Dr. rer. nat. Michael Obersteiner

Background

• Started stuides in Biochemistry in 2010
• Switched to Computer Science in 2011 and finished M.Sc. in 2016
• Currently working on my dissertation in Computer Science with a focus on high dimensional numerics and the Sparse Grid Combination Technique

Research interests

• Molecular Dynamics
• HPC
• Combination technique
• Sparse Grids Methods
• Machine Learning

Projects

DisCoTec (distributed Combination Technique):

A C++ code that targets exascale computing with the Combination Technique for time-dependent PDEs. It was created in the course of the EXAHD project founded by the DFT and was part of SPPEXA. The code is parallelized via MPI and offers massively parallel execution of PDEs with arbitrary black box solvers that are provided by the user and integrated via adapters. The code is open-source and can be found on Github:

https://github.com/SGpp/DisCoTec

sparseSpACE (the Sparse Grid spatially adaptive combination environment):

A python code that implements the spatially adaptive variants of the Combination Technique developed during my disseration. It is designed to support arbitrary operations on the generated grids so that it potentially targets all Sparse Grid and Combination Technique applications. Currently there are implementations for numerical integration and interpolation, Uncertainty Quantification, Machine Learning with Density Estimation and PDE simulations. The code is open-source and published on Github:

https://github.com/obersteiner/sparseSpACE

Open and Running Student Projects

Open Hiwi Positions

Open Student Projects

see: https://www5.in.tum.de/wiki/index.php/Projects_in_Sparse_Grids_and_High_Dimensional_Approximation

You can also come to my office and discuss possible topics.

Running Student Projects

• P. Resch: Adaptive Romberg-Quadrature for the Sparse Grid Combination Technique, Bachelor's Thesis, Fakultät für Informatik, Informatics, since May 2020