[lammps-users] Questions about new and old version

Dear all,

 I have a question about new and old version software\.

 1\) Recently, I have found that the same input for relaxation of

dislocation in new and old version LAMMPS, will obtain different relaxation
results. In old version, the atoms are arranged well, while in new version
atoms are in wrong sites but maintain right configuration, I am puzzled

what are "wrong" sites.

please keep in mind, that a minimization will
only find local minima and particularly if you
have a configuration sitting on a transition between
two minima even very small changes in the code
(or using a different number of processors, changing
the order in which forces are summed up) can
decide which of the available minima your system
will go into. have you compared the two resulting
energies? if they are equal, have you checked if
the two results are related by symmetry?

about these results.

 The input file is the following:

this is totally useless.

units metal

dimension 3

boundary s s s

 atom\_style         molecular

read_data data.type1

pair_style sw

pair_coeff * * GaN.sw Ga N

compute 2 all stress/atom

compute 3 all pe/atom

fix 2 all nve

# run

thermo 100

minimize 1.0e-4 1.0e-6 100 1000

#timestep 0.001ps

dump 2 all xyz 100 dump.crystal

run 10000000

 2\) If I only use nve command without langevin , How does the LAMMPS

determine the temperature of the system?

ouch! this is a text book question. please look it up yourself!
you don't need a temperature control to determine a system's
temperature. and to be more precise, we really don't know the
temperature, only its kinetic energy, but there are ways to relate
the kinetic energy to the temperature of a(n infinitely large) body,
which we apply to finite bodies as well.