[Pw_forum] Symmetries and Pseudopotentials

[Daniel Cellucci]

TO The Quantum Espresso Experts,

I was performing a convergence test for the cutoff energy of the SnO2 unit cell (Rutile) when I stumbled upon something odd during the 'vc-relax; runs. At some wfc cutoff energies QE does not find the proper symmetries, giving this error:
warning: symmetry operation #  N not allowed.   fractional translation:
       0.5000000 -0.5000000  0.5000000  in crystal coordinates
The program then proceeds to optimize the cell, but does not recognize its orthogonality and I end up with a parameter matrix that contains very small- but very present- off-diagonal elements.

Even stranger, the vc-relax calculation recognizes the symmetries and runs perfectly for a wfc cutoff of 30,35,40, and 55 Ry, but at 25, 45, and 50 Ry the symmetries aren't found. Also, my rho cutoff is 10*wfc for each of the calculations. (i.e. 30->300, 35->350, etc.)

I tried simplifying my coordinates, thinking that my precision was overkill, but that didn't work, and I am thinking it has to do with my pseudopotentials. I am using the O.pw91-van_ak.UPF included with QE for Oxygen, and Sn.pw91-n-van.UPF for Tin, which is downloadable from the QE site at: http://www.quantum-espresso.org/pseudo/1.3/html/Sn.html

Right now I am continuing my studies with a wfc cutoff of 30Ry and a rho cutoff of 300Ry, since at this energy there was no issue with symmetries or bfgs convergence and the final parameters weren't that much different from the 55Ry calculation. However I am curious to know why these energies are causing problems with the symmetry finder.

Thank you for your time, 

Daniel Cellucci
University of Georgia 
Physics Department
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20110131/d8ea4daa/attachment.html>