Improving Computational Workflows and Representations in Dense Polymer Simulations
Faculty Mentor Information
Dr. Eric Jankowski, Boise State University; and Stephanie McCallum, Boise State University
Presentation Date
7-15-2026
Abstract
Initializing polymers is a universal problem for molecular dynamics simulations, especially dense systems. We implement dissipative particle dynamics to generate non-overlapping polymer chains and measure performance versus system sizes across a wide range of parameters to identify conditions of a faster, more reliable, and less computationally expensive initialization. We focus on the highest packing fraction (1.4 unit-sphere simulation elements per unit volume) and perform upwards of 1,000 simulations on both laptops and graphics processing units. We find the present methods reliably initialize one-million particle systems 3500x more efficiently than previous work. We discuss use of these configurations as starting points for simulation of Kremer-Grest chains, whose higher sensitivity to topological error demonstrate that our protocol can be used today for production workflows studying entangled polymer structure and dynamics.
Improving Computational Workflows and Representations in Dense Polymer Simulations
Initializing polymers is a universal problem for molecular dynamics simulations, especially dense systems. We implement dissipative particle dynamics to generate non-overlapping polymer chains and measure performance versus system sizes across a wide range of parameters to identify conditions of a faster, more reliable, and less computationally expensive initialization. We focus on the highest packing fraction (1.4 unit-sphere simulation elements per unit volume) and perform upwards of 1,000 simulations on both laptops and graphics processing units. We find the present methods reliably initialize one-million particle systems 3500x more efficiently than previous work. We discuss use of these configurations as starting points for simulation of Kremer-Grest chains, whose higher sensitivity to topological error demonstrate that our protocol can be used today for production workflows studying entangled polymer structure and dynamics.