[Wien] precision problem

[Jorissen Kevin]

Equivalency is determined by organizing the chemical environment of each atom into shells (where one shell contains all neighbours of the same kind (Z) at the same distance to the particular atom in consideration).  Then these shell structures are compared; all atoms within the unit cells that have the same set of shells are considered by nn to be equivalent (this equivalency may be reduced later on by sgroup for reasons of symmetry).
 
The accuracy for comparison of interatomic distances in nn is, I think, 10^-4, while the output is only written with three decimals.  I remember having a difficult case once where I tuned these two things (the accuracy is in the parameter dlimit at the beginning of dstart, the output format is at the line with label 553 (change f6.3 to, eg., f 7.4 to see four decimals)) to see what was going on.  
This may help you to see what nn is actually doing.  With default settings, it's possible that atom distances (and thus also shell structures) differ by, say, 2 10^-4, and atoms are made inequivalent, but you don't see why in the summary at the bottom of case.outputnn.
 
Another point to consider is that in this overview at the end of case.outputnn, by default only a few shells are written (I think 4).  But nn considers often many more shells to determine equivalencies !
To see all the shells considered by nn, I think I once changed : 
        write(66,553)index,((shdist(index,i),icnt(index,i,j), &
                     shellz(index,i,j),j=1,iz(index,i)),i=1,4)
to 
        write(66,553)index,((shdist(index,i),icnt(index,i,j), &
                     shellz(index,i,j),j=1,iz(index,i)),i=1,ishell(index))
and then recompiled.
 
I admit that all this was usefulonly in the extreme case of a C60 molecule, where there are many interatomic distances that are rather big and almost (but not exactly) equal.
And most of the information is also in the part before the overview, where information is printed for every atom position in the unit cell.
 
Of course, you must also make sure that the accuracy of your atomic coordinates is better than dlimit.
 
 
Kevin.
 
 

 
 

	-----Oorspronkelijk bericht----- 
	Van: stargmoon [mailto:stargmoon at yahoo.com] 
	Verzonden: di 11/4/2003 10:24 
	Aan: wien at zeus.theochem.tuwien.ac.at 
	CC: 
	Onderwerp: [Wien] precision problem
	
	
	Dear wien2k user,
	 
	I met a problem in doing 'x nn'. My input structure is the structure with space group P-3m1, and one type of atoms occupy the "6i" positions.I think I can be sure that my input structure is correct, because the space group analysis with wien2k and other software all indicate the right space group. However, after the 'x nn' finished, I always got the error information, that is, the Mult not equal and ityp not equal. I checked the number and the distance of the nearest neighbors printed out in the file case.outputnn, but I can't find out any difference between those atoms which was declared to be not equal by 'x nn'. Have you ever met the same problem, and can I ignore those error (because the space group is the same as expected)?
	 
	Thanks for your reply!
	 
	Best,
	 
	stargmoon

	
  _____  

	Do you Yahoo!?
	Protect your identity with Yahoo! Mail AddressGuard <http://antispam.yahoo.com/whatsnewfree> 

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/ms-tnef
Size: 6810 bytes
Desc: not available
Url : http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20031105/77a169d0/attachment.bin