Weired phonon spectrum in EPW.5.1.0

[liu xinbao]

I try EPW 5.1.0 to calculate the electron phonon coupling in grahene.
And I do the wannier fit firstly, the band fit well with the DFT band but phonon spectrum is very weired like following
https://imgur.com/a/ZAv2f5Q
and I calculate the phonon self energy and then plot linewidth.phself with third column which is eigenvalue of phonon but still get the same result with above.
But when I repeat this calculation with EPW 5.0.0, I got the correct phonon spectrum.
Besides, When I calculate the phonon self-energy and a2f of graphene with interpolated K-mesh, I often get

Code: Select all

*** Error in `/home/xbliu/src/qe-6.4/bin/epw.x': malloc(): memory corruption: 0x0000000002e10de0 ***
*** Error in `/home/xbliu/src/qe-6.4/bin/epw.x': malloc(): memory corruption: 0x0000000003615de0 ***
*** Error in `/home/xbliu/src/qe-6.4/bin/epw.x': malloc(): memory corruption: 0x00000000038cae10 ***
*** Error in `/home/xbliu/src/qe-6.4/bin/epw.x': free(): invalid next size (normal): 0x00000000044d7c10 ***
*** Error in `/home/xbliu/src/qe-6.4/bin/epw.x': free(): invalid next size (normal): 0x0000000003d71bf0 ***

or

Code: Select all

forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image              PC                Routine            Line        Source
epw.x              0000000000EF698D  Unknown               Unknown  Unknown
libpthread-2.17.s  00002B98BAA3F6D0  Unknown               Unknown  Unknown
epw.x              0000000000485AE9  selfen_phon_q_            406  selfen_phon.f90
epw.x              00000000004230C2  ephwann_shuffle_         1309  ephwann_shuffle.f90
epw.x              0000000000412CE7  elphon_shuffle_wr         767  elphon_shuffle_wrap.f90
epw.x              0000000000407E34  MAIN__                    150  epw.f90
epw.x              00000000004071DE  Unknown               Unknown  Unknown
libc-2.17.so       00002B98BAF70445  __libc_start_main     Unknown  Unknown
epw.x              00000000004070E9  Unknown               Unknown  Unknown

when I increase the interpolated K mesh
My input file was list as following

Code: Select all

--
&inputepw
  prefix      = 'graphene',
  amass(1)    = 12.011
  amass(2)    = 12.011
  outdir      = './'
  dvscf_dir   = '../phonons/save'
  etf_mem = 1
  iverbosity = 1

  elph        = .true.
  kmaps       = .t.
  epbwrite    = .f.
  epbread     = .f.
  epwwrite    = .f.
  epwread     = .t.

  wannierize  = .f.
  nbndsub     =  5
  nbndskip    =  0
  num_iter    = 500
  dis_win_max = 7.3
  dis_froz_max= 0.3
  proj(1)     = 'C1:sp2;pz'
  proj(2)     = 'C2:pz'
  wdata(1) = 'bands_plot = .true.'
  wdata(2) = 'begin kpoint_path'
  wdata(3) = 'G 0.0 0.0 0.0 M 0.5 0.0 0.0'
  wdata(4) = 'M 0.5 0.0 0.0 K 0.333333 0.333333 0.0'
  wdata(5) = 'K 0.333333 0.333333 0.0 G 0.0 0.0 0.0'
  wdata(6) = 'end kpoint_path'
  wdata(7) = 'bands_num_points = 100'

  band_plot = .t.
!  delta_approx = .t.
!  elecselfen  = .false.
!  phonselfen  = .t.
!  a2f         = .f.

  fsthick     = 1 ! eV
  eptemp      = 300 ! K
  degaussw    = 0.001 ! eV

! nkf1         = 100
! nkf2         = 100
! nkf3         = 1
! nqf1         = 50
! nqf2         = 50
! nqf3         = 1
 filqf = 'graphene_band.kpt'
 filkf = 'graphene_band.kpt'
!   rand_k = .t.
!   rand_q = .t.
!   rand_nk = 125000
!   rand_nq = 500


  nk1         = 20
  nk2         = 20
  nk3         = 1
  nq1         = 10
  nq2         = 10
  nq3         = 1
 /
     14 cartesian
 0.000000000   0.000000000   0.000000000
 0.000000000   0.115470054   0.000000000
 0.000000000   0.230940108   0.000000000
 0.000000000   0.346410162   0.000000000
 0.000000000   0.461880215   0.000000000
 0.000000000  -0.577350269   0.000000000
 0.100000000   0.173205081   0.000000000
 0.100000000   0.288675135   0.000000000
 0.100000000   0.404145188   0.000000000
 0.100000000   0.519615242   0.000000000
 0.200000000   0.346410162   0.000000000
 0.200000000   0.461880215   0.000000000
 0.200000000   0.577350269   0.000000000
 0.300000000   0.519615242   0.000000000

And my working condition is intelmpi2018 + mpiifort+gcc compiler.
I don't know if it was because my input file was irrational.
Any suggestion is appreciated.

[sponce]

Hello,

Your input looks ok to me.
Are you sure you are plotting the data correctly ?

At the end of EPW, you should get a file name "phband.freq"
You can convert this file with plotband.x from QE/bin/ into a gnuplot format.

Best wishes,
Samuel

[liu xinbao]

Thank you for your reply.
I plot it just as what you said using plotband.x on phband.freq. I also open the phband.freq and found the eigenvalue of phonon at different q point looks weired as well, they are discrete at some q point and some frequency are incrediblely high.
Part of phband.freq are attached following.

Code: Select all

        1.9504       22.4566       36.0198      111.3675      198.8882      204.3479
            0.090000  0.051962  0.000000
        2.2615       23.7343       38.1117      111.2734      198.6862      204.6126
            0.095000  0.054848  0.000000
        2.5845       25.0046       40.1981      111.1735      198.4749      204.8602
            0.100000  0.057735  0.000000
        2.9199       26.2668       42.2786      111.0678      198.2546      205.0887
            0.105000  0.060622  0.000000
        3.2678       27.5204       44.3527      110.9562      198.0255      205.2959
            0.110000  0.063509  0.000000
        3.6284       28.7652       46.4201      110.8385      197.7881      205.4806
            0.115000  0.066395  0.000000
     1564.6553        0.0000        0.0000     1564.6553        0.0000        0.0000
            0.120000  0.069282  0.000000
     1632.6838        0.0000        0.0000     1632.6838        0.0000        0.0000
            0.125000  0.072169  0.000000
     1700.7123        0.0000        0.0000     1700.7123        0.0000        0.0000
            0.130000  0.075056  0.000000
     1768.7408        0.0000        0.0000     1768.7408        0.0000        0.0000
            0.135000  0.077942  0.000000
     1836.7692        0.0000        0.0000     1836.7692        0.0000        0.0000

And I am sure my k point path is correct, because both the electronic band and the EPW5.0.0 version with the same input file give the same correct result.