yumin qian wrote: > I need to write a small program to do some post-process > calculation , and I need to read in the wave function and > charge density
modify one of the many codes in PP/ that read from the data file > But I don't know the data structure of the wave > function evc(npwx,nbnd) or evc(npwx*npol,nbnd) in noncolinear case evc(i,ibnd): ibnd is the band index, i is the index of plane waves for the noncolinear case: i=1,npw are plane waves with up spin (polarization 1) i=npwx+1,npwx+npw are PW with down spin (polarization 2) npw depends on the k-point, npwx is the maximum npw > In LSDA case what about the spin part of wavefunction in LSDA you have only spin-up or spin-down components. The number of k-points is doubled and the first set contains spin-up states, the second set spin-down states. The different approach between LSDA and noncolinear approaches is somewhat confusing. LSDA was implemented this way because it was easy (causing minimal disruption to the codes, originally written for unpolarized systems) and efficient Paolo -- Paolo Giannozzi, Democritos and University of Udine, Italy
