Steffen Seckler, M.Sc. (hons)

Address:

TU München

Institut für Informatik

Boltzmannstr. 3

85748 Garching b. München

Office:

MI 02.05.036

Email:

Phone:

(089) 289 18 616

Fax:

(089) 289 18 607

Office Hours:

by arrangement

Student Projects

Open Student Projects

Topics change (frequently) with my ongoing work and depend mostly on the interests and experience of the student. It is best to contact me directly if you are interested in a thesis or student project. Nevertheless, here is a (potentially outdated) list of topics:

• Load Balancing 1
  • Flexible Partitioning Schemes -> Adaption to non-cubic partitions
  • Diffusive Load Balancing (HPC, Software Engineering) 1 2, 3
  • Advanced Heterogeneous Load Balancing for MD Simulations (HPC, Math) 1, 2, 3
  • Adaption of the Fast-Multipole-Method for KD-Trees (Math, Software Engineering)
  • Adaptive Communication Schemes Between Different Zonal Methods (Software Engineering, Load Balancing), Papers see below
• Zonal Methods 1, 2, 3
  • Flexible Software Design for Zonal Methods (Software Engineering)
  • Adaptive Communication Schemes Between Different Zonal Methods (Math, Software Engineering, Load Balancing)
• Overlapping Computation and Communication:
  • Developing new Overlapping Schemes (HPC, MPI)
  • Overlapping of Global Communication (HPC, MPI)
• GPU-Kernels for Molecular Dynamics (HPC, Cuda, OpenCL)
• Parallel in Time Molecular Dynamics (HPC, Math) 1
• Invasic Molecular Dynamics (HPC, MPI)

(Note: topics marked with "Math" can usually be used for an IDP.)

To get an idea of further possible topics, have a look at the running and finished projects below, as well as the following links: Molecular Dynamics Simulations, High Performance Computing, Nikola Tchipev, M.Sc..

Finished Student Projects

• S. Sauermann: Implementation and Optimization of the Midpoint Method in ls1-mardyn [BibTeX].
  Bachelor's thesis, Institut für Informatik, Technische Universität München, September 2017.
• T. Schilling: Neighbor Lists in Molecular Dynamics Simulations: Implementation and Parallelization [BibTeX].
  Bachelor's thesis, August 2017.
• M. Müller: Neighborhood Lists in Molecular Dynamics Simulations: SIMD-Vectorization [BibTeX].
  Bachelor's thesis, August 2017.
• S. Griebel: Optimizing Load Balancing of Heterogeneous Particle Simulations on Heterogeneous Systems [pdf] [BibTeX].
  Bachelor's thesis, Institut für Informatik, Technische Universität München, July 2017.
• A. Schmelz: Optimization of Molecular Dynamics Simulations Using One-Sided MPI-Directives [pdf] [BibTeX].
  Master's thesis, Institut für Informatik, Technische Universität München, June 2017.
• J. Spahl: Evaluation of Zonal Methods for Small Molecular Systems [pdf] [BibTeX].
  Bachelor's thesis, Institut für Informatik, Technische Universität München, November 2016.

Teaching

• Modellbildung und Simulation SS16
• Scientific Computing 1 WS16/17
• Modellbildung und Simulation SS17

FMM Short Workshop

These files have been used at the chair retreet 2017. It uses the pyfmm code published on http://pyfmm.googlecode.com/ under the MIT license. It has been modified for the 2-D charge potential. In the workshop the participants were required to add the translation operators.

• https://www5.in.tum.de/mardyn/pyfmm-nbody.zip
• https://www5.in.tum.de/mardyn/pyfmm-nbody-solution.zip

Publications

• J. Vrabec, M. Bernreuther, H.-J. Bungartz, W.-L. Chen, W. Cordes, R. Fingerhut, C. W. Glass, J. Gmehling, R. Hamburger, M. Heilig, M. Heinen, M. T. Horsch, C.-M. Hsieh, M. Hülsmann, P. Jäger, P. Klein, S. Knauer and Ködder: SkaSim – Skalierbare HPC-Software für molekulare Simulationen in der chemischen Industrie [BibTeX].
  In Chemie Ingenieur Technik, Volume 90(3), p. 281, 295–306. WILEY-VCH Verlag GmbH & Co. KGaA, March 2018.
• S. Seckler, N. Tchipev, H.-J. Bungartz and P. Neumann: Load Balancing for Molecular Dynamics Simulations on Heterogeneous Architectures [pdf] [BibTeX].
  In 2016 IEEE 23rd International Conference on High Performance Computing, p. 101–110, December 2016. This work represents IEEE-copyrighted material.
• S. Seckler: The Fault Tolerant Combination Technique in an Iterative Refinement Framework [pdf] [BibTeX].
  Master's thesis, Institut für Informatik, Technische Universität München, October 2015.
• R. Arora, T. Borbath, D. D'Angella, N. Hoppe, P.-C. Sarbu, S. Seckler and S. Stemmer: SAPIENS - A New Generation MR Spectroscopy Processing, Analysis and Visualization Software (BGCE Honours Project) [BibTeX].
  Studienarbeit/SEP/IDP, Institut für Informatik, Technische Universität München, April 2015.

Talks

• F. Gratl, N. Tchipev, S. Seckler, P. Neumann and H.-J. Bungartz: Node-Level Performance for Partile Simulations [BibTeX].
  TaLPas Project Meeting 09 May 2018, VISUS, Stuttgart, May 2018.
• F. Gratl, N. Tchipev, S. Seckler, P. Neumann and H.-J. Bungartz: Task-Based Approaches for Molecular Dynamics Simulations [pdf] [BibTeX].
  89th GAMM Annual Meeting, Munich, March 2018.
• S. Seckler, S. Griebel, N. Tchipev, P. Neumann and H.-J. Bungartz: Fully Heterogeneous Load Balancing in ls1 MarDyn [BibTeX].
  SIAM Conference on Parallel Processing for Scientific Computing, Tokyo, Japan, March 2018.
• N. Tchipev, S. Seckler, H.-J. Bungartz and P. Neumann: Node-Level Performance for Partile Simulations [BibTeX].
  TaLPas Project Meeting and Colloquium 26/27 October 2017, DKRZ, Hamburg, October 2017.
• N. Tchipev, A. Costinescu, S. Seckler, P. Neumann and H.-J. Bungartz: Towards Autotuning Between OpenMP Schemes for Molecular Dynamics on Intel Xeon Phi [BibTeX].
  SIAM Conference on Computational Science and Engineering, Atlanta, Georgia, March 2017. MS182 Challenges and Approaches for Enabling Large-Scale Molecular Simulation - Part II of II.
• N. Tchipev, S. Seckler, A. Costinescu, P. Neumann and H.-J. Bungartz: Optimizations and design changes in ls1-mardyn [BibTeX], Kaiserslautern, October 2016.
• N. Tchipev, A. Costinescu, S. Seckler, P. Neumann and H.-J. Bungartz: Molecular Dynamics on KNL with ls1-mardyn [BibTeX].
  Intel Collaboration Hub, ISC High Performance 2016, Frankfurt, June 2016.
• N. Tchipev, S. Seckler, P. Neumann and H.-J. Bungartz: Optimizing ls1-mardyn for Xeon Phi [BibTeX].
  IXPUG/EMEA Workshop, Ostrava, Czech Republic, March 2016.