[gmx-users] eigenrmsf in COMBINED ESSENTIAL DYNAMICS

[Tsjerk Wassenaar]

Hi Sridhar

On 2/8/07, sridhar at cdfd.org.in <sridhar at cdfd.org.in> wrote:
> Dear Dr. Tsjerk,
>
> Thankyou for the suggession. I am however not clear with the explanation.
> Did you mention the calculation of variance using the mean projection for
> the eigenrmsf calculations?
>

Yes.

> The projections of the two trajectories on the combined eigenvectors are
> different. But, the eigen rmsf plots with give a residue wise rmsf
> fluctuations along individual eigenvectors are also same for both the
> trajectories.
>

This is due to the way the principal components are defined. I suggest
you read a statistics textbook on principal component analysis.

> Definetly the fluctuations along the combined eigen vectors should be
> different for separate trajectories.
>

By no means.

> Can you explain in a bit detail?
>

It will help you more to read a text on PCA first. I did give it my
best shot before. The main thing that seems to be bothering you is
what PCA is doing exactly, with which you do have to note that you're
dealing with PCA on a combined data set. To understand that may not be
trivial, certainly not without understanding the basic theory from
PCA.

Cheers,

Tsjerk

-- 
Tsjerk A. Wassenaar, Ph.D.
Junior UD (post-doc)
Biomolecular NMR, Bijvoet Center
Utrecht University
Padualaan 8
3584 CH Utrecht
The Netherlands
P: +31-30-2539931
F: +31-30-2537623