lj

Lennard-Jones force terms for reduced-unit simulations.

import mlx_atomistic.lj

Classes

LennardJonesPotential

class LennardJonesPotential
    def __init__(epsilon: float = 1.0, sigma: float = 1.0, cutoff: float | None = 2.5, shift: bool = True, topology: Topology | None = None, one_four_scale: float = 1.0, backend: NonbondedBackend = 'auto', tile_size: int = 512, memory_budget_bytes: int | None = DEFAULT_DENSE_MEMORY_BUDGET_BYTES, name: str = 'lj', supports_virial: bool = True, analytic_virial_supported: bool = False, use_fused_kernel: bool = False)

Naive all-pairs Lennard-Jones potential in reduced units.

Parameters

Name	Type	Default	Description
epsilon	float	1.0
sigma	float	1.0
cutoff	float | None	2.5
shift	bool	True
topology	Topology | None	None
one_four_scale	float	1.0
backend	NonbondedBackend	'auto'
tile_size	int	512
memory_budget_bytes	int | None	DEFAULT_DENSE_MEMORY_BUDGET_BYTES
name	str	'lj'
supports_virial	bool	True
analytic_virial_supported	bool	False
use_fused_kernel	bool	False

Methods

energy_forces

def energy_forces(positions: mx.array, cell: Cell | None = None, pairs: object | None = None) -> tuple[mx.array, mx.array]

Return potential energy and forces for positions with shape (n_particles, 3).

Parameters

Name	Type	Default	Description
positions	mx.array		Particle coordinates, shape (n_particles, 3).
cell	Cell | None	None	Optional periodic cell for minimum-image distances. Defaults to None.
pairs	object | None	None	Optional neighbor list, neighbor blocks, or dense pair array; the nonbonded backend is chosen automatically. Defaults to None.

Returns

• tuple[mx.array, mx.array] — An (energy, forces) tuple: scalar LJ energy and forces of shape (n_particles, 3).

Raises

• ValueError — If positions is not (n_particles, 3) or a lazy topology is used without a runtime pair provider.