Garching Campus of the Technical University of Munich. Sprungk will receive the award and deliver his talk, “Noise-Level Robust Sampling Methods for Bayesian Inverse Problems,” on March 25, 2020.
The SIAM Activity Group on Uncertainty Quantification (SIAG/UQ) awards the SIAG/UQ Early Career Prize every two years to an individual in their early career for outstanding research contributions in the field of uncertainty quantification in the three calendar years prior to the award year. The award recognizes Sprungk for his fundamental works on the convergence and robustness of Monte Carlo methods for Bayesian inversion problems.
Björn Sprungk is an Assistant Professor in applied mathematics at the Technical University Bergakademie Freiberg. He received his PhD in mathematics from the Technical University Chemnitz (TUC) under the supervision of Oliver Ernst in 2017. His PhD thesis, "Numerical Methods for Bayesian Inference in Hilbert Spaces," was awarded the Dr.-Klaus-Körper-Prize of the Society of Applied Mathematics and Mechanics (GAMM) and the University Award of TUC. He continued his work as a postdoctoral researcher first at the University of Mannheim and then at the Institute for Mathematical Stochastics of the Georg-August-University Göttingen before he joined TU Bergakademie Freiberg in February 2020. Sprungk’s research focuses on numerical methods for uncertainty quantification and Bayesian inference in high-dimensional spaces. In particular, he is interested in the analysis of sampling methods such as Markov chain Monte Carlo algorithms.
Q: Why are you excited to be awarded the SIAG/UQ Early Career Prize?
A: I am greatly honored to receive the SIAG/UQ Early Career Prize and very grateful to the colleagues who supported my nomination. It is pleasing and motivating that the work I've done with my collaborators is being recognized and highlighted, especially, in such an active and fast-growing field of research as uncertainty quantification.
Q: Could you tell us a bit about the research that won you the prize?
A: My research is about efficient sampling methods for posterior probability distributions occurring in Bayesian inverse problems. In particular, I studied Markov chain Monte Carlo algorithms in function spaces which are suitable for Bayesian inference for partial differential equations. We improved an existing algorithm by incorporating approximative covariance information and proved a dimension-independent exponential convergence of the resulting Markov chain. Moreover, in numerical experiments our algorithm displayed a remarkable qualitative property: a robust performance with regard to the concentration of the posterior. This is particularly desirable for Bayesian inverse problems with highly informative data. Basic sampling methods usually perform very poorly in such situations. We started analyzing this robustness which also led to improvements for several other sampling methods.
Q: What does your research mean to the public?
A: With the advent of data-based decisions and simulations using large available data sets, the development of efficient algorithms, which allow to take into account all (or most of) the available information, becomes a crucial task. Our work contributes to that challenge and provides faster algorithms for statistical inference and decision-making with informative data.
Q: What does participation in SIAM mean to you?
A: The SIAM conferences with their stimulating atmosphere allow me to stay in touch with many colleagues from various fields, to start new collaborations, and to keep up with recent developments in applied mathematics. Moreover, SIAM journals and books provide high quality research and offer a great platform with large visibility and high impact for publishing one's own work.