[gmx-users] subspaces / cov. matrices overlap

[Jose D Faraldo-Gomez]

Hi,

I'm trying to compare the two methods to calculate the overlap between 
simulations available in g_anaeig. In particular I'm comparing two different 
simulations, as well as different time windows within each simulation.

When I use the method based on the trace of the covariance matrix I get in 
general low values, and these increase with the sampling window lenght in a 
similar way as for Hpr in Berk's last paper in PRE. So far so good.

However, when I use the method based in eigenvectors, I always get a final 
overlap of 1, no matter which pair of windows/simulation I use. I am using 
all the eigenvectors in the trr's for the overlap calculation, i.e. I do:

g_anaeig_d -v $trr -v2 $trr2 -over -last 402 (for 134 C-alphas)

and I always get that the 402 eigenvectors in $trr2 overlap completely wiht 
the 402 eienvetors in $trr (the overlap increases linearly), independently 
of the choice of these sets.

So again: what am I missing? Is this related to this definition of overlap 
based on eigenvectors, to the properties of the eigenvectors themselves, or 
to me being totally useless?

Thanks again,

Jose


_________________________________________________________________
Send and receive Hotmail on your mobile device: http://mobile.msn.com