The Journal of Chemical Physics, Journal Year: 2025, Volume and Issue: 162(7)
Published: Feb. 18, 2025
We compare three model-free numerical methods for inverting structural data to obtain interaction potentials, namely, iterative Boltzmann inversion (IBI), test-particle insertion (TPI), and a machine-learning (ML) approach called ActiveNet. Three archetypal models of two-dimensional colloidal systems are used as test cases: Weeks–Chandler–Anderson short-ranged repulsion, the Lennard-Jones potential, repulsive shoulder with two length scales. Additionally, on an experimental suspension spheres acquired by optical microscopy methods. The have different merits. IBI is only choice when radial distribution function known but particle coordinates unavailable. TPI requires snapshots positions can extract both pair- higher-body potentials without need simulation. ML be particles tracked in time it returns force rather than potential. However, unravel pair interactions from any one-body forces (such drag or propulsion) does not rely equilibrium distributions its derivation. Our results may serve guide method needed application reference further development methodology itself.
Language: Английский