unique spline question

[Axel]

On Nov 29, 1:37 am, Teodoro Laino <teodor... at gmail.com> wrote:
> hi axel,

hi teo,

i'll try to get you the input (not my call),  but perhaps i should ask
differently. if i use unique spline, do i _have_ to change the spline
epsilon to a much smaller to get the same accuracy than without?
or would the default epsilon still be sufficient?

axel.

> this should not be the case.. as usual I need to see the input file
> for FIST.
> Can't say anything else without it.
>
> cheers
> teo
>
> On 29 Nov 2007, at 00:52, Axel wrote:
>
> > hi all!
>
> > i'm currently playing around with options to save memory
> > for a large QM/MM calculation and i noticed, that when
> > using the default spline settings or using unique splines
> > with the recommended epsilon of 1.e-12 the memory
> > consumption is about the same.
>
> > am i comparing apples and oranges here?
>
> > since the default epsilon is 1e-7 and of course
> > if i use a 1.e-12 epsilon without unique splines
> > enabled memory use explodes.
>
> > is there any recommendation what would be the best
> > way to continue?