[gmx-users] [gmx-users]g_velacc bug?

[Claudio J. Margulis]

Sorry if this appears posted more than once
This is the third time I send it and it seems like it is not getting
posted.
C.
------------------------------------------------------------------------
-----------------------
Dear David, I did try changing the code as you suggested. But it gave
the wrong results. Let me explain,
 
I wrote a bunch of  programs to take trajectories in .gro format (which
I obtained by trjconv -nojump - . from the original trr ). For each of
the PF6- molecules in the box I computed the trajectory of the center of
mass and proceeded to compute the averaged time correlation function of
it with another little program I wrote.
 
I compared this result to
1) unchanged g_velacc
2) changing XX XX for YY ZZ
3) changing XX XX for YY ZZ and modifying the loop as you suggested.
 
My pedestrian computation gives exactly the same result as in 2, 1 is a
little different, and 3 is totally different.
 
This is my understanding of the reason why the cm vac and the p vac are
so widely different:
 
The reason why the velocity autocorrelation function for P and for the
center of mass are way different is that there is a lot of very high
frequency movement of P about the CM.
If one computes the trajectory of one particular P and compares it to
that of the center of mass of its corresponding PF6- ion, one obtains
basically identical trajectories (as one would expect). The only
difference is very high frequency (but low intensity) oscillations in
the case of P. Therefore trajectories look nearly identical but
velocities are widely different due to the constant change of direction
of velocity of P.
 
Anyway this is what I have so far, I felt that those VAC should have
been similar, but really CM and P only should share the low frequency
part, P is in average at the CM but it's fluctuating like crazy around
it and the vac displays these high frequency oscillations.
 
Therefore what you are saying in your last posting puzzles me. Are you
really sure?
I did write my own code to check the results .
 
Thanks!!
Claudio.
 
 
------------------------------------------------------------------------
-----------------
Claudio J. Margulis
Assistant Professor
Department of Chemistry
University of Iowa
319 CB
319-335-0615 (office)
http://www.uiowa.edu/~chemdept/faculty/margulis/index.html
------------------------------------------------------------------------
-----------------
 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20031204/05736c67/attachment.html>