Re: [AMBER] Units from quasiharmonic analysis of Principal Components

From: David A Case <>
Date: Sun, 19 Jul 2020 20:57:58 -0400

On Sun, Jul 19, 2020, German P. Barletta wrote:
>I know from Case's paper on crambin, and this mailing list, that when
>running this kind of analysis with cpptraj (that is, using the matrix
>mwcovar command), eigenvalues are reported in units of cm^-1. Why is that?
>It's supposed to be a diagonalization of a covariance matrix, so the
>eigenvalues should be in units of distance^2 (say, cm^2). Right? Is Amber
>just taking the inverse square and reporting those values? If so, why? Or
>is there something else I'm missing?

The key is the "mw" (mass-weighted) part: if you just ask to diagonalize
the covariance matrix, you will get what you describe above. We took
a request to do mass-weighting as an indication that one would most
likely be comparing to either normal mode calculations or experimental
vibrational frequencies.

Note that "wavenumbers" (cm^-1) is a conventional unit, giving
1/wavelength for the fundamental vibrational excitation in the harmonic
limit.: This is proportional to energy, and wavenumbers are often treated
as a unit of energy:

     E = h*nu = hc/lambda = hc(1/lambda).

They do *not* correspond to just taking the inverse square-root of a
covariance matrix eigenvalue: mass weighting would make the units all
wrong if you tried to do that.

...hope this helps...dac

AMBER mailing list
Received on Sun Jul 19 2020 - 18:00:02 PDT
Custom Search