Current Diploma / PhD topics
For detailed explanations, possible other topics, and personal advice, please consult Bernd Hartke directly.
global optimization:
- method development:
- avoiding re-discovery of known minima (tabu search) (also allows for an estimate of search space covered)
- merger between an extended crossover operator and “cultural algorithm” ideas, for better prediction of new candidate solutions
- discovery of links between building block characteristics (valence, degree of directionality, range) and cluster structures, using simple model potentials
- improvement of “numerical” evolutionary algorithms (for system-specific reparametrization of semiempirics, or for potential fitting)
- applications:
- large clusters of Kanamycin A molecules with physiological cations (collaboration with Prof. Schröder, Dermatology)
- heterogeneous clusters: ligand hulls, hydration clusters, clathrate hydrates
- solvent clusters around reacting species, and their influence on the reaction (with DFG funding)
quantum-mechanical reaction dynamics:
- on-the-fly PES generation, interpolation instead of re-calculation of PES points
- further development of an adaptive-basis representation with collocation
- application to high-D examples, real-life applications e.g. within SFB677
classical-mechanical (direct) dynamics:
- refinement of existing approaches to system-specific reparametrization of semiempirical electronic structure methods (with inclusion of analytic 1st derivative formulae)
- surface-hopping (photo)chemical dynamics of molecular switches, without and with external forces (SFB 677)
- design of improved molecular switches and molecular motors
- extraction of “reaction mechanisms” from trajectory data
current and past diploma/PhD research topics:
aggregation of organic molecules: (collaboration project with Prof. Lüning): molecules interacting via multiple H-bonds
- validation of several quantum chemical approaches, in comparison to experimental data
- explanation of existing data on different binding strengths of several molecule pairs
- development of design recommendations for new molecule pairs with improved binding strength and specifity
modeling of molecular rotors: (collaboration with Prof. Herges)
- classical-mechanical models:
- optimization of rotor performance with parameter tests in one-dimensional models
- full-dimensional molecular dynamics of rotor molecules in action
- electronic structure calculations:
- ab-initio and semiempirical calculations of potential energy surfaces for rotor molecules
- adjustment of simple 1D models and/or new force field terms to ab-initio calculations
quantum-mechanical reaction dynamics:
- sparse wavefunction representation:
- simplification of existing 1D proof-of-principle code, replacing numerical quadrature by collocation
- extension from 1D to high-D
- combination with existing high-end propagation code (Frank von Horsten)
- application to high-D examples
- real-life applications: Prof. Rauhut, Prof. Temps, …
- direct collaboration with numerical mathematics (via Center for Numerical Simulations)
protein folding:
- method development:
- development of a statistical, united-atom potential of a new type
- improvements and applications of an existing folding code
- applications:
- real-life proteins from the PDB data bank
large-scale molecular dynamics of hydrated oligosaccharides: (collaboration project with Prof. Lindhorst)
- method development:
- force field validation/improvement/extension
- improvement and extension of analysis tools
- applications:
- influence of oligosaccharides on the surrounding water, and vice versa
- changes of hydration properties of ions in the neighborhood of oligosaccharides
- binding free energies between FimH and oligosaccharides, in physiological water
- inclusion of membrane models
global optimization:
- method development:
- improvement of existing codes: extension to arbitrary mixtures of arbitrary molecules, with rigid and flexible degrees of freedom in arbitrary mixture
- avoiding re-discovery of known minima (tabu search)
- improvement of “numerical” evolutionary algorithms (for system-specific reparametrization of semiempirics, or for potential fitting)
- design of new molecules for specific tasks, via global optimization of substituent patterns
- applications:
- benchmark functions
- large, heterogeneous Lennard-Jones clusters
- small Kanamycin A clusters
Last updated on Mon Aug 23, 2010