Hi everybody,
I am studying the stability of defects in graphene at finite temperature. To carry out this, I usually simulate the periodic cell for 2000 ps. Does anyone know if this time is enough? How can I know that this time is enough? How long would it be appropriate in this cases? Because, any changes can occur after this time, so I am losing information, however I cannot simulate during a long time (i.e. seconds) due to reasonable reasons.
Many thanks in advance.
Juan Pedro.
I think you just have to try it with longer and longer simulations
and see what happens.
Steve
2011/9/19 Juan Pedro Méndez Granado <[email protected]...>:
Hi everybody,
I am studying the stability of defects in graphene at finite temperature. To
carry out this, I usually simulate the periodic cell for 2000 ps. Does
anyone know if this time is enough? How can I know that this time is enough?
How long would it be appropriate in this cases? Because, any changes can
occur after this time, so I am losing information, however I cannot simulate
during a long time (i.e. seconds) due to reasonable reasons.
this is one of the elementary problems of MD (or rather statistical mechanics).
a lot of research effort has been dedicated to the study of rare event with
finite size system and finite time and there are multiple ways to approach
the subject. simply running until something happens, is the worst. it is in
the very nature of statistics that you cannot predict a specific event.
since what you are looking for is effectively a free energy, the
easiest way to approximate the timing issue is to run simulations
at elevated temperatures, and try to reweight the distribution of
the "reaction" time for a lower temperature. check out the TAD
method that is implemented in LAMMPS, for example:
http://lammps.sandia.gov/doc/tad.html
cheers,
axel.