Dear Professor Case,
Thank you for your comments.
> PCA is independent of time, so it no notion of "fast" and "slow".
I am afraid that I made a over-interpretation of PCA.
My earlier idea is below:
An eigen value denotes an amplitude of mode.
Then, if the amplitude can be considered as the wavelength of the mode,
Frequency of the mode should be proportional to eigen value
(Although I assume the same velocity for all modes).
Slow and fast mean in this sense.
I think now that I should perform fourier transform for the time course of
Coordinates projected onto an eigen vector to know frequency of the eigen mode.
Thank you for your comments, again.
It really help me to notice that point.
Yours sincerely,
Ikuo KURISAKI
-----Original Message-----
From: case [mailto:case.biomaps.rutgers.edu]
Sent: Sunday, February 17, 2013 12:04 AM
To: AMBER Mailing List
Subject: Re: [AMBER] What algorithm does ptraj use for solvation of eigen-value
problem.
On Sat, Feb 16, 2013, kurisaki wrote:
>
> In the contrast, can we consider the eigen vector with smaller eigen value
> represent fast mode, such like bond stretching between atoms?
> They could be ignored if we focus on global dynamics of protein.
> If so, it is clear that I should ignore such a fast mode.
PCA is independent of time, so it no notion of "fast" and "slow". It the
eigenvalue of the 306th vector isolated from other eigenvectors? If you
really want to investigate this, you should look at the character of this
eigenvector, and see if you learn anything from what atoms are moving.
It would also be of use to know how many atoms (or Ca atoms) are in your
system.
....dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat Feb 16 2013 - 23:00:02 PST
