Project B2: Many-body effects and optimized mapping schemes for systematic coarse-graining

The first goal of the B2 project is to provide the consortium with a platform for systematic coarse-graining via the open-source software package “Versatile Object-oriented Toolkit for Coarse-graining Applications” (VOTCA). Projects requiring swift parameterizations of coarse-grained models have already benefited from using this toolkit. The second goal is the development of coarse-grained potentials that capture more accurately many-body effects, by going beyond standard pair-wise interactions. To this end, we develop and test various coarse-graining strategies based on short-range three-body, local-density-dependent, and local-conformation-dependent potentials. Further, we devise optimized mapping schemes for coarse-grained representations using machine-learning techniques: In the previous funding period, we trained artificial neural networks for structural coarse-graining, and kernel-based methods to develop a general model for three-body potentials. Building upon our previous research, we will advance our coarse-graining strategies to better reproduce conformational details and dynamics, and also expand our scope to the automatic detection of coarse-grained variables and improved back-mapping schemes. Both projects contribute to the core development of the VOTCA package in the direction of the many-body coarse-grained potentials, justifying the collaborative proposal of two PIs. The developed techniques will be tested on a set of small organic molecules in a collaborative effort with MERCK Darmstadt, targeting in silico prediction of morphological, optical, and electronic properties of vacuum-deposited thin organic films.

Employing Artificial Neural Networks to Identify Reaction Coordinates and Pathways for Self-Assembly
Jörn H. Appeldorn, Simon Lemcke, Thomas Speck, Arash Nikoubashman
The Journal of Physical Chemistry B 126 (27), 5007-5016, (2022)
see publication

Ultra-coarse-graining of homopolymers in inhomogeneous systems
Fabian Berressem, Christoph Scherer, Denis Andrienko, Arash Nikoubashman
Journal of Physics: Condensed Matter 33 (25), 254002 (2021)
see publication

Computing inelastic neutron scattering spectra from molecular dynamics trajectories
Thomas F. Harrelson, Makena Dettmann, Christoph Scherer, Denis Andrienko, Adam J. Moulé, Roland Faller
Scientific Reports 11 (1), (2021)
see publication

BoltzmaNN: Predicting effective pair potentials and equations of state using neural networks
F. Berressem and A. Nikoubashman
Journal of Chemical Physics 154, 124123 (2021)
see publication

Kernel-Based Machine Learning for Efficient Simulations of Molecular Liquids
Christoph Scherer, René Scheid, Denis Andrienko, Tristan Bereau
Journal of Chemical Theory and Computation 16 (5), 3194-3204 (2020)
see publication

Understanding three-body contributions to coarse-grained force fields
Christoph Scherer, Denis Andrienko
Physical Chemistry Chemical Physics 20 (34), 22387-22394 (2018)
see publication

The PCPDTBT Family: Correlations between Chemical Structure, Polymorphism, and Device Performance
G. L. Schulz, F. S. U. Fischer, D. Trefz, A. Melnyk, A. Hamidi-Sakr, M. Brinkmann, D. Andrienko, S. Ludwigs
Macromolecules 50 (4), 1402-1414 (2017)
see publication

Computational materials discovery in soft matter
T. Bereau, D. Andrienko, K. Kremer
APL Materials 4, 053101 (2016)
see publication

Comparison of systematic coarse-graining strategies for soluble conjugated polymers
Christoph Scherer and Denis Andrienko
The European Physical Journal Special Topics 225, 1441-1461, (2016)
see publication