[Openmd-users] Update

[Dan Gezelter]

Brett,

  The LJ potential truncation depends on how you specified things in the md file.

In general, OpenMD will use a cubic switching function, s(r), with these parameters setting the upper and lower bounds for the switching function:

cutoffRadius = 9.0;
switchingRadius = 7.65;

V(r) = V_{LJ}(r) * s(r)   (r < rcut)
V(r) = 0                  (r > rcut)

If you set both to the same value, e.g.

cutoffRadius = 15.0;
switchingRadius = 15.0;

then the LJ behavior will default to shifted potential,  V(r) = V_{LJ}(r) - V_{LJ}(rcut).

You can explicitly override the default cutoff behavior using electrostaticSummationMethod.  In this version of OpenMD, we're tethering the LJ shifting behavior to the electrostatic behavior:

electrostaticSummationMethod = "SHIFTED_FORCE";      // guarantees shifted frc
electrostaticSummationMethod = "SHIFTED_POTENTIAL";  // guarantees shifted pot

I.e. 

  shifted Force:  V(r) = V_{LJ}(r) - V_{rcut} -  V'_{LJ}(rcut) * (r - rcut)
  shifted Pot:    V(r) = V_{LJ}(r) - V_{LJ}(rcut)

As for the interaction matrix, we're in the process of adding "explicit non-bonded interactions" to the next version, and this would allow you to specify sigma and epsilon values (or even other functional forms) that could override the default mixing rules.  The necessary infrastructure isn't completely in place yet, but this is certainly on our radar.

 --Dan

On Jul 16, 2010, at 6:12 AM, Brett Donovan wrote:

> Dear Dan
> 
> I've been busy running on our cluster here, all seems to be working
> well on up to a few hundred processors. I have a few questions etc
> regarding the potentials and LJ interaction matrix.
> 
> Potential - what is the exact form of the potential employed? I think
> you said these are cut and shifted potentials, if there somewhere this
> is documented?
> 
> Interaction Matrix - there might be need to employ a slight deviation
> away from the LB mixing rules. Is there a smart and easy way to employ
> this? i.e. SSD and AMIDE has a scaling factor rather than just
> \epsilon = \sqrt(\epsilon_A\epsilon_B), we could have \epsilon =
> \alpha \sqrt(\epsilon_A\epsilon_B)
> 
> Best wishes
> 
> Brett

***********************************************
  J. Daniel Gezelter
  Associate Professor of Chemistry
  Director of Graduate Admissions
  Department of Chemistry and Biochemistry
  251 Nieuwland Science Hall
  University of Notre Dame
  Notre Dame, IN 46556-5670

  phone:  +1 (574) 631-7595
  fax:    +1 (574) 631-6652
  e-mail: gezelter at nd.edu
  web:    http://www.nd.edu/~gezelter
************************************************