Section: Scientific Foundations

Overview

The adaptive simulation algorithms we develop typically consist in two main components. The first one determines which degrees of freedom are simulated at a give time step, based on the current system's state, as well as user-defined precision or cost thresholds. The second component incrementally updates the system's state based on the set of active degrees of freedom. In particular, incremental algorithms update the system's potential energy and forces. This allows the user to smoothly trade between precision and cost.

We detail this approach in two important types of simulations: Cartesian quasi-statics and torsion-angle dynamics. A novel, very general approach for adaptive dynamics simulations of particles — that has a number of important benefits over previous approaches — is mentioned in more detail in Section 6.1 .