Re: [potfit-users] potfit-users@lists.sourceforge.net

[Daniel S. <sch...@gm...>]

Hi Sacha,

after quickly looking at your issue I woul assume there actually is 
a short distance. This means that potfit tries to calculate the potential
for a distance where no potential is defined.

Especially the error indicates this, because the code for this error is pretty
simple:

  /* check for distances shorter than minimal distance in table */
  rr = r - pt->begin[col];
  if (rr < 0)
    error(1, "%f %f %d\nShort distance", r, pt->begin[col], col);

In your case, potfit tries to calculate the value of the transfer function
at r = 0 while the potential table starts at r = 1.922195.

Did you modify the code which produces this error?
If not, would it be possible to provide me the input files
which produce this error (in a private mail, not to the list)?

Best regards,
Daniel

Am Freitag, 8. Mai 2015, 22:55:47 schrieb Koch, Sascha:
> Dear Potfit team,
> 
> at first thanks for sharing your great code, after having got started a bit
> it looks very promosing.
> 
> I am currently trying to improve an existing Fe-O EAM potential by Xiao Wang
> Zhou and Haydn N G Wadley (J Phys.: Condens. Matter 17 (2005) p. 3619-3635)
> with regard to interstitialk formation energies.
> I have constructed an 16x16x16 bcc Fe matrix and out there with 2-3 cells
> spacing 9x terahedral O interstitials (I am doing this because the Fe_O
> potential is quite good at the basic stuff, Ec, lattice constant bulk
> moduli, Ev etc, but not for the interstitial which are improtant when I try
> to merge an Fe-Y-O).
> I have got all the 52 parametres in the 7 functions from the article and I
> allowed a range of app. 20% and this works quite okay. When I start an
> optimasationw with the simulated annealing method I get amn F of arround
> 4000 and both valiues get similar.
> 
> However whenever I want to perform a powell minimazation afterwards (since F
> is below 10000 should be okay) I get this error:
> 
> "Starting powell minimization ...
> loops		error_sum	force calculations
>     0	      3870.592615	2220242
> [ERROR] 0.000000 1.922195 3
> Short distance"
> 
> I am startled because I use an analytical eam potential and cannot see how
> this can come from an analytical (thos seems like your webpage states
> typical for tabulated ones...).
> 
> I could only imagine that the emebediing functions have a too low cutoff but
> it is not that. The second thing too close spacing of the atoms is
> unlikely, because I relaxed the configuration in lammps already with the
> potential......
> 
> I am curios about any hints and many
> Thanks for the help in advance
> 
> Sascha Koch
> 
> (KIT IAM-AWP)