Re: Ewald error estimation

From: darden <>
Date: Mon 21 May 2001 15:58:17 -0500 (CDT)

The basis of this term discussed in Essmann et al (J Chem Phys 103,8577)
is the following 3 observations:
1) The rms force error is estimated by the error in the virial trace
2) The energy + trace(virial) should add to zero for exact Ewald
3) The energy is 10X more accurate than forces or virial
Thus |energy + trace(virial)|/|energy| estimates the error in the virial
trace and thus rms force error

In practice this approximately works much of the time but is not really
precise. There are precise estimates for the case of force interpolated
pme (frc_int=1 in amber6 and beyond) see e.g. Darden et al. J Chim Phys.
94, 1346 or Deserno and Helm J Chem Phys 109: 7678, but no one has got one
for the smooth pme (frc_int=0)

Hope this helps
Tom Darden
 On Mon, 21 May 2001, Okimasa Okada wrote:

> Dear AMBER,
> Would you please let me know the meanings of "Ewald error estimation",
> which is printed out when I use PME method. I would be appreciated,
> if you give me the formula or articles related to this.
> Thank you for your help.
> Sincerely yours,
> Okimasa Okada
> Fuji Xerox Co., Ltd.,
> IT Device Business Development,
> New Business Center,
> 430 Sakai, Nakai-machi, Ashigarakami-gun,
> Kanagawa 259-0157, Japan
Received on Mon May 21 2001 - 13:58:17 PDT
Custom Search