How to restart EPW calculations in the following two cases?

[hellolori]

Dear all,

I have a question about how to restart the calculation in the following two cases:
(1) Calculations of prefix.epb*. I do not find any related input tags about this topic.
(2) Interpolations on the fine k- and q-grid. I try to restart from a crashed interpolation calculation, but I failed.
First, I start a new calculation by setting the following tags:

Code: Select all

epbwrite='false'
epbread='false'
epwwrite='true'
epwread='false'
etf_mem     = 1
ep_coupling = .true.
elph        = .true.
wannierize  = .true.
restart=.true.
restart_freq = 10

The calculation stop at the process of interpolation. Then, I change the related tags to restart the calculation

Code: Select all

epbwrite='false'
epbread='false'
epwwrite='false'
epwread='true'
etf_mem     = 1
ep_coupling = .true.
elph        = .true.
wannierize  = .false.
restart=.true.
!restart_freq = 10

But the restarting calculation always failed, and the last several lines read:

Code: Select all

141      Number of ep-matrix elements per pool :        84375 ~=  659.18 Kb (@ 8 bytes/ DP)
142      Restart from:         41/     18000
143
144      ismear =     1 iq =      41 coord.:   0.00000  0.06667  0.00000 wt:   0.00006
145      -------------------------------------------------------------------

The standard output reads:

Code: Select all

forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image              PC                Routine            Line        Source
epw  000000000050B517  selfen_phon_q_            348  selfen_phon.f90
epw  0000000000494B5C  ephwann_shuffle_         1004  ephwann_shuffle.f90
epw  000000000047A5CD  elphon_shuffle_wr         766  elphon_shuffle_wrap.f90
epw  0000000000441571  MAIN__                    150  epw.f90
epw  000000000044062C  Unknown               Unknown  Unknown
libc.so.6          00000039C9A1ECDD  Unknown               Unknown  Unknown
epw_debug_2018022  0000000000440529  Unknown               Unknown  Unknown

Can anyone help me?

[hellolori]

Can anyone help me? thanks

[carla.verdi]

Hi,

The prefix.epb* files are calculated almost at the beginning of the EPW run. To store them on disk you use epbwrite=.true.; when restarting from existing prefix.epb* files you can set epbwrite=.false. and epbread=.true.
You cannot break the calculation of the prefix.epb* files in multiple restart sets.

Regarding the second point, using epbwrite=.false., epbread=.false, epwwrite=.false., epbread=.true.
allows you to restart an EPW calculation by interpolating all quantities on the fine k,q grids. I believe the input 'restart' and 'restart_freq' are not relevant for a phonon self-energy calculation (from the output, I think that's what you're doing?)

Best,
Carla

[liu xinbao]

I also want to ask the same question about restart the crashed calculation because of exceeding time limitation of PBS, the output stop with following

Code: Select all

     
     Progression iq (fine) =      25450/     27000
     Progression iq (fine) =      25500/     27000
     Progression iq (fine) =      25550/     27000
     Progression iq (fine) =      25600/     27000
     Progression iq (fine) =      25650/     27000
     Progression iq (fine) =      25700/     27000
     Progression iq (fine) =      25750/     27000
     Progression iq (fine) =      25800/     27000
     Progression iq (fine) =      25850/     27000
     Progression iq (fine) =      25900/     27000
     Progression iq (fine) =      25950/     27000
     Progression iq (fine) =      26000/     27000
     Progression iq (fine) =      26050/     27000
     Progression iq (fine) =      26100/     27000
     Progression iq (fine) =      26150/     27000
     Progression iq (fine) =      26200/     27000

as you can see, it is very close to the end.
My input tags are listed as following

Code: Select all

  ep_coupling = .t.
  elph        = .t.
  kmaps       = .t.
  epbwrite    = .f.
  epbread     = .t.

  epwwrite = .f.
  epwread  = .t.

  etf_mem     =  1
  ephwrite    = .f.
  eliashberg  = .t.
  max_memlt = 5.0d0
  laniso = .true.
  limag = .true.
  lpade = .true.

Any help will be appreciated

[Zhe_L]

Hi

I think you also need kmaps = .true.

to liu
I'm afraid you can't restart from here.