tiger_eye
Re:NetworkX 3.0
on Wednesday January 11, 2023 @11:58AM (#34340811)
Great talk!
This slide shows some speedups when using the GraphBLAS [1] backend [2]:
https://i.postimg.cc/mr3mkKtx/100x-Faster-Network-X.png
[1] https://github.com/python-graphblas/python-graphblas
[2] https://github.com/python-graphblas/graphblas-algorithms
To learn the latest about GraphBLAS, join the free online GraphBLAS session at HPEC next Tuesday:
https://graphblas.org/hpec_bof.html https://ieee-hpec.org/
Also, these are probably better intro pages to GraphBLAS:
https://graphblas.org/ https://graphblas.org/GraphBLAS-Pointers/
GraphBLAS is underrated and underused IMHO. If you use e.g. scipy.sparse, NetworkX, or similar, you should check out GraphBLAS. It is really fast even compared to scipy.sparse, and more capable in many ways.
They've actually started implementing the NetworkX API
https://github.com/python-graphblas/graphblas-algorithms
with python-graphblas