Re: [AMBER] error in nmode calculations

From: Jason Swails <jason.swails.gmail.com>
Date: Wed, 28 Mar 2012 17:46:54 -0400

On Wed, Mar 28, 2012 at 4:11 PM, Cannon, John F. <
CannonJ.health.missouri.edu> wrote:

> >From your input, you are expecting a nmode calculation using only two
> frames (1 to 10 every 5 frames).
> How could any vibration frequencies be calculated with only two
> conformations?
>

It's calculated from the Hessian matrix of the system. The vibrational
frequency of each normal mode is just the eigenvalue of that mode
(eigenvector).

If you're referring to using the mass-weighted covariance matrix
(essentially PCA) to get your modes and frequencies, then you certainly
need more than 2 frames to get any kind of meaningful results. Implemented
in MMPBSA.py, that's referred to as the quasi-harmonic approximation, which
is distinct from what we're calling normal mode analysis here.

I'm not suggesting that it's good practice to use only 2 conformations for
a normal mode calculation, but it does give you vibrational frequencies,
and it's a good approach for debugging an issue with normal mode
calculations.

All the best,
Jason

-- 
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Mar 28 2012 - 15:30:04 PDT
Custom Search