API Reference

• DGLGraph – Untyped graph with node/edge features
  • Adding nodes and edges
  • Querying graph structure
  • Removing nodes and edges
  • Transforming graph
  • Converting from/to other format
  • Using Node/edge features
  • Computing with DGLGraph
• DGLHeteroGraph – Typed graph with node/edge features
  • Conversion to and from heterogeneous graphs
  • Querying metagraph structure
  • Querying graph structure
  • Using Node/edge features
  • Transforming graph
  • Computing with DGLHeteroGraph
• Feature Initializer
  • dgl.init.base_initializer
  • dgl.init.zero_initializer
• BatchedDGLGraph – Enable batched graph operations
  • Merge and decompose
  • Query batch summary
  • Graph Readout
• Builtin functions
  • Message functions
  • Reduce functions
• Graph Traversal
  • dgl.traversal.bfs_nodes_generator
  • dgl.traversal.bfs_edges_generator
  • dgl.traversal.topological_nodes_generator
  • dgl.traversal.dfs_edges_generator
  • dgl.traversal.dfs_labeled_edges_generator
• Message Propagation
  • dgl.propagate.prop_nodes
  • dgl.propagate.prop_edges
  • dgl.propagate.prop_nodes_bfs
  • dgl.propagate.prop_nodes_topo
  • dgl.propagate.prop_edges_dfs
• User-defined function related data structures
  • EdgeBatch
  • NodeBatch
• Graph samplers
  • Distributed sampler
• Dataset
  • Utils
  • Dataset Classes
  • Molecular Graphs
• Transform – Graph Transformation
  • dgl.transform.line_graph
  • dgl.transform.reverse
  • dgl.transform.to_simple_graph
  • dgl.transform.to_bidirected
  • dgl.transform.khop_adj
  • dgl.transform.khop_graph
  • dgl.transform.laplacian_lambda_max
  • dgl.transform.knn_graph
  • dgl.transform.segmented_knn_graph
  • dgl.transform.add_self_loop
  • dgl.transform.remove_self_loop
  • dgl.transform.metapath_reachable_graph
• NN Modules
  • NN Modules (PyTorch)
  • NN Modules (MXNet)
• DGLSubGraph – Class for subgraph data structure
  • Mapping between subgraph and parent graph
  • Synchronize features between subgraph and parent graph
• Graph Store – Graph for multi-processing and distributed training
  • Querying the distributed setting
  • Using Node/edge features
  • Computing with Graph store
  • Construct a graph store
• NodeFlow – Graph sampled from a large graph
  • Querying graph structure
  • Converting to other format
  • Using Node/edge features
  • Mapping between NodeFlow and parent graph
  • Synchronize features between NodeFlow and parent graph
  • Computing with NodeFlow
• DGL Random Number Generator Controls
  • dgl.random.seed
• Model Zoo
  • Chemistry