[gmx-developers] Problems with artificial minima when using soft-core potentials

[Alan E. Mark]

All of this comes down to choice and whether as developers we want to to 
provide flexibility or try to prevent users doing something which may 
cause their simulations to fail.

Enforcing the use of a single parameter for both the LJ and 
electrostatics prevented some people from making a bad choice. At the 
same time it removed the flexibility inherent in the original algorithm 
leading to the need for the various work arounds outlined.

In regard to Michael's and Bert's comments, as the alpha parameter for 
the electrostatics becomes large the algorithm effectively becomes 
equivalent to switching off the electrostatics first and thus these 
minima disappear. My concern Bert with your approach is simply that is 
that it is more complex than is needed and, as the interaction is 
reduced, that it may become possible to sample the discontinuity at r=0.

Alan

-- 
--------------------------------------------------------------------
This message was composed using voice recognition software. Please forgive any strange words I fail to correct.
____________________________________________________________________
Prof. Alan E. Mark

BRISBANE
Molecular Dynamics Group
School of Chemistry and Molecular Biosciences (SCMB)
University of Queensland
Brisbane, QLD 4072,
Australia

http://scmb.uq.edu.au/staff/alan-mark
http://compbio.biosci.uq.edu.au

Physical address:
Rm 331 Level 3, Molecular Biosciences Building (#76)

Email 		a.e.mark at uq.edu.au
Secretary:    	md.secretary at uq.edu.au
Phone:          +61-7-3365 4180
FAX:            +61-7-3365 3872
Secr:      	+61-7-3365 3821
                  
Unless stated otherwise, this e-mail represents only the views of the Sender and not the views of The University of Queensland