ur-whitelab / hoomd-tf

A plugin that allows the use of Tensorflow in Hoomd-Blue for GPU-accelerated ML+MD
https://hoomd-tf.readthedocs.io
MIT License
30 stars 8 forks source link

[JOSS] Example Usage #197

Closed RMeli closed 4 years ago

RMeli commented 4 years ago

I think the Jupyter Notebooks showcasing the functionality of the code are great! They are an extremely valuable addition which will facilitate the use and adoption of hoomd-tf. To make them even better, I'd suggest to add some text explaining a bit more what the notebook is about and guiding the user through the most complicated cells, if possible.

Running the notebooks with the latest version of the code (be8b299cc8c62add8fff2ff719713b41814cc6ff) I encountered the following problems:

I think the solution for the ValueError: Found no dependence on positions or neighborsso forces cannot be computed is the following (see https://github.com/ur-whitelab/hoomd-tf/pull/183#pullrequestreview-436907018 and PR #186 ):

-    force = graph.compute_forces(energy)
+    force = graph.compute_forces(energy, positions=True)

JOSS review

whitead commented 4 years ago

Thanks for pointing out these issues. Some of these notebooks are indeed out of date and not that descriptive. Hopefully I've addressed this in PR #198

malramsay64 commented 4 years ago

Changes are great. LGTM