Compare the g vertex between the EPW and DFPT
Posted: Wed Nov 19, 2025 4:13 am
Dear,
I aim to compare the electron-phonon coupling matrix elements (g-vertex) computed by EPW and DFPT. For the DFPT calculations performed with ph.x, I note from the source code elphon.f90 that the dynamical matrices (dyn) used to construct the g-vertex are obtained without applying the Acoustic Sum Rule (ASR). In contrast, the EPW calculation incorporates ASR correction. This difference in treatment leads to discrepancies in the resulting dynamical matrices and phonon eigenvalues between the two approaches. Consequently, the g-vertex obtained from EPW is expected to differ from that of DFPT when these factors are considered. Given these systematic differences, how should we assess the consistency between the EPW and DFPT results?
The input file of ph.x is shown as below. As an examples, the phonon frequncey near the gamma points is negative in DFPT, while positive in EPW, due to the implementation of ASR.
**************
&inputph
prefix = 'blue'
outdir = './'
fildyn = 'blue.dyn.xml'
fildvscf = 'dvscf'
tr2_ph = 1.0d-19
q_in_band_form = .true.
qplot = .true.
electron_phonon = 'prt'
kx = 0.0
ky = 0.0
kz = 0.0
/
3
0.500000 0.288675 0.000000 63
0.00000000 0.00000000 0.00000000 63
0.333333 0.577350 0.000000 1
*****************
I aim to compare the electron-phonon coupling matrix elements (g-vertex) computed by EPW and DFPT. For the DFPT calculations performed with ph.x, I note from the source code elphon.f90 that the dynamical matrices (dyn) used to construct the g-vertex are obtained without applying the Acoustic Sum Rule (ASR). In contrast, the EPW calculation incorporates ASR correction. This difference in treatment leads to discrepancies in the resulting dynamical matrices and phonon eigenvalues between the two approaches. Consequently, the g-vertex obtained from EPW is expected to differ from that of DFPT when these factors are considered. Given these systematic differences, how should we assess the consistency between the EPW and DFPT results?
The input file of ph.x is shown as below. As an examples, the phonon frequncey near the gamma points is negative in DFPT, while positive in EPW, due to the implementation of ASR.
**************
&inputph
prefix = 'blue'
outdir = './'
fildyn = 'blue.dyn.xml'
fildvscf = 'dvscf'
tr2_ph = 1.0d-19
q_in_band_form = .true.
qplot = .true.
electron_phonon = 'prt'
kx = 0.0
ky = 0.0
kz = 0.0
/
3
0.500000 0.288675 0.000000 63
0.00000000 0.00000000 0.00000000 63
0.333333 0.577350 0.000000 1
*****************