Re: AW: Energy minimization

From: Vermaas, Joshua (Joshua.Vermaas_at_nrel.gov)
Date: Wed Aug 17 2016 - 10:55:14 CDT

If two atoms are initially close together, the VDW interactions, as
calculated via a Lennard Jones potential, are ludicrously strong. This
can lead to humongous forces, which in turn lead to humongous velocities
that violate some of the underlying assumptions in molecular dynamics.
Similarly, a stretched bond can have the same effect. Minimization lets
the particles move around to eliminate these defects, but doesn't result
in dynamics, since no velocities are added into the position integrator.
Usually, if I need more than 1000-2000 minimization steps before getting
something I can simulate normally without crashing, there is something
wrong with my system. At that point, the rate of energy change is pretty
slow, and you'll be near some sort of minimum.

-Josh

On 08/17/2016 04:14 AM, d.ronika_at_iitg.ernet.in wrote:
> I do not understand what you meant by blow apart. Can you please explain
> why do we actually do minimization?
>> From what I understood, you mean that his minimization seems to not have
>> converged even after 1.000.000 steps. This is somewhat untypical but may
>> be related to your system. Can u explain what your system consists of?
>> But, btw. you don’t need a fully converged minimization if you plan to
>> do molecular dynamics afterwards, since the minimization usually has the
>> purpose, just to remove initial conditions that would blow apart the
>> system otherwise.
>>
>>
>>
>> Norman Geist
>>
>>
>>
>>
>>
>> Von: owner-namd-l_at_ks.uiuc.edu [mailto:owner-namd-l_at_ks.uiuc.edu] Im Auftrag
>> von Aron Broom
>> Gesendet: Dienstag, 16. August 2016 16:57
>> An: namd-l_at_ks.uiuc.edu; d.ronika_at_iitg.ernet.in
>> Betreff: Re: namd-l: Energy minimization
>>
>>
>>
>> are you doing energy minimization or dynamics? you say here 1 ns and talk
>> about time-steps, that sounds like dynamics. Energy minimization would
>> just be steps, one can't really assume any particular time.
>>
>>
>>
>> Assuming it really is minimization, I think you should expect a continuous
>> decrease in energy over time. I don't really see why it would need to be
>> straight. Maybe I'm missing something, can you explain why you think it
>> ought to be straight?
>>
>>
>>
>> On Mon, Aug 15, 2016 at 1:05 AM, <d.ronika_at_iitg.ernet.in
>> <mailto:d.ronika_at_iitg.ernet.in> > wrote:
>>
>>
>> In my models used for energy minimization in NAMD, after 1ns i.e., 1000000
>> time-steps. I am not getting a completely straight line. Is it compulsary
>> to get a totally straight line in the energy minimization step? what are
>> the reasons for not getting a completely straight line?
>>
>>
>>
>>
>>
>>
>>
>> --
>>
>> Aron Broom M.Sc
>> PhD Student
>> Department of Chemistry
>> University of Waterloo
>>
>>
>
>

This archive was generated by hypermail 2.1.6 : Sun Dec 31 2017 - 23:20:38 CST