Re: Calculation of distance traveled by an atom

From: Aron Broom (broomsday_at_gmail.com)
Date: Tue Dec 06 2011 - 02:43:34 CST

My first message was written before I read what Axel had responded with. I
admit that I'm not able to completely follow what he has written (likely on
account of an inadequate mathematics background), but I imagine that if you
google "Verlet integrator", you will get some good clues as to why knowing
the position and force does not mean that you know the velocity.

Also, while this is a moot point given the above statement, do you have
SHAKE turned on? Because in addition to the above, if SHAKE was on this
would clearly lead to discrepancies. But as above, I think you should
likely not expect this approach of comparing displacement from velocity and
trajectory to work out.

~Aron

On Tue, Dec 6, 2011 at 12:22 AM, Axel Kohlmeyer <akohlmey_at_gmail.com> wrote:

>
>
> On Tue, Dec 6, 2011 at 12:11 AM, M.Raviprasad <M.Raviprasad_at_my.ndsu.edu>wrote:
>
>> -----Original Message-----****
>>
>> From: Axel Kohlmeyer [mailto:akohlmey_at_gmail.com] ****
>>
>> Sent: Monday, December 05, 2011 5:37 PM****
>>
>> To: M.Raviprasad****
>>
>> Cc: namd-l_at_ks.uiuc.edu****
>>
>> Subject: Re: namd-l: Calculation of distance traveled by an atom****
>>
>> ** **
>>
>> On Mon, Dec 5, 2011 at 3:54 PM, M.Raviprasad <M.Raviprasad_at_my.ndsu.edu>
>> wrote:****
>>
>> > To NAMD users,****
>>
>> >** **
>>
>> > I calculated distance traveled by an Oxygen atom (gas molecule) in the
>> ****
>>
>> > polymeric system using DCD file. In order to make sure the ****
>>
>> > calculation, I used velocity to calculate the distance (distance =
>> Velocity*time step).****
>>
>> > Velocity was taken from the velocity DCD file. However, these two ****
>>
>> > values are not matching. I believe that results should be the same ****
>>
>> > (not exact but****
>>
>> > close) irrespective of the methods.****
>>
>> ** **
>>
>> i disagree. if it was possible to even approximately predict the
>> trajectory of atoms this way, we wouldn't have these issues with requiring
>> so much computer time to do MD simulations of big systems.****
>>
>> ** **
>>
>> I understand that we cannot predict the trajectory by simply doing the
>> calculation. But velocities stored in the VEL.DCD file were calculated
>> based on the atomic positions stored in the coordinate DCD file. So I
>>
> no, they are not! the *forces* are computed from the coordinates.
> velocities are computed from previous velocities and the force.
> please look up newton's laws.
>
>> would expect to have approximately same distance between two time frames
>> in both ways (Coordinate DCD and Velocity DCD).
>>
> this is nonsense.
>
>> ****
>>
>> ** **
>>
>> > What am I doing wrong?****
>>
>> ** **
>>
>> you are completely ignoring the fact how time integration works****
>>
>> ** **
>>
>> I understand that the time integration for coordinates and velocities are
>> different. But stored coordinates and velocities are for the same time
>> step. So I would think that integration time would not increase the error
>> in the approximate calculation.
>>
> sorry, your reasoning would _only_ work, if there were *no* forces
> at all acting between the atoms. again, see newton's laws.
> we are talking *elementary* classical mechanics stuff here.
>
>> and the fact that atomic velocities are significantly impacted by high
>> frequency vibrations and librations, that would integrate out on long time
>> scales.****
>>
>> ** **
>>
>> But, these vibrations due to the system temperature will be adjusted by
>> rescaling the velocities. So I do not think much of a problem with this in
>> the calculation of distances.
>>
> huh? this statement makes even less sense than the previous ones.
>
>> > Is there any other way to calculate and verify the distance travelled
>> ****
>>
>> > by an atom?****
>>
>> ** **
>>
>> i don't see why you need to "verify" the coordinate data. ****
>>
>> the coordinate dcd file is the perfect(!) record of how each atom has
>> traveled during your simulation and the distance between any two reference
>> points is as accurate as it gets (outside of quantum effects that don't
>> show up in classical MD in the first place).
>>
>> I agree that the coordinate dcd is perfect for distance calculation.
>>
> so why do you need to "verify" it?
>
> axel.
>
>
>> Thanks****
>>
>> Ravi ****
>>
>> ** **
>>
>> ** **
>>
>> ** **
>>
>> cheers,****
>>
>> axel.****
>>
>> ** **
>>
>> > Any suggestion is appreciated****
>>
>> >** **
>>
>> > Thanks****
>>
>> >** **
>>
>> > Ravi****
>>
>> >** **
>>
>> >** **
>>
>> ** **
>>
>> ** **
>>
>> ** **
>>
>> --****
>>
>> Dr. Axel Kohlmeyer****
>>
>> akohlmey_at_gmail.com http://goo.gl/1wk0****
>>
>> ** **
>>
>> College of Science and Technology****
>>
>> Temple University, Philadelphia PA, USA.****
>>
>> ** **
>>
>> ** **
>>
>
>
>
> --
> Dr. Axel Kohlmeyer
> akohlmey_at_gmail.com http://goo.gl/1wk0
>
> College of Science and Technology
> Temple University, Philadelphia PA, USA.
>

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:21:03 CST