Page 1 of 1

ERROR FEXCP: supplied Exchange-correletion table is too small

Posted: Mon Nov 22, 2021 1:58 am
by musanna_galib
I am trying to relax alucone structure on the Zn surface. However, the bonds are breaking. I was informed of the error message :

ERROR FEXCP: supplied Exchange-correletion table is too small, maximal index : 3850

What should I do to solve this problem? Could anyone give some advice ? Thanks

Re: ERROR FEXCP: supplied Exchange-correletion table is too small

Posted: Mon Nov 22, 2021 7:03 am
by henrique_miranda
I cannot reproduce the issue because the POSCAR file is missing.
However, I found this warning in your OUTCAR file:

Code: Select all

 -----------------------------------------------------------------------------
|                                                                             |
|           W    W    AA    RRRRR   N    N  II  N    N   GGGG   !!!           |
|           W    W   A  A   R    R  NN   N  II  NN   N  G    G  !!!           |
|           W    W  A    A  R    R  N N  N  II  N N  N  G       !!!           |
|           W WW W  AAAAAA  RRRRR   N  N N  II  N  N N  G  GGG   !            |
|           WW  WW  A    A  R   R   N   NN  II  N   NN  G    G                |
|           W    W  A    A  R    R  N    N  II  N    N   GGGG   !!!           |
|                                                                             |
|     The distance between some ions is very small. Please check the          |
|     nearest-neighbor list in the OUTCAR file.                               |
|     I HOPE YOU KNOW WHAT YOU ARE DOING!                                     |
|                                                                             |
 -----------------------------------------------------------------------------
Please have a look if this is the case.
You are doing an ionic relaxation, if the ions are too close you might get too large forces which might affect the convergence of the subsequent steps.
Looking at the CONTCAR (the intermediate structure), it seems that something is wrong in the positions of the atoms.

In general, before posting it is recommended that you search if the question was already answered sometime in the forum.
Here are a few related posts about the same issue:
forum/viewtopic.php?f=4&t=18042
forum/viewtopic.php?f=4&t=1491

Those refer to cell relaxation but there it is recommended to increase ENCUT to suppress this error.
If this does not solve the issue in your case then please report back to us.

Re: ERROR FEXCP: supplied Exchange-correletion table is too small

Posted: Mon Nov 22, 2021 6:51 pm
by musanna_galib
Hello Henrique,

Thanks for the reply. I increased the ENCUT in a trial. But it didn't help! I also tried with bond lengths from previous papers. In this case, also bonds are breaking. The errors here are

Code: Select all

 -----------------------------------------------------------------------------

|                                                                             
|     internal error in: us.F  at line: 1339                                  
|                                                                             
|     internal ERROR SETYLM_AUG: 6 -8 24 56 5 24 -9                           
|                                                                             
|     If you are not a developer, you should not encounter this problem.      
|     Please submit a bug report.       

                       
Note: The following floating-point exceptions are signalling: IEEE_UNDERFLOW_FLAG IEEE_DENORMAL
STOP 1
STOP 1
--------------------------------------------------------------------------
mpirun detected that one or more processes exited with non-zero status, thus causing
the job to be terminated. The first process to do so was:

  Process name: [[40062,1],15]
  Exit code:    1
--------------------------------------------------------------------------
I am adding all files here. Hope you will be able to reproduce now! Looking forward for your reply. Thanks!

Re: ERROR FEXCP: supplied Exchange-correletion table is too small

Posted: Mon Nov 22, 2021 9:20 pm
by henrique_miranda
Ok, now I see the structure that you started your ionic relaxation with in the POSCAR file and it looks better than the CONTCAR from the previous set of files that you've sent.
But if you look at the CONTCAR file of this run you will see that the structure is all over the place.

This might be because your structure is still far away from equilibrium, you get large forces and the POTIM is possibly too large and then you get a nonsensical structure.
Try reducing POTIM and see if that helps

Also, have a look at the documentation of the IBRION INCAR tag:
wiki/index.php/IBRION
There it is suggested to try IBRION=2 for cases where you are far away from equilibrium.

Re: ERROR FEXCP: supplied Exchange-correletion table is too small

Posted: Thu Nov 25, 2021 9:55 pm
by musanna_galib
Hello Henrique,

Thanks for the suggestion. I changed POTIM to 0.015 from 0.5 (default) and IBRION to 2. But these didn't help to get the relaxed structure (CONTCAR attached). I am adding the files here. Can you kindly give some suggestion on it? Thanks.

got core.14376 file. The error.txt gives -

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:
mlx5: se101: got completion with error:
00000000 00000000 00000000 00000000
00000000 00000000 00000000 00000000
00000022 00000000 00000000 00000000
00000000 00008813 08000662 cd9c52d2
[[9278,1],95][btl_openib_component.c:3641:handle_wc] from se101 to: se188 error polling LP CQ with status REMOTE ACCESS ERROR status number 10 for wr_id a82d680 opcode 1 vendor error 136 qp_idx 1
#0 0x2aaab38933ff in ???
#1 0x2aaab4d7cc00 in ???
#2 0x2aaab4d7db1a in ???
#3 0x2aaab4d1573b in ???
#4 0x2aaab2435c14 in ???
#5 0x2aaab2491502 in ???
#6 0x2aaab244b507 in ???
#7 0x2aaab21a9d98 in ???
#8 0x440f77 in ???
#9 0x4462b5 in ???
#10 0x903e1d in ???
#11 0x943348 in ???
#12 0xed0c05 in ???
#13 0xeb7796 in ???
#14 0xed1269 in ???
#15 0x2aaab387f554 in ???
#16 0x41d4da in ???
#17 0xffffffffffffffff in ???

Re: ERROR FEXCP: supplied Exchange-correletion table is too small

Posted: Fri Nov 26, 2021 7:13 am
by henrique_miranda
I am sorry you are experiencing this problem.
This might be an internal issue in VASP or some issue with the compilation/hardware.
I will try to have a look into it.

In the meantime:
It seems that the relaxation is proceeding correctly and a new CONTCAR is being produced after each ionic step.
You can restart the calculation from this newly relaxed structure by copying CONTAR to POSCAR and running again.

Notice however that there are very large differences between the initial POSCAR and CONTCAR which likely means that your initial structure is not very good.
You might be able to relax it regardless of that but it might take unnecessarily many ionic steps.

Re: ERROR FEXCP: supplied Exchange-correletion table is too small

Posted: Fri Nov 26, 2021 7:53 pm
by musanna_galib
Thanks for the reply.
ISTART =1 requires reading orbitals from wavecar. But as the job crushes, the wavecar is not written. How can I restart in this case?

Also, can you give some idea how can I get a more stable structure to start relaxation? Is there anything I am missing here?

Re: ERROR FEXCP: supplied Exchange-correletion table is too small

Posted: Fri Nov 26, 2021 8:59 pm
by henrique_miranda
You can restart without the WAVECAR using ISTART=0.

You can try to set up the structure based on experimental parameters if you have them.

Re: ERROR FEXCP: supplied Exchange-correletion table is too small

Posted: Thu Dec 23, 2021 12:23 am
by musanna_galib
Hello Henrique,

Have you looked into the problem ---Program received signal SIGSEGV: Segmentation fault - invalid memory reference? I am still facing this.

Code: Select all

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:
mlx5: se199: got completion with error:
00000000 00000000 00000000 00000000
00000000 00000000 00000000 00000000
00000022 00000000 00000000 00000000
00000000 00008813 08000662 cfe25fd2
[[41037,1],95][btl_openib_component.c:3641:handle_wc] from se199 to: se200 error polling LP CQ with status REMOTE ACCESS ERROR status number 10 for wr_id 9d2fb80 opcode 1  vendor error 136 qp_idx 1
#0  0x2aaab38933ff in ???
#1  0x2aaab4d7cc00 in ???
#2  0x2aaab4d7db1a in ???
#3  0x2aaab4d1573b in ???
#4  0x2aaab2435c14 in ???
#5  0x2aaab24906a1 in ???
#6  0x2aaab249195d in ???
#7  0x2aaab244b507 in ???
#8  0x2aaab21a9d98 in ???
#9  0x440f77 in ???
#10  0x4462b5 in ???
#11  0x903e1d in ???
#12  0x943348 in ???
#13  0xed0c05 in ???
#14  0xeb7796 in ???
#15  0xed1269 in ???
#16  0x2aaab387f554 in ???
#17  0x41d4da in ???
#18  0xffffffffffffffff in ???