Source code for jittor_geometric.nn.conv.cheb_conv

from typing import Optional
from jittor_geometric.typing import OptVar

import jittor as jt
from jittor import Var, nn
from jittor_geometric.nn.conv import MessagePassing
from jittor_geometric.utils import get_laplacian

from ..inits import glorot, zeros


[docs] class ChebConv(MessagePassing): r"""The chebyshev spectral graph convolutional operator from the `"Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering" <https://arxiv.org/abs/1606.09375>`_ paper .. math:: \mathbf{X}^{\prime} = \sum_{k=1}^{K} \mathbf{Z}^{(k)} \cdot \mathbf{\Theta}^{(k)} where :math:`\mathbf{Z}^{(k)}` is computed recursively by .. math:: \mathbf{Z}^{(1)} &= \mathbf{X} \mathbf{Z}^{(2)} &= \mathbf{\hat{L}} \cdot \mathbf{X} \mathbf{Z}^{(k)} &= 2 \cdot \mathbf{\hat{L}} \cdot \mathbf{Z}^{(k-1)} - \mathbf{Z}^{(k-2)} and :math:`\mathbf{\hat{L}}` denotes the scaled and normalized Laplacian :math:`\frac{2\mathbf{L}}{\lambda_{\max}} - \mathbf{I}`. Args: in_channels (int): Size of each input sample, or :obj:`-1` to derive the size from the first input(s) to the forward method. out_channels (int): Size of each output sample. K (int): Chebyshev filter size :math:`K`. normalization (str, optional): The normalization scheme for the graph Laplacian (default: :obj:`"sym"`): 1. :obj:`None`: No normalization :math:`\mathbf{L} = \mathbf{D} - \mathbf{A}` 2. :obj:`"sym"`: Symmetric normalization :math:`\mathbf{L} = \mathbf{I} - \mathbf{D}^{-1/2} \mathbf{A} \mathbf{D}^{-1/2}` 3. :obj:`"rw"`: Random-walk normalization :math:`\mathbf{L} = \mathbf{I} - \mathbf{D}^{-1} \mathbf{A}` :obj:`\lambda_max` should be a :class:`jittor.Var` of size :obj:`[num_graphs]` in a mini-batch scenario and a scalar/zero-dimensional tensor when operating on single graphs. bias (bool, optional): If set to :obj:`False`, the layer will not learn an additive bias. (default: :obj:`True`) **kwargs (optional): Additional arguments of :class:`jittor_geometric.nn.conv.MessagePassing`. Shapes: - **input:** node features :math:`(|\mathcal{V}|, F_{in})`, edge indices :math:`(2, |\mathcal{E}|)`, edge weights :math:`(|\mathcal{E}|)` *(optional)*, batch vector :math:`(|\mathcal{V}|)` *(optional)*, maximum :obj:`lambda` value :math:`(|\mathcal{G}|)` *(optional)* - **output:** node features :math:`(|\mathcal{V}|, F_{out})` """
[docs] def __init__(self, in_channels: int, out_channels: int, K: int, normalization: Optional[str] = 'sym', bias: bool = True, **kwargs): kwargs.setdefault('aggr', 'add') super(ChebConv, self).__init__(**kwargs) assert K > 0 assert normalization in [None, 'sym', 'rw'], 'Invalid normalization' self.in_channels = in_channels self.out_channels = out_channels self.normalization = normalization self.lins = nn.ModuleList([ nn.Linear(in_channels, out_channels, bias=False) for _ in range(K) ]) if bias: self.bias = jt.zeros(out_channels) else: self.bias = None self.reset_parameters()
[docs] def reset_parameters(self): for lin in self.lins: glorot(lin.weight) zeros(self.bias)
def __norm__(self, edge_index: Var, num_nodes: Optional[int], edge_weight: OptVar, normalization: Optional[str], lambda_max: OptVar = None, dtype: Optional[int] = None, batch: OptVar = None): edge_index, edge_weight = get_laplacian(edge_index, edge_weight, normalization, dtype, num_nodes) assert edge_weight is not None if lambda_max is None: lambda_max = 2.0 * edge_weight.max() elif not isinstance(lambda_max, Var): lambda_max = jt.array([lambda_max], dtype=dtype) assert lambda_max is not None if batch is not None and lambda_max.numel() > 1: lambda_max = lambda_max[batch[edge_index[0]]] edge_weight = (2.0 * edge_weight) / lambda_max inf_mask = edge_weight == float('inf') edge_weight = jt.where(inf_mask, jt.zeros_like(edge_weight), edge_weight) loop_mask = edge_index[0] == edge_index[1] edge_weight = jt.where(loop_mask, edge_weight - 1, edge_weight) return edge_index, edge_weight
[docs] def execute(self, x: Var, edge_index: Var, edge_weight: OptVar = None, batch: OptVar = None, lambda_max: OptVar = None) -> Var: edge_index, norm = self.__norm__( edge_index, x.size(self.node_dim), edge_weight, self.normalization, lambda_max, dtype=x.dtype, batch=batch, ) Tx_0 = x Tx_1 = x out = self.lins[0](Tx_0) # propagate_type: (x: Var, norm: Var) if len(self.lins) > 1: Tx_1 = self.propagate(edge_index, x=x, norm=norm) out = out + self.lins[1](Tx_1) for lin in self.lins[2:]: Tx_2 = self.propagate(edge_index, x=Tx_1, norm=norm) Tx_2 = 2. * Tx_2 - Tx_0 out = out + lin(Tx_2) Tx_0, Tx_1 = Tx_1, Tx_2 if self.bias is not None: out = out + self.bias return out
[docs] def forward(self, x: Var, edge_index: Var, edge_weight: OptVar = None, batch: OptVar = None, lambda_max: OptVar = None) -> Var: """PyTorch Geometric style forward method alias for execute""" return self.execute(x, edge_index, edge_weight, batch, lambda_max)
[docs] def message(self, x_j: Var, norm: Var) -> Var: return norm.view(-1, 1) * x_j
def __repr__(self) -> str: return (f'{self.__class__.__name__}({self.in_channels}, ' f'{self.out_channels}, K={len(self.lins)}, ' f'normalization={self.normalization})')