School of Mathematics and Natural Sciences

Mr PD Dr. Marcel Schweitzer

Lecturer

Mr PD Dr. Marcel Schweitzer

Contact

G.15.28

+49 202 439-3773

marcel[at]uni-wuppertal.de

Position Description

I am a lecturer in the "Scientific Computing and High Performance Computing" group. I am responsible for teaching "Software Engineering" and the corresponding practical course for Bachelor students. Additionally, I occassionally teach specialized Master courses on topics in applied mathematics and graph theory. My main research area is numerical linear algebra.

Office hour

Wednesday, 13:00-14:00 (during teaching term, otherwise only by arrangement)

Research interests

  • Approximation of matrix functions
  • Krylov subspace methods, in particular restarted methods
  • Randomized numerical linear algebra
  • Fréchet derivatives and condition numbers of matrix functions
  • Exploiting low-rank structures in numerical computations
  • Decay behavior in matrix functions
  • Applications of numerical linear algebra techniques in network analysis
  • Implicit/matrix-free trace estimation algorithms

List of publications

As principal investigator of the DFG-funded project “Matrix functions via randomized sketching” I work on randomized Krylov methods for matrix functions, together with my PhD student Emil Krieger.

Teaching

Teaching in summer semester 2026:

Link to all of my courses and supervised theses

Software

Software related to my research:

  • ks-res-ode: Randomized Krylov methods for ordinary differential equations
  • laplace_restarting: Implementation of a restarted Arnoldi method for Laplace transforms and Bernstein functions (MATLAB)
  • t-frechet: Toolbox for computing t-functions (and their Fréchet derivatives) of third-order tensors (MATLAB)
  • sketched_fAb: Randomized Krylov methods for the approximation of matrix functions (MATLAB)
  • FUNM_QUAD: An implementation of a stable quadrature-based restarted Arnoldi method for matrix functions (MATLAB)

Software developed by students:

  • VSWorkbench: Extensible VS Code Plugin for incorporating Developer Tools (Bachelor thesis Sufyan Dahalan)

The f(A)bulous scientific network for matrix functions and exponential integrators

I am coordinator of the DFG-funded scientific network “The f(A)bulous scientific network for matrix functions and exponential integrators” which brings together experts in matrix functions, exponential integrators and model order reduction and aims to improve

  • the efficient and reliable implementation of algorithms for matrix functions, especially in high-performance computing;
  • cross-pollination with exponential integrator techniques;
  • benchmarking and FAIR principles for matrix functions and exponential integrators; and
  • connections with matrix equations and model order reduction.

Communications in NLA

To allow young researchers in numerical linear algebra to also present their work during the Covid-19 pandemic, I organized the online seminar series "Communications in NLA" from September 2020 to May 2021, together with a few colleagues.

All talks have been recorded and can be watched on our Youtube channel.

Membership in editorial boards

I am a managing editor of Electronic Transactions on Numerical Analysis (ETNA).

CV