Hierarchical Tucker (H-Tucker) format and Support for Parallelization

[i3s93]

Hello,

I am working on a project concerning tensor decompositions and recently came across this library in my survey of existing software tools. I am very much interested in using TensorLy in my research, but I have several questions I would like to ask, which are as follows:

1. I was looking at the different tensor formats supported by the library, and the one I am looking for is the Hierarchical Tucker or H-Tucker format. I see that standard Tucker is supported, but it does not look like H-Tucker is supported. Is there an interest in supporting this format? If new features are developed "as needed" I would be happy to help contribute on this effort.

2. Does TensorLy support multi-GPU parallelism or distributed memory parallelism? Are these features backend-dependent or do you have a recommended tool for this, e.g., mpi4py or Dask?

Thank you in advance for taking the time to review my questions and providing any clarification of the above topics.