Dear all,
After my initial structure (For exammple polymer PE) is well annealed at 300k, could I continue to use the “minimize” command to equilibrate the mentioned structure obtained in 300k?
If it’s possible. Based on the minimized structure, I want to operate the pulling test (Ensemble is not adopted), in which “minimize” is performed in every steps (to simulate the static pulling).
Then I want to know the pulling process is under 0 K or under 300K? (some physical colleague believed that the pulling proces it’s under 0K, while the engineering background colleague thought it is under 300K).
Thanks a lot,
Best wishes,
redstone.
Dear all,
After my initial structure (For exammple polymer PE) is well
annealed at 300k, could I continue to use the "minimize" command to
equilibrate the mentioned structure obtained in 300k?
a minimization quenches the structure to a low temperature configuration.
if you want to equilibrate for 300K then a minimization will undo part of the
equilibration.
If it's possible. Based on the minimized structure, I want to
operate the pulling test (Ensemble is not adopted), in which "minimize" is
performed in every steps (to simulate the static pulling).
technically, that is possible, yes.
Then I want to know the pulling process is under 0 K or under
300K? (some physical colleague believed that the pulling proces it's under
0K, while the engineering background colleague thought it is under 300K).
neither. while a minimization is strictly speaking "operating" your system
at 0K, you will get trapped in some local minimum and never have a real 0K
structure. the repeated minimization will thus result in some undefined
state, and each minimization after a pull will probably bring you to a different
(lower) potential energy surface. but you also are not at 300K, since you *do*
quench your system. a pull at 300K would require doing a slow pull with
MD and corresponding dissipative termalization, e.g. through fix langevin.
axel.
Dear Axel,
Thanks for your rapid reply, your explanation is quite clear, actually I want to get one quastic static pulling process at 300k.
Concerning to your reply “a pull at 300K would require doing a slow pull with
MD and corresponding dissipative termalization, e.g. through fix langevin.”, I still have two questions:
-
For the slow pull corresponding to 6nm workpiece, which kind of speed I should give for MD. From the traditional mechanics point of view, the quastic static strain loading rate is around 0.01 per second (corrsponding to 6E-13 A/ps), it seems not possible for MD simulation. or there are some other methods?
-
I checked “lammps help” again, we can use “fix langevin” to keep system temperature at 300K, could I use also combine with “fix nve”? besides how I understand “the implicit solvent” with respect to “quastic static operation” I needed?
(sorry for my poor theorical background)
All the best,
redstone
Dear all,
After my initial structure (For exammple polymer PE) is well
annealed at 300k, could I continue to use the “minimize” command to
equilibrate the mentioned structure obtained in 300k?
a minimization quenches the structure to a low temperature configuration.
if you want to equilibrate for 300K then a minimization will undo part of the
equilibration.
If it’s possible. Based on the minimized structure, I want to
operate the pulling test (Ensemble is not adopted), in which “minimize” is
performed in every steps (to simulate the static pulling).
technically, that is possible, yes.
Then I want to know the pulling process is under 0 K or under
300K? (some physical colleague believed that the pulling proces it’s under
0K, while the engineering background colleague thought it is under 300K).
neither. while a minimization is strictly speaking “operating” your system
at 0K, you will get trapped in some local minimum and never have a real 0K
structure. the repeated minimization will thus result in some undefined
state, and each minimization after a pull will probably bring you to a different
(lower) potential energy surface. but you also are not at 300K, since you do
quench your system. a pull at 300K would require doing a slow pull with
MD and corresponding dissipative termalization, e.g. through fix langevin.
axel.
Dear Axel,
Thanks for your rapid reply, your explanation is quite clear,
actually I want to get one quastic static pulling process at 300k.
Concerning to your reply "a pull at 300K would require doing a slow
pull with
MD and corresponding dissipative termalization, e.g. through fix langevin.",
I still have two questions:1. For the slow pull corresponding to 6nm workpiece, which kind of speed I
should give for MD. From the traditional mechanics point of view, the
quastic static strain loading rate is around 0.01 per second (corrsponding
to 6E-13 A/ps), it seems not possible for MD simulation. or there are some
other methods?
you cannot have an exact match, but you can get very close.
1) you can follow the protocol that you initially mentioned, i.e.
take a reasonably equilibrated work piece. define a group
of atoms at each end of the workpiece and use fix setforce
to zero out any forces on them. displace one of the groups
by a (small) increment and relax the rest with a minimization
and repeat.
2) do the same as with 2), but this time use MD and run until
the system is re-equilibrated (total energy is fluctuating around
a constant value)
3) use fix move and displace the group in a continuous and
a reasonably slow motion so that any energy added to the
system through the motion can be dissipated and the workpiece
effectively stays in equilibrium. what is a good speed would
need to be determined empirically. with too fast a motion, the
system cannot relax well enough.
each variant has benefits and disadvantages.
axel.
Dear Axel,
Thanks for your rapid reply, your explanation is quite clear,
actually I want to get one quastic static pulling process at 300k.
Concerning to your reply "a pull at 300K would require doing a slow
pull with
MD and corresponding dissipative termalization, e.g. through fix langevin.",
I still have two questions:1. For the slow pull corresponding to 6nm workpiece, which kind of speed I
should give for MD. From the traditional mechanics point of view, the
quastic static strain loading rate is around 0.01 per second (corrsponding
to 6E-13 A/ps), it seems not possible for MD simulation. or there are some
other methods?2. I checked "lammps help" again, we can use "fix langevin" to keep system
temperature at 300K, could I use also combine with "fix nve"? besides how I
understand "the implicit solvent" with respect to "quastic static
operation" I needed?
it has no relevance. but one side effect of those random motions that
mimic a solvent is, that it also dissipates energy better/faster in most
cases. that is helpful in your case. other thermostat methods would
work, too but it would take longer to dissipate the energy. in any case,
you have to make sure that the time constant of the thermostat is
large enough so that it doesn't interfere with your dynamics too much,
but also not too large.
(sorry for my poor theorical background)
that is easy to fix, too. 
axel.
Dear Axel,
I really thank you very much for your careful and detailed explanation.
thanks for your time.
Best,
Redstone
Dear Axel,
Thanks for your rapid reply, your explanation is quite clear,
actually I want to get one quastic static pulling process at 300k.
Concerning to your reply “a pull at 300K would require doing a slow
pull with
MD and corresponding dissipative termalization, e.g. through fix langevin.”,
I still have two questions:
- For the slow pull corresponding to 6nm workpiece, which kind of speed I
should give for MD. From the traditional mechanics point of view, the
quastic static strain loading rate is around 0.01 per second (corrsponding
to 6E-13 A/ps), it seems not possible for MD simulation. or there are some
other methods?
you cannot have an exact match, but you can get very close.
-
you can follow the protocol that you initially mentioned, i.e.
take a reasonably equilibrated work piece. define a group
of atoms at each end of the workpiece and use fix setforce
to zero out any forces on them. displace one of the groups
by a (small) increment and relax the rest with a minimization
and repeat. -
do the same as with 2), but this time use MD and run until
the system is re-equilibrated (total energy is fluctuating around
a constant value) -
use fix move and displace the group in a continuous and
a reasonably slow motion so that any energy added to the
system through the motion can be dissipated and the workpiece
effectively stays in equilibrium. what is a good speed would
need to be determined empirically. with too fast a motion, the
system cannot relax well enough.
each variant has benefits and disadvantages.
axel.