[gmx-users] GROMACS and CNT/graphene (probably for the 1000th time)

[Alex]

Okay, I am learning here. Here's what I did:

1. Copied the entire oplsaa folder in the local directory.

2. Added the following to ffnonbonded.itp:
 opls_995   CJ      6    12.01100     0.000       A    3.40000e-01  3.61200e-01
 opls_996   CJ      6    12.01100     0.000       A    3.40000e-01  3.61200e-01
 opls_997   CJ      6    12.01100     0.000       A    3.40000e-01  3.61200e-01
 
3. Added the following to ffbonded.itp
under dihedral types:
  CJ     CJ     CJ     CJ      3     17.30770   0.00000 -17.30770   0.00000   0.00000   0.00000 ; From opt. Brenner
under angle types:
  CJ     CJ     CJ      2   120.000    776.923   ; G96 angle, param from opt. Brenner
under bond types:
  CJ    CJ      1    0.14200   420420.0   ; from opt. Brenner 698.129 N/m

4. Added the following to atomname2type.n2t:
CJ   opls_995    0      12.011  2    C  0.142  C 0.142
CJ   opls_996    0      12.011  3    C  0.142  C 0.142  C 0.142
CJ   opls_997    0      12.011  1    C  0.142

After all of this g_x2top with -noparam does almost everything right (the angle type
needs to be manually changed to 2, although it is explicitly 2 above).
But whatever, I can run the in vacuo dynamic simulation, and the sheet
behaves beautifully and respects my PBC (the box has precise in-plane
dimensions).

I have solvated my sheet with quite a bit of SPC water (~32K molecules), about to test some
energy minimization followed by a dynamic simulation.

Does the above sound reasonable?

Thanks,

Alex

p.s. The structure alone was working right from that tutorial, it's
the solvation part that sucked.

JL> If you're introducing a new [defaults] directive, then whatever you're doing is
JL> breaking the required force field format.  Chapter 5 is your best friend here.
JL> You say you want some custom parameters and that you're having issues with
JL> implementing new parameter into existing force fields, so I'm telling you that
JL> the best approach (in my mind) is to do something simple.  Something that you
JL> can write in a few lines (for a CNT you have a single C type and I can count the
JL> bonded parameters on one hand).  Make sure that works.  Then worry about making
JL> it more complicated.  But at that point, you're just adding more [atomtypes] and
JL> [bond|angle|dihedraltypes], nothing more.  You only ever have one [defaults] -
JL> these are the master instructions for the shape of the functional form and how
JL> the force field is interpreted.

JL> Canary in the mineshaft - get the CNT working alone.  Make sure you understand
JL> everything about that.  Because at that point, you've done most of the hard work
JL> and everything is just an addition to it.

JL> -Justin

JL> -- 
JL> ==================================================

JL> Justin A. Lemkul, Ph.D.
JL> Ruth L. Kirschstein NRSA Postdoctoral Fellow

JL> Department of Pharmaceutical Sciences
JL> School of Pharmacy
JL> Health Sciences Facility II, Room 629
JL> University of Maryland, Baltimore
JL> 20 Penn St.
JL> Baltimore, MD 21201

JL> jalemkul at outerbanks.umaryland.edu | (410) 706-7441
JL> http://mackerell.umaryland.edu/~jalemkul

JL> ==================================================