Jérôme Hénin

If you try this algorithm, you'll see that it is reversible (that is, qunwrap reverses qwrap). I think we're on the same page now. I hope you don't mind if...

I rebased this to clean it up and then realized I could not push to your fork, @EzryStIago. Now in PR #708

Thanks, those seem valid to me. 1. I think this is promising because calling rotate() many times for the same quaternion does redundant work, especially compared with caching the rotation...

Couldn't the same purpose be achieved by a more portable test? Namely try to open the file, and if not `good()` we give up. Second question: will this play nicely...

Hi Haohao, The NAMD implementation still needs some refinement: it comes with performance limitations, namely, you cannot use MTS and you need energy computation every time step. Other than that,...

In alchemical simulations, I usually scale down the timestep, because one of the end points has a molecule in gas phase. Depending on the nature of the ligand, you may...

> Hi, @jhenin , I suggest also make available the Colvars harmonic restraints as the alchemical CVs, which will let us completely give up FEP/TI in binding free energy calculations....

The main use case for distributing CVCs over threads is that of multiple, expensive CVCs of similar computational cost. It is hard to know how widespread it is, but in...

I see no problem with dropping it.

@zwpku Any chance that libtorch could be linked dynamically at run time? That would let us support it with minimally invasive changes to the MD engine build process.