xnn.common.models

Modules

base	The contract every model obeys.
charge_solve	Global charge solve for LES: charge equilibration with the latent Ewald kernel.
d3	Grimme DFT-D3 London dispersion: a geometry-dependent correction for any xnn model.
d4	DFT-D4 London dispersion: a charge-dependent correction for any xnn model.
dispersion	Shared machinery of the dispersion add-ons (DFT-D3, DFT-D4).
dispersion_fast	Fast paths of the dispersion terms (DFT-D3, DFT-D4).
eeq	Large-system EEQ charges: matrix-free operator, solvers, implicit differentiation.
electrostatics	Point-charge electrostatics shared by the charge-predicting models.
fast	Fast paths: optimized kernels behind the same parameters as the reference code.
hub	Model hub: one-line from_pretrained() loading of pre-trained models.
les	Latent Ewald Summation (LES): long-range interactions for any xnn model.
ops	Tensor operations shared across model families (gnn / cnn / dnn).
outputs	Differentiate energy to get conservative forces and the stress tensor.
pme	Smooth particle-mesh Ewald reciprocal sums (Essmann and co-workers, 1995).
reciprocal	Reciprocal-space sums through factorized phases (the Ewald fast paths).
recompute	Recomputation blocks whose memory stays bounded through a double backward.
registry	Name -> class registry so new models are added without touching the core.