[gmx-users] ERROR: Twin-range neighbour searching (NS) with simple NS algorithm notimplemented

[Mehmet Suzen]

Anton Feenstra wrote:

> Selecting rlist!=rvdw or rlist!=rcoulomb (you have both!) will
> (obviously?) get you a twin-range setup. As the message says, that
> is not implemented. AFAI understand, that is not impossible, in fact
> it would be almost trivial to implement, but it would add another
> combinatorial increment to the number of different inner-loop types
> that we already have (ask Erik Lindahl how much we like that!)...

For programming conveneince, I could understand why it is not
implemented due 
to sophisticated and  powerful inner-loop approach. But maybe one needs
to really
use a twin-range cut-off for vacuum simulations. That fas the case for
me.
On the other hand, it is nice to have a no cut-off option for the
interactions, which I also
need.

> By the way, do you really want a coulomb cut-off of 15 (!) nanometer,
> or did you have 15 Angstrom in mind (i.e. 1.5 nm)?

No no, it is 15 nm. It was a test case.

Cheerz
Mehmet