Re: reversing constant velocity SMD

From: Peter Freddolino (petefred_at_ks.uiuc.edu)
Date: Thu Apr 24 2008 - 08:24:55 CDT

Hi Jeff,
the way namd calculates the current reference position is by multiplying
the timestep by the SMD velocity. So, if (for example) you pull at 0.001
A/ps for 1 ns along the x axis, reach 1 A, and then switch the sign of
the pulling direction, namd "thinks" that the target atom should be at
x=-1, and applies a force accordingly. The way to fix this is to make a
new SMDfile with the coordinates as they are immediately when you
reverse the pulling direction, and restart your timestep at 0. This can
be inferred from eq. 3 from
http://www.ks.uiuc.edu/Research/namd/2.6/ug/node33.html and following it
through to its logical conclusion if you flip the sign of v after a
nonzero number of timesteps.
Best,
Peter

Jeff Forbes wrote:
> I have performed a constant velocity SMD of a protein with
> a velocity of 0.000005A/step. I would now like to push the
> SMD atom back towards its original position at the same velocity.
> Simply changing the sign of the SMDDir parameter does not work as
> expected.
> The SMD atom moves in the correct direction, but the velocity is much
> faster than expected.
> The original position of the SMD atom is reached in ~ 70 ps, where as
> the stretched starting position was reached
> after ~42 ns. The average reverse velocity is ~ 3e-3A/step, quite a
> bit faster.
> Any suggestion on how to get the expected behavior would be welcomed.
>
>
> Jeffrey G. Forbes, Ph.D.
> Muscle Proteomics and Nanotechnology Section
> Laboratory of Muscle Biology
> National Institute of Arthritis and
> Musculoskeletal and Skin Diseases
> National Institutes of Health
> Department of Health and Human Services
> 50 South Drive
> Building 50, Room 1139
> Bethesda, MD 20892-8024
> Voice: 301-451-9535
> FAX: 301-402-0009

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:49:24 CST