Molecular dynamics simulations play a crucial role in understanding and designing new materials, particularly in cases where experimental observations are impossible and multiple design parameters are relevant. However, many interesting processes—e.g., chemical reactions, conformational changes, phase transitions, etc.—require prohibitively long simulation times in standard simulations. This is known as the “rare event” problem.
To overcome it, one can use different kinds of enhanced sampling algorithms to drive the system away from already visited configurations and toward the desired product. By applying such accelerated simulations in combination with evolutionary algorithms, we can efficiently design the self-assembly of novel materials with tailor-made structures and properties (see https://doi.org/10.1021/acsnano.4c17597, https://doi.org/10.1063/5.0210034). Moreover, we can quantify how likely the target phase is to be successfully formed under different conditions in terms of free energy, i.e., log probability, and assess possible competing byproduct phases.
In this project, your goal will be to integrate different enhanced sampling algorithms and evolution strategies in order to design complex materials. You will have the opportunity to become familiar with multiple state-of-the-art methods in simulation, optimization, machine learning, etc. Experience with optimization methods and/or molecular simulation is desirable.
