[Feature Request] DPLR in PyTorch backend

[ChiahsinChu]

Summary

Dear developers,

I have developed DPLR in PyTorch backend in the plugin mode. This implementation has passed tests and been ready to use. The reason I have not started a PR to the main repository is that:

1. This implementation depends on two PRs (feat(pt): add plugin for data modifier #4661 and feat(pt): add atomic_weight to dipole model inference #4628).
2. This implementation depends on a third-party package torch-admp for PME calculation.

I would like to know your suggestions about this feature, keeping it as a stand-alone plugin or merging it to the main repository.

Best regards.

Detailed Description

See above.

Further Information, Files, and Links

No response