EPW Forum

I also find the same issue. I am using QE-6.1 currently and in the process of upgrading to the latest. Meanwhile with QE-6.1 I get a similar inconsistency
as "gbrunin". My input file looks like:

epbwrite = .false.
epbread = .false.
epwwrite = .false.
epwread = .true.

elph = .true.
kmaps = .true.
etf_mem = .false.

nbndsub = 12
nbndskip = 0

dis_win_max = 29
dis_froz_max= 14
dis_froz_min= -8

restart = .true.
restart_freq= 500

nk1 = 6
nk2 = 6
nk3 = 6

nq1 = 6
nq2 = 6
nq3 = 6

nkf1 = 50
nkf2 = 50
nkf3 = 50

nqf1 = 50
nqf2 = 50
nqf3 = 50



And the output always ends up starting from the beginning.

Number of ep-matrix elements per pool : 106272 ~= 830.25 Kb (@ 8 bytes/ DP)
Restart from: 3001/ 125000

===================================================================
Electron (Imaginary) Self-Energy in the Migdal Approximation
===================================================================


Fermi Surface thickness = 25.000000 eV

Golden Rule strictly enforced with T = 0.025852 eV
Gaussian Broadening: 0.010000 eV, ngauss= 1
DOS = -0.000000 states/spin/eV/Unit Cell at Ef= 5.658487 eV

Progression iq (fine) = 100/ 125000
Progression iq (fine) = 200/ 125000
Progression iq (fine) = 300/ 125000
Progression iq (fine) = 400/ 125000
Progression iq (fine) = 500/ 125000
Creation of a restart point
Progression iq (fine) = 600/ 125000



Does the fine grid have to be defined as randk/q for the "restart" to work?

Nandan.

Hi everyone,

It is working with the last version of QE. I simply installed it and used it with the same input file, no need to rerun anything.



Guillaume

Hello,

The version 4.3 of EPW and QE 6.2 is out : http://epw.org.uk/Main/DownloadAndInstall

Best,
Samuel

Hi,

I installed QE-6.2 and restarted EPW calculation from scratch.
I wanted to test whether the restart option works.

With the new version, the job crashes with segmentation fault
everytime.

This is where the job was killed so that the restart file was created:
==========================================
Golden Rule strictly enforced with T = 0.025852 eV
Gaussian Broadening: 0.010000 eV, ngauss= 1

Progression iq (fine) = 50/ 125000
Creation of a restart point
Progression iq (fine) = 100/ 125000
Creation of a restart point
Progression iq (fine) = 150/ 125000
Creation of a restart point
==========================================


The error message upon restart

==========================================


ibndmin = 1 ebndmin = -0.574
ibndmax = 12 ebndmax = 1.597


Number of ep-matrix elements per pool : 54432 ~= 425.25 Kb (@ 8 bytes/ DP)
Restart from: 151/ 125000
rank 25 in job 2 triana_34496 caused collective abort of all ranks
exit status of rank 25: return code 174
===========================================

Error:

forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line Source
epw.x 0000000000F2AEA1 Unknown Unknown Unknown
epw.x 0000000000F28FDB Unknown Unknown Unknown
epw.x 0000000000EC3134 Unknown Unknown Unknown
epw.x 0000000000EC2F46 Unknown Unknown Unknown
epw.x 0000000000E45739 Unknown Unknown Unknown
epw.x 0000000000E4DC36 Unknown Unknown Unknown
libpthread-2.17.s 00007F2A1D4065E0 Unknown Unknown Unknown
epw.x 00000000004A7531 selfen_elec_q_ 303 selfen_elec.f90
epw.x 000000000044EF57 ephwann_shuffle_ 983 ephwann_shuffle.f90
epw.x 000000000043F447 elphon_shuffle_wr 764 elphon_shuffle_wrap.f90
epw.x 0000000000407285 MAIN__ 150 epw.f90
epw.x 00000000004065DE Unknown Unknown Unknown
libc-2.17.so 00007F2A1CD53C05 __libc_start_main Unknown Unknown
epw.x 00000000004064E9 Unknown Unknown Unknown


What may be the reason for this?

Nandan.

Hello Nandan,

I might have overlooked the mode-resolved thing when I coded the restart.

Could you try without:


And tell me if it works. If so I'll correct it.

Thanks,
Samuel

Hi Samuel,

You are right, it works without

iverbosity = 3


Thanks. Can you suggest the way forward since I would like to have "iverbosity = 3".

Nandan.

Hello,

I've added it on my to-do list. Will come back to you when done.

Best,
Samuel

Great.

Thanks.

Nandan.

Hi,

I also get the same error if I leave

but it works when I set them to false. Is there a way to obtain a2f with a restart calculation ?

Thanks,
Guillaume

Hello,

At the moment only electron self-energy can be restarted. I will eventually code a general restart.

If you want it sooner, you can try to implement it yourself by copying what was done for self_elec (follow the variable "restart").

Best,
Samuel