AMBER Archive (2006)

Subject: AMBER: Re: Confusion re AMBER SMD output

From: Alik Widge (alik_at_cs.cmu.edu)
Date: Mon Jul 31 2006 - 17:54:59 CDT

As a followup, mainly so this gets archived:

I've talked to Dr. Adrian Roitberg, who's listed in the manual as one of
the contacts for the SMD code, and he has been able to reproduce my
problem. It turns out to be the inherent behavior of the NMR/restraints
code, which underlies the SMD implementation.

The easy way to prevent the issue is that instead of following the example
in the manual and writing, e.g.,

        &rst iat=485,134, r2=15.0,r2a=20.0, rk2=5000.0 /

one must also set the "rk2a" variable to the same value as "rk2":

        &rst iat=485,134, r2=15.0,r2a=20.0, rk2=5000.0, rk2a=5000.0 /

My thanks to Dr. Roitberg for iterating on this with me over email and
helping me get my project back on track!

Alik

On Mon, 24 Jul 2006, Alik Widge wrote:

> Hello again,
>
> I have been exploring the application of the AMBER9 steered MD code to my
> system, and have discovered something that I do not quite understand.
> According to the manual, when I specify that the SMD history should be
> dumped to a file, I should get four columns:
>
> x0 x_actual force work
>
>
> If I further understand this correctly, the "force" column should tell me
> the current value of the restraint force on the atom(s) being steered,
> according to the usual restraint force F = K*(x - x_0)**2 . However, that
> is not consistent with the values I am observing in the output. For
> instance, if I run the test given under /test/jar , the first few lines of
> the output are:
>
> 15.00000 15.12396 -1239.55482 0.00000
> 15.00500 14.75768 2470.68119 3.07782
> 15.01000 15.13490 -1246.46571 6.13835
> 15.01500 15.15041 -1350.03026 -0.35289
> 15.02000 14.77085 2481.56731 2.47596
> 15.02500 15.12423 -987.34073 6.21152
> 15.03000 15.18296 -1520.41603 -0.05787
>
> I am fairly certain this output is correct, because it's similar to what
> the test's "save" file contains, and it also matches the example in the
> AMBER9 manual. However, the "force" column is clearly not K*(x-x0)**2. In
> this part of the file, the relation is more like 2*K*(x0-x).
>
> What's even more strange is that, towards the end of the run, the forces
> appear to decay to zero even though there is still substantial error
> between the commanded and actual positions. Again looking at the /test/jar
> file, I find:
>
> 19.96500 19.90832 3.96732 66.99162
> 19.97000 19.89329 4.60255 67.01305
> 19.97500 19.87926 4.78696 67.03652
> 19.98000 19.86629 4.54839 67.05986
> 19.98500 19.85980 3.75589 67.08062
> 19.99000 19.86077 2.58457 67.09647
> 19.99500 19.86732 1.27678 67.10612
>
>
> Just looking at the last line, with a force constant of 5000 kcal/mol
> (which is what's specified in the dist.RST file), the force should be in
> the hundreds. In actuality, the force is 10*(x0-x).
>
>
> I can precisely compute the forces in the entire file with the equation:
>
> F = 10*(# steps remaining)*(x0 - x)
>
> Could some wiser person please explain to me what I am seeing? Does the
> dist_vs_t file simply not contain the information I think it does? Or, if
> I am right, why are the forces being computed in this very strange way?
> Are the forces in dist_vs_t the actual forces that are being applied to
> the steered atoms?
>
> Alik Widge
> The Robotics Institute
> Carnegie Mellon University
>
>
>

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu