[gmx-users] my own routine

[Berk Hess]

>I'm  calculating geometrical axes at every frame using ql77 for the
>diagonalization. The problem is not the diagonalization.
>The problem is the .xtc file because everything is OK when I check the
>program on .gro formats while I get problems using the .xtc
>trajectory. In fact everything is OK except for a couple of frames where
>the simulation has been restarted
>to continue after interruption. At these two frames I obtain wrong
>results for the eigenvectors, for example the modulus is different from
>the unity.
>Is there a function to overcome the problem of these interrupted frames
>using a .xtc trajectory?

This all seems very strange.
I have never heard of some frames being wrong and later frames being ok.

The problem can only be in the xtc if some coordinates are really wrong,
such as all coordinates being identical.
You should be able to see this with ngmx.
But then the gro files can not be ok.
But how did you obtain the gro files then?
I assume with trjconv, but with which options?

Does the system you analyze consist of multiple molecules?

BTW There is a routine in gmxlib/princ.c called principal_comp that 
calculates
the geometrical axes, not using ql77, but using jacobi.

Berk.

_________________________________________________________________
Play online games with your friends with MSN Messenger 
http://messenger.msn.nl/