rsmith

A Reverse Monte Carlo (RMC) structure refinement tool written in Rust. Refines atomic structures against experimental X-ray and neutron scattering data while enforcing physical constraints and pair potentials.

Features

• S(Q) and g(r) fitting -- simultaneous refinement against structure factor and pair distribution function
• Hybrid RMC -- pair potentials (Buckingham, Pedone, Coulomb DSF, tabulated) bias refinement toward energetically favorable configurations
• Hard constraints -- minimum interatomic distances and coordination number bounds
• Simulated annealing -- exponential cooling schedule with adaptive step size
• Structural analysis -- coordination numbers and bond angle distributions for validation
• Incremental updates -- O(N_neighbors) per move via cell lists and precomputed lookup tables

Quick Start

cargo build --release
./target/release/rsmith config.toml

Minimal config.toml:

[system]
structure = "glass.data"
format = "lammps"

[system.types]
1 = "Ca"
2 = "Si"
3 = "O"

[data.xray_sq]
file = "experimental.sq"
sigma = 0.02

[rmc]
max_moves = 500_000

[sq]
rdf_cutoff = 11.0

[constraints.min_distance]
"Si-O" = 1.2
"O-O" = 2.0
"Ca-O" = 1.8

Modes

rsmith config.toml                      # RMC refinement
rsmith config.toml --compute-sq-only    # Compute S(Q) only
rsmith config.toml --analyze            # Structural analysis

Documentation

Full documentation is available at rsmith book or build locally:

cd docs && mdbook serve

References

• McGreevy, R.L. & Pusztai, L. (1988). Reverse Monte Carlo Simulation. Mol. Simul., 1, 359-367.
• Pedone, A. et al. (2006). A New Self-Consistent Empirical Interatomic Potential Model for Oxides. J. Phys. Chem. B, 110, 11780-11795.