Jérôme Hénin
Jérôme Hénin
> If you choose to just back up the VMD view with Tcl in the Dashboard Done, keeping the issue open.
Another benefit of that would be debugGradients on non-scalar cvcs.
This is a likely cause behind a long-standing bug reported by Chris Chipot, FE calculations on a scripted coordinate based on a vector cvc.
It would be nice to define a shortest path towards implementing this, if possible without all the changes envisioned in #61. But if necessary, that can be a good reason...
I can try, but I only expect a larger fittingGroup to distribute the counter-force on more atoms. Wouldn't the net counter-force would stay the same, as long as the fittingGroup...
It also affects pmf calculations on scalar variables that are scripted functions of vector components. That is a combination of documented features.
To avoid undetected incorrect behavior, the fit gradients should be available only when atomic gradients are calculated in the first place. I see two levels at which it can be...
While working on this, it becomes clear to me that it the moving frame of reference system, in all applications I'm aware of, is CVC-wide. Whenever there are several atom...
Commit 31a232a13a611f062bb5ff0bec156eeb9776fe90, solves the case where fit gradients were silently missing, unbeknownst to the user. Therefore, downgrading this issue from bug to enhancement.