[gmx-users] PME accuracy

[Erik Lindahl]

Hi Vincent,

First - the accuracy estimates are those of Darden and coworkers; check  
out the Smooth PME paper mentioned in the references.

Second, I'll have a look both at PME and the normal (non-PME) fourier  
method. We haven't used that for 1-2 years, so there might be a bug  
there rather than in PME...

Cheers,

Erik


On Tuesday, May 6, 2003, at 03:21 America/Los_Angeles, Vincent  
Ballenegger wrote:

> Vincent Ballenegger wrote:
>
>> * Ewald method with fourier_nx=25:
>>               force = -10.824      "exact result"
>> * PME method with fourier_spacing=0.1, pme_order=4:
>>               force = -7.789         error: 28% !!!
>
> I should have mentionned that the above figures were obtained with  
> rcoulomb = 0.9.
> (The measured force is of course the projection of the force vector  
> along the diagonal of the box.)
>
> Changing rcoulomb gives (for PME with fourier_spacing=0.1 and  
> pme_order=4):
>
> rcoulomb=1.74 (~ half the box size)  => force = -10.728       error: 1%
> rcoulomb=1.30                        => force = -10.394       error: 4%
> rcoulomb=1.00                        => force =  -9.105       error:  
> 16%
>
> Vincent
>
> _______________________________________________
> gmx-users mailing list
> gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please don't post (un)subscribe requests to the list. Use the www  
> interface or send it to gmx-users-request at gromacs.org.
>
------------------------------------------------------------------------ 
-----
Erik Lindahl, MSc, PhD     <lindahl at stanford.edu>
D109, Fairchild Building
Dept. Structural Biology, Stanford University School of Medicine
Tel. 650-7250754    Fax. 650-7238464