[gmx-users] Re: Re:unexpected large difference in potential energy calculation by different method: cut-off and switch

[David van der Spoel]

Dongsheng Zhang wrote:
> Dear David,
> 
> Thank you for your reply. When I read your email at the first time. I
> think you have answered my question. But when I think it again today, I
> find I still have a question:
> 
> I can understand that there will be some artifacts to calculate forces
> by using switch function in MD simulation because switch function only
> smooths the potential. My question is: If I calculate the total
> potential energy of a system by rerunning a trajectory file with two
> different methods (Cut-off and switch), should I get close results
> because I don't need to calculate the force to determine the next
> configuration of the system? 
Interesting question. Please try it.
It depends on the options etc.
> 
> 
> 
> Dongsheng Zhang wrote:
> 
>>Hi, > > I am comparing different methods to calculate the coulomb and VDW
>>potential energy. I think the results for cut-off and switch should be
>>close. However, the difference is very large. Could anyone tell me what
>>happen? Detailed information is as follows:
> 
> 
> Switched potentials should be avoided. In GROMACS the switch is atom 
> based which gives huges artifacts, in other codes you can choose group 
> based switches which has less artifacts but introduces a horrible 
> arbitrariness in the definition of the groups. For more info see:
> 
> The Origin of Layer Structure Artifacts in Simulations of Liquid Water
> David van der Spoel and Paul J. van Maaren, in press
> J. Chem. Theor. Comput. http://dx.doi.org/10.1021/ct0502256
> 
> 


-- 
David.
________________________________________________________________________
David van der Spoel, PhD, Assoc. Prof., Molecular Biophysics group,
Dept. of Cell and Molecular Biology, Uppsala University.
Husargatan 3, Box 596,  	75124 Uppsala, Sweden
phone:	46 18 471 4205		fax: 46 18 511 755
spoel at xray.bmc.uu.se	spoel at gromacs.org   http://xray.bmc.uu.se/~spoel
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++