components

Components for the ML models.

Modules

atom_ref	Atomic reference module from [M-OFDFT].
density_coeff_embedding	Module for embedding density coefficients in an 'atom-hot' tensor.
dimension_wise_rescaling	Dimension-wise Rescaling is one of the enhancement modules to cope with vast gradient ranges as described in the second paragraph of section 4.3 in [M-OFDFT].
g3d_layer	Graphormer layer with scalar distance features as attention bias.
g3d_layer_tf	Graphormer layer with scalar distance features as attention bias.
g3d_stack	Module to wrap a stack of G3DLayer.
gbf_module	Compute edge features by applying Gaussian basis functions to the distances between nodes.
graphformer	Implements the full model as described in [M-OFDFT].
initial_guess_delta_module	Module for the projMINAO guess from [M-OFDFT].
local_frames_module	Calculates the local frames of the atoms in the molecule.
loss_function	Module containing the loss functions for the ML-DFT model.
mlp
natural_reparametrization	Natural reparametrization of the LCAB Ansatz for the electron density.
node_embedding	The NodeEmbedding module initializes the hidden node features \(h\).
sample_weighers	This module contains classes for weighing samples in a batch.
shrink_gate_module	Shrink gate.
toy_net	Toy architecture for testing purposes.
training_metrics	MAE of energy, MAE for gradient, MAE for proj_minao.