Adaptive Resolution Approach in Simulation of Complex Polymer Structures

The study of polymer-based composites is a challenging multiscale problem, involving multiple time and length scales. 
From specific interface interactions to macroscopically observable mechanical failure, the time and length scales range from femtosecond and sub-angstrom to years and metres, respectively. 
Computer simulations predicting the behaviour at the largest scales from the finest atomistic details remain impossible in the foreseeable future. 
Therefore, coarse-grained models are usually used, which approximate the material behaviour at the level of interest, but at the expense of losing information.
However, to understand processing-structure-property relationships of the new materials, also atomistic/molecular structure information is required.

This thesis tackles two of the main problems in the molecular simulation of polymeric materials: simulation of chemical reactions at the coarse-grained level and reverse mapping to the fine-grained level.

The importance of simulating chemical reactions is especially visible in composite materials where the processes at the interface can influence the macroscopical properties.
The reactive processes can already be simulated, in great details, using methods known from computational chemistry.
However, the most significant drawback is the computational cost required to run the simulation.
The classical molecular dynamics, using so-called non-reactive force-fields, proved its ability to predict properties of large systems at long timescales.
For few decades, significant efforts have been put to incorporate chemical processes into the classical MD simulation, both into the fine-grained and the coarse-grained representations.
The first part of this thesis is devoted to this effort.
We propose a general method that allows simulating a step-growth polymerization process with the possibility of splitting the water molecules.

Although the coarse-grained simulation can already predict certain properties of the simulated systems, still the fine-grained representation can be necessary to calculate specific properties such as permeability or the glass transition temperature.
The second part of this work is devoted to a method that allows seamlessly transit from the coarse-grained to fine-grained representation.
We proposed a generic reverse map approach that can be used for various polymeric systems, from simple polymer melts of linear chains to complex polymer networks.

The solutions proposed in both of the parts are firmly influenced by the adaptive resolution approach, which was developed to simulate systems composed of two representations: a coarse-grained and a fine-grained.