Re: [QE-users] problem of plottiing band structure based on hybrid HSE functional
Hi Lorenzo, Thanks for your useful suggestions! I reselect the projections of atomic orbitals around the fermi level and increase the iterative numbers to achieve the convergence so that I get a good band structure which is similar with that from pbe calculation. Thank you very much! Best regards, Jingjing --M.Sc. Jingjing YuUniversität LeipzigFelix-Bloch-Institut für FestkörperphysikHalbleiterphysikLinnéstraße 504103 Leipzig, Germany On 7/14/2022 14:05,jy95vowa wrote: Dear Lorenzo, Thanks for your positive reply. I didn't check the convergence of the MLWF spread. I will check it firstly and find a better starting guess to see if the problems could be solved. Thank you very much! Best regards, Jingjing --M.Sc. Jingjing YuUniversität LeipzigFelix-Bloch-Institut für FestkörperphysikHalbleiterphysikLinnéstraße 504103 Leipzig, Germany On 7/13/2022 09:30,Lorenzo Paulatto wrote: Dear Jingjing Yu, from a quick look at the HSE bands, it looks like you have strong k-depend oscillations, which may indicate that the maximally localized wannier function are not particularly localized. This may require a better starting guess, or much more iterations in wannier90. Did you check the convergence of the MLWF spread? You can use projwfc to get an idea of the initial guess. hth On 12/07/2022 18:42, jy95vowa wrote: Dear all, I am a new beginner to do the calculation by using quantum espresso. But now I meet some problems in plotting the band structure based on HSE hybrid functional calculation using the band interpolation method based on MLWF implemented in wannier90 package. When I used HSE functional to do the scf calculation for the first step, it works normally and the related scf file is shown below: &CONTROL calculation= 'scf' prefix= 'Ga4GeO8' outdir= './outdir' pseudo_dir= '/home/yyb/qe-7.0/pseudo' verbosity= 'high' etot_conv_thr = 1.0d-4 forc_conv_thr = 1.0d-4 / &SYSTEM ibrav= 0 nat= 26 ntyp= 3 nosym= .false. occupations = 'fixed' input_dft='HSE' ecutwfc = 100 ecutrho = 400 nbnd = 150 nqx1=1, nqx2=1, nqx3=1 exx_fraction = 0.4 / &ELECTRONS electron_maxstep = 100 conv_thr = 1.0e-11 mixing_mode = 'plain' mixing_beta = 0.4d0 / ATOMIC_SPECIES Ga 69.723 Ga_ONCV_PBE_FR-1.0.upf Ge 72.640 Ge_ONCV_PBE_FR-1.0.upf O 15.999 O_ONCV_PBE_FR-1.0.upf ATOMIC_POSITIONS {crystal} Ga 0.0366214274 0.50 0.7161107456 Ga 0.9633785726 0.50 0.2838892544 Ga 0.2415567616 -0.00 0.6398948438 Ga 0.7584432384 -0.00 0.3601051562 Ge 0.00 0.00 0.00 O 0.1101079368 0.00 0.6956952128 O 0.8898920632 -0.00 0.3043047872 O 0.0585469042 0.50 0.9272169933 O 0.9414530958 0.50 0.0727830067 O 0.8537435172 0.50 0.5370718416 O 0.6611349265 0.50 0.2154100471 O 0.3388650735 0.50 0.7845899529 O 0.1462564828 0.50 0.4629281584 Ga 0.5366214307 0.00 0.7161107535 Ga 0.4633785693 0.00 0.2838892465 Ga 0.7415567530 0.50 0.6398948380 Ga 0.2584432470 0.50 0.3601051620 Ge 0.50 0.50 0.00 O 0.6101079541 0.50 0.6956952567 O 0.3898920459 0.50 0.3043047433 O 0.5585469109 0.00 0.9272170028 O 0.4414530891 0.00 0.0727829972 O 0.3537435256 -0.00 0.5370718546 O 0.1611349374 0.00 0.2154100463 O 0.8388650626 0.00 0.7845899537 O 0.6462564744 -0.00 0.4629281454 K_POINTS crystal 27 0. 0. 0. 3.703704e-02 0. 0. 0. 3.703704e-02 0. 0. 0.6667 3.703704e-02 0. 0. 0. 3.703704e-02 0.
Re: [QE-users] problem of plottiing band structure based on hybrid HSE functional
Dear Lorenzo, Thanks for your positive reply. I didn't check the convergence of the MLWF spread. I will check it firstly and find a better starting guess to see if the problems could be solved. Thank you very much! Best regards, Jingjing --M.Sc. Jingjing YuUniversität LeipzigFelix-Bloch-Institut für FestkörperphysikHalbleiterphysikLinnéstraße 504103 Leipzig, Germany On 7/13/2022 09:30,Lorenzo Paulatto wrote: Dear Jingjing Yu, from a quick look at the HSE bands, it looks like you have strong k-depend oscillations, which may indicate that the maximally localized wannier function are not particularly localized. This may require a better starting guess, or much more iterations in wannier90. Did you check the convergence of the MLWF spread? You can use projwfc to get an idea of the initial guess. hth On 12/07/2022 18:42, jy95vowa wrote: Dear all, I am a new beginner to do the calculation by using quantum espresso. But now I meet some problems in plotting the band structure based on HSE hybrid functional calculation using the band interpolation method based on MLWF implemented in wannier90 package. When I used HSE functional to do the scf calculation for the first step, it works normally and the related scf file is shown below: &CONTROL calculation= 'scf' prefix= 'Ga4GeO8' outdir= './outdir' pseudo_dir= '/home/yyb/qe-7.0/pseudo' verbosity= 'high' etot_conv_thr = 1.0d-4 forc_conv_thr = 1.0d-4 / &SYSTEM ibrav= 0 nat= 26 ntyp= 3 nosym= .false. occupations = 'fixed' input_dft='HSE' ecutwfc = 100 ecutrho = 400 nbnd = 150 nqx1=1, nqx2=1, nqx3=1 exx_fraction = 0.4 / &ELECTRONS electron_maxstep = 100 conv_thr = 1.0e-11 mixing_mode = 'plain' mixing_beta = 0.4d0 / ATOMIC_SPECIES Ga 69.723 Ga_ONCV_PBE_FR-1.0.upf Ge 72.640 Ge_ONCV_PBE_FR-1.0.upf O 15.999 O_ONCV_PBE_FR-1.0.upf ATOMIC_POSITIONS {crystal} Ga 0.0366214274 0.50 0.7161107456 Ga 0.9633785726 0.50 0.2838892544 Ga 0.2415567616 -0.00 0.6398948438 Ga 0.7584432384 -0.00 0.3601051562 Ge 0.00 0.00 0.00 O 0.1101079368 0.00 0.6956952128 O 0.8898920632 -0.00 0.3043047872 O 0.0585469042 0.50 0.9272169933 O 0.9414530958 0.50 0.0727830067 O 0.8537435172 0.50 0.5370718416 O 0.6611349265 0.50 0.2154100471 O 0.3388650735 0.50 0.7845899529 O 0.1462564828 0.50 0.4629281584 Ga 0.5366214307 0.00 0.7161107535 Ga 0.4633785693 0.00 0.2838892465 Ga 0.7415567530 0.50 0.6398948380 Ga 0.2584432470 0.50 0.3601051620 Ge 0.50 0.50 0.00 O 0.6101079541 0.50 0.6956952567 O 0.3898920459 0.50 0.3043047433 O 0.5585469109 0.00 0.9272170028 O 0.4414530891 0.00 0.0727829972 O 0.3537435256 -0.00 0.5370718546 O 0.1611349374 0.00 0.2154100463 O 0.8388650626 0.00 0.7845899537 O 0.6462564744 -0.00 0.4629281454 K_POINTS crystal 27 0. 0. 0. 3.703704e-02 0. 0. 0. 3.703704e-02 0. 0. 0.6667 3.703704e-02 0. 0. 0. 3.703704e-02 0. 0. 0. 3.703704e-02 0. 0. 0.6667 3.703704e-02 0. 0.6667 0. 3.703704e-02 0. 0.6667 0. 3.703704e-02 0. 0.6667 0.6667 3.703704e-02 0. 0. 0. 3.703704e-02 0. 0. 0. 3.703704e-02 0. 0. 0.6667 3.703704e-02 0. 0. 0. 3.703704e-02 0. 0. 0. 3.703704e-02 0. 0. 0.6667 3.703704e-02 0.
Re: [QE-users] problem of plottiing band structure based on hybrid HSE functional
Dear Jingjing Yu, from a quick look at the HSE bands, it looks like you have strong k-depend oscillations, which may indicate that the maximally localized wannier function are not particularly localized. This may require a better starting guess, or much more iterations in wannier90. Did you check the convergence of the MLWF spread? You can use projwfc to get an idea of the initial guess. hth On 12/07/2022 18:42, jy95vowa wrote: Dear all, I am a new beginner to do the calculation by using quantum espresso. But now I meet some problems in plotting the band structure based on HSE hybrid functional calculation using the band interpolation method based on MLWF implemented in wannier90 package. When I used HSE functional to do the scf calculation for the first step, it works normally and the related scf file is shown below: &CONTROL calculation= 'scf' prefix= 'Ga4GeO8' outdir= './outdir' pseudo_dir= '/home/yyb/qe-7.0/pseudo' verbosity= 'high' etot_conv_thr = 1.0d-4 forc_conv_thr = 1.0d-4 / &SYSTEM ibrav= 0 nat= 26 ntyp= 3 nosym= .false. occupations = 'fixed' input_dft='HSE' ecutwfc = 100 ecutrho = 400 nbnd = 150 nqx1=1, nqx2=1, nqx3=1 exx_fraction = 0.4 / &ELECTRONS electron_maxstep = 100 conv_thr = 1.0e-11 mixing_mode = 'plain' mixing_beta = 0.4d0 / ATOMIC_SPECIES Ga 69.723 Ga_ONCV_PBE_FR-1.0.upf Ge 72.640 Ge_ONCV_PBE_FR-1.0.upf O 15.999 O_ONCV_PBE_FR-1.0.upf ATOMIC_POSITIONS {crystal} Ga 0.0366214274 0.50 0.7161107456 Ga 0.9633785726 0.50 0.2838892544 Ga 0.2415567616 -0.00 0.6398948438 Ga 0.7584432384 -0.00 0.3601051562 Ge 0.00 0.00 0.00 O 0.1101079368 0.00 0.6956952128 O 0.8898920632 -0.00 0.3043047872 O 0.0585469042 0.50 0.9272169933 O 0.9414530958 0.50 0.0727830067 O 0.8537435172 0.50 0.5370718416 O 0.6611349265 0.50 0.2154100471 O 0.3388650735 0.50 0.7845899529 O 0.1462564828 0.50 0.4629281584 Ga 0.5366214307 0.00 0.7161107535 Ga 0.4633785693 0.00 0.2838892465 Ga 0.7415567530 0.50 0.6398948380 Ga 0.2584432470 0.50 0.3601051620 Ge 0.50 0.50 0.00 O 0.6101079541 0.50 0.6956952567 O 0.3898920459 0.50 0.3043047433 O 0.5585469109 0.00 0.9272170028 O 0.4414530891 0.00 0.0727829972 O 0.3537435256 -0.00 0.5370718546 O 0.1611349374 0.00 0.2154100463 O 0.8388650626 0.00 0.7845899537 O 0.6462564744 -0.00 0.4629281454 K_POINTS crystal 27 0. 0. 0. 3.703704e-02 0. 0. 0. 3.703704e-02 0. 0. 0.6667 3.703704e-02 0. 0. 0. 3.703704e-02 0. 0. 0. 3.703704e-02 0. 0. 0.6667 3.703704e-02 0. 0.6667 0. 3.703704e-02 0. 0.6667 0. 3.703704e-02 0. 0.6667 0.6667 3.703704e-02 0. 0. 0. 3.703704e-02 0. 0. 0. 3.703704e-02 0. 0. 0.6667 3.703704e-02 0. 0. 0. 3.703704e-02 0. 0. 0. 3.703704e-02 0. 0. 0.6667 3.703704e-02 0. 0.6667 0. 3.703704e-02 0. 0.6667 0. 3.703704e-02 0. 0.6667 0.6667 3.703704e-02 0.6667 0. 0. 3.703704e-02 0.6667 0. 0. 3.703704e-02 0.6667 0. 0.6667 3.703704e-02 0.6667 0. 0. 3.703704e-02 0.6667 0. 0. 3.703704e-02 0.6667 0. 0.6667 3.703704e-02 0.6667 0.6667 0. 3.703704e-02 0.6667 0.6667 0. 3.703704e-02 0.6667 0.6667 0.6667 3.703704e-02 CELL_PARAMETERS {angstrom} 12.418376121 0.0 0.012838255 0.0 3.047685209 0.0 -5.492030502 0.0 7.786621268 Then I executed pw2wannier90.x in qe-7.0 package and wannier90.x located in the wannier package, the execution has no problem but the band structure looks very weird attached below. I know that the shape of band structure calculated from the hybrid function should be similar with that of pbe functional , but there is a rigid shift between two of them (the calculated bandgap based on HSE is about 4.5 eV from scf calculation, the energy gap is about 2.28 eV based on PBE ). But in my case, the plotting band shape lying on the bottom of conduction band and bottom of valence band changes, and the the lowest point of the conduction band is not located at Gamma point compared with that of pbe calculation attached below. I don't know why the band shapes changes and how to solve it? The input win.file has been attached. num_bands = 150 num_wann = 10 dis