General Information

The package in kgcnn contains several layer classes to build up graph convolution models in Keras with Tensorflow, PyTorch or Jax as backend. Some models are given as an example in literature. Focus of kgcnn is (batched) graph learning for molecules kgcnn.molecule and materials kgcnn.crystal. Below you can find explanations and information on how to use kgcnn . See Reference under Package Content for code documentation.

General:

• Introduction
• Installation
• Data
  • Graph Dict
  • Graph List
  • Datasets
    • Special Datasets
• Models
  • Functional API
  • Subclassing Model
  • Templates
  • Loading options
    • 1. Padded Tensor
    • 2. Ragged input
    • 3. Direct disjoint input via data loader.
• Layers
  • Implementaion details
    • Casting
    • Gather
    • Convolution
    • Aggregation
    • Pooling
• Literature
  • Training Scripts
    • Data hyperparameter
    • Model hyperparameter
    • Training hyperparameter
    • Info
  • Benchmarks
• Molecules
  • Training example
• Forces
  • Fit Force Model
  • Molecular dynamics simulation
    • Keras model in MolDynamicsModelPredictor
    • Use ASE compatible KgcnnSingleCalculator

Package Content

Reference:

• kgcnn package
  • Subpackages
    • kgcnn.backend package
    • kgcnn.crystal package
    • kgcnn.data package
    • kgcnn.graph package
    • kgcnn.initializers package
    • kgcnn.io package
    • kgcnn.layers package
    • kgcnn.literature package
    • kgcnn.losses package
    • kgcnn.metrics package
    • kgcnn.models package
    • kgcnn.molecule package
    • kgcnn.ops package
    • kgcnn.optimizers package
    • kgcnn.training package
    • kgcnn.utils package
  • Module contents

Indices and tables

• Index

• Module Index

• Search Page