Hydragnn

Summary
New or improved capabilities included in **v3.0rc1** release are as follows:

1. Enhancement in message passing layers through generalization of the class inheritance to enable the inclusion of a broader set of message passing policies
2. Inclusion of equivariant message passing layers from the original implementations of:

- SchNet [https://pubs.aip.org/aip/jcp/article/148/24/241722/962591/SchNet-A-deep-learning-architecture-for-molecules](url)
- DimeNet++ [https://arxiv.org/abs/2011.14115](url)
- EGNN [https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&ved=2ahUKEwjB_trwsrqCAxVUMlkFHRWjAmkQFnoECBcQAQ&url=https%3A%2F%2Farxiv.org%2Fpdf%2F2102.09844&usg=AOvVaw3Ey_bFHJxpm9_xFxbSqWAq&opi=89978449](url) models

3. Restructuring of class inheritance for data management
4. Support of DDStore [https://github.com/ORNL/DDStore](url) capabilities for improved distributed data parallelism on large volumes of data that cannot fit on intra-node memory capacities
5. Large-scale system support for OLCF-Crusher and OLCF-Frontier

Summary
New or improved capabilities included in v2.0.0 release are as follows:
* Enhancement in message passing layers through class inheritance
* Adding transformation to ensure translation and rotation invariance
* Supporting various optimizers
* Atomic descriptors
* Integration with continuous CI test
* Distributed printouts and timers
* Profiling
* Support of ADIOS2 for scalable data loading
* Large-scale system support, including Summit (ORNL) and Perlmutter (NERSC)

Capabilities provided in v1.0.0 release (Oct 2021)
Major capabilities included in the previous release v1.0.0 are as follows:
* Multi-task graph neural network training with enhanced message passing layers
* Distributed Data Parallelism (DDP) support