HECATE

Hybrid Electrochemical Computation for Atomistic Transport and Evolution

HECATE is a research framework for hybrid atomistic simulation of electrochemical interfaces. It combines charge-equilibration molecular dynamics, machine-learned interatomic potentials, on-the-fly local event discovery, kinetic Monte Carlo, and post-event molecular dynamics in one reproducible workflow.

The current implementation targets solid-electrolyte interphase evolution in lithium batteries, while its core event catalogue, clock, local-search, and potential interfaces are system-independent.

Workflow

One chemical epoch contains four phases:

1. rigid classical QEq-MD for electrolyte transport;
2. full-degree-of-freedom DP-QEq MD;
3. local DP-QEq event discovery, endpoint screening, CI-NEB, and KMC;
4. full-degree-of-freedom DP-QEq MD after the selected event.

The event-preparation phases share one kinetic clock and a common state representation. Post-event MD adds physical time without reusing the consumed event threshold. Reactive MD transitions invalidate state-dependent event data before the next search.

Highlights

• constant-potential charge equilibration with a JAX/DMFF PME backend;
• Deep Potential support with optional multi-model uncertainty screening;
• LAMMPS backend for rigid classical transport;
• local spherical searches with bond-directed or random-displacement candidates;
• live-frame GPU batching without duplicated or padded candidates;
• exact-QEq endpoint checks and climbing-image NEB;
• global event catalogue, BKL selection, and exact phase-type superbasin exits;
• one JSON file for all run-time simulation parameters.

Installation

HECATE requires Python 3.10 or newer.

python -m pip install -e .

The minimal installation provides configuration, structure, clock, catalogue, and analytic-potential functionality. Production DP-QEq calculations also require a compatible DeepMD-kit installation and a GPU-enabled JAX/DMFF stack. Rigid transport requires the LAMMPS Python module; CUDA/Kokkos is optional.

Model files, fitted force-field parameters, initial charge files, and atomistic input structures are not distributed in this repository.

Quick start

Copy and edit the supplied JSON template. All relative paths are resolved from the directory containing the JSON file.

cp examples/parameters.example.json parameters.json
hecate check parameters.json
hecate classify POSCAR
hecate run parameters.json

See Configuration for the parameter groups and input requirements.

Repository contents

hecate/                         Python implementation
examples/parameters.example.json  sanitized run template
docs/CONFIGURATION.md           user-facing configuration guide
CHANGELOG.md                    public release history

Research status

HECATE is research software under active development. Before using generated rates for scientific conclusions, users should validate the potential, charge-equilibration model, event-search convergence, barrier window, and timescale treatment for their own system.

No software license has been assigned to this initial public snapshot.