[gmx-users] XYZ-Direction of eigenvector

[Mark Abraham]

On 14/12/2011 11:20 PM, Linder Tobias wrote:
> Dear gromacs users,
>
> I used g_covar in order to get the eigenvectors of a MD simulation. Analysis of the eigenvalue plot showed the high eigenvalue of the first eigenvector. I also used gmxdump in order to analyse eigenvec.trr. Frame 1 of the eigenvec.trr shows the eigenvector direction of each atom of eigenvector 1.
> I was wondering if there is a single XYZ-Direction (only one general direction, not for each atom) of this first eigenvector because to my (basic) knowledge, there should be an eigenvector direction corresponding to the eigenvalue in the eigenvalue plot.

You removed the global motion when you fitted to a reference structure, 
so all the remaining motion is internal. So you need to expect that the 
set of orthonormal vectors, in whose directions the mean-squared 
fluctuations are independent of each other (i.e. the eigenvectors of the 
eigenvalues you found), will generally have contributions from every 
remaining degree of freedom.

Mark