<br/>Condensed phase chemical or biomolecular systems often involve thousands of atoms and even more degrees of freedom. Every description or model of such a complex system is caught in a dilemma of resolution. On the one hand a description needs to be very detailed to capture all aspects of the behavior of a system accurately. On the other hand, detail comes with a number of drawbacks. Detailed models are more expensive to evaluate and therefore the accessible system sizes and time scales are limited. Furthermore, detailed descriptions are more complicated to analyze and important insights tend to be hidden from human observers. To address this dilemma, it is inevitable to travel through different levels of resolution. Detailed models are required for an accurate description while coarser models can sample larger parts of phase pace, and even coarser representations are required to give human observers insights into the relevant features otherwise hidden in high dimensional data. We use different machine learning methods to establish links and to seamlessly bridge between levels of resolution which allows for an efficient framework to describe systems on multiple scales.

Travelling through levels of resolution with machine-learning methods

2019 Spring National Meeting

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.