[gmx-developers] Fwd: [gmx-revision] check for rcoulomb=rlist with PME, Ewald and PPPM

[Berk Hess]

Michael Shirts wrote:

>>grompp now check that rcoulomb is equal to rlist with PME, Ewald and PPPM.
>>    
>>
>
>I'm a little confused as to why this change was made.  Does this mean
>that there will be an error printed if  rlist > rcoulomb?  In my
>experience, it was impossible to get good energy conservation  unless
>rlist > rcoulomb, and that a specific change was made a couple
>versions ago to -change-  from the very grompp behavior described
>before.  As I understand it, with rlist = rcoulomb, particles drift in
>and out, and things that should get treated with short-range
>interations aren't . . .
>  
>
The behavior of Gromacs for the PME electrostatics has never been different.
It always used all interactions up to rlist.
It is always better to use everything up to rlist.
Drifting out of 1 nm gives a smaller error than out of 0.8 nm.

So if you had rcoulomb<rlist,
you should set rcoulomb=rlist and decrease ewald_rtol
to obtain the same beta as before.

Berk.