[gmx-developers] Re: The development of a new implementation

[Erik Lindahl]

Hi Ramon,
>
> On the other hand, how do you handle constraints? Adjusting
> velocities so that they do not modify constraints is not difficult,
> and does not require iteration. But it leads to drifts. One must
> constraint positions, and doing so would require either dropping
> reversibility or iterating. Perhaps there is a method in the  
> bibligraphy
> to apply constraints in a reversible way.

Unfortunately, when we digged through papers a couple of years ago  
the only solution for velocity verlet + constraints + PR coupling was  
iterating. However, IIRC, you only need to update the position  
constraints in the first iteration since the relative velocity over  
the bond vanishes after that.

The non-conservativeness of anisotropic pressure scaling is  
interesting, but also problematic since one of the premier  
applications would be membranes, where you obviously want separate  
coupling in the XY plane and Z normal :-)

Cheers,

Erik

PS: Great to see some progress and discussion about this - we're  
quite interested, it's just that we've been too busy with other stuff!

-----------------------------------------------------------
Erik Lindahl  <lindahl at sbc.su.se>
Assistant Professor, Stockholm Bioinformatics Center
Stockholm University, SE 106 91 Stockholm
Phone: +46 8 5537 8564     Fax: +46 8 5537 8214




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-developers/attachments/20051012/a5a4b0ba/attachment.html>