[gmx-users] box size changing during isotropic pressure coupling

[Chris Neale]

>Here tau_p=4, which is already quite large.
>But there are no "real" problems in the system, the results should be fine.
>But the relative ling tau_p might be the reason for the strange scaling.
>This would make the scaling factor always close to 1 plus or minus one bit.
>I have not done the binary math, so I can not see if somehow this would
>make numbers slightly below 20 scale down and above 20 scale up.

While this makes some sense, the triclinic box data that I have posted 
http://www.pomeslab.com/CN_GromacsPost_071808.html (bottom of page)
suggests that there is no such split at 20 nm but that this may be more of an X+Y vs Z issue.

David mentioned that the website was not available? Sorry, it does appear to be visible now though.
Please let me know if you can not see the images and I will post them to photobucket or some such site.

Thanks again,
Chris.