[gmx-users] Constraining N-H and H-H bonds in ammonium via SETTLE?

[Erik Lindahl]

Hi John,

On May 23, 2007, at 9:09 PM, John D. Chodera wrote:

> Hi all,
>
> Can the SETTLE implementation in gromacs be used to constrain N-H  
> and H-H bond lengths in ammonium (NH4+) molecules?
>
> We have prepared an .itp file (attached below) -- based on the  
> Pande group ffamber port -- for ammonium (NH4+).  If gromacs could  
> settle N-H bonds too, would this involve simply adding the  
> following line?
>
> [ settles ]
> ; OW    funct    doh      dhh
> 1       1        0.101    0.166

No, it won't fly that way. The reason Settle works (and is special)  
is that the three atoms form an isolated molecule, so knowing the  
forces on all three sites we can solve for the resulting rigid-body  
rotation analytically.  You can't just add another bond to that  
molecule that is constrained the normal way, although in theory you  
could create a 5-body SETTLE (probably too expensive).

Cheers,

Erik