Some analysis techniques

[thangckt]

Dear Developers,

I wonder if you can:

1. give a little guide on how to implement a per-atom collective variable. Some things are similar to MultiColvar in PLUMED.

2. Add some analysis techniques, such as contact matrix, clustering analysis,...

3. Implement an Enhanced Sampling technique called LogMFD

It will be very helpful if you have example docs for a case implementation

Thank you so much

[p-zubieta]

Hi @thangckt. Thanks for the interest in the library. The answer to all of your questions is, in principle, yes, except possibly for the the clustering analysis, which sound a bit tricky to get to work with jax.

And yes, we need an example in the documentation showing the user how to define custom CVs.

[thangckt]

Dear @pabloferz

I see an article in Plumed's reference Analyzing and Driving Cluster Formation in Atomistic Simulations that describes in detail the implemented algorithm of DFS clustering. I hope it can work within PySAGES, since the size of the largest cluster is a commonly used CV to compute free energy profile.

Also, the Advanced Sampling LogFMD which is based on the mean-force dynamic scheme is worth considering.
I wish I could do something to contribute, but my coding skill is so bad for that :(