Research

Application

Our research ranges from the development of highly efficient quantum-chemical methods to the application thereof on large-scale research projects. The latter include the computation of small organic systems, biochemical macromolecules such as enzymes, DNA, and RNA (also in context of SFB1309 on epigenetic mechanisms), as well as, covalent organic (COF) and metal organic frameworks (MOF) in the context of energy-conversion processes (excellence-cluster e-conversion). For these challenging systems, we predict reaction mechanisms, molecular interactions, and molecular properties, including hyperfine coupling constants or nuclear magnetic resonance (NMR) shifts. Past and ongoing application projects include:

• characterization of chemical and biochemical reactions using multiscale modelling and extensive sampling
• plausible reaction pathways leading to the first building blocks of life
• NMR chemical shifts in extended chemical and biochemical systems
• application of enhanced sampling techniques, like umbrella sampling and adaptive biasing force simulations
• currently largest MP2 calculation on a DNA repair system with 2025 atoms and 20 371 basisfunctions
• structure determination in the solid state by assignment of MAS-NMR spectra

Development

To enable these large-scale projects, where calculations on systems with more than 1000 quantum chemically treated atoms are routinely performed, we develop our own quantum chemistry program package FermiONs++. The focus of our development efforts are thereby on the development of efficient linear- and sublinear-scaling methods for applications on large molecules with 1000 and more atoms. Past and ongoing development projects include:

• highly parallel, linear scaling methods to compute Fock-exchange using seminumerical integration (sn-LinK)
• tight and rigorous integral partition bounds for fast and effective screening of general electron repulsion integrals
• efficient formulations of the random phase approximation (RPA) and extension thereof to gradients and beyond RPA methods, such as second order screened exchange RPA (SOSEX-RPA)
• low scaling formulations of second-order Møller-Plesset perturbation theory (MP2) and extension thereof to first derivative properties, like molecular gradients and hyperfine coupling constants, as well as second derivative properties, like nuclear magnetic resonance shifts
• large-scale ab initio non-adiabatic molecular dynamics (NAMD) on hybrid CPU/GPU architectures
• methods for identifying free energy hot-spots and assigment of contributions by specific atoms or groups to the vibrational free energy