global_add_pool() missing default batch argument

[qihuazhong]

🐛 Describe the bug

According to the doc (https://pytorch-geometric.readthedocs.io/en/stable/modules/nn.html#torch_geometric.nn.pool.global_add_pool) and the code. The batch argument should be optional. However, the default None is not provided in the function signature and would raise an error if not provided.

Should be an easy fix by chaning

def global_add_pool(x: Tensor, batch: Optional[Tensor],
                    size: Optional[int] = None) -> Tensor:

to

def global_add_pool(x: Tensor, batch: Optional[Tensor] = None,
                    size: Optional[int] = None) -> Tensor:

Environment

• PyG version:
• PyTorch version:
• OS:
• Python version:
• CUDA/cuDNN version:
• How you installed PyTorch and PyG (conda, pip, source):
• Any other relevant information (e.g., version of torch-scatter):

[rusty1s]

Yes, it is optional (meaning you can pass in None) but it does not have a default argument (which is intended).

[eurunuela]

Yes, it is optional (meaning you can pass in None) but it does not have a default argument (which is intended).

It is actually a required argument. If you try to call global_add_pool(x) it will raise an error saying you didn't pass the batch argument. It should be given a default value or the docs should be updated to say that it is a required argument.