Postdoc in Large-Scale Computational Uncertainty Quantification for Inverse Problems – DTU Compute

Do you want to work in an interdisciplinary research team and contribute to the development of and methods for uncertainty quantification (UQ) for inverse problems? We invite applications for a two-year postdoc with focus on the development of new techniques and software for large-scale uncertainty quantification for inverse problems. The position is part of the research initiative CUQI: Computational Uncertainty Quantification for Inverse problems: https://sites.dtu.dk/cuqi funded by the Villum Foundation and headed by Professor Per Christian Hansen. The group currently consists of 6 permanent staff members, 4 postdocs, and 9 PhD students. We consider inverse problems (such as image deblurring, tomographic imaging, source reconstruction, and fault inspection) and we apply methods from Bayesian inference to determine the solution’s sensitivity to errors and inaccuracies in the data, the models, etc. We develop both the mathematical foundations and a computational UQ platform, CUQIpy (https://cuqi-dtu.github.io/CUQIpy/) to enable intuitive and extensive application of UQ techniques to a range of inverse problems in academia and industry. Responsibilities and qualifications You will join the CUQIpy subgroup led by Senior Researcher Jakob Sauer Jørgensen and focus on developing new large-scale methods enabling Bayesian inference for high-dimensional inverse problems with large amounts of data. Some possible research directions include (but are not limited to) the development of application of: • Variational Bayesian methods for approximating high-dimensional, complicated posterior distributions enabling approximate inference at much reduced cost compared to classic MCMC methods. • Scalable optimization-based samplers such as randomize-then-optimize (RTO) techniques. • Sampling methods exploiting gradient and higher-order derivative information, for example through automatic differentiation techniques, and/or employing a multi-level approach. You are expected to publish research papers and give presentations at scientific conferences. An integral part of the role will be to develop the new methods within the CUQIpy Python package as well as contribute substantially to the general continued development of CUQIpy. Our working style is highly collaborative and supportive – and through your involvement in CUQIpy you will have lots of opportunities for joint computational research projects in collaboration with other members of the CUQI team. You will also be expected to contribute to training and support of CUQIpy users and to co-supervise MSc and PhD students. The position offers rich opportunities to build a strong profile in computational UQ for inverse problems and scientific software development. The ideal candidate has: • A strong background in computational inverse problems in the Bayesian setting. • Experience with one or more of the above-mentioned approaches or other approaches for large-scale computational UQ. • Excellent programming skills (ideally Python) and communication skills. • Experience with practical applications of inverse problems and real data. It is essential to be self-motivated and to thrive on collaboration and teamwork. Furthermore, good command of the English language is required. Candidates must have a PhD degree or equivalent in scientific computing, computational science & engineering, applied mathematics, or equivalent academic qualifications. If you do not have your diploma at the time of application, please provide a statement from your supervisor. Further information Further information may be obtained from Professor Per Christian Hansen (pcha@dtu.dk) and Senior Researcher Jakob Sauer Jørgensen: http://www.imm.dtu.dk/~jakj/ (mailto:jakj@dtu.dk). Application procedure To apply, please read the full job advertisement here: Postdoc in Large-Scale Computational Uncertainty Quantification for Inverse Problems: https://efzu.fa.em2.oraclecloud.com/hcmUI/CandidateExperience/en/sites/CX_1/job/1650/?utm_medium=jobshare Application deadline: 1 June 2023

Last updated: 27 April 2023