energy drift magnitude in ab initio molecular dynamics
Posted: Sun Jul 24, 2005 2:46 pm
Dear Vasp users,
I'm performing molecular dynamics simulations with 96 lead atoms in hcp structure. My timestep (POTIM) is equal to 2.0 fs and I sample the Brillouin zone by 1 kpoint centered at the gamma point. First I perform 2000 steps in which I scale the particle velocities every 2 time-steps (SMASS=-1) then I keep constant the total energy for others 2000 time_steps (SMASS=-3). The problem
is that during the energy conservation run I observe a energy drift around 25meV/atom that seems too big to me since I'm used to use classical molecular dynamics methods. Is it normal an energy drift of this magnitude in ab initio molecular dynamics? Shall I try to reduce it choosing more kpoints or a bigger timestep (also if the computer time will increase in a dramatical way...)
I'm performing molecular dynamics simulations with 96 lead atoms in hcp structure. My timestep (POTIM) is equal to 2.0 fs and I sample the Brillouin zone by 1 kpoint centered at the gamma point. First I perform 2000 steps in which I scale the particle velocities every 2 time-steps (SMASS=-1) then I keep constant the total energy for others 2000 time_steps (SMASS=-3). The problem
is that during the energy conservation run I observe a energy drift around 25meV/atom that seems too big to me since I'm used to use classical molecular dynamics methods. Is it normal an energy drift of this magnitude in ab initio molecular dynamics? Shall I try to reduce it choosing more kpoints or a bigger timestep (also if the computer time will increase in a dramatical way...)