Open VolodyaCO opened 1 year ago
To whom might be interested, I have been maintaining a forked version of the package under the name "tensornetwork-ng" which can be found at this GitHub repository: https://github.com/refraction-ray/TensorNetwork-NG. The main motivation behind this fork is to ensure the robustness of the downstream project, TensorCircuit, and to ensure its compatibility with future breaking changes in Python, NumPy, and JAX.
I must admit that due to limited resources and effort, I have only been able to maintain certain submodules while temporarily pruning others (irrelevant from tensorcircuit). However, I would be more than happy to welcome and encourage anyone interested to participate and contribute and I look forward to any potential collaborations and contributions :)
Thanks for your answer @refraction-ray. Indeed, my main interest is tensorcircuit. I see that in tensorcircuit people have been overriding some parts of the JAX backend from tensornetwork, but I was wondering why tensorcircuit was relying on an unmaintained package.
@VolodyaCO , TensorNetwork is a great package and I really appreciate the ML backend agnostic design. The main reason I chose to use this package is because, at the time I began working on TensorCircuit in 2020, TensorNetwork was being actively maintained and well-supported.
I am interested in using this repository, but there are no contributions in the past two years. It seems as though this repository is not maintained anymore. Are you planning to support this repo any time in the future? If not, do you recommend a fork or any other tensor network library in Python that is actively developed and supported? Thanks. CC @mganahl