QE 7.1 pw2bgw run errors

[Hong Tang]

Dear BerkeleyGW team,

I met errors when using pw2bgw.x convert wavefunction into BGW form. 

The crash error is:

 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

     task #       660

     from diropn : error #         3

     wrong record length

 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

     task #       453

     from diropn : error #         3

     wrong record length

 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

Here is the job submission script: 

#!/bin/bash

#SBATCH -A xxxxx

#SBATCH -C cpu

#SBATCH --qos=regular

#SBATCH --time=12:00:00

#SBATCH --nodes=128

# # #SBATCH --ntasks-per-node=16

#SBATCH --mail-type=begin,end,fail

#SBATCH --mail-user=xxxxxx

#OpenMP settings:

export OMP_NUM_THREADS=2

export OMP_PLACES=threads

export OMP_PROC_BIND=spread

module load espresso/7.1-libxc-6.1.0-cpu

PW="pw.x"

MPIRUN="srun -n 1024  -c 32 --cpu_bind=cores"

POOLS="-npools 64" #We only use pools for calculations with many k-points

cd ./01-scf

$MPIRUN $PW   -ndiag 2   -pd .true.   -in     ./in     &>     ./out

 rm ./*.wfc*

cd ..

#

PW2BGW="/global/common/software/nersc/pm-2021q4/sw/qe/7.1/pm-cpu/B/bin/pw2bgw.x"

cd ./02-wfn

$MPIRUN $PW   -ndiag 2  -pd .true.   $POOLS      -in ./in &> ./out

$MPIRUN $PW2BGW     -pd .true.      -in ./pp_in &> ./pp_out

 rm ./*.wfc*

cd ..

______________________________________

Please note that I used    "    -pd .true.     "   to avoid the error       " Error in routine  fft_type_set (6):

  there are processes with no planes. Use pencil decomposition (-pd .true.)  "

01-scf folder ran correctly. 

For 02-wfn,   pw.x run correctly.  The errors appear when running pw2bgw.x   .    

Please check all the files in folder 02-wfn, which I attached below. 

The files in  .save folder were removed in order to reduce the size of the attached file. 

I am sure that the .xml file is correct and is linked to the one in 01-scf folder. 

Please help me what is the problem and how to solve this.

Best,

Hong

[Hong Tang]

Let me add some things:

1. if you check file "out" and file "pp_out", in the "Parallelization info " section, there are very different. is this the reason?

2. the system has "fractional translation" system, as you can see in file "out",      " 12 Sym. Ops., with inversion, found (10 have fractional translation) ", will this matter for BerkeleyGW? 

Best,

Hong

[Zhenglu Li]

Dear Hong,

Maybe you can try running pw2bgw with 1 core in serial to see if you are using too many cores?  BerkeleyGW works with fractional translations.

Best,

Zhenglu

---

From: Hong Tang <tangh...@gmail.com>
Sent: Tuesday, August 29, 2023 6:46 PM
To: BerkeleyGW Help <he...@berkeleygw.org>
Cc: Hong Tang <tangh...@gmail.com>
Subject: Re: QE 7.1 pw2bgw run errors

 

--
You received this message because you are subscribed to the Google Groups "BerkeleyGW Help" group.
To unsubscribe from this group and stop receiving emails from it, send an email to help+uns...@berkeleygw.org.
To view this discussion on the web visit https://groups.google.com/a/berkeleygw.org/d/msgid/help/308a6852-10f8-4272-94a2-bacad55f5716n%40berkeleygw.org.

[Hong Tang]

Dear Zhenglu,

Thanks for helping!

I just tried 1 core and got error:

 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
     Error in routine  splitwf (9767):
      wrong size for pwt
 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

I noted that  there is a warning "   Check: negative core charge=   -0.000099  " . This may related to small charge density cut I used.

Does it matter and cause this issue?

Best,

Hong

[Zhenglu Li]

Dear Hong,

The little deviation in charge density should be fine.  Maybe you can also try the latest version of QE and the associated pw2bgw.

Best,

Zhenglu

---

From: Hong Tang <tangh...@gmail.com>
Sent: Tuesday, August 29, 2023 9:24 PM
To: BerkeleyGW Help <he...@berkeleygw.org>
Cc: Zhenglu Li <zhen...@usc.edu>; Hong Tang <tangh...@gmail.com>

[Hong Tang]

Also I checked with 1 core run 

     G-vector sticks info
     --------------------
     sticks:   dense  smooth     PW     G-vecs:    dense   smooth      PW
     Sum        1483    1483    421                78109    78109   11731

However, my first error run has 

    Parallelization info
     --------------------
     sticks:   dense  smooth     PW     G-vecs:    dense   smooth      PW
     Min           1       1      0                   63       63       0
     Max           2       2      1                   90       90      43
     Sum        1483    1483    451                78109    78109   13177

Please note that 

column 4 PW   421 vs 451.

[Hong Tang]

Dear Zhenglu,

I used this QE7.1 version installed by nersc team on nersc for other small systems. They are ok, on problem.
But why for this one having problem?

Best,
Hong

[Zhenglu Li]

Dear Hong,

Not sure why it doesn't work in this case.  The error message looks like the files are not opened properly, see QE forum: https://lists.quantum-espresso.org/pipermail/users/2009-March/011730.html  There are a few solutions following the QE thread by either modifying the source file or recompiling with a specific flag.  It looks like this is possibly associated with an integer overflow, which could be system dependent.

Best,

Zhenglu

---

From: Hong Tang <tangh...@gmail.com>
Sent: Tuesday, August 29, 2023 9:36 PM

To view this discussion on the web visit https://groups.google.com/a/berkeleygw.org/d/msgid/help/edf44226-4f4c-4c89-a5ab-e92b44fa7d33n%40berkeleygw.org.

[Hong Tang]

Dear Zhenglu,

That solutions are too complicated to me. How to do it in practice?

Do you know the code writers of pw2bgw.x ?   Please let me know. Or forward this email to them. 

Maybe the developers know better about this issue. 

I recall that during my experience of using pw2bgw.x I met occasionally this issue (I have an old email-question in BerkelyGW forum relating the same issue, I am just wondering whether other BGW users had the same issue or not). But sometimes, this error disappears if I just rerun pw.x and pw2bgw. But this time, I tried several times, but it is still there.  

Please help.

[Hong Tang]

Problem solved. The .xml file in 02- folder was accidentally changed. it caused the issue. 

Best,

Hong