[gmx-developers] Re: Cubic splines - again

[Berk Hess]

Mathias PUETZ wrote:

>
> Hi Berk,
>
> I don't really understand what you are getting at with your "as well" 
> statement below.
>
I meant that one could supply the potential at the table points
and then cubic spline interpolate the potential.
That would give continuous V, V' and V''.
The force at table points would then not exactly match the real force,
but that would not be the case at intermediate points with any kind of 
interpolation.

Thus we can leave the innerloops unchanged.

It would also make supplying user potentials easier,
but one probably has to take care that the potential values are accurate,
otherwise the splines will fluctuate rapidly.
Also the interpolated values might depend on the boundary values at
the ends of the table. I would have to look up some details on splines
to see how we could solve this best.

Berk.