mirror of
https://github.com/triqs/dft_tools
synced 2024-09-02 01:03:42 +02:00
1406 lines
59 KiB
Fortran
1406 lines
59 KiB
Fortran
|
|
c ******************************************************************************
|
|
c
|
|
c TRIQS: a Toolbox for Research in Interacting Quantum Systems
|
|
c
|
|
c Copyright (C) 2011 by L. Pourovskii, V. Vildosola, C. Martins, M. Aichhorn
|
|
c
|
|
c TRIQS is free software: you can redistribute it and/or modify it under the
|
|
c terms of the GNU General Public License as published by the Free Software
|
|
c Foundation, either version 3 of the License, or (at your option) any later
|
|
c version.
|
|
c
|
|
c TRIQS is distributed in the hope that it will be useful, but WITHOUT ANY
|
|
c WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
|
|
c FOR A PARTICULAR PURPOSE. See the GNU General Public License for more
|
|
c details.
|
|
c
|
|
c You should have received a copy of the GNU General Public License along with
|
|
c TRIQS. If not, see <http://www.gnu.org/licenses/>.
|
|
c
|
|
c *****************************************************************************/
|
|
|
|
SUBROUTINE outqmc(elecn,qbbot)
|
|
C %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|
|
C %% %%
|
|
C %% This subroutine creates the output files : %%
|
|
C %% - case.ctqmcout, with all the informations necessary for a %%
|
|
C %% CTQMC computation. %%
|
|
C %% - case.symqmc, describing all the symmetries of the system %%
|
|
c %% (necessary for CTQMC computation too). %%
|
|
C %% - case.parproj which gives the partial charge projectors for %%
|
|
C %% orbitals and for all atoms. %%
|
|
C %% - case.sympar which contains the symmetry matrices for all %%
|
|
C %% the included orbitals (for partial charge analysis) %%
|
|
C %% %%
|
|
C %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|
|
|
|
C Definition of the variables :
|
|
C -----------------------------
|
|
USE almblm_data
|
|
USE common_data
|
|
USE file_names
|
|
USE prnt
|
|
USE reps
|
|
USE symm
|
|
USE projections
|
|
IMPLICIT NONE
|
|
C
|
|
INTEGER :: iorb, icrorb, irep, isrt
|
|
INTEGER :: l, m, is, i1, i2, isym
|
|
INTEGER :: ik, il, ib, ir, n
|
|
INTEGER :: ind1, ind2, iatom
|
|
INTEGER :: timeinvflag
|
|
REAL(KIND=8) :: qbbot, elecn, factor
|
|
COMPLEX(KIND=8) :: ephase
|
|
COMPLEX(KIND=8),DIMENSION(:,:), ALLOCATABLE :: hk
|
|
COMPLEX(KIND=8), DIMENSION(:,:), ALLOCATABLE :: spinrot
|
|
COMPLEX(KIND=8), DIMENSION(:,:), ALLOCATABLE :: densprint
|
|
COMPLEX(KIND=8), DIMENSION(:,:), ALLOCATABLE :: time_op
|
|
C
|
|
C
|
|
C =======================================
|
|
C Writing the file case.ctqmcout :
|
|
C =======================================
|
|
C
|
|
WRITE(buf,'(a)')'Writing the file case.ctqmcout...'
|
|
CALL printout(0)
|
|
C
|
|
C Definition of 1 electron-Volt
|
|
WRITE(ouctqmc,'(a)') '13.605698'
|
|
C
|
|
C ---------------------------------------
|
|
C General informations about the system :
|
|
C ---------------------------------------
|
|
C
|
|
C Number of k-points in the I-BZ
|
|
WRITE(ouctqmc,'(i6)') nk
|
|
C Definition of the spin-polarized flag ifSP
|
|
IF (ifSP) THEN
|
|
WRITE(ouctqmc,'(i6)') 1
|
|
ELSE
|
|
WRITE(ouctqmc,'(i6)') 0
|
|
ENDIF
|
|
C Definition of the Spin-orbit flag ifSO
|
|
IF (ifSO) THEN
|
|
WRITE(ouctqmc,'(i6)') 1
|
|
ELSE
|
|
WRITE(ouctqmc,'(i6)') 0
|
|
ENDIF
|
|
C The only possible combinations are :
|
|
C - (0,0) which stands for a paramagnetic computation without SO.
|
|
C - (1,0) which stands for computation without SO using spin-polarized input files.
|
|
C - (1,1) which stands for computation with SO using spin-polarized input files.
|
|
C
|
|
C Writing the total charge below the lower limit of the energy window (variable "qbbot")
|
|
WRITE(ouctqmc,*) qbbot
|
|
C Writing the total number of electrons (valence band+semicore) (variable "elecn").
|
|
C It is also the charge upto the Fermi level.
|
|
WRITE(ouctqmc,*) elecn
|
|
C
|
|
C ------------------------------------------
|
|
C Description of all the included orbitals :
|
|
C ------------------------------------------
|
|
C
|
|
C Definition of the number of included orbitals "norb"
|
|
WRITE(ouctqmc,'(i6)') norb
|
|
C Description of each orbital "iorb"
|
|
DO iorb=1,norb
|
|
IF (ifSO) THEN
|
|
WRITE(ouctqmc,'(4(i6,x))') orb(iorb)%atom, orb(iorb)%sort,
|
|
& orb(iorb)%l, 2*(2*orb(iorb)%l+1)
|
|
ELSE
|
|
WRITE(ouctqmc,'(4(i6,x))') orb(iorb)%atom, orb(iorb)%sort,
|
|
& orb(iorb)%l, 2*orb(iorb)%l+1
|
|
ENDIF
|
|
ENDDO
|
|
C an orbital "iorb" is described by :
|
|
C - the associated atom
|
|
C - the corresponding atomic sort
|
|
C - the considered orbital number l
|
|
C - the size of the corresponding matrices : 2*l+1 without SO ; 2*(2*l+1) with SO
|
|
C
|
|
C ----------------------------------------
|
|
C Description of the correlated orbitals :
|
|
C ----------------------------------------
|
|
C
|
|
C Definition of the number of correlated orbitals "ncrorb"
|
|
WRITE(ouctqmc,'(i6)') ncrorb
|
|
C Description of each correlated orbital "icrorb"
|
|
DO icrorb=1,ncrorb
|
|
l=crorb(icrorb)%l
|
|
isrt=crorb(icrorb)%sort
|
|
C
|
|
C The case l=0 is a particular case of "non-mixing" basis.
|
|
C --------------------------------------------------------
|
|
IF (l==0) THEN
|
|
C For the s-orbitals, the only irep possible is the matrix itself.
|
|
IF (ifSP.AND.ifSO) THEN
|
|
C If SO is taken into account, spinor rotation matrix is considered.
|
|
C The spin is not a good quantum number, so the whole representation is used.
|
|
WRITE(ouctqmc,'(6(i6,x))') crorb(icrorb)%atom, isrt, 0,
|
|
& 2, crorb(icrorb)%ifSOat, 1
|
|
ELSE
|
|
C Without SO, only the rotation matrix in orbital space is necessary.
|
|
WRITE(ouctqmc,'(6(i6,x))') crorb(icrorb)%atom, isrt, 0,
|
|
& 1, crorb(icrorb)%ifSOat, 1
|
|
ENDIF
|
|
C
|
|
C If the basis representation needs a complete spinor rotation approach (basis with "mixing" ).
|
|
C ---------------------------------------------------------------------------------------------
|
|
ELSEIF(reptrans(l,isrt)%ifmixing) THEN
|
|
C In this case, the SO is necessary considered, spinor rotation matrices are used.
|
|
IF (crorb(icrorb)%ifsplit) THEN
|
|
C If only an irep is correlated
|
|
DO irep=1,reptrans(l,isrt)%nreps
|
|
IF (crorb(icrorb)%correp(irep)) THEN
|
|
WRITE(ouctqmc,'(6(i6,x))') crorb(icrorb)%atom,
|
|
& isrt, l, reptrans(l,isrt)%dreps(irep),
|
|
& crorb(icrorb)%ifSOat, irep
|
|
ENDIF
|
|
ENDDO
|
|
ELSE
|
|
C If no particular irep is correlated
|
|
WRITE(ouctqmc,'(6(i6,x))') crorb(icrorb)%atom, isrt, l,
|
|
& 2*(2*l+1), crorb(icrorb)%ifSOat, 1
|
|
ENDIF
|
|
C
|
|
C If the basis representation can be reduce to the up/up block (basis without "mixing").
|
|
C --------------------------------------------------------------------------------------
|
|
ELSE
|
|
IF (ifSP.AND.ifSO) THEN
|
|
C If SO is taken into account, spinor rotation matrices are considered.
|
|
C The spin is not a good quantum number, so the whole representation is used.
|
|
WRITE(ouctqmc,'(6(i6,x))') crorb(icrorb)%atom, isrt, l,
|
|
& 2*(2*l+1), crorb(icrorb)%ifSOat, 1
|
|
ELSE
|
|
C Without SO, only the rotation matrix in orbital space is necessary.
|
|
IF (crorb(icrorb)%ifsplit) THEN
|
|
C If only an irep is correlated
|
|
DO irep=1,reptrans(l,isrt)%nreps
|
|
IF (crorb(icrorb)%correp(irep)) THEN
|
|
WRITE(ouctqmc,'(6(i6,x))') crorb(icrorb)%atom,
|
|
& isrt, l, reptrans(l,isrt)%dreps(irep),
|
|
& crorb(icrorb)%ifSOat, irep
|
|
ENDIF
|
|
ENDDO
|
|
ELSE
|
|
C If no particular irep is correlated
|
|
WRITE(ouctqmc,'(6(i6,x))') crorb(icrorb)%atom, isrt, l,
|
|
& (2*l+1), crorb(icrorb)%ifSOat, 1
|
|
END IF ! End of the ifsplit if-then-else
|
|
END IF ! End of the ifSO if-then-else
|
|
END IF ! End of the ifmixing if-then-else
|
|
END DO ! End of the icrorb loop
|
|
C an orbital "iorb" is described by :
|
|
C - the associated atom
|
|
C - the corresponding atomic sort
|
|
C - the considered orbital number l
|
|
C - the size of the "correlated" submatrix (can be the whole matrix)
|
|
C - the flag ifSOat which states that SO is considered for this orbital
|
|
C - the number of the irep
|
|
C
|
|
C ------------------------------------------------------------------------------------------------
|
|
C Description of the global to local coordinates transformation Rloc for each correlated orbital :
|
|
C ------------------------------------------------------------------------------------------------
|
|
C Description of each transformation Rloc
|
|
DO icrorb=1,ncrorb
|
|
l=crorb(icrorb)%l
|
|
isrt=crorb(icrorb)%sort
|
|
iatom=crorb(icrorb)%atom
|
|
C
|
|
C The case l=0 is a particular case of "non-mixing" basis.
|
|
C --------------------------------------------------------
|
|
IF(l==0) THEN
|
|
C For the s-orbitals, the only irep possible is the matrix itself.
|
|
IF (ifSP.AND.ifSO) THEN
|
|
C If SO is taken into account, spinor rotation matrix must be considered.
|
|
ALLOCATE(spinrot(1:2,1:2))
|
|
spinrot(:,:)=0.d0
|
|
C The spinor-rotation matrix is directly calculated from the Euler angles a,b and c.
|
|
IF (rotloc(iatom)%timeinv) THEN
|
|
factor=(rotloc(iatom)%a+rotloc(iatom)%g)/2.d0
|
|
spinrot(2,1)=EXP(CMPLX(0.d0,factor))*
|
|
& DCOS(rotloc(iatom)%b/2.d0)
|
|
spinrot(1,2)=-CONJG(spinrot(2,1))
|
|
C Up/dn and Dn/up terms
|
|
factor=-(rotloc(iatom)%a-rotloc(iatom)%g)/2.d0
|
|
spinrot(2,2)=-EXP(CMPLX(0.d0,factor))*
|
|
& DSIN(rotloc(iatom)%b/2.d0)
|
|
spinrot(1,1)=CONJG(spinrot(2,2))
|
|
ELSE
|
|
factor=(rotloc(iatom)%a+rotloc(iatom)%g)/2.d0
|
|
spinrot(1,1)=EXP(CMPLX(0.d0,factor))*
|
|
& DCOS(rotloc(iatom)%b/2.d0)
|
|
spinrot(2,2)=CONJG(spinrot(1,1))
|
|
C Up/dn and Dn/up terms
|
|
factor=-(rotloc(iatom)%a-rotloc(iatom)%g)/2.d0
|
|
spinrot(1,2)=EXP(CMPLX(0.d0,factor))*
|
|
& DSIN(rotloc(iatom)%b/2.d0)
|
|
spinrot(2,1)=-CONJG(spinrot(1,2))
|
|
ENDIF
|
|
C Printing the transformation informations
|
|
DO m=1,2
|
|
WRITE(ouctqmc,*) REAL(spinrot(m,1:2))
|
|
ENDDO
|
|
DO m=1,2
|
|
WRITE(ouctqmc,*) AIMAG(spinrot(m,1:2))
|
|
ENDDO
|
|
DEALLOCATE(spinrot)
|
|
C Without SO, only the rotation matrix in orbital space is necessary.
|
|
ELSE
|
|
C In this case, the Rloc matrix is merely identity.
|
|
WRITE(ouctqmc,*) 1.d0
|
|
WRITE(ouctqmc,*) 0.d0
|
|
ENDIF
|
|
C Printing the time inversion flag if the calculation is spin-polarized (ifSP=1)
|
|
IF (ifSP) THEN
|
|
timeinvflag=0
|
|
IF (rotloc(iatom)%timeinv) timeinvflag=1
|
|
WRITE(ouctqmc,'(i6)') timeinvflag
|
|
ENDIF
|
|
C
|
|
C If the basis representation needs a complete spinor rotation approach (basis with "mixing" ).
|
|
C ---------------------------------------------------------------------------------------------
|
|
ELSEIF (reptrans(l,isrt)%ifmixing) THEN
|
|
C In this case, the calculation is necessary spin-polarized with SO, spinor rotation matrices are used.
|
|
IF(crorb(icrorb)%ifsplit) THEN
|
|
C If only an irep is correlated
|
|
ind1=1
|
|
DO irep=1,reptrans(l,isrt)%nreps
|
|
IF(crorb(icrorb)%correp(irep)) THEN
|
|
ind2=ind1+reptrans(l,isrt)%dreps(irep)-1
|
|
DO m=ind1,ind2
|
|
WRITE(ouctqmc,*) REAL(rotloc(iatom)%
|
|
& rotrep(l)%mat(m,ind1:ind2))
|
|
ENDDO
|
|
DO m=ind1,ind2
|
|
WRITE(ouctqmc,*)AIMAG(rotloc(iatom)%
|
|
& rotrep(l)%mat(m,ind1:ind2))
|
|
ENDDO
|
|
ENDIF
|
|
ind1=ind1+reptrans(l,isrt)%dreps(irep)
|
|
ENDDO
|
|
ELSE
|
|
C If no particular irep is correlated
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(ouctqmc,*) REAL(rotloc(iatom)%
|
|
& rotrep(l)%mat(m,1:2*(2*l+1)))
|
|
ENDDO
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(ouctqmc,*) AIMAG(rotloc(iatom)%
|
|
& rotrep(l)%mat(m,1:2*(2*l+1)))
|
|
ENDDO
|
|
ENDIF
|
|
C Printing if the transformation included a time-reversal operation.
|
|
timeinvflag =0
|
|
IF (rotloc(iatom)%timeinv) timeinvflag=1
|
|
WRITE(ouctqmc,'(i6)') timeinvflag
|
|
C
|
|
C If the basis representation can be reduce to the up/up block (basis without "mixing").
|
|
C --------------------------------------------------------------------------------------
|
|
ELSE
|
|
IF (ifSP.AND.ifSO) THEN
|
|
C If SO is taken into account, spinor rotation matrices are considered.
|
|
C The spin is not a good quantum number, so the whole representation is used.
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(ouctqmc,*) REAL(rotloc(iatom)%
|
|
& rotrep(l)%mat(m,1:2*(2*l+1)))
|
|
ENDDO
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(ouctqmc,*) AIMAG(rotloc(iatom)%
|
|
& rotrep(l)%mat(m,1:2*(2*l+1)))
|
|
ENDDO
|
|
ELSE
|
|
C The calculation is either spin-polarized without SO or paramagnetic.
|
|
C The spin is a good quantum number and irep are possible.
|
|
IF(crorb(icrorb)%ifsplit) THEN
|
|
C If only an irep is correlated
|
|
ind1=-l
|
|
DO irep=1,reptrans(l,isrt)%nreps
|
|
IF(crorb(icrorb)%correp(irep)) THEN
|
|
ind2=ind1+reptrans(l,isrt)%dreps(irep)-1
|
|
DO m=ind1,ind2
|
|
WRITE(ouctqmc,*) REAL(rotloc(iatom)%
|
|
& rotrep(l)%mat(m,ind1:ind2))
|
|
ENDDO
|
|
DO m=ind1,ind2
|
|
WRITE(ouctqmc,*) AIMAG(rotloc(iatom)%
|
|
& rotrep(l)%mat(m,ind1:ind2))
|
|
ENDDO
|
|
ENDIF
|
|
ind1=ind1+reptrans(l,isrt)%dreps(irep)
|
|
ENDDO
|
|
ELSE
|
|
C If no particular irep is correlated
|
|
DO m=-l,l
|
|
WRITE(ouctqmc,*) REAL(rotloc(iatom)%
|
|
& rotrep(l)%mat(m,-l:l))
|
|
ENDDO
|
|
DO m=-l,l
|
|
WRITE(ouctqmc,*) AIMAG(rotloc(iatom)%
|
|
& rotrep(l)%mat(m,-l:l))
|
|
ENDDO
|
|
END IF ! End of the ifsplit if-then-else
|
|
END IF ! End of the ifSO if-then-else
|
|
C Printing if the transformation included a time-reversal operation.
|
|
IF (ifSP) THEN
|
|
timeinvflag =0
|
|
IF (rotloc(iatom)%timeinv) timeinvflag=1
|
|
WRITE(ouctqmc,'(i6)') timeinvflag
|
|
END IF
|
|
END IF ! End of the ifmixing if-then-else
|
|
END DO ! End of the icrorb loop
|
|
C for each correlated orbital icrorb, the transformation Rloc is described by :
|
|
C - the real part of the submatrix associated to "icrorb"
|
|
C - the imaginary part of the submatrix associated to "icrorb"
|
|
C - a flag which states if a time reversal operation is included in the transformation (if SP only )
|
|
C
|
|
C -------------------------------------------------------------------------------------------------------------
|
|
C Description of the transformation from complex harmonics to the basis associated to each correlated orbital :
|
|
C -------------------------------------------------------------------------------------------------------------
|
|
C
|
|
DO icrorb=1,ncrorb
|
|
l=crorb(icrorb)%l
|
|
isrt=crorb(icrorb)%sort
|
|
C The transformation is printed only for the first (representative) atom of each sort.
|
|
IF (crorb(icrorb)%first) THEN
|
|
C
|
|
C The case l=0 is a particular case of "non-mixing" basis.
|
|
C --------------------------------------------------------
|
|
IF (l==0) THEN
|
|
C For the s-orbitals, the only basis considered is the complex basis.
|
|
IF (ifSP.AND.ifSO) THEN
|
|
C If SO is taken into account, orbital+spin space is considered.
|
|
ALLOCATE(spinrot(1:2,1:2))
|
|
spinrot(:,:)=0.d0
|
|
spinrot(1,1)=1.d0
|
|
spinrot(2,2)=1.d0
|
|
C Printing the number of irep in the considered basis and the size of each irep
|
|
C in the considered basis (in this case, it's 1 and 2.)
|
|
WRITE(ouctqmc,*) 1, 2
|
|
C Printing the transformation matrix
|
|
DO m=1,2
|
|
WRITE(ouctqmc,*) REAL(spinrot(m,1:2))
|
|
ENDDO
|
|
DO m=1,2
|
|
WRITE(ouctqmc,*) AIMAG(spinrot(m,1:2))
|
|
ENDDO
|
|
C
|
|
DEALLOCATE(spinrot)
|
|
ELSE
|
|
C Without SO, only the matrix in orbital space is necessary.
|
|
C Printing the number of irep in the considered basis and the size of each irep
|
|
C in the considered basis (in this case, it's 1 and 1.)
|
|
WRITE(ouctqmc,'(2(i6,x))') 1, 1
|
|
C Printing the transformation matrix
|
|
WRITE(ouctqmc,*) 1.d0
|
|
WRITE(ouctqmc,*) 0.d0
|
|
ENDIF
|
|
C
|
|
C If the basis representation needs a complete spinor rotation approach (basis with "mixing" ).
|
|
C ---------------------------------------------------------------------------------------------
|
|
ELSEIF (reptrans(l,isrt)%ifmixing) THEN
|
|
C In this case, the SO is necessary considered, spinor rotation matrices are used.
|
|
IF (crorb(icrorb)%ifsplit) THEN
|
|
WRITE(ouctqmc,'(15(i6,x))') reptrans(l,isrt)%nreps,
|
|
& reptrans(l,isrt)%dreps(1:reptrans(l,isrt)%nreps)
|
|
ELSE
|
|
WRITE(ouctqmc,'(2(i6,x))') 1, 2*(2*l+1)
|
|
ENDIF
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(ouctqmc,*) REAL(reptrans(l,isrt)%
|
|
& transmat(m,1:2*(2*l+1)) )
|
|
ENDDO
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(ouctqmc,*) AIMAG(reptrans(l,isrt)%
|
|
& transmat(m,1:2*(2*l+1)) )
|
|
ENDDO
|
|
C
|
|
C If the basis representation can be reduce to the up/up block (basis without "mixing").
|
|
C --------------------------------------------------------------------------------------
|
|
ELSE
|
|
IF (ifSP.AND.ifSO) THEN
|
|
C If SO is taken into account, orbital+spin space is considered.
|
|
ALLOCATE(spinrot(1:2*(2*l+1),1:2*(2*l+1)))
|
|
spinrot(:,:)=0.d0
|
|
spinrot(1:2*l+1,1:2*l+1)=
|
|
& reptrans(l,isrt)%transmat(-l:l,-l:l)
|
|
spinrot(2*l+2:2*(2*l+1),2*l+2:2*(2*l+1))=
|
|
& reptrans(l,isrt)%transmat(-l:l,-l:l)
|
|
C Printing the number of irep in the considered basis and the size of each irep
|
|
C in the considered basis (in this case, it's 1 and 2*(2*l+1).)
|
|
WRITE(ouctqmc,*) 1, 2*(2*l+1)
|
|
C Printing the transformation matrix
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(ouctqmc,*) REAL(spinrot(m,1:2*(2*l+1)) )
|
|
ENDDO
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(ouctqmc,*) AIMAG(spinrot(m,1:2*(2*l+1)) )
|
|
ENDDO
|
|
C
|
|
DEALLOCATE(spinrot)
|
|
C Without SO, only the rotation matrix in orbital space is necessary.
|
|
ELSE
|
|
IF (crorb(icrorb)%ifsplit) THEN
|
|
WRITE(ouctqmc,'(8(i6,x))') reptrans(l,isrt)%nreps,
|
|
& reptrans(l,isrt)%dreps(1:reptrans(l,isrt)%nreps)
|
|
ELSE
|
|
WRITE(ouctqmc,'(2(i6,x))') 1, 2*l+1
|
|
ENDIF
|
|
DO m=-l,l
|
|
WRITE(ouctqmc,*) REAL(reptrans(l,isrt)%
|
|
& transmat(m,-l:l) )
|
|
ENDDO
|
|
DO m=-l,l
|
|
WRITE(ouctqmc,*) AIMAG(reptrans(l,isrt)%
|
|
& transmat(m,-l:l) )
|
|
END DO
|
|
END IF ! End of the ifSO if-then-else
|
|
END IF ! End of the ifmixing if-then-else
|
|
END IF ! End of the iffirst if-then-else
|
|
END DO ! End of the icrorb loop
|
|
C for each correlated orbital icrorb, the basis transformation is described by :
|
|
C - the number of irep in the new basis
|
|
C - the dimension of each irep in the new basis
|
|
C - the real part of the basis transformation
|
|
C - the imaginary part of the basis transformation
|
|
C
|
|
C -------------------------------------------------------------------------
|
|
C Description of the number of bands in the energy window at each k_point :
|
|
C -------------------------------------------------------------------------
|
|
C
|
|
DO is=1,ns
|
|
C If SO is considered, the number of up and dn bands are the same.
|
|
IF ((ifSP.AND.ifSO).and.(is.eq.2)) cycle
|
|
C Printing the number of included bands in the window for each k_point
|
|
DO ik=1,nk
|
|
WRITE(ouctqmc,'(i6)')
|
|
& ABS(kp(ik,is)%nb_top-kp(ik,is)%nb_bot+1)
|
|
ENDDO
|
|
ENDDO
|
|
C
|
|
C -----------------------------------------------------------
|
|
C Description of the projectors for the correlated orbitals :
|
|
C -----------------------------------------------------------
|
|
DO ik=1,nk
|
|
DO icrorb=1,ncrorb
|
|
l=crorb(icrorb)%l
|
|
isrt=crorb(icrorb)%sort
|
|
C
|
|
C The case l=0 is a particular case of "non-mixing" basis.
|
|
C --------------------------------------------------------
|
|
IF (l==0) THEN
|
|
C For the s-orbitals, the only irep possible is the matrix itself.
|
|
C if the calculation is spin-polarized, up and dn projectors are written the one above the other (orbital+spin-space of size 2)
|
|
DO is=1,ns
|
|
WRITE(ouctqmc,*)
|
|
& REAL(pr_crorb(icrorb,ik,is)%mat_rep(1,
|
|
& kp(ik,is)%nb_bot:kp(ik,is)%nb_top))
|
|
ENDDO
|
|
DO is=1,ns
|
|
WRITE(ouctqmc,*)
|
|
& AIMAG(pr_crorb(icrorb,ik,is)%mat_rep(1,
|
|
& kp(ik,is)%nb_bot:kp(ik,is)%nb_top))
|
|
ENDDO
|
|
C
|
|
C If the basis representation needs a complete spinor rotation approach (basis with "mixing" ).
|
|
C ---------------------------------------------------------------------------------------------
|
|
ELSEIF (reptrans(l,isrt)%ifmixing) THEN
|
|
C In this case, the calculation is necessary spin-polarized with SO, spinor rotation matrices are used.
|
|
IF(crorb(icrorb)%ifsplit) THEN
|
|
C If only 1 irep is correlated
|
|
ind1=1
|
|
DO irep=1,reptrans(l,isrt)%nreps
|
|
IF(crorb(icrorb)%correp(irep)) THEN
|
|
ind2=ind1+reptrans(l,isrt)%dreps(irep)-1
|
|
DO m=ind1,ind2
|
|
WRITE(ouctqmc,*)
|
|
& REAL(pr_crorb(icrorb,ik,1)%mat_rep(m,
|
|
& kp(ik,1)%nb_bot:kp(ik,1)%nb_top))
|
|
ENDDO
|
|
DO m=ind1,ind2
|
|
WRITE(ouctqmc,*)
|
|
& AIMAG(pr_crorb(icrorb,ik,1)%mat_rep(m,
|
|
& kp(ik,1)%nb_bot:kp(ik,1)%nb_top))
|
|
ENDDO
|
|
ENDIF
|
|
ind1=ind1+reptrans(l,isrt)%dreps(irep)
|
|
ENDDO
|
|
ELSE
|
|
C If no particular irep is correlated
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(ouctqmc,*)
|
|
& REAL(pr_crorb(icrorb,ik,1)%mat_rep(m,
|
|
& kp(ik,1)%nb_bot:kp(ik,1)%nb_top))
|
|
ENDDO
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(ouctqmc,*)
|
|
& AIMAG(pr_crorb(icrorb,ik,1)%mat_rep(m,
|
|
& kp(ik,1)%nb_bot:kp(ik,1)%nb_top))
|
|
ENDDO
|
|
ENDIF
|
|
C
|
|
C If the basis representation can be reduce to the up/up block (basis without "mixing").
|
|
C --------------------------------------------------------------------------------------
|
|
ELSE
|
|
IF ((.not.(ifSP.AND.ifSO)).AND.crorb(icrorb)%ifsplit) THEN
|
|
C If only 1 irep is correlated (case without SO)
|
|
ind1=-l
|
|
DO irep=1,reptrans(l,isrt)%nreps
|
|
IF(crorb(icrorb)%correp(irep)) THEN
|
|
ind2=ind1+reptrans(l,isrt)%dreps(irep)-1
|
|
DO is=1,ns
|
|
DO m=ind1,ind2
|
|
WRITE(ouctqmc,*)
|
|
& REAL(pr_crorb(icrorb,ik,is)%mat_rep(m,
|
|
& kp(ik,is)%nb_bot:kp(ik,is)%nb_top))
|
|
ENDDO
|
|
ENDDO
|
|
DO is=1,ns
|
|
DO m=ind1,ind2
|
|
WRITE(ouctqmc,*)
|
|
& AIMAG(pr_crorb(icrorb,ik,is)%mat_rep(m,
|
|
& kp(ik,is)%nb_bot:kp(ik,is)%nb_top))
|
|
ENDDO
|
|
ENDDO
|
|
ENDIF
|
|
ind1=ind1+reptrans(l,isrt)%dreps(irep)
|
|
ENDDO
|
|
ELSE
|
|
C If no particular irep is correlated (case with and without SO)
|
|
DO is=1,ns
|
|
DO m=-l,l
|
|
WRITE(ouctqmc,*)
|
|
& REAL(pr_crorb(icrorb,ik,is)%mat_rep(m,
|
|
& kp(ik,is)%nb_bot:kp(ik,is)%nb_top))
|
|
ENDDO
|
|
ENDDO
|
|
DO is=1,ns
|
|
DO m=-l,l
|
|
WRITE(ouctqmc,*)
|
|
& AIMAG(pr_crorb(icrorb,ik,is)%mat_rep(m,
|
|
& kp(ik,is)%nb_bot:kp(ik,is)%nb_top))
|
|
ENDDO
|
|
ENDDO
|
|
END IF ! End of the ifsplit if-then-else
|
|
END IF ! End of the ifmixing if-then-else
|
|
END DO ! End of the icrorb loop
|
|
END DO ! End of the ik loop
|
|
C for each k-point and each correlated orbital, the corresponding projector is described by :
|
|
C - the real part of the "correlated" submatrix
|
|
C - the imaginary part of the "correlated" submatrix
|
|
C
|
|
C ----------------------------------------------------------------------------
|
|
C Description of the weight of each k_point for the simple point integration :
|
|
C ----------------------------------------------------------------------------
|
|
DO ik=1,nk
|
|
WRITE(ouctqmc,*) kp(ik,1)%weight
|
|
C This is a geometrical factor independent of the spin value.
|
|
ENDDO
|
|
C
|
|
C -----------------------------------------------------
|
|
C Description of the Hamiltonian H(k) at each k_point :
|
|
C -----------------------------------------------------
|
|
DO is=1,ns
|
|
DO ik=1,nk
|
|
C If the calculation is spin-polarized with SO, the numbers for up and dn bands are the same.
|
|
IF ((ifSP.AND.ifSO).AND.(is.eq.2)) cycle
|
|
DO ib=kp(ik,is)%nb_bot,kp(ik,is)%nb_top
|
|
WRITE(ouctqmc,*) kp(ik,is)%eband(ib)
|
|
ENDDO
|
|
ENDDO
|
|
ENDDO
|
|
C for each spin value is and each k-point,
|
|
C - the energies of the band with spin is at point k
|
|
C
|
|
C
|
|
C ======================================
|
|
C Writing the file case.symqmc :
|
|
C ======================================
|
|
C
|
|
WRITE(buf,'(a)')'Writing the file case.symqmc...'
|
|
CALL printout(0)
|
|
C
|
|
C ----------------------------------------------------------
|
|
C Description of the general informations about the system :
|
|
C ----------------------------------------------------------
|
|
WRITE(ousymqmc,'(2(i6,x))') nsym, natom
|
|
C nysm is the total number of symmetries in the system
|
|
C natom is the total number of atom in the unit cell
|
|
C
|
|
C -------------------------------------------------------------------
|
|
C Description of the permutation matrix associated to each symmetry :
|
|
C -------------------------------------------------------------------
|
|
DO isym=1,nsym
|
|
WRITE(ousymqmc,'(100(i6,x))') srot(isym)%perm(1:natom)
|
|
ENDDO
|
|
C
|
|
C ------------------------------------------------------------
|
|
C Description of the time-reversal property for each symmetry :
|
|
C ------------------------------------------------------------
|
|
IF (ifSP) THEN
|
|
ALLOCATE(timeflag(nsym))
|
|
timeflag=0
|
|
DO isym=1,nsym
|
|
IF (srot(isym)%timeinv) timeflag(isym)= 1
|
|
ENDDO
|
|
WRITE(ousymqmc,'(100(i6,x))') timeflag(1:nsym)
|
|
DEALLOCATE(timeflag)
|
|
C When the calculation is spin-polarized (with SO), a flag which states
|
|
C if a time reversal operation is included in the transformation isym is written.
|
|
ENDIF
|
|
C
|
|
C -----------------------------------------------------------------------------------------
|
|
C Description of the representation matrices of each symmetry for each correlated orbital :
|
|
C -----------------------------------------------------------------------------------------
|
|
DO isym=1,nsym
|
|
DO icrorb=1,ncrorb
|
|
isrt=crorb(icrorb)%sort
|
|
l=crorb(icrorb)%l
|
|
C
|
|
C The case l=0 is a particular case of "non-mixing" basis.
|
|
C --------------------------------------------------------
|
|
IF(l==0) THEN
|
|
C For the s-orbitals, the only irep possible is the matrix itself.
|
|
IF (ifSP.AND.ifSO) THEN
|
|
C If SO is taken into account, spinor rotation matrix is considered.
|
|
ALLOCATE(spinrot(1:2,1:2))
|
|
spinrot(:,:)=0.d0
|
|
IF (srot(isym)%timeinv) THEN
|
|
C In this case, the Euler angle Beta is Pi. The spinor rotation matrix is block-diagonal,
|
|
C since the time-reversal operator is included in the definition of the transformation.
|
|
C
|
|
factor=srot(isym)%phase/2.d0
|
|
C We remind that the field phase is (g-a) if beta=Pi.
|
|
C Up/up and Dn/dn terms
|
|
spinrot(1,1)=EXP(CMPLX(0.d0,factor))
|
|
spinrot(2,2)=CONJG(spinrot(1,1))
|
|
C spinrot(1,1) = -exp(+i(alpha-gamma)/2) ; spinrot(2,2) = -exp(-i(alpha-gamma)/2)
|
|
C in good agreement with Wien conventions for the definition of this phase factor.
|
|
ELSE
|
|
C In this case, the Euler angle Beta is 0. The spinor rotation matrix is block-diagonal,
|
|
C No time-reversal treatment was applied to the transformation.
|
|
C
|
|
factor=srot(isym)%phase/2.d0
|
|
C We remind that the field phase is (a+g) if beta=0.
|
|
C Up/up and Dn/dn terms
|
|
spinrot(1,1)=EXP(CMPLX(0.d0,factor))
|
|
spinrot(2,2)=CONJG(spinrot(1,1))
|
|
C the field phase is 2pi-(alpha+gamma) in this case.
|
|
C spinrot(1,1) = -exp(-i(alpha+gamma)/2) ; spinrot(2,2) = -exp(i(alpha-gamma)/2)
|
|
C in good agreement with Wien conventions for the definition of this phase factor.
|
|
END IF
|
|
C Printing the transformation informations
|
|
DO m=1,2
|
|
WRITE(ousymqmc,*) REAL(spinrot(m,1:2))
|
|
ENDDO
|
|
DO m=1,2
|
|
WRITE(ousymqmc,*) AIMAG(spinrot(m,1:2))
|
|
ENDDO
|
|
C
|
|
DEALLOCATE(spinrot)
|
|
C Without SO, only the rotation matrix in orbital space is necessary.
|
|
ELSE
|
|
C In this case, the Rloc matrix is merely identity.
|
|
WRITE(ousymqmc,*) 1.d0
|
|
WRITE(ousymqmc,*) 0.d0
|
|
ENDIF
|
|
C
|
|
C If the basis representation needs a complete spinor rotation approach (basis with "mixing" ).
|
|
C ---------------------------------------------------------------------------------------------
|
|
ELSEIF (reptrans(l,isrt)%ifmixing) THEN
|
|
C In this case, the SO is necessary considered, spinor rotation matrices are used.
|
|
IF(crorb(icrorb)%ifsplit) THEN
|
|
C If only an irep is correlated
|
|
ind1=1
|
|
DO irep=1,reptrans(l,isrt)%nreps
|
|
IF(crorb(icrorb)%correp(irep)) THEN
|
|
ind2=ind1+reptrans(l,isrt)%dreps(irep)-1
|
|
DO m=ind1,ind2
|
|
WRITE(ousymqmc,*) REAL(srot(isym)%
|
|
& rotrep(l,isrt)%mat(m,ind1:ind2))
|
|
ENDDO
|
|
DO m=ind1,ind2
|
|
WRITE(ousymqmc,*)AIMAG(srot(isym)%
|
|
& rotrep(l,isrt)%mat(m,ind1:ind2))
|
|
ENDDO
|
|
ENDIF
|
|
ind1=ind1+reptrans(l,isrt)%dreps(irep)
|
|
ENDDO
|
|
ELSE
|
|
C If no particular irep is correlated
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(ousymqmc,*) REAL(srot(isym)%
|
|
& rotrep(l,isrt)%mat(m,1:2*(2*l+1)))
|
|
ENDDO
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(ousymqmc,*) AIMAG(srot(isym)%
|
|
& rotrep(l,isrt)%mat(m,1:2*(2*l+1)))
|
|
ENDDO
|
|
ENDIF
|
|
C
|
|
C If the basis representation can be reduce to the up/up block (basis without "mixing").
|
|
C --------------------------------------------------------------------------------------
|
|
ELSE
|
|
IF (ifSP.AND.ifSO) THEN
|
|
C If the calculation is spin-polarized with SO, spinor rotation matrices are considered.
|
|
C The spin is not a good quantum number, so the whole representation is used.
|
|
ALLOCATE(spinrot(1:2*(2*l+1),1:2*(2*l+1)))
|
|
spinrot(:,:)=0.d0
|
|
IF (srot(isym)%timeinv) THEN
|
|
C In this case, the Euler angle Beta is Pi. The spinor rotation matrix is block-diagonal,
|
|
C since the time-reversal operator is included in the definition of the transformation.
|
|
C
|
|
factor=srot(isym)%phase/2.d0
|
|
C We remind that the field phase is (g-a) in this case.
|
|
C Up/up block :
|
|
ephase=EXP(CMPLX(0.d0,factor))
|
|
C As a result, ephase = -exp(i(alpha-gamma)/2)
|
|
spinrot(1:2*l+1,1:2*l+1)=
|
|
= ephase*srot(isym)%rotrep(l,isrt)%mat(-l:l,-l:l)
|
|
C Dn/dn block :
|
|
ephase=CONJG(ephase)
|
|
C Now, ephase = -exp(-i(alpha-gamma)/2)
|
|
spinrot(2*l+2:2*(2*l+1),2*l+2:2*(2*l+1))=
|
|
= ephase*srot(isym)%rotrep(l,isrt)%mat(-l:l,-l:l)
|
|
ELSE
|
|
C In this case, the Euler angle Beta is 0. The spinor rotation matrix is block-diagonal,
|
|
C No time-reversal treatment was applied to the transformation.
|
|
C
|
|
factor=srot(isym)%phase/2.d0
|
|
C We remind that the field phase is (a+g) in this case.
|
|
C Up/up block :
|
|
ephase=EXP(CMPLX(0.d0,factor))
|
|
C As a result, ephase = -exp(-i(alpha+gamma)/2)
|
|
spinrot(1:2*l+1,1:2*l+1)=
|
|
= ephase*srot(isym)%rotrep(l,isrt)%mat(-l:l,-l:l)
|
|
C Dn/dn block :
|
|
ephase=CONJG(ephase)
|
|
C Now, ephase = -exp(i(alpha+gamma)/2)
|
|
spinrot(2*l+2:2*(2*l+1),2*l+2:2*(2*l+1))=
|
|
= ephase*srot(isym)%rotrep(l,isrt)%mat(-l:l,-l:l)
|
|
END IF
|
|
C Printing the transformation informations
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(ousymqmc,*) REAL(spinrot(m,1:2*(2*l+1)) )
|
|
ENDDO
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(ousymqmc,*) AIMAG(spinrot(m,1:2*(2*l+1)) )
|
|
ENDDO
|
|
C
|
|
DEALLOCATE(spinrot)
|
|
C In the other cases (spin-polarized without SO or paramagnetic), only the rotation matrix in orbital space is necessary.
|
|
ELSE
|
|
IF(crorb(icrorb)%ifsplit) THEN
|
|
C If only an irep is correlated
|
|
ind1=-l
|
|
DO irep=1,reptrans(l,isrt)%nreps
|
|
IF(crorb(icrorb)%correp(irep)) THEN
|
|
ind2=ind1+reptrans(l,isrt)%dreps(irep)-1
|
|
DO m=ind1,ind2
|
|
WRITE(ousymqmc,*) REAL(srot(isym)%
|
|
& rotrep(l,isrt)%mat(m,ind1:ind2))
|
|
ENDDO
|
|
DO m=ind1,ind2
|
|
WRITE(ousymqmc,*) AIMAG(srot(isym)%
|
|
& rotrep(l,isrt)%mat(m,ind1:ind2))
|
|
ENDDO
|
|
ENDIF
|
|
ind1=ind1+reptrans(l,isrt)%dreps(irep)
|
|
ENDDO
|
|
ELSE
|
|
C If no particular irep is correlated
|
|
DO m=-l,l
|
|
WRITE(ousymqmc,*) REAL(srot(isym)%
|
|
& rotrep(l,isrt)%mat(m,-l:l))
|
|
ENDDO
|
|
DO m=-l,l
|
|
WRITE(ousymqmc,*) AIMAG(srot(isym)%
|
|
& rotrep(l,isrt)%mat(m,-l:l))
|
|
ENDDO
|
|
END IF ! End of the ifsplit if-then-else
|
|
END IF ! End of the ifSO if-then-else
|
|
END IF ! End of the ifmixing if-then-else
|
|
END DO ! End of the icrorb loop
|
|
END DO ! End of the isym loop
|
|
C for each symmetry and each correlated orbital icrorb, is described :
|
|
C - the real part of the submatrix associated to "isym" for the "icrorb"
|
|
C - the imaginary part of the submatrix associated to "isym" for the "icrorb"
|
|
C
|
|
C -----------------------------------------------------------------------
|
|
C Description of the time reversal operator for each correlated orbital :
|
|
C -----------------------------------------------------------------------
|
|
C This description occurs only if the computation is paramagnetic. (ifSP=0)
|
|
IF (.not.ifSP) THEN
|
|
DO icrorb=1,ncrorb
|
|
l=crorb(icrorb)%l
|
|
isrt=crorb(icrorb)%sort
|
|
C
|
|
C The case l=0 is a particular case of "non-mixing" basis.
|
|
C --------------------------------------------------------
|
|
IF (l==0) THEN
|
|
C For the s-orbitals, the only basis considered is the complex basis.
|
|
WRITE(ousymqmc,*) 1.d0
|
|
WRITE(ousymqmc,*) 0.d0
|
|
C
|
|
C If the basis representation can be reduce to the up/up block (basis without "mixing").
|
|
C --------------------------------------------------------------------------------------
|
|
ELSE
|
|
C Calculation of the time-reversal operator
|
|
ALLOCATE(time_op(-l:l,-l:l))
|
|
time_op(:,:)=0.d0
|
|
DO m=-l,l
|
|
time_op(m,m)=1.d0
|
|
ENDDO
|
|
C time_op is Identity.
|
|
CALL timeinv_op(time_op,(2*l+1),l,isrt)
|
|
C time_op is now the time-reversal operator in the desired basis ({new_i})
|
|
C
|
|
IF(crorb(icrorb)%ifsplit) THEN
|
|
C If only an irep is correlated
|
|
ind1=-l
|
|
DO irep=1,reptrans(l,isrt)%nreps
|
|
IF(crorb(icrorb)%correp(irep)) THEN
|
|
ind2=ind1+reptrans(l,isrt)%dreps(irep)-1
|
|
DO m=ind1,ind2
|
|
WRITE(ousymqmc,*) REAL(time_op(m,ind1:ind2))
|
|
ENDDO
|
|
DO m=ind1,ind2
|
|
WRITE(ousymqmc,*) AIMAG(time_op(m,ind1:ind2))
|
|
ENDDO
|
|
ENDIF
|
|
ind1=ind1+reptrans(l,isrt)%dreps(irep)
|
|
ENDDO
|
|
ELSE
|
|
C If no particular irep is correlated
|
|
DO m=-l,l
|
|
WRITE(ousymqmc,*) REAL(time_op(m,-l:l))
|
|
ENDDO
|
|
DO m=-l,l
|
|
WRITE(ousymqmc,*) AIMAG(time_op(m,-l:l))
|
|
ENDDO
|
|
END IF ! End of the ifsplit if-then-else
|
|
DEALLOCATE(time_op)
|
|
END IF ! End of the l if-then-else
|
|
END DO ! End of the icrorb loop
|
|
END IF ! End of the ifsp if-then-else
|
|
C for each correlated orbital icrorb, the time-reversal operator is described by :
|
|
C - its real part
|
|
C - its imaginary part
|
|
C
|
|
C
|
|
C ===============================
|
|
C Writing the file case.parproj :
|
|
C ===============================
|
|
C
|
|
WRITE(buf,'(a)')'Writing the file case.parproj...'
|
|
CALL printout(0)
|
|
C
|
|
C ----------------------------------------------------------------
|
|
C Description of the size of the basis for each included orbital :
|
|
C ----------------------------------------------------------------
|
|
DO iorb=1,norb
|
|
WRITE(oupartial,'(3(i6))') norm_radf(iorb)%n
|
|
ENDDO
|
|
C There is not more than 1 LO for each orbital (hence n < 4 )
|
|
C
|
|
C The following descriptions are made for each included orbital.
|
|
DO iorb=1,norb
|
|
l=orb(iorb)%l
|
|
isrt=orb(iorb)%sort
|
|
C ------------------------------------
|
|
C Description of the Theta projector :
|
|
C ------------------------------------
|
|
C
|
|
C The case l=0 is a particular case of "non-mixing" basis.
|
|
C --------------------------------------------------------
|
|
IF (l==0) THEN
|
|
DO ik=1,nk
|
|
DO ir=1,norm_radf(iorb)%n
|
|
DO is=1,ns
|
|
WRITE(oupartial,*)
|
|
& REAL(pr_orb(iorb,ik,is)%matn_rep(1,
|
|
& kp(ik,is)%nb_bot:kp(ik,is)%nb_top,ir))
|
|
ENDDO
|
|
DO is=1,ns
|
|
WRITE(oupartial,*)
|
|
& AIMAG(pr_orb(iorb,ik,is)%matn_rep(1,
|
|
& kp(ik,is)%nb_bot:kp(ik,is)%nb_top,ir))
|
|
ENDDO
|
|
ENDDO ! End of the ir loop
|
|
ENDDO ! End of the ik loop
|
|
C
|
|
C If the basis representation needs a complete spinor rotation approach (basis with "mixing" ).
|
|
C ---------------------------------------------------------------------------------------------
|
|
ELSEIF (reptrans(l,isrt)%ifmixing) THEN
|
|
C In this case, the calculation is necessary spin-polarized with SO, spinor rotation matrices are used.
|
|
DO ik=1,nk
|
|
DO ir=1,norm_radf(iorb)%n
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(oupartial,*)
|
|
& REAL(pr_orb(iorb,ik,1)%matn_rep(m,
|
|
& kp(ik,1)%nb_bot:kp(ik,1)%nb_top,ir))
|
|
ENDDO
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(oupartial,*)
|
|
& AIMAG(pr_orb(iorb,ik,1)%matn_rep(m,
|
|
& kp(ik,1)%nb_bot:kp(ik,1)%nb_top,ir))
|
|
ENDDO
|
|
ENDDO ! End of the ir loop
|
|
ENDDO ! End of the ik loop
|
|
C
|
|
C If the basis representation can be reduce to the up/up block (basis without "mixing").
|
|
C --------------------------------------------------------------------------------------
|
|
ELSE
|
|
DO ik=1,nk
|
|
DO ir=1,norm_radf(iorb)%n
|
|
DO is=1,ns
|
|
DO m=-l,l
|
|
WRITE(oupartial,*)
|
|
& REAL(pr_orb(iorb,ik,is)%matn_rep(m,
|
|
& kp(ik,is)%nb_bot:kp(ik,is)%nb_top,ir))
|
|
ENDDO
|
|
ENDDO ! End of the is loop
|
|
DO is=1,ns
|
|
DO m=-l,l
|
|
WRITE(oupartial,*)
|
|
& AIMAG(pr_orb(iorb,ik,is)%matn_rep(m,
|
|
& kp(ik,is)%nb_bot:kp(ik,is)%nb_top,ir))
|
|
ENDDO
|
|
ENDDO ! End of the is loop
|
|
ENDDO ! End of the ir loop
|
|
ENDDO ! End of the ik loop
|
|
ENDIF ! End of the ifmixing if-then-else
|
|
C for each included orbital, for each k-point and each |phi_j> elmt,
|
|
C the corresponding Thetaprojector is described by :
|
|
C - the real part of the matrix
|
|
C - the imaginary part of the matrix
|
|
C
|
|
C -------------------------------------------------------------------------------
|
|
C Description of the density matrices below the lower limit e_bot of the window :
|
|
C -------------------------------------------------------------------------------
|
|
C
|
|
C The case l=0 is a particular case of "non-mixing" basis.
|
|
C --------------------------------------------------------
|
|
IF (l==0) THEN
|
|
C With SO, the density matrix is printed for the complete spin+orbital subspace.
|
|
C (in this case, this means a matrix of size 2)
|
|
IF (ifSP.AND.ifSO) THEN
|
|
ALLOCATE(densprint(2,2))
|
|
densprint(1,1)=densmat(1,iorb)%mat(1,1)
|
|
densprint(2,2)=densmat(2,iorb)%mat(1,1)
|
|
densprint(1,2)=densmat(3,iorb)%mat(1,1)
|
|
densprint(2,1)=densmat(4,iorb)%mat(1,1)
|
|
DO m=1,2
|
|
WRITE(oupartial,*) REAL(densprint(m,1:2))
|
|
ENDDO
|
|
DO m=1,2
|
|
WRITE(oupartial,*) AIMAG(densprint(m,1:2))
|
|
ENDDO
|
|
DEALLOCATE(densprint)
|
|
C In the other cases the density matrix is printed for each spin subspace independently.
|
|
C (in this case, this means two matrices of size 1)
|
|
ELSE
|
|
DO is=1,ns
|
|
WRITE(oupartial,*) REAL(densmat(is,iorb)%mat(1,1))
|
|
ENDDO
|
|
DO is=1,ns
|
|
WRITE(oupartial,*) AIMAG(densmat(is,iorb)%mat(1,1))
|
|
ENDDO
|
|
ENDIF ! End of the ifSO if-then-else
|
|
C
|
|
C If the basis representation needs a complete spinor rotation approach (basis with "mixing" ).
|
|
C ---------------------------------------------------------------------------------------------
|
|
ELSEIF (reptrans(l,isrt)%ifmixing) THEN
|
|
C In this case, the calculation is necessary spin-polarized with SO, spinor rotation matrices are used.
|
|
C As a result, the density matrix is printed for the complete spin+orbital subspace.
|
|
C (this means a matrix of size 2*(2*l+1))
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(oupartial,*)
|
|
& REAL(densmat(1,iorb)%mat(m,1:2*(2*l+1)))
|
|
ENDDO
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(oupartial,*)
|
|
& AIMAG(densmat(1,iorb)%mat(m,1:2*(2*l+1)))
|
|
ENDDO
|
|
C
|
|
C If the basis representation can be reduce to the up/up block (basis without "mixing").
|
|
C --------------------------------------------------------------------------------------
|
|
ELSE
|
|
C With SO, the density matrix is printed for the complete spin+orbital subspace.
|
|
C (this means a matrix of size 2*(2*l+1))
|
|
IF (ifSP.AND.ifSO) THEN
|
|
ALLOCATE(densprint(1:2*(2*l+1),1:2*(2*l+1)))
|
|
densprint(1:2*l+1,1:2*l+1)=
|
|
& densmat(1,iorb)%mat(-l:l,-l:l)
|
|
densprint(2*l+2:2*(2*l+1),2*l+2:2*(2*l+1))=
|
|
& densmat(2,iorb)%mat(-l:l,-l:l)
|
|
densprint(1:2*l+1,2*l+2:2*(2*l+1))=
|
|
& densmat(3,iorb)%mat(-l:l,-l:l)
|
|
densprint(2*l+2:2*(2*l+1),1:2*l+1)=
|
|
& densmat(4,iorb)%mat(-l:l,-l:l)
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(oupartial,*) REAL(densprint(m,1:2*(2*l+1)))
|
|
ENDDO
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(oupartial,*) AIMAG(densprint(m,1:2*(2*l+1)))
|
|
ENDDO
|
|
DEALLOCATE(densprint)
|
|
C In the other cases, the density matrix is printed for each spin subspace independently.
|
|
C (this means two matrices of size 2*l+1)
|
|
ELSE
|
|
DO is=1,ns
|
|
DO m=-l,l
|
|
WRITE(oupartial,*)
|
|
& REAL(densmat(is,iorb)%mat(m,-l:l))
|
|
ENDDO
|
|
ENDDO
|
|
DO is=1,ns
|
|
DO m=-l,l
|
|
WRITE(oupartial,*)
|
|
& AIMAG(densmat(is,iorb)%mat(m,-l:l))
|
|
ENDDO
|
|
ENDDO
|
|
ENDIF ! End of the ifSO if-then-else
|
|
ENDIF ! End of the ifmixing if-then-else
|
|
C for each included orbital, the corresponding density matrix is described by :
|
|
C - the real part of the matrix
|
|
C - the imaginary part of the matrix
|
|
C
|
|
C --------------------------------------------------------------------
|
|
C Description of the global to local coordinates transformation Rloc :
|
|
C --------------------------------------------------------------------
|
|
C Description of each transformation Rloc
|
|
iatom=orb(iorb)%atom
|
|
C
|
|
C The case l=0 is a particular case of "non-mixing" basis.
|
|
C --------------------------------------------------------
|
|
IF(l==0) THEN
|
|
C For the s-orbitals, the only irep possible is the matrix itself.
|
|
IF (ifSP.AND.ifSO) THEN
|
|
C If SO is taken into account, spinor rotation matrix must be considered.
|
|
ALLOCATE(spinrot(1:2,1:2))
|
|
spinrot(:,:)=0.d0
|
|
C The spinor-rotation matrix is directly calculated from the Euler angles a,b and c.
|
|
IF (rotloc(iatom)%timeinv) THEN
|
|
factor=(rotloc(iatom)%a+rotloc(iatom)%g)/2.d0
|
|
spinrot(2,1)=EXP(CMPLX(0.d0,factor))*
|
|
& DCOS(rotloc(iatom)%b/2.d0)
|
|
spinrot(1,2)=-CONJG(spinrot(2,1))
|
|
C Up/dn and Dn/up terms
|
|
factor=-(rotloc(iatom)%a-rotloc(iatom)%g)/2.d0
|
|
spinrot(2,2)=-EXP(CMPLX(0.d0,factor))*
|
|
& DSIN(rotloc(iatom)%b/2.d0)
|
|
spinrot(1,1)=CONJG(spinrot(2,2))
|
|
ELSE
|
|
factor=(rotloc(iatom)%a+rotloc(iatom)%g)/2.d0
|
|
spinrot(1,1)=EXP(CMPLX(0.d0,factor))*
|
|
& DCOS(rotloc(iatom)%b/2.d0)
|
|
spinrot(2,2)=CONJG(spinrot(1,1))
|
|
C Up/dn and Dn/up terms
|
|
factor=-(rotloc(iatom)%a-rotloc(iatom)%g)/2.d0
|
|
spinrot(1,2)=EXP(CMPLX(0.d0,factor))*
|
|
& DSIN(rotloc(iatom)%b/2.d0)
|
|
spinrot(2,1)=-CONJG(spinrot(1,2))
|
|
ENDIF
|
|
C Printing the transformation informations
|
|
DO m=1,2
|
|
WRITE(oupartial,*) REAL(spinrot(m,1:2))
|
|
ENDDO
|
|
DO m=1,2
|
|
WRITE(oupartial,*) AIMAG(spinrot(m,1:2))
|
|
ENDDO
|
|
DEALLOCATE(spinrot)
|
|
C Without SO, only the rotation matrix in orbital space is necessary.
|
|
ELSE
|
|
C In this case, the Rloc matrix is merely identity.
|
|
WRITE(oupartial,*) 1.d0
|
|
WRITE(oupartial,*) 0.d0
|
|
ENDIF
|
|
C Printing the time inversion flag if the calculation is spin-polarized (ifSP=1)
|
|
IF (ifSP) THEN
|
|
timeinvflag=0
|
|
IF (rotloc(iatom)%timeinv) timeinvflag=1
|
|
WRITE(oupartial,'(i6)') timeinvflag
|
|
ENDIF
|
|
C
|
|
C If the basis representation needs a complete spinor rotation approach (basis with "mixing" ).
|
|
C ---------------------------------------------------------------------------------------------
|
|
ELSEIF (reptrans(l,isrt)%ifmixing) THEN
|
|
C In this case, the calculation is necessary spin-polarized with SO, spinor rotation matrices are used.
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(oupartial,*) REAL(rotloc(iatom)%
|
|
& rotrep(l)%mat(m,1:2*(2*l+1)))
|
|
ENDDO
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(oupartial,*) AIMAG(rotloc(iatom)%
|
|
& rotrep(l)%mat(m,1:2*(2*l+1)))
|
|
ENDDO
|
|
C Printing if the transforamtion included a time-reversal operation.
|
|
timeinvflag =0
|
|
IF (rotloc(iatom)%timeinv) timeinvflag=1
|
|
WRITE(oupartial,'(i6)') timeinvflag
|
|
C
|
|
C If the basis representation can be reduce to the up/up block (basis without "mixing").
|
|
C --------------------------------------------------------------------------------------
|
|
ELSE
|
|
IF (ifSP.AND.ifSO) THEN
|
|
C If SO is taken into account, spinor rotation matrices are considered.
|
|
C The spin is not a good quantum number, so the whole representation is used.
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(oupartial,*) REAL(rotloc(iatom)%
|
|
& rotrep(l)%mat(m,1:2*(2*l+1)))
|
|
ENDDO
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(oupartial,*) AIMAG(rotloc(iatom)%
|
|
& rotrep(l)%mat(m,1:2*(2*l+1)))
|
|
ENDDO
|
|
ELSE
|
|
C The calculation is either spin-polarized without SO or paramagnetic.
|
|
C The spin is a good quantum number and irep are possible.
|
|
DO m=-l,l
|
|
WRITE(oupartial,*) REAL(rotloc(iatom)%
|
|
& rotrep(l)%mat(m,-l:l))
|
|
ENDDO
|
|
DO m=-l,l
|
|
WRITE(oupartial,*) AIMAG(rotloc(iatom)%
|
|
& rotrep(l)%mat(m,-l:l))
|
|
ENDDO
|
|
END IF ! End of the ifSO if-then-else
|
|
C Printing if the transformation included a time-reversal operation.
|
|
IF (ifSP) THEN
|
|
timeinvflag =0
|
|
IF (rotloc(iatom)%timeinv) timeinvflag=1
|
|
WRITE(oupartial,'(i6)') timeinvflag
|
|
END IF
|
|
END IF ! End of the ifmixing if-then-else
|
|
C for each included orbital iorb, the transformation Rloc is described by :
|
|
C - the real part of the submatrix associated to "iorb"
|
|
C - the imaginary part of the submatrix associated to "iorb"
|
|
C - a flag which states if a time reversal operation is included in the transformation (if SP only )
|
|
C
|
|
END DO ! End of the iorb loop
|
|
C
|
|
C
|
|
C ==============================
|
|
C Writing the file case.sympar :
|
|
C ==============================
|
|
C
|
|
WRITE(buf,'(a)')'Writing the file case.sympar...'
|
|
CALL printout(0)
|
|
C
|
|
C ----------------------------------------------------------
|
|
C Description of the general informations about the system :
|
|
C ----------------------------------------------------------
|
|
WRITE(ousympar,'(2(i6,x))') nsym, natom
|
|
C nysm is the total number of symmetries of the system
|
|
C natom is the total number of atom in the unit cell
|
|
C
|
|
C -------------------------------------------------------------------
|
|
C Description of the permutation matrix associated to each symmetry :
|
|
C -------------------------------------------------------------------
|
|
DO is=1,nsym
|
|
WRITE(ousympar,'(100(i6,x))') srot(is)%perm
|
|
ENDDO
|
|
C
|
|
C ------------------------------------------------------------
|
|
C Description of the time-reversal property for each symmetry :
|
|
C ------------------------------------------------------------
|
|
IF (ifSP) THEN
|
|
ALLOCATE(timeflag(nsym))
|
|
timeflag=0
|
|
DO isym=1,nsym
|
|
IF (srot(isym)%timeinv) timeflag(isym)= 1
|
|
ENDDO
|
|
WRITE(ousympar,'(100(i6,x))') timeflag(1:nsym)
|
|
DEALLOCATE(timeflag)
|
|
C When the calculation is spin-polarized (with SO), a flag which states
|
|
C if a time reversal operation is included in the transformation isym is written.
|
|
ENDIF
|
|
C
|
|
C -------------------------------------------------------------------------------------------
|
|
C Description of the representation matrices of each symmetry for all the included orbitals :
|
|
C -------------------------------------------------------------------------------------------
|
|
DO isym=1,nsym
|
|
DO iorb=1,norb
|
|
isrt=orb(iorb)%sort
|
|
l=orb(iorb)%l
|
|
C
|
|
C The case l=0 is a particular case of "non-mixing" basis.
|
|
C --------------------------------------------------------
|
|
IF(l==0) THEN
|
|
C For the s-orbitals, the only irep possible is the matrix itself.
|
|
IF (ifSP.AND.ifSO) THEN
|
|
C If SO is taken into account, spinor rotation matrix is considered.
|
|
ALLOCATE(spinrot(1:2,1:2))
|
|
spinrot(:,:)=0.d0
|
|
IF (srot(isym)%timeinv) THEN
|
|
C In this case, the Euler angle Beta is Pi. The spinor rotation matrix is block-diagonal,
|
|
C since the time-reversal operator is included in the definition of the transformation.
|
|
C
|
|
factor=srot(isym)%phase/2.d0
|
|
C We remind that the field phase is (g-a) if beta=Pi.
|
|
C Up/up and Dn/dn terms
|
|
spinrot(1,1)=EXP(CMPLX(0.d0,factor))
|
|
spinrot(2,2)=CONJG(spinrot(1,1))
|
|
C spinrot(1,1) = -exp(i(alpha-gamma)/2) ; spinrot(2,2) = -exp(-i(alpha-gamma)/2)
|
|
C in good agreement with Wien conventions for the definition of this phase factor.
|
|
ELSE
|
|
C In this case, the Euler angle Beta is 0. The spinor rotation matrix is block-diagonal,
|
|
C No time-reversal treatment was applied to the transformation.
|
|
C
|
|
factor=srot(isym)%phase/2.d0
|
|
C We remind that the field phase is (a+g) if beta=0.
|
|
C Up/up and Dn/dn terms
|
|
spinrot(1,1)=EXP(CMPLX(0.d0,factor))
|
|
spinrot(2,2)=CONJG(spinrot(1,1))
|
|
C spinrot(1,1) = -exp(-i(alpha+gamma)/2) ; spinrot(2,2) = -exp(i(alpha-gamma)/2)
|
|
C in good agreement with Wien conventions for the definition of this phase factor.
|
|
END IF
|
|
C Printing the transformation informations
|
|
DO m=1,2
|
|
WRITE(ousympar,*) REAL(spinrot(m,1:2))
|
|
ENDDO
|
|
DO m=1,2
|
|
WRITE(ousympar,*) AIMAG(spinrot(m,1:2))
|
|
ENDDO
|
|
C
|
|
DEALLOCATE(spinrot)
|
|
C Without SO, only the rotation matrix in orbital space is necessary.
|
|
ELSE
|
|
C In this case, the Rloc matrix is merely identity.
|
|
WRITE(ousympar,*) 1.d0
|
|
WRITE(ousympar,*) 0.d0
|
|
ENDIF
|
|
C
|
|
C If the basis representation needs a complete spinor rotation approach (basis with "mixing" ).
|
|
C ---------------------------------------------------------------------------------------------
|
|
ELSEIF (reptrans(l,isrt)%ifmixing) THEN
|
|
C In this case, the SO is necessary considered, spinor rotation matrices are used.
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(ousympar,*) REAL(srot(isym)%
|
|
& rotrep(l,isrt)%mat(m,1:2*(2*l+1)))
|
|
ENDDO
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(ousympar,*) AIMAG(srot(isym)%
|
|
& rotrep(l,isrt)%mat(m,1:2*(2*l+1)))
|
|
ENDDO
|
|
C
|
|
C If the basis representation can be reduce to the up/up block (basis without "mixing").
|
|
C --------------------------------------------------------------------------------------
|
|
ELSE
|
|
IF (ifSP.AND.ifSO) THEN
|
|
C If the calculation is spin-polarized with SO, spinor rotation matrices are considered.
|
|
C The spin is not a good quantum number, so the whole representation is used.
|
|
ALLOCATE(spinrot(1:2*(2*l+1),1:2*(2*l+1)))
|
|
spinrot(:,:)=0.d0
|
|
IF (srot(isym)%timeinv) THEN
|
|
C In this case, the Euler angle Beta is Pi. The spinor rotation matrix is block-diagonal,
|
|
C since the time-reversal operator is included in the definition of the transformation.
|
|
C
|
|
factor=srot(isym)%phase/2.d0
|
|
C We remind that the field phase is (g-a) in this case.
|
|
C Up/up block :
|
|
ephase=EXP(CMPLX(0.d0,factor))
|
|
C AS a result, ephase = -exp(i(alpha-gamma)/2)
|
|
spinrot(1:2*l+1,1:2*l+1)=
|
|
= ephase*srot(isym)%rotrep(l,isrt)%mat(-l:l,-l:l)
|
|
C Dn/dn block :
|
|
ephase=CONJG(ephase)
|
|
C Now, ephase = -exp(-i(alpha-gamma)/2)
|
|
spinrot(2*l+2:2*(2*l+1),2*l+2:2*(2*l+1))=
|
|
= ephase*srot(isym)%rotrep(l,isrt)%mat(-l:l,-l:l)
|
|
ELSE
|
|
C In this case, the Euler angle Beta is 0. The spinor rotation matrix is block-diagonal,
|
|
C No time-reversal treatment was applied to the transformation.
|
|
C
|
|
factor=srot(isym)%phase/2.d0
|
|
C We remind that the field phase is 2pi-(alpha+gamma) in this case.
|
|
C Up/up block :
|
|
ephase=EXP(CMPLX(0.d0,factor))
|
|
C As a result, ephase = -exp(-i(alpha+gamma)/2)
|
|
spinrot(1:2*l+1,1:2*l+1)=
|
|
= ephase*srot(isym)%rotrep(l,isrt)%mat(-l:l,-l:l)
|
|
C Dn/dn block :
|
|
ephase=CONJG(ephase)
|
|
C Now, ephase = -exp(i(alpha+gamma)/2)
|
|
spinrot(2*l+2:2*(2*l+1),2*l+2:2*(2*l+1))=
|
|
= ephase*srot(isym)%rotrep(l,isrt)%mat(-l:l,-l:l)
|
|
END IF
|
|
C Printing the transformation informations
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(ousympar,*) REAL(spinrot(m,1:2*(2*l+1)) )
|
|
ENDDO
|
|
DO m=1,2*(2*l+1)
|
|
WRITE(ousympar,*) AIMAG(spinrot(m,1:2*(2*l+1)) )
|
|
ENDDO
|
|
C
|
|
DEALLOCATE(spinrot)
|
|
C In the other cases (spin-polarized without SO or paramagnetic), only the rotation matrix in orbital space is necessary.
|
|
ELSE
|
|
DO m=-l,l
|
|
WRITE(ousympar,*) REAL(srot(isym)%
|
|
& rotrep(l,isrt)%mat(m,-l:l))
|
|
ENDDO
|
|
DO m=-l,l
|
|
WRITE(ousympar,*) AIMAG(srot(isym)%
|
|
& rotrep(l,isrt)%mat(m,-l:l))
|
|
ENDDO
|
|
END IF ! End of the ifSO if-then-else
|
|
END IF ! End of the ifmixing if-then-else
|
|
END DO ! End of the iorb loop
|
|
END DO ! End of the isym loop
|
|
C for each symmetry and each included orbital iorb, is described :
|
|
C - the real part of the matrix associated to "isym" for the "iorb"
|
|
C - the imaginary part of the matrix associated to "isym" for the "iorb"
|
|
C
|
|
C ---------------------------------------------------------------------
|
|
C Description of the time reversal operator for each included orbital :
|
|
C ---------------------------------------------------------------------
|
|
C This description occurs only if the computation is paramagnetic (ifSP=0)
|
|
IF (.not.ifSP) THEN
|
|
DO iorb=1,norb
|
|
l=orb(iorb)%l
|
|
isrt=orb(iorb)%sort
|
|
C
|
|
C The case l=0 is a particular case of "non-mixing" basis.
|
|
C --------------------------------------------------------
|
|
IF (l==0) THEN
|
|
C For the s-orbitals, the only basis considered is the complex basis.
|
|
WRITE(ousympar,*) 1.d0
|
|
WRITE(ousympar,*) 0.d0
|
|
C
|
|
C If the basis representation can be reduce to the up/up block (basis without "mixing").
|
|
C --------------------------------------------------------------------------------------
|
|
ELSE
|
|
C Calculation of the time-reversal operator
|
|
ALLOCATE(time_op(-l:l,-l:l))
|
|
time_op(:,:)=0.d0
|
|
DO m=-l,l
|
|
time_op(m,m)=1.d0
|
|
ENDDO
|
|
C time_op is Identity.
|
|
CALL timeinv_op(time_op,(2*l+1),l,isrt)
|
|
C time_op is now the time-reversal operator in the desired basis ({new_i})
|
|
C
|
|
DO m=-l,l
|
|
WRITE(ousympar,*) REAL(time_op(m,-l:l))
|
|
ENDDO
|
|
DO m=-l,l
|
|
WRITE(ousympar,*) AIMAG(time_op(m,-l:l))
|
|
ENDDO
|
|
DEALLOCATE(time_op)
|
|
END IF ! End of the l if-then-else
|
|
END DO ! End of the icrorb loop
|
|
END IF ! End of the ifsp if-then-else
|
|
C for each included orbital iorb, the time-reversal operator is described by :
|
|
C - its real part
|
|
C - its imaginary part
|
|
C
|
|
RETURN
|
|
END
|
|
|
|
|
|
|