curvedSpaceSim: A framework for simulating particles interacting along geodesics

Toler H. Webb,Daniel M. Sussman

2024-05-01

Abstract:A large number of powerful, high-quality, and open-source simulation packages exist to efficiently perform molecular dynamics simulations, and their prevalence has greatly accelerated discoveries across a wide range of scientific domains. These packages typically simulate particles in free (Euclidean) space, with options to specify a variety of boundary conditions. While more exotic, many physical systems are constrained to and interact across curved surfaces, such as organisms moving across the landscape, colloids pinned at curved fluid-fluid interfaces, and layers of epithelial cells forming highly curved tissues. The calculation of distances and the updating of equations of motion in idealized geometries (namely, on surfaces of constant curvature) can be done analytically, but it is much more challenging to efficiently perform molecular-dynamics-like simulations on arbitrarily curved surfaces. This article discusses a simulation framework which combines tools from particle-based simulations with recent work in discrete differential geometry to model particles that interact via geodesic distances and move on an arbitrarily curved surface. We present computational cost estimates for a variety of surface complexities with and without various algorithmic specializations (e.g., restrictions to short-range interaction potentials, or multi-threaded parallelization). Our flexible and extensible framework is set up to easily handle both equilibrium and non-equilibrium dynamics, and will enable researchers to access time- and particle-number-scales previously inaccessible.

Soft Condensed Matter

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to efficiently perform particle dynamics simulations on arbitrary surfaces. Specifically, existing molecular dynamics simulation software usually runs in flat (Euclidean) space, while many physical systems actually interact on surfaces. For example, organisms move on landscapes, colloids are fixed at curved - liquid interfaces, and epithelial cells form highly curved tissue layers. The distance calculation and the update of the equations of motion in these systems can be analytically completed in idealized geometries (such as surfaces of constant curvature), but it is very challenging on arbitrary surfaces.

To meet this challenge, the author has developed a framework named `curvedSpaceSim`, which combines the latest advances in particle simulation tools and discrete differential geometry to simulate particles that interact through geodesic distances and move on arbitrary surfaces. This framework can handle both equilibrium and non - equilibrium dynamics, and can access previously unreachable time scales and particle number scales. In addition, the paper also discusses the estimation of computational costs under different surface complexities and the impact of various algorithm optimizations (such as the limitation of short - range interaction potentials, multi - threading parallelization, etc.).