RTG 2450 / GRK 2450

Associated Projects

The main projects P1-7 are accompanied by a selection of associated projects funded mainly by external sources. The topics range from machine learning and philosophy of computer simulation to method development in computational catalysis and protein research.

 

Philosophy of Computer Simulation

"In the materials sciences [... the] use of computer simulations raises a series of philosophical questions that are also highly relevant in other fields of scientific research: Can the validation and verification of computer simulations be decoupled? Are computer simulations epistemically opaque? [...] What role do computer simulations play in understanding physical processes on different time and length scales? [...] What epistemic goals are related to the use of computer simulations? [...] What is, e.g., the epistemic role of visualizations?" Read more at  https://www.itas.kit.edu/english/projects_schw19_diss.php.

Contact: Julie Schweer

 

Applications of Machine Learning Methods to Quantum Chemical Simulations

Modern methods of statistical learning such as artificial neural networks are used in various scientific fields to approximate complex relationships between variables. In the field of quantum-chemical simulations, where parametrization of functions to empirical data is frequently used as a method for improving the accuracy of a method or reducing its computational cost, these methods can shine with their flexibility and ability to learn correlations for which no explicit functional form is known. The goal of this project is to apply machine learning methods in order to improve or speed up simulations of charge and exciton transfer, biochemical systems and chemical reactivity.

Contact: Mila Krämer

 

Method Development for Computational Catalysis

Reaction energetics in (ab initio) computational catalysis are typically derived from local harmonic approximations of the potential energy surface. However, this Harmonic Approximation (HA) can become inaccurate for some systems (e.g. at elevated temperatures or for weakly interacting systems during adsorption processes). This project combines molecular dynamics simulation, coordinate transformations and thermodynamic integration to compute anharmonic corrections to the HA.

Contact: Jonas Amsler

 

Deriving Protein Structures by Integrating Experimental Information into Biomolecular Simulations

Biomolecular simulations are a powerful tool to complement and interpret ambiguous experimental data on proteins to obtain structural models. This project revolves around developing a molecular dynamics based method to integrate small-angle X-ray scattering data into protein simulations using structure-based models. Such data-assisted simulations critically rely on parameters, where most importantly experimental information needs to be balanced with respect to the underlying physical model. To overcome this parameter selection problem, a self-adapting variant of dynamic particle swarm optimization is developed and tested.

Contact: Marie Weiel-Potyagaylo

 

Exciton Transfer Simulations in Organic and Biological Systems

Simulations of exciton transfer processes in huge molecular systems such as organic crystals and biological light-harvesting complexes are computationally challenging. The goal of this project is the development and application of a program for direct exciton transfer to study transport mechanisms and to calculate physical observables without any prior assumptions. This is realised utilizing non-adiabatic dynamics methods to simulate the dynamics of the coupled electronic and nuclear degrees of freedom in a QM/MM scheme (DFTB+/GROMACS). The electronic degrees of freedom are treated with a coarse-grained Frenkel Hamiltonian that is parametrized on-the-fly with quantum or machine learning approaches.

Contact: Philipp Dohmen

 

Contact
Name Title Group Contact
M.Sc. Prof. Felix Studt, KIT IKFT
M.Sc. Prof. Marcus Elstner, KIT IPC
M. Sc Prof. Marcus Elstner, KIT IPC
M.A. Prof. Rafaela Hillerbrand, KIT ITAS
  Dr. Alexander Schug, KIT SCC