On Feb 11, 2007, at 17:30 , Marcel Mohr wrote: > I am calculating wurtzite CdSe with the PBE Vanderbilt pseudos > provided by the pwscf page. Lattice constant is very good. I am > using a cutoff of 50 Ry, (that is even far beyond convergence)
you may need a very high cutoff on the charge density in order to get good phonons with ultrasoft pseudopotentials. Unfortunately this more than offsets the speed advantage of a lower cutoff on the wavefunctions > tr2_ph=1.0d-14 use a smaller threshold: 10^-18 or so > Maybe the electric field has to be included in the calculation the LO-TO splitting IS calculated: it is not added to the output frequencies, but it can be added later using an auxiliary code Paolo --- Paolo Giannozzi, Democritos and University of Udine, Italy
