Re: Add repulsive potential with dummy atom

From: Ropon-Palacios G. (
Date: Tue Mar 16 2021 - 17:45:01 CDT

Dear Giacomo,


I have not had any rapprochement with tclBC, I don't know if you could give me an example that I can start with please.


On the other hand, I found in the example of colvar one called "", basically what I want is that the molecules (84 ligands placed around my protein at random), I don't know how to add, forming a cluster of itself, already I have a high number and many authors have mentioned this phenomenon. This script ( that I found in the examples might help me.






From: Giacomo Fiorin <>
Date: Tuesday, March 16, 2021 at 5:01 PM
To: NAMD list <>, David Hardy <>
Subject: Re: namd-l: Add repulsive potential with dummy atom


Hi Geo, the documentation at the Colvars website is a bit richer than what is included in the NAMD manual:;!!DZ3fjg!rTvkcxHZMdgD-cSj2vJjRDsVB3nInQ8zcezbiQHEdvI4YOC9Ey0P3AABAShYYhq8VA$

and there are specific examples listed on the top page:;!!DZ3fjg!rTvkcxHZMdgD-cSj2vJjRDsVB3nInQ8zcezbiQHEdvI4YOC9Ey0P3AABASinxQtBcw$


However, there isn't likely to be an input file that is just ready to be reused, though, for several reasons:

1. There are details of your system that you didn't describe, but would be needed to prepare a good input.

2. The version of NAMD that you are using is important, as it affects the Colvars features available to you. In principle it is possible to build a more recent version of Colvars with older versions of NAMD, but this is not supported (meaning, it will likely work but taking more work on your part).

3. The Lennard-Jones potential is not present as a pre-compiled function for Colvars, although there are provisions to add custom functions at runtime listed in the Colvars doc.

4. If I understand correctly what Dave also understood, you have many pairs of molecules, which would mean a lot of variables that aren't really "collective".


Without resorting to advanced features/packages, why not simply pick one atom for each molecule, and apply forces between them with tclBC? It is slower than C++ code (like e.g. the force field functions from NAMD or the collective variables from Colvars), but it would be at least parallelized over nodes, and probably result in small overhead.



This archive was generated by hypermail 2.1.6 : Fri Dec 31 2021 - 23:17:11 CST