[Wannier] wannier90 & vasp: Incorrect Cartesian Structures

[Conn ORourke]

Hi there, 

I was wondering if anyone could explain a problem I am seeing with the wannier90 library compiled in VASP.
The problem is that the Cartesian coords in the seedname.win file generated by the library are different to that of the original POSCAR. Likewise the bond lengths from the wannier90_centres.xyz file are different to that of the POSCAR file.

For example:
For a POSCAR file with the following lattice:
15.932414    0.0000000   0.00000000
 0.0000000   16.005113   0.00000000
 0.0000000   0.0000000   15.9324110
with the two first atoms:
Direct
  0.1665762551037838  0.2052544038323533  0.1664989385925252
  0.1796712991481728  0.1930070499806985  0.6768832070032641

and equivalent Cartesian coords:
 2.653962        3.285120        2.652730
 2.862597        3.089100       10.784382
the wannier90.win file produced has:
begin unit_cell_cart
    15.9324140     0.0000000     0.0000000
     0.0000000    16.0051130     0.0000000
     0.0000000     0.0000000    15.9324110
end unit_cell_cart

begin atoms_cart
Mg       2.6899086     3.2554542     2.6535155
Mg       2.8794998     3.0745452    10.7780014

If I generate the wannier90.win file by hand, and explicitly enter the correct Cartesian positions of the atoms into the atoms_cart block, those in the wannier90.wout file are again different. I also noticed that in the output file it says:
 Ignoring <atoms_cart> in input file
Is there some reason why the atoms_cart block doesn't get read, or a flag to set the code to read it.

Are these problems as a result of the compilation, or something else?
Many thanks!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/wannier/attachments/20160518/76c21fd9/attachment.html>