On Mon, 2014-04-28 at 11:17 +0400, James Starlight wrote:
> Forgot to mention
>
>
> In any "entropy outputs" I've not detected any errors besides warnings
> in the
> _MMPBSA_cpptraj_entropy.out
>
> ACTION OUTPUT:
> Warning: In covariance matrix 'comp.matrix', # of frames 100 is less
> than
> Warning: # of columns 14160. If diagonalized, the max # of non-zero
> Warning: eigenvalues will be 100
> Warning: In covariance matrix 'rec.matrix', # of frames 100 is less
> than
> Warning: # of columns 13998. If diagonalized, the max # of non-zero
> Warning: eigenvalues will be 100
> Warning: In covariance matrix 'lig.matrix', # of frames 100 is less
> than
> Warning: # of columns 162. If diagonalized, the max # of non-zero
> Warning: eigenvalues will be 100
>
>
> or many many warnings like
>
> warning: setting vibrational entropy to zero for mode 9413 with vtemp
> = -14945443.762680
> warning: setting vibrational entropy to zero for mode 9414 with vtemp
> = -14939332.231481
> warning: setting vibrational entropy to zero for mode 9415 with vtemp
> = -14933726.365450
> warning: setting vibrational entropy to zero for mode 9416 with vtemp
> = -14926595.749537
> warning: setting vibrational entropy to zero for mode 9417 with vtemp
> = -14922297.649364
> warning: setting vibrational entropy to zero for mode 9418 with vtemp
> = -14915258.643458
When I said that quasi-harmonic calculations take a long time to
converge what I meant was that you need a long simulation with many
snapshots. 100 frames is not enough. As the warning message tells you,
the covariance matrix has, at MOST, only as many orthogonal modes as you
have snapshots.
Since you would expect 3N-6 distinct normal modes (where N is the number
of atoms of your system), you will need at least that many frames just
to get all of the eigenvectors. You will probably need quite a bit more
than that to get converged answers.
You really should have a more thorough understanding of the
quasi-harmonic approximation before trying to use it and interpret the
results.
Good luck,
Jason
--
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Apr 28 2014 - 05:00:03 PDT
