[gmx-users] nonbonded LJ-parameters

[Stephan Dammer]

Hi,

I have a remark/question concerning the use of nonbonded LJ-parameters in
GROMACS.

Assume you simulate a mixture of "artificial" gas and liquid atoms. The
beginning of your topology file could look like this:

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
[ defaults ]
; nbfunc 1:LJ   comb-rule: provide LJ parameters sigma and epsilon
1               3

[ atomtypes ]
; name  mass             charge    ptype      sigma     epsilon
GAS     40.000           0.000     A          0.34      1.0
LIQ     20.00            0.000     A          0.34      3.0
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

Due to combination rule = 3 the LJ-parameters are sigma and epsilon, (and
not C_6 and C_12). If you do not want to use the combination rule 3 to
determine the LJ-parameters for interactions between GAS and LIQ atoms,
you have to provide interaction parameters yourself in the
"nonbonded_params" part of the topology file. As far as I understood the
manual, using combination rule 3 determines that these interaction
parameters are sigma_GAS_LIQ and epsilon_GAS_LIQ. However, it seems that
if you provide these LJ-parameters as sigma and epsilon (as was done for
the GAS-GAS and LIQ-LIQ interactions) this will NOT work (at least
according to a simple check I performed). Hence, something like

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
[ nonbond_params]
; i     j       func    sigma   epsilon, func=1=LJ
  GAS   LIQ     1       0.34  0.01
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

will NOT lead to interaction parameters sigma_GAS_LIQ=0.34
epsilon_GAS_LIQ=0.01. Instead, (as far as I could check it) one has to
provide the LJ-parameters C_6 and C_12 for the GAS-LIQ interactions

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
[ nonbond_params]
; i     j       func    C_6             C_12
  GAS   LIQ     1       6.179217664e-5  9.545682736e-8
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

although for GAS-GAS and LIQ-LIQ interactions epsilon and sigma are used.

I posted this, because I do not know if this behavior is wanted (and I
simply misunderstood the manual), or whether this perhaps should be fixed.
Moreover, maybe others have stepped upon the same problem.

Best regards,
Stephan

ps: I use GROMACS 3.2.1
--------------------------------------------------------------------------
Stephan M. Dammer                             University of Twente
e-mail: s.m.dammer at tnw.utwente.nl             Physics of Fluids
web: http://stilton.tnw.utwente.nl/~dammer/   P.O. Box 217, 7500 AE Enschede
telefone: [+31] (53) 489 2486                 The Netherlands
fax:      [+31] (53) 489 8068                 room: CT-A 124