SCA with CMPS.
- class topomodelx.nn.simplicial.sca_cmps.SCACMPS(in_channels_all, complex_dim, n_layers=2, att=False)[source]#
SCA with CMPS.
- Parameters:
- in_channels_alllist[int]
Dimension of features on each node, edge, simplex, tetahedron,… respectively
- complex_dimint
Highest dimension of simplicial complex feature being trained on.
- n_layersint, default = 2
Amount of message passing layers.
- attbool
Whether to use attention.
- forward(x, laplacian_down_list, incidence_t_list)[source]#
Forward computation through layers, then linear layers, then avg pooling.
- Parameters:
- xlist[torch.Tensor]
Tensor inputs for each dimension of the complex (nodes, edges, etc.).
- laplacian_down_listlist[torch.Tensor]
List of the down laplacian matrix for each dimension in the complex starting at edges.
- incidence_t_listlist[torch.Tensor]
List of the transpose incidence matrices for the edges and faces.
- Returns:
- torch.Tensor, shape = (1)
Label assigned to whole complex.