EDIFF can't be recognized

Question on input files/tags, interpreting output, etc.

Please check whether the answer to your question is given in the VASP online manual or has been discussed in this forum previously!

Moderators: Global Moderator, Moderator

Locked
Message
Author
gu301purdue
Newbie
Newbie
Posts: 19
Joined: Fri Jul 02, 2021 12:57 am

EDIFF can't be recognized

#1 Post by gu301purdue » Thu Nov 04, 2021 1:47 am

Dear colleagues,
I tried to submit a job, however, my EDIFF can't be recognized. The input file is as follows. I marked the parameters that can't be recognized. The error message is
Error reading item EDIFF from file INCAR.
Error code was IERR= 5 ... .

Can anyeone help?

Thanks!
Tina



SYSTEM = sio2
#NWRITE = 1

#Startparameter for this Run (values of parameters are equalt to default):
ISTART = 0 0=new, 1=wave, 2=same cutoff&latt (1&2 need WAVECAR file)
#ICHARG = 2 0=charge from wave, 1=charge from CHGCAR, 2=atom superpos.
LWAVE=.FALSE.

Electronic Relaxation 1
ENCUT=800 planewave cutoff
NELM= 50 max number of electronic steps
#EDIFF=1E-6
EDIFFG=-1E-4 force stopping-criterion for geometry steps
#NELMIN=4
PREC=Normal

Ionic Relaxation
NSW=1000 max number of geometry steps
IBRION=2 ionic relax: 0-MD 1-quasi-New 2-CG
ISIF=4 (2:force=y stress=y ions=y shape=n volume=n,
ISYM=1 1=use symmetry, 0 = no symmetry
POTIM=0.3 initial time step for geo-opt (increase for soft sys)


Electronic Relaxation 2
IALGO=48 algorithm (8=CG for small, 48=RMM for big systems)
LREAL=A

#LPLANE = .TRUE.
#NPAR = 8
#NSIM = 1
~

martin.schlipf
Global Moderator
Global Moderator
Posts: 458
Joined: Fri Nov 08, 2019 7:18 am

Re: EDIFF can't be recognized

#2 Post by martin.schlipf » Thu Nov 04, 2021 8:30 am

Sorry I cannot reproduce this issue. Please add a complete list of input files.

Note that is often easy to reduce this problem to a simple case. You could for example use an INCAR file that only contains the line that doesn't work. If that fixes the issue, then it is related to the interplay of the different tags. Then you can systematically add the lines back in until it fails and often that gives you a clue as to why it doesn't work.

gu301purdue
Newbie
Newbie
Posts: 19
Joined: Fri Jul 02, 2021 12:57 am

Re: EDIFF can't be recognized

#3 Post by gu301purdue » Thu Nov 04, 2021 9:29 pm

Thanks for the suggestion!

Here are more input files, it's a super cell for SiO2 (quartz)

Thanks a lot!

**CONTCAR**
Untitled (Generated by CrystalMaker X fo
1.00000000000000
8.5147617699999998 -4.9160000000000004 0.0000000000000000
0.0000000000000000 9.8320000000000007 0.0000000000000000
0.0000000000000000 0.0000000000000000 10.8108000000000004
Si O
24 48
Direct
0.2346809653573951 0.0000000000000000 0.0833333332999970
0.2346809653573951 0.0000000000000000 0.5833333332999970
0.2346809653573951 0.5000000000000000 0.0833333332999970
0.2346809653573951 0.5000000000000000 0.5833333332999970
0.7346809653573952 0.0000000000000000 0.0833333332999970
0.7346809653573952 0.0000000000000000 0.5833333332999970
0.7346809653573952 0.5000000000000000 0.0833333332999970
0.7346809653573952 0.5000000000000000 0.5833333332999970
-0.0000000000000000 0.2346809653573951 0.4166666667000030
-0.0000000000000000 0.2346809653573951 0.9166666667000030
-0.0000000000000000 0.7346809653573952 0.4166666667000030
-0.0000000000000000 0.7346809653573952 0.9166666667000030
0.5000000000000000 0.2346809653573951 0.4166666667000030
0.5000000000000000 0.2346809653573951 0.9166666667000030
0.5000000000000000 0.7346809653573952 0.4166666667000030
0.5000000000000000 0.7346809653573952 0.9166666667000030
0.2653190346426050 0.2653190346426050 0.2500000000000000
0.2653190346426050 0.2653190346426050 0.7500000000000000
0.2653190346426050 0.7653190346426048 0.2500000000000000
0.2653190346426050 0.7653190346426048 0.7500000000000000
0.7653190346426048 0.2653190346426050 0.2500000000000000
0.7653190346426048 0.2653190346426050 0.7500000000000000
0.7653190346426048 0.7653190346426048 0.2500000000000000
0.7653190346426048 0.7653190346426048 0.7500000000000000
0.2038160586340258 0.1341071500547586 0.1411810263541583
0.2038160586340258 0.1341071500547586 0.6411810263541583
0.2038160586340258 0.6341071500547588 0.1411810263541583
0.2038160586340258 0.6341071500547588 0.6411810263541583
0.7038160586340256 0.1341071500547586 0.1411810263541583
0.7038160586340256 0.1341071500547586 0.6411810263541583
0.7038160586340256 0.6341071500547588 0.1411810263541583
0.7038160586340256 0.6341071500547588 0.6411810263541583
0.3658928499452413 0.0697089085792742 0.4745143597541572
0.3658928499452413 0.0697089085792742 0.9745143597541571
0.3658928499452413 0.5697089085792741 0.4745143597541572
0.3658928499452413 0.5697089085792741 0.9745143597541571
0.8658928499452412 0.0697089085792742 0.4745143597541572
0.8658928499452412 0.0697089085792742 0.9745143597541571
0.8658928499452412 0.5697089085792741 0.4745143597541572
0.8658928499452412 0.5697089085792741 0.9745143597541571
0.4302910914207259 0.2961839413659742 0.3078476930541614
0.4302910914207259 0.2961839413659742 0.8078476930541613
0.4302910914207259 0.7961839413659744 0.3078476930541614
0.4302910914207259 0.7961839413659744 0.8078476930541613
0.4302910914207259 0.7961839413659744 0.8078476930541613
0.9302910914207259 0.2961839413659742 0.3078476930541614
0.9302910914207259 0.2961839413659742 0.8078476930541613
0.9302910914207259 0.7961839413659744 0.3078476930541614
0.9302910914207259 0.7961839413659744 0.8078476930541613
0.1341071500547586 0.2038160586340258 0.3588189736458416
0.1341071500547586 0.2038160586340258 0.8588189736458417
0.1341071500547586 0.7038160586340256 0.3588189736458416
0.1341071500547586 0.7038160586340256 0.8588189736458417
0.6341071500547588 0.2038160586340258 0.3588189736458416
0.6341071500547588 0.2038160586340258 0.8588189736458417
0.6341071500547588 0.7038160586340256 0.3588189736458416
0.6341071500547588 0.7038160586340256 0.8588189736458417
0.0697089085792742 0.3658928499452413 0.0254856402458428
0.0697089085792742 0.3658928499452413 0.5254856402458429
0.0697089085792742 0.8658928499452412 0.0254856402458428
0.0697089085792742 0.8658928499452412 0.5254856402458429
0.5697089085792741 0.3658928499452413 0.0254856402458428
0.5697089085792741 0.3658928499452413 0.5254856402458429
0.5697089085792741 0.8658928499452412 0.0254856402458428
0.5697089085792741 0.8658928499452412 0.5254856402458429
0.2961839413659742 0.4302910914207259 0.1921523069458387
0.2961839413659742 0.4302910914207259 0.6921523069458387
0.2961839413659742 0.9302910914207259 0.1921523069458387
0.2961839413659742 0.9302910914207259 0.6921523069458387
0.7961839413659744 0.4302910914207259 0.1921523069458387
0.7961839413659744 0.4302910914207259 0.6921523069458387
0.7961839413659744 0.9302910914207259 0.1921523069458387
0.7961839413659744 0.9302910914207259 0.6921523069458387

**KPoints**
Automatic mesh
0
Gamma
2 2 2
0. 0. 0.
~
**POTCAR** (just list the title here)

TITEL = PAW_PBE Si 05Jan2001
TITEL = PAW_PBE O 08Apr2002

martin.schlipf
Global Moderator
Global Moderator
Posts: 458
Joined: Fri Nov 08, 2019 7:18 am

Re: EDIFF can't be recognized

#4 Post by martin.schlipf » Fri Nov 05, 2021 7:30 am

Sorry, I still cannot reproduce the issue. Can you please attach a zip file containing all the relevant information (see below)? Please submit the setup that fails and tell me if there is anything special in your setup (e.g. OpenMP) that might be relevant.
admin wrote: Mon Sep 24, 2018 1:44 pm What are the Rules and Expectations for this forum?
The VASP team will be taking an active role in moderating the content of this forum. To help you and other users with your post we require some pertinent information about your problem in form of a zip file. Any threads not following the rules of the forum will be deleted, closed, or moved to an appropriate forum.

N.B.: new threads will only appear in the forum after they have been approved by a forum moderator. Please be patient, this may take some time (up to a day).

A thread belongs to this forum if:
  • you experience a runtime error (i.e., the code crashes) or hangup.
  • VASP exits with an "internal error" message.
  • you have a strong suspicion that the output of VASP for a certain calculation is "wrong".
Please read the following guidelines before opening a new thread.

Posting Rules:
  • Note that this forum is only used for bugreports. Before reporting any problems please ensure that you have
    • browsed or searched previously posted problems. Some questions may have been answered already.
    • searched our growing wiki that contains the documentation of VASP.
    • checked the POSCAR file (duplication of ions?, meaningful structure in general?). This is the most common source of reported problems. More information about the POSCAR is found here.
    • checked the INCAR and KPOINTS file for typos. Are the settings plausible? Read more about all INCAR tags on our wiki.
    • checked the POTCAR file for consistency with the POSCAR, (mixing PBE and LDA potentials?, potential of a species is missing?)
  • Provide a report in form of a zip-file in the attachment of your post that contains:
    • all input files of the job, that is POSCAR, INCAR, KPOINTS, POTCAR
    • OUTCAR and stdout of the run
    • your jobscript, if you use one
Please note that zip files smaller than 8 MB are accepted only.

gu301purdue
Newbie
Newbie
Posts: 19
Joined: Fri Jul 02, 2021 12:57 am

Re: EDIFF can't be recognized

#5 Post by gu301purdue » Sun Nov 07, 2021 7:46 pm

Thanks a lot for checking it. I attached a zip file with all the input files. I used ISIF2/4/5 as the input incar file and I attached them there. Right now, EDIFF is marked as "#". The run file was also attached. I don't think there is any special setup. Can you reproduce the problem this time?

Thanks a lot!
You do not have the required permissions to view the files attached to this post.

martin.schlipf
Global Moderator
Global Moderator
Posts: 458
Joined: Fri Nov 08, 2019 7:18 am

Re: EDIFF can't be recognized

#6 Post by martin.schlipf » Mon Nov 08, 2021 8:01 am

Okay, with this I could reproduce the issue. The reason for this error is that the line ending uses Windows format. In Linux a line ending is a single character, whereas in Windows it is two. VASP reads it in the Linux format splitting at the line break and therefore has an excess character from the Windows line break. Then the string is not a number anymore, but a number plus this extra character.

The solution to this issue is to either change to a Linux line ending; if you use vi, you can do this by set ff=unix. Alternatively, you can add some extra characters (whitespace of comment) after the number, so that the line end character is not directly next to the number.

gu301purdue
Newbie
Newbie
Posts: 19
Joined: Fri Jul 02, 2021 12:57 am

Re: EDIFF can't be recognized

#7 Post by gu301purdue » Sun Nov 14, 2021 6:55 pm

Thanks a lot! I edited the input file and now the software can read the parameters. However, I still get the error message like this, and the job didn't run very smoothly. What should I do? Right now, EDIFF is 1e-6, should I change it to a smaller value? What value would be recommended?

Thanks!

|
| EEEEEEE RRRRRR RRRRRR OOOOOOO RRRRRR ### ### ### |
| E R R R R O O R R ### ### ### |
| E R R R R O O R R ### ### ### |
| EEEEE RRRRRR RRRRRR O O RRRRRR # # # |
| E R R R R O O R R |
| E R R R R O O R R ### ### ### |
| EEEEEEE R R R R OOOOOOO R R ### ### ### |
| |
| ZBRENT: fatal error in bracketing |
| please rerun with smaller EDIFF, or copy CONTCAR |
| to POSCAR and continue |
| |
| ----> I REFUSE TO CONTINUE WITH THIS SICK JOB ... BYE!!! <----

martin.schlipf
Global Moderator
Global Moderator
Posts: 458
Joined: Fri Nov 08, 2019 7:18 am

Re: EDIFF can't be recognized

#8 Post by martin.schlipf » Mon Nov 15, 2021 7:01 am

Could you be more specific, what you mean by "didn't run very smoothly", please?

I would try copying the CONTCAR file to POSCAR and continuing. What can also help is to visualize the structure to see if it looks like what you would expect.
Please also check if all of your individual electronic iterations converge. If you see in the OSZICAR file that some reach the maximum number of iterations (NELM = 60 by default), then your forces can be screwed up and your calculation won't converge.

More generally, you need to understand that we try to optimize the ionic positions of many atoms here, so the mixer tries to minimize the forces based on the previous data. If that data contains iterations with spurious results, you will contaminate the path. So you want to make sure that the structures that are visited are at least somewhat reasonable.

Other things you could try include changing POTIM to reduce the initial step size or changing to IBRION = 1 for a different mixer to determine the optimal positions.

gu301purdue
Newbie
Newbie
Posts: 19
Joined: Fri Jul 02, 2021 12:57 am

Re: EDIFF can't be recognized

#9 Post by gu301purdue » Fri Nov 19, 2021 12:58 am

Thanks for the suggestions! We updated a few parameters, but still get the error message that "ZBRENT: fatal error in bracketing, please rerun with smaller EDIFF, or copy CONTCAR to POSCAR and continue". I set up scripts in the run file so CONTCAR can be copied to POSCAR when the first round of job is finished (the next round of job is based on a different ISIF value). However, usually the job will be stopped in the middle. If I manually copy CONTCAR to POSCAR and run the job with this new ISIF value, the job will not stop in the middle and can continue until the error message is there. Not sure what's the reason there.

We also try to see if more CPU cores utilized by the job can solve the problem. However, right now, we can only utilize one core, although we try to set up for 24 cores in our node. The input file is attached as follows. Is there any other parameters that we should include if we would like to use 24 cores? Such as NBANDS?

Input file:

SYSTEM = sio2
NWRITE = 1

#Startparameter for this Run (values of parameters are equalt to default):
ISTART = 0 0=new, 1=wave, 2=same cutoff&latt (1&2 need WAVECAR file)
ICHARG = 2 0=charge from wave, 1=charge from CHGCAR, 2=atom superpos.
LWAVE=.FALSE.

Electronic Relaxation 1
ENCUT=800 planewave cutoff
NELM= 80 max number of electronic steps
EDIFF=1E-8
EDIFFG=-1E-7 force stopping-criterion for geometry steps
NELMIN=4
PREC=Normal

Ionic Relaxation
NSW=1000 max number of geometry steps
IBRION=2 ionic relax: 0-MD 1-quasi-New 2-CG
ISIF=2 (2:force=y stress=y ions=y shape=n volume=n,
ISYM=1 1=use symmetry, 0 = no symmetry
POTIM=0.05 initial time step for geo-opt (increase for soft sys)


Electronic Relaxation 2
IALGO=48 algorithm (8=CG for small, 48=RMM for big systems)
LREAL=A

LPLANE = .TRUE.
NPAR=1
NCORE=24
NSIM=1
~

martin.schlipf
Global Moderator
Global Moderator
Posts: 458
Joined: Fri Nov 08, 2019 7:18 am

Re: EDIFF can't be recognized

#10 Post by martin.schlipf » Tue Nov 23, 2021 7:03 am

I realize that you use a large number of steps (NSW). How many steps does it take for you to run into this problem?

martin.schlipf
Global Moderator
Global Moderator
Posts: 458
Joined: Fri Nov 08, 2019 7:18 am

Re: EDIFF can't be recognized

#11 Post by martin.schlipf » Tue Nov 23, 2021 7:16 am

I moved your question about the parallel setup into a new topic

Locked