Hello,
Langevin dynamics avoids a common problem with the berendsen thermostat
called the "flying block of ice," which essentially drains vibrational
energy into translational/rotational energy more or less. Wikipedia has a
decent explanation of it
http://en.wikipedia.org/wiki/Flying_ice_cube .
ntt=1 simply re-scales the velocities of all of the particles to reach a
target temperature. ntt=3 adds a random "kick" to simulate solvent
collisions, which is far less susceptible to non-physical artifacts such as
the flying block of ice.
Now keep in mind that the simulations are chaotic (since it is a many-body
problem), so even the tiniest deviations in initial conditions can cause
simulations to diverge very very quickly. Since one of the simulations is
actually adding a random "kick" to the atoms, you can bet that these two
simulations will become very different very quickly. Only the ensemble
equilibrium properties calculated at the end should match as long as both
simulations properly sample the same ensemble (that is, if ntt=1 doesn't
fall into one of its traps). All that said, I always use ntt=3.
Hope this helps,
Jason
On Thu, Jul 1, 2010 at 12:05 AM, <missing_grace.sina.com> wrote:
> Dear amber users,
> When I follow the amber toturial and run MD ,the toturial1 recommend us to
> use the new Langevin temperature equilibration scheme(NTT=3) for maintaining
> the system temperature,other than NTT=1which is not very efficient at
> ensuring the temperature even across the system .
> But when I turn to the tutrail3 (folding trp-cage peptide),I found when
> heating the system ,it use NTT=1 and tautp=1.0 to heat the system .
> So I am confused ,why we should use NTT=1,not NTT=3 here. Are there many
> differences between these two schemes ?
> Then I try to fold a small peptide(implicit water) , but I find there
> are big rmsd differences when use NTT=1 and NTT=3 respectively.
> I thought that the method we use to heat the system should not influence
> the final result so much.I don't why it is ?
> could you explain it to me ?
> Thank you very much !
>
>
> yours
>
> grace
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Graduate Student
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Jun 30 2010 - 23:30:05 PDT
