JAX-ReaxFF: A Gradient Based Framework for Extremely Fast Optimization of Reactive Force Fields Traditional methods for optimizing ReaxFF parameters rely on slow, stochastic techniques like genetic algorithms or Monte Carlo methods. We present JAX-Reaxff, a new optimizer that leverages modern machine ...
from jax_md import util from jax_md import io from jax_md import rigid_body from jax_md import reaxff_helper from jax_md import reaxff_interactions from jax_md import reaxff_energy from jax_md import reaxff_forcefield from ._reaxff import reaxff_energy from ._reaxff import reaxff_int...
5 changes: 4 additions & 1 deletion 5 jax_md/reaxff/reaxff_interactions.py Original file line numberDiff line numberDiff line change @@ -573,7 +573,10 @@ def reaxff_inter_list(displacement: DisplacementFn, version of the torsion interactions solver_model: Control variable for the solver...
4 changes: 0 additions & 4 deletions 4 jax_md/__init__.py Original file line numberDiff line numberDiff line change @@ -25,10 +25,6 @@ from jax_md import util from jax_md import io from jax_md import rigid_body from ._reaxff import reaxff_energy from ._reaxff import reaxff...
2 changes: 1 addition & 1 deletion 2 jax_md/_energy/electrostatics.py Original file line numberDiff line numberDiff line change @@ -129,7 +129,7 @@ def energy_fn(position, **kwargs): S2 = jnp.abs(structure_factor(g, position, charge))**2 fn = lambda g2: jnp.exp(-g2 / (...
from jax_md import rigid_body try: # Attempt to load colab_tools if IPython is installed. from jax_md import colab_tools 4 changes: 4 additions & 0 deletions 4 jax_md/reaxff/__init__.py Original file line numberDiff line numberDiff line change @@ -0,0 +1,4 @@ from . import...