Numerical Mathematics Research Group
The Numerical Mathematics Research Group is one of the research groups within the Department of Mathematics. We focus on the development and mathematical analysis of algorithms for the computer-aided solution of practical problems in the natural sciences, medicine, and engineering. In doing so, we represent the mathematical subdiscipline of numerical mathematics in both research and teaching.
Our Research Profile
The research topics of the Numerical Methods Research Group are wide-ranging and, on the application side, cover a variety of problems in the natural sciences, engineering, and medicine.
From a mathematical perspective, we investigate efficient numerical solution methods for partial differential equations as well as for related inverse parameter identification problems. The solution methods are basis-oriented and draw on finite element, Fourier, or wavelet systems.
Of particular interest to us are the convergence and complexity properties of adaptive discretization methods, which can autonomously compute an efficient and reliable numerical approximation based on a posteriori error estimators. For the mathematical analysis of such algorithms, we employ methods from functional analysis in Banach and Hilbert spaces, as well as from nonlinear approximation theory.
Efficient optimization algorithms also play an important role in the concrete numerical solution of inverse problems. Here, we are interested in methods of numerical nonsmooth optimization, which allow structural constraints of the sought solution to be better captured than with classical optimization strategies. In practice, for example in parametric regularization methods, multiscale optimization problems also arise, such as bilevel optimization problems. We are investigating novel numerical solution methods for these problems.
In some applications in the natural sciences, one also encounters the analysis of high-dimensional numerical fields, known as tensors. To this end, we investigate various modern methods of numerical multilinear algebra, e.g., for tensor decomposition and tensor eigenvalue problems.
Research Focus Areas
- Numerical Methods for PDEs
- Numerical Methods for Inverse Problems
- Convergence and Complexity of Adaptive Discretization Methods
- Numerical Fourier and Wavelet Methods
- Redundant Basis Systems
- Nonsmooth Numerical Optimization
- Bilevel Optimization
- Numerical Multilinear Algebra
Contact the Numerical Analysis Research Group
Postal address
University of Siegen
Department of Mathematics
Numerical Methods Group
Walter-Flex-Straße 3
57068 Siegen
Visitor address
University of Siegen
Department of Mathematics
Numerical Methods Research Group
ENC-B, Level 3
Walter-Flex-Straße 3
57072 Siegen
Secretariat
Please contact the office.