Corin Wagen

I've been seeing the exact same error, using a modified version of the ``ReplicaExchange`` class from [this tutorial](https://openmmtools.readthedocs.io/en/0.18.1/gettingstarted.html). I tried increasing ``n_restart_attempts`` to 100 but it didn't help at all!...

Right, we've been thinking about that as well - initial guess Hessian can make a big difference, cf. [recent Merz work](https://chemrxiv.org/engage/chemrxiv/article-details/64e37e3700bbebf0e68dd9c4). Even a cheap semiempirical/ML Hessian probably beats a Schlegel-type...

I've only used the API interface, so I don't have a strong opinion about CLI arguments, but that seems reasonable. I suppose Hessian engine could default to the regular (gradient)...

+1 to more prebuilt distributions, it would be nice to have some Python 3.11 support as well. (Thanks for such a great project!)

Yeah, I agree w/ @kaedroho that this could be useful. Sometimes it's nice to have mixed-backend queues where you can add serverless workers to meet demand spikes while keeping long-running...

Agreed -- for instance, if you have a bunch of text in the collection with "author" stored in the metadata, it would be very nice to be able to filter...

Any headway here? This is a pretty frustrating error.

Ditto here, I get some weird issues on CPU. The pre-trained models worked well for me, but I had to manually override a variety of "device" settings to force PyTorch...

Note that similar issues have been seen before: #794

Similarly when I compute f(-) for fluorobenzene, I get ortho = -0.049, meta = -0.045, and para = -0.081. The [literature values](https://www.sciencedirect.com/science/article/abs/pii/016612809189021R) are ortho = 0.047, meta = 0.026, and...