Re: [AMBER] units of eigenvalues from pca

From: case <case.biomaps.rutgers.edu>
Date: Wed, 15 Sep 2010 08:07:14 -0400

On Wed, Sep 15, 2010, moitrayee.mbu.iisc.ernet.in wrote:
>
> I have calculated the eigenvalues from covariance matrix using
> Ambertools. I am confused about the unit of the eigenvalues. It is
> mentioned in the documentation that the eigenvalues obtained from a
> mass-weighted covariance matrix is cm-1. However, i do not understand
> what is the unit in my case.

If you use mass-weighted coordinates, the eigenvalues are treated as
quasiharmonic frequencies, and reported in cm**-1. See

%A M. Karplus
%A J.N. Kushick
%T Method for estimating the configurational entropy of macromolecules
%J Macromolecules
%V 14
%P 325-332
%D 1981

For other types of PCA analysis, the eigenvalues have arbitrary units; I guess
they are in Ang.**2, but only relative values have much meaning, as far as I
know. Note that quasiharmonic frequencies are inversely related to PCA
eigenvalues, so that small frequencies correspond to large PCA amplitudes and
vice-versa.

...dac


_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Sep 15 2010 - 05:30:05 PDT
Custom Search