I'm simulating the deformation of a nanotubes' system, including 1020
carbon atoms, by LAMMPS released recently. While shifting between
successive several-thousand-steps' NVT integration and the
intermediate energy minimization with style hftn, an approximately
constant difference of total potential energy between the two
processes appears at every equilibrium, which seems to be
unreasonable. In fact, the total potential energy decreases by
approximately 35(metal units) during every minimization, from which an
energy increment of this constant amount follows in the very begining
of the several-thousand-steps' NVT integration. Considering that,
compared to this quantity, the energy change caused by deformation
during every several-thousand-steps' NVT integration is rather
small(about 1 with metal units), I think the situation can hardly be
explained by the configuration change of the system, or any other
reason I could come up with. Is there anything special about the
compute of etotl? Coincidentally, I found the kinetic energy of this
300K-thermostat system to be approximately the same as the above
quantity, but according to the manual, there may not be any relevance
between them.
Your question isn't specific enough to answer. You need
to explain something you think LAMMPS is doing wrong,
with data to support it, if that's what you're after.
Steve