I've used igraph. While it's much faster, for me at least, modifying the graph once it's constructed is harder compared to network. Haven't worked with cugraph though. As always use the right tool for the job

/? "networkx" "igraph" "cugraph" site:github.com inurl:awesome https://www.google.com/search?q=%22networkx%22+%22igraph%22+... :

- https://github.com/johnhany/awesome-list#graph lists a few Tensorflow and Pytorch + graphs applications

CuGraph docs > List of Supported and Planned Algorithms: https://docs.rapids.ai/api/cugraph/stable/graph_support/algo...

https://github.com/rapidsai/cugraph#news :

> NEW! nx-cugraph, a NetworkX backend that provides GPU acceleration to NetworkX with zero code change. :

  pip install nx-cugraph-cu11 --extra-index-url https://pypi.nvidia.com
  export NETWORKX_AUTOMATIC_BACKENDS=cugraph

I would be interested in one that is properly type annotated. None of the options here (NetworkX, igraph, cugraph) are.

https://news.ycombinator.com/item?id=36922924 :

> pytype (Google) [1], PyAnnotate (Dropbox) [2], and MonkeyType (Instagram) [3] all do dynamic / runtime PEP-484 type annotation type inference [4] to generate type annotations.

Hypothesis generates tests from type annotations; and icontract and pycontracts do runtime type checking.