Klaus Leppkes

Contact Details:

  • Phone: +49 (0)241/80 20127
  • Mail: leppkes@stce.rwth-aachen.de
  • Office:  Room 240, Seffenter Weg 23

Work at STCE

  • Development of dco including various frontends like Fortran, .NET, python and matlab.
  • Development of GPGPU adjoint software (dco/map)
  • Development of the NAG AD Library.
  • Thinking about numerical optimization and other problems...:)

Area of expertise

  • Automatic Differentiation
  • Applying dco to non-trivial C++ and Fortran Codes.
  • HPC
  • Coding in C++, Fortran, python
  • x86 ecosystem and how to program it efficiently.
  • Numerical Optimization

Publications

  • U. Naumann, J. Lotz, K. Leppkes, M. Towara: Algorithmic differentiation of numerical methods: Tangent and adjoint solvers for parameterized systems of nonlinear equations. ACM Transactions on Mathematical Software (TOMS) 41.4 (2015).
  • Sagebaum, M., Gauger, N. R., Naumann, U., Lotz, J., & Leppkes, K. Algorithmic differentiation of a complex C++ code with underlying librariesProcedia Computer Science18, 208-217 (2013).
  • J. Lotz, K. Leppkes, J. Ungermann, and U. Naumann. Efficient discrete adjoint computation in a spectral simulation code. Technical Report AIB 2011-05, RWTH Aachen, 2011.
  • Ungermann, J., Blank, J., Lotz, J., Leppkes, K., Guggenmoser, T., Kaufmann, M., Preusse, P., Naumann, U., and Riese, M.: A 3-D tomographic trajectory retrieval for the air-borne limb-imager GLORIA, Atmospheric Measurement Techniques Discussions 4(3), 3805–3859, 2011
  • Leppkes, K. and Naumann, U.: DCO/C – Efficient Derivative Code by Overloading in C++, 10 Tech. Rep. AIB-2011-06, RWTH Aachen, 2011
  • Leppkes, Klaus: Effiziente Berechnung von Adjungierten bei Fixpunktiterationen, GI-Edition - Lecture Notes in Informatics (LNI), volume S-10, Bonner Köllen Verlag, 23–26, Eds: Gesellschaft für Informatik , 2011
  • Rauser, Florian, Riehme, Jan, Leppkes, Klaus, Korn, Peter, and Naumann, Uwe: On the use of discrete adjoints in goal error estimation for shallow water equations, Procedia Computer Science 1(1), volume 1, 107–115, 2010, ICCS 2010
  • Leppkes, K., and Naumann, U.: On the use of direct vs. iterative sparse linear solvers in adjoint computations, RWTH Aachen University No. AIB-2009-22, September 2009

Talks

  • Algorithmic Differentiation by Overloading: dco/c++, STCE retreat, 2012, Simonskall, Germany
  • On Call Tree Reversal in adjoint mode AD by overloading, Twelfth Euro AD Workshop, 2011, Berlin, Germany
  • Effiziente Berechnung von Adjungierten bei Fixpunktiterationen, Informatiktage 2011, 2011, Bonn, Germany

Presentations

  • K. Leppkes and U. Naumann: Optimization of Support Structures, The 5th International Conference on Automatic Differentiation, 2008, Bonn, Germany
  • K. Leppkes and U. Naumann: Direct vs. iterative sparse linear solvers in adjoint computations, SIAM Workshop on Combinatorial Scientific Computing (CSC09) , 2009, Monterey Bay (California), United States
  • K. Leppkes: Effiziente Berechnung von Adjungierten bei Fixpunktiterationen, Informatiktage 2011, 2011, Bonn, Germany

Short CV

  • born December 1982 in Kleve, Germany
  • Education
    • 1993-1999 Johanna Sebus Gymnasium: Fachoberschulreife (equi. Specialized High School Maturity)
    • 1999-2002 GEOSOFT Vermessungssysteme GmbH: Ausbildung Fachinformatiker (equi. job training IT specialist)
    • 2004-2006 Abendgymnasium Kempen: Abitur (equi.A-Level)
    • 2006-today Rhein Rhein-Westfälische Technische Hochschule: studying Computational Engineering Science
  • Employment
    • 2002-2006 GEOSOFT Vermessungssysteme GmbH: software developer for geodetic calculation programs
    • 2007-2009 Fraunhofer Institute for Laser Technology ILT: student researcher
    • 2009-today LuFG Informatik 12: Software and Tools for Computational Engineering: student researcher