[lammps-users] New to lammps... need tips on dangerous builds etc..

Dear LAMMPS users,

I am interested in running molecular dynamics simulations using lammps. For a start I am trying to simulate a simple SPC/E water system (831 molecules in a 3nm x 3nm x 3nm cube, periodic in x,y,z) for 500ps with 2fs timestep. I am using the lj/cut/coul/long pair potential with a 1.1 nm cut-off. For implementing the SPC/E water model, I use the fix/SHAKE option with appropriate tolerance definitions. I would like to seek some clarifications regarding a few observations... (I am attaching my input file and log file along with this message)

1. Although my simulations run for the entire length of 500 ps, I get a message which reports of 6462 dangerous builds. What could be the reason for that, and what are the consequences of such "dangerous builds"?

2. In the SHAKE algorithm, the manual suggests that we implement eh SPC or SPC/E model for water by means of harmonic bond and bond angle potentials of the form k1(r-r0)^2 and k2(theta - theta0)^2 resp. However, the values of k1 and k2 seem to have little effect on how the bonds and angles are being constrained by the SHAKE algorithm. What would be a suitable choice for these two constants?

Further, I would like to know whether the calculations performed are with double precision. If not what necessary changes are to be effected in order to obtain double precision?

Thank you,


Vikhram V.S.

in.water (760 Bytes)

log.lammps (598 KB)

See Section 2.7 of the doc pages for a discussion of dangerous builds.
it means the first time LAMMPS checked for re-neighboring, one or more
atoms had already moved half the skin distance. So it is possible
you missed a pair of atoms that were already close enough to be interacting.

SHAKE constrains the bond/angle to the value set by the potential. So
the K factors
are ignored by SHAKE.

All calcs in LAMMPS are double precision.