[gmx-users] Buckingham potential

[Stefano Piana]

Hallo,
I am trying to perform a calculation with a Buckingham potential,
however I always get Nan in the energy of my system.
I also tried to put all the parameters of the Buck pot to 0. I still get
Nan. On the other hand if I set my nbfunct to 1 (that is LJ) with all
the parameters set to 0 (which should be the same as Buck pot with all
parameters to 0) I get "reasonable" numbers out of the calculation.

these are the input and outputs:
--------------- Input Buckingham ------------------------
#define _FF_GROMOS
#define _FF_GROMOS96



[ defaults ]
; nbfunc comb-rule gen-pairs fudgeLJ fudgeQQ
       2       1        yes      0.0     0.0



[ atomtypes ]
C         1.00000       0.000       A 0.00000E+00 0.00000E+00 0.00000E+00
O         1.00000       0.000       A 0.00000E+00 0.00000E+00 0.00000E+00
Mg        1.00000       0.000       A 0.00000E+00 0.00000E+00 0.00000E+00
Ox        1.00000       0.000       A 0.00000E+00 0.00000E+00 0.00000E+00



[ nonbond_params ]
C    C         2   0.000000E+00   0.000000E+00   0.000000E+00
C    O         2   0.000000E+00   0.000000E+00   0.000000E+00
C    Mg        2   0.000000E+00   0.000000E+00   0.000000E+00
C    Ox        2   0.000000E+00   0.000000E+00   0.000000E+00
O    O         2   0.000000E+00   0.000000E+00   0.000000E+00
O    Mg        2   0.000000E+00   0.000000E+00   0.000000E+00
O    Ox        2   0.000000E+00   0.000000E+00   0.000000E+00
Mg   Mg        2   0.000000E+00   0.000000E+00   0.000000E+00
Mg   Ox        2   0.000000E+00   0.000000E+00   0.000000E+00
Ox   Ox        2   0.000000E+00   0.000000E+00   0.000000E-00

--------------------- Output Buckingham ----------------------------

Steepest Descents did not converge in 2 steps
    Potential Energy  =          nan
Maximum force:  1.38452e+04

---------------------Input LJ --------------------------------------
#define _FF_GROMOS
#define _FF_GROMOS96



[ defaults ]
; nbfunc comb-rule gen-pairs fudgeLJ fudgeQQ
       1       1        yes      0.0     0.0



[ atomtypes ]
C         1.00000       0.000       A 0.00000E+00 0.00000E+00
O         1.00000       0.000       A 0.00000E+00 0.00000E+00
Mg        1.00000       0.000       A 0.00000E+00 0.00000E+00
Ox        1.00000       0.000       A 0.00000E+00 0.00000E+00

------------------- Output LJ ---------------------------------------

Steepest Descents did not converge in 2 steps
Potential Energy  = -1.26825e+06
Maximum force:  1.97743e+04

---------------------------------------------------------------------

I have tried all the possible combination of Coulomb interactions and
also NS search and periodic-nonperiodic system.
It really looks like the problem is switching on the Buckingham
potential. I noticed that in the logfile these lines appear:

Determining largest Buckingham b parameter for table
Buckingham b parameters, min: 0, max: 83.3333
Generated table with 500 data points for COUL.
Tabscale = 500 points/nm
Generated table with 500 data points for LJ6.
Tabscale = 500 points/nm
Generated table with 500 data points for EXPMIN.
Tabscale = 0 points/nm

It looks like the table for the Buck potential is not generated at all!
This might be related to a change in the code that has been done time
ago to save some memory and that was reported on the mailing list. I
don't know if this is the reason for the weird behaviour...
I would like to know if anybody in the list that has been successful in
running calculations with a Buckingham potential with Gromacs 3.1.4
could send me some input files with coordinates in such a way that I can
test it on my Linux box.
Any kind of suggestion is warmly welcome!

Stefano