mirror of https://gitlab.com/QEF/q-e.git
109 lines
4.3 KiB
Plaintext
109 lines
4.3 KiB
Plaintext
!
|
|
! Spectra correction tool by Oana Bunau and Matteo Calandra, Agust 2015
|
|
!
|
|
! This small code allows to
|
|
!
|
|
! 1) Remove states below a certain energy from the spectrum
|
|
! In large gap insulator it should work as the usual
|
|
! cut_occupied_states option but it is much much faster
|
|
! In metals/semimetals the normal and more time consuming
|
|
! option cut_occupied_states=.true. used in XSpectra
|
|
! should work better. The procedure use here can sometimes
|
|
! work for metals but care is needed.
|
|
!
|
|
! 2) For L23 edges, this code generates the full L23 spectrum from
|
|
! the L2 edge only. the L3 edge is obtained multiplying by two the
|
|
! L2 and shifting it by the spin-orbit splitting between 2p1/2
|
|
! and 2p3/2. Thus the spectrum in input must be the L2 edge.
|
|
!
|
|
! 3) Convolute a given spectrum known on a regular energy grid with
|
|
! a constant lorentz brodening gamma_hole, or with an atan
|
|
! broadening. In more details (s_in, s_out are cross sections):
|
|
!
|
|
! s_out(W)=integral s_in(om)*Gamma(W)/(Gamma(W)**2+(om-W)**2)/pi
|
|
!
|
|
! and the integral is over all the space (-infty, infty).
|
|
!
|
|
! If conv_type='lorentz' then Gamma(W)=gamma_hole constant
|
|
!
|
|
! If conv_type='lorentz_atan' then
|
|
! Gamma(W)=gamma_hole for W < xe0
|
|
! =Gamma_MAX*(0.5+atan(e-1/(e*e))) otherwise
|
|
! where e=(W-xe0)/(Ectr-xe0)
|
|
! and Ectr is the inflection point of the atan and xe0 is
|
|
! the pre-edge onset.
|
|
! For the definition of e see Eq. 7 and below in
|
|
! O. Bunau and M. Calandra, PRB 87, 205105 (2013)
|
|
!
|
|
! Finally emin_conv, emax_conv and nenergy_conv define
|
|
! the energy grid of the convoluted spectrum (energies labeled W above).
|
|
! If omitted, the energy grid is the same as in the input cross section.
|
|
!
|
|
! Important, please note that the convolution adds on TOP of the
|
|
! width of your original spectrum. So you should run xspectra with a
|
|
! tiny xgamma.
|
|
!------------------------------------------------------------------------------------
|
|
! How to use:
|
|
!
|
|
! 1) Use in single processor, no parallelization
|
|
!
|
|
! 2) Input file description:
|
|
!
|
|
! case (i): removal of occupied states from the spectrum
|
|
!
|
|
! &input_manip
|
|
! cross_section_file='xanes.dat',
|
|
! option='cut_occ_states',
|
|
! xe0=13.0,
|
|
! shift_spectrum=.false.,
|
|
! &end
|
|
!
|
|
! if shift_spectrum=.true. the spectrum is shifted in xe0
|
|
!
|
|
! case (ii): removal of occupied states from the spectrum
|
|
!
|
|
! &input_manip
|
|
! cross_section_file='xanes.dat',
|
|
! option='add_L2_L3',
|
|
! element='Cu',
|
|
! &end
|
|
!
|
|
! element is used to read the spin orbit splitting 2p1/2, 2p3/2
|
|
!
|
|
! The cross_section_file is the xanes.dat file with the cross section.
|
|
!
|
|
! case (iii) - lorentz
|
|
!
|
|
! &input_manip
|
|
! cross_section_file='xanes.dat.L23',
|
|
! option='convolution',
|
|
! shift_spectrum=.false.,
|
|
! element='Cu',
|
|
! conv_type='lorentz',
|
|
! gamma_hole=1.0,
|
|
! emin_conv=-50.0,
|
|
! emax_conv=100.0,
|
|
! nenergy_conv=1000,
|
|
! &end
|
|
!
|
|
! emin_conv, emax_conv and nenergy_conv can be omitted
|
|
!
|
|
!
|
|
! case (iii) - lorentz_atan
|
|
!
|
|
! &input_manip
|
|
! cross_section_file='xanes.dat.L23',
|
|
! option='convolution',
|
|
! shift_spectrum=.false.,
|
|
! element='Cu',
|
|
! conv_type='lorentz_atan',
|
|
! gamma_hole=0.05,
|
|
! gamma_max=6.0,
|
|
! ectr=-8.0,
|
|
! xe0=-19.0,
|
|
! &end
|
|
!
|
|
!
|
|
!
|
|
!----------------------------------------------------------------------------
|