Adding of two edge index

[shrimonmuke0202]

How can I add two edge_index and feed this new edge_index in the GCN layer?

How can I add my original edge_index with another sparse matrix which is generated from random gaussian distribution?

[adhikarirsr]

After adding the sparse random matrix, will you get a weighted adjacency matrix?

[shrimonmuke0202]

I want to add random Gaussian noise in the edge index ie in the adjacency matrix of a graph.

[adhikarirsr]

Navie way may look like this:
You can convert your edge_index to Adjacency matrix using to_scipy_sparse_matrix and then add your sparse random Gaussian matrix to it. If I understood correctly then it will give you weighted adjacency matrix. Then you can convert that into edge index as follows:

row, col = adj.nonzero()
edge_index = torch.stack([row, col], dim=0)
edge_weight = adj[row, col]

edge_weight will contain the weight of the edges.

[rusty1s]

That is correct. If you are looking for a sparse version, you can also sample a fixed amount of edges via torch_geometric.utils.negative_sampling, create corresponding edge weights via torch.randn(num_additional_edges), and concatenate both tensors to your original sparse adjacency matrix.