torch_geometric.nn.conv.SimpleConv

class SimpleConv(aggr: Optional[Union[str, List[str], Aggregation]] = 'sum', combine_root: Optional[str] = None, **kwargs)[source]

Bases: MessagePassing

A simple message passing operator that performs (non-trainable) propagation.

\[\mathbf{x}^{\prime}_i = \bigoplus_{j \in \mathcal{N(i)}} e_{ji} \cdot \mathbf{x}_j\]

where \(\bigoplus\) defines a custom aggregation scheme.

Parameters:

  • aggr (str or [str] or Aggregation, optional) – The aggregation scheme to use, e.g., "add", "sum" "mean", "min", "max" or "mul". In addition, can be any Aggregation module (or any string that automatically resolves to it). (default: "sum")

  • combine_root (str, optional) – Specifies whether or how to combine the central node representation (one of "sum", "cat", "self_loop", None). (default: None)

  • **kwargs (optional) – Additional arguments of torch_geometric.nn.conv.MessagePassing.

Shapes:

  • inputs: node features \((|\mathcal{V}|, F)\) or \(((|\mathcal{V_s}|, F), (|\mathcal{V_t}|, *))\) if bipartite, edge indices \((2, |\mathcal{E}|)\)

  • outputs: node features \((|\mathcal{V}|, F)\) or \((|\mathcal{V_t}|, F)\) if bipartite

forward(x: Union[Tensor, Tuple[Tensor, Optional[Tensor]]], edge_index: Union[Tensor, SparseTensor], edge_weight: Optional[Tensor] = None, size: Optional[Tuple[int, int]] = None) → Tensor[source]

Runs the forward pass of the module.

reset_parameters() → None

Resets all learnable parameters of the module.