Added GKAN

[danielshin1]

Added GKAN: Graph Kolmogorov-Arnold Networks as introduced in paper by Mehrdad Kiamari, and Bhaskar Krishnamachari. and the Kolmogorov-Arnold Network comes from the work https://arxiv.org/pdf/2404.19756 and implementation of kan is here https://github.com/KindXiaoming/pykan.

[puririshi98]

@danielshin1 do you have any experiments showing that this model outperforms existing GNNs?
It would make it much easier for me to merge this if you could test this model on papers100
https://github.com/pyg-team/pytorch_geometric/blob/1157d80b6926b9772c51cd73386ad72315194c20/examples/ogbn_train_scheduler.py
see this example which will soon be merged into main branch. if you can copy this code and add your GKAN model to it and name the new file ogbn_train_kolmogorov_arnolds.py. if you can produce significant lift over the this would be a better example to have than the CORA example which is basiclly toy data.
I am very interesting in this work and had been wondering how KAN effects GNNs. I look forward to the results and working with you to get this merged in

[puririshi98]

also try to get the Ci to be all green

[puririshi98]

@danielshin1, this is a great project, checking in if you still want to get it merged, see above for next steps

[puririshi98]

@danielshin1 circling back