[Ifeffit] P 1 space group

Matt Newville newville at cars.uchicago.edu
Sun Oct 2 16:04:36 CDT 2005

Hi Anatoly,

I agree that
   a) it's unfortunate that Artemis is too clever in noticing that a
P1 space group has some symmetry and prevents the Feff calculation
when this happens.   I'd suggest a space group called 'Cluster' or
'Molecule' or something that would indicate that there should be no
space filling procedures done, and that all atoms in the
molecule/cluster are specified, but still let you enter atomic
coordinates as relative to a 'lattice constant'.   It is possible to
just enter the xyz coords in the feff.inp, but Atoms/Artemis does such
a nice job at filling out the rest of feff.inp that using a P1-like
space group is a very convenient choice.

   b) using a = 99.9, b=100.0, c=100.1 is a good way to fool Artemis/Atoms.

> For molecular substances, the accuracy in atomic coordinates that are obtained by one
> of the three most popular methods (e.g., in Protein Data Bank, the sources are X-ray
> crystallography, solution NMR, or theoretical modeling, here is a good review:
> http://www.rcsb.org/pdb/experimental_methods.html) is much worse than 0.1 A.

That seems a little unfair.  Certainly most macromolecular (>5kD)
diffraction data as is typically found in the PDB often has a
resolution worse than 1A.  But powder diffraction measurements and
theoretical modeling of crystallized small molecules (and the
definition of 'small' seems to be growing) and clusters often have
precisions better than 0.01A.

If 0.1A really were the accuracy needed, then you could use
   a = 90, b=100, c=110
in your example.  That would give a very noticeable difference in any
decent powder diffraction data and in the EXAFS.  OTOH, using '99.9,
100.0, 100.1'  would be more challenging to see with powder
diffraction and very difficult to see with EXAFS.  In both cases, the
distances could easily be refined from that "off by 0.1%" starting
point.  For EXAFS at least, the same would not be true if the starting
distances were off by 10%.

Anyway, paraphrasing Juraj's suggestion, using '99.99, 100.00, 100.01'
will certainly be OK and is not any harder to type than '90, 100,


More information about the Ifeffit mailing list