Dear Andrei, Thank you for your kind reply, and I shall have some tests according your suggestion.
Best regards, Zhongli On 04/06/2009, apost...@uni-osnabrueck.de <apost...@uni-osnabrueck.de> wrote: >> Dear Andrei, >> You said in your SIESTA-tuto that the frozen phonon calculations usually >> proceed after CG relaxation. > > Dear Zhongli, > this is not obligatory, techniclly speaking, but any book on phonons > tells you how they are introduced from Taylor expansion arond > the equilibrium. So, this equilibrium first has to be found. > If your system does not have internal coordiates you don't have > to worry (say, for pure NaCl you can calculate phonons right away > for any lattice constant without a need to relax), but if the internal > cordinates are present (as say in wurtzite or rutile) - yes, > the relaxation in principle should be done first. > Oterwise, you have a chance to get imaginary modes > with eigenector pointing towards the relaxed > geometry, where the system would be able to lower its energy. > >> Is that means we should >> first set "MD.TypeOfRun CG" to relax > > yes > >> and then change it to "MD.TypeOfRun FC" >> with "UseSaveData T" > > yes > >> and "MD. UseSaveXV T" to calculate the force constants? > > No, in fact I suggested to copy the relaxed coordinates > into the .fdf file and then proceed phonons with > MD. UseSaveXV F > But it is just a suggestion. > > Best regards > > Andrei > -- PhD. student of IAMP, Sichuan University, China E-mail: zl.liu.c...@gmail.com
