[gmx-users] pme_order and fourierspacing

[Erik Lindahl]

Hi,

We haven't really tested this in detail (the way to do it is to check 
the force RMSD vs. a very accurate force calculation), but in general
I'd try to increase pme_order to 6 and increase the spacing to 
0.15-0.18, or use pme_order 12 with 0.2.

Again, this might be overkill since I've never tried larger 
fourierspacings :-) The size of the grid depends on the third power
of the spacing though, so if you increase it by 50% you decrease the
size of the grid by almost a factor 4.

I'm pretty sure 1.2 is way too large since that would mean you only have
4*4*4 cells in a typical box...

Cheers,

Erik


>>
>>pme_order = 8 / fourierspacing = 0.48
>>
>>vs
>>
>>pme_order =10 / fourierspacing = 1.20
> 
> 
> try comparing to the the default:
> pme_order = 4 / fourierspacing = 0.12
> 
>