Skip to main content

About the program

The scientific topic

Scientific machine learning (SciML) for simulation and inverse modelling refers to the development of theory/algorithms for combining data-driven models from machine learning with domain-specific handcrafted models from applied mathematics and scientific computing with applications to simulation and inverse modelling. Some key sub-topics that we hope to explore during the program are domain awareness, interpretability, and robustness, all in the context of SciML for simulation and inverse modelling.

SciML offers new ways for addressing grand challenges in science/engineering that have been difficult to address with methods from traditional scientific computing and where purely data-driven machine learning approaches are unlikely to succeed. In addition, the hybrid modelling paradigm that is at the core of SciML may not only yield discoveries in science and new methodologies in engineering but it is also expected to transform research in applied mathematics.

Format of the program

The main part of the program is organised as a "research retreat” that can be viewed as version of the research-in-groups  format offered by the International Centre for Mathematical Sciences (ICMS)  in Edinburgh but with larger groups. Alternatively, one can also view the program as a shorter variant of the semester-long research programs offered by many institutes, like Institut Mittag-Leffler , Institute for Pure & Applied Mathematics (IPAM) , Institut Henri Poincaré , Oberwolfach Research Institute for Mathematics , and the Isaac Newton Institute for Mathematical Sciences .

  • Scientific activities: There will be a mid-term conference/workshop (3-5 June) that gathers most core participants. The specific format and speakers for the mid-term workshop are not yet decided, but expect a mix of talks related to deep learning for PDEs with talks on deep learning for inverse problems. This will be complemented with informal research seminars at regular intervals organised within the program.
  • Social activities: There will be a smaller ”opening event” second week into the program and a similar ”closing event” towards the end of the program. The mid-term conference/workshop will include a joint dinner for all participants. Finally, we plan to have daily joint morning coffee/breakfast (arranged by the program) and a weekly joint dinner (arranged by the program) for invited participants.
Belongs to: Scientific Machine Learning for Simulation and Inverse Modelling
Last changed: Mar 13, 2024