Hello,
I am using OPLS-AA parameters generated using LUNAR (all2lmp_oplsaa_tinker.frc) for a LAMMPS simulation with
pair_style lj/cut/coul/long 12.0.
I noticed that the generated LJ sigma values appear to be systematically smaller than the original Tinker OPLS-AA values by a factor of 2^{1/6}. For example:
-
ether O (OS_122): Tinker OPLS-AA = 2.900 Å, LUNAR output = 2.583606 Å
-
CT_123: 3.500 Å → 3.118146 Å
-
carbonyl O (O_223): 2.960 Å → 2.637060 Å
In each case,
σLUNAR=σTinker/2^{1/6}.
The epsilon values, however, remain unchanged.
Since the Tinker OPLS-AA parameter file specifies radiustype SIGMA, and LAMMPS lj/cut also expects the LJ zero-crossing sigma, I am wondering whether this 2^{-1/6} scaling is intentional or whether it represents an Rmin-to-sigma conversion issue.
Could someone clarify whether the LUNAR-generated sigma values should be used directly with lj/cut/coul/long, or whether the original OPLS sigma values should be used?
Thank you.
