Charlles Abreu
Charlles Abreu
The Python API of OpenMM's [Library Layer](http://docs.openmm.org/latest/api-python/library.html) is a wrapper of compiled C++ objects. The functionality you mention doesn't exist, but it would significantly increase the flexibility of custom OpenMM...
> I'm not sure I completely understand. It sounds like it would be really expensive to compute, since it would require multiple trips back and forth between different forces? It...
@wayuer19, as @peastman mentioned, the angle between the two planes equals the angle between their normal vectors. Therefore, if you evaluate `S=[(a2-a1)x(a3-a1)].[(b2-b1)x(b3-b1)]`, you can divide it by a product of...
> I think you also need to update the bounds in `metrics/QuantileLoss`. I did it in the first commit, but it involved changing the `bounds` variable (see below), which gets...
This is ready for a new review. As explained in https://github.com/chemprop/chemprop/pull/1223#issuecomment-2902823433, this PR should get priority over #1223 due to a required rebase.
> This PR does two things. First it registers quantile loss as a metric. Second it allows for quantile metric/loss function arguments in a v1 model to be converted to...
I needed to git-rebase on #1229 so the conversion could work correctly in the quantile regression case.
This PR is ready for a new review.
@pearlDingzhen, take a look at [CVPack](https://github.com/RedesignScience/cvpack).
It is true that biasing `rg^2` instead of `rg` can be good enough. But when one actually needs to transform a CV before applying the Metadynamics bias, wrapping it in...