Error: SHAKE algorithm did not converge!

[duccoi]

Hi everyone,
I am running MD calculations with the blue moon sampling method to predict substitution reaction for a large system of 120 atoms with one k-point. However, the job runs 200 steps, then the error ''SHAKE algorithm did not converge!" appears. Can you suggest to me what to do in this situation?

Code: Select all

NCORE=8
AMIN=0.01
PREC=Low
EDIFF=1e-4
LWAVE=.FALSE.
LCHARG=.FALSE.
NELMIN=4
NELM=500
LREAL=.FALSE.
ALGO=VeryFast
LREAL= Auto
ISYM=0
ISMEAR=-1
SIGMA=0.1
   
############################# MD setting ##################################
IBRION=0                                 # MD simulation
NSW=10000                                 # number of steps
POTIM=2                                  # integration step
TEBEG=600                                # simulation temperature
MDALGO=11                                # Andersen thermostat
ANDERSEN_PROB=0.10                       # collision probability
LBLUEOUT=.TRUE.                          # write down output needed to 
                                         # compute free-energy gradient
##############################################################################]



Error: SHAKE algorithm did not converge! err=  5.299201546233423E-002
 -----------------------------------------------------------------------------
|                                                                             |
|     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     ###     ###     ###     |
|                                                                             |
|     Error too large, I have to terminate this calculation!                  |
|                                                                             |
|       ---->  I REFUSE TO CONTINUE WITH THIS SICK JOB ... BYE!!! <----       |
|                                                                             |
 -----------------------------------------------------------------------------

[fabien_tran1]

Hi,

Can you please upload the POSCAR and OUTCAR files.

[duccoi]

Sorry, I can not share the POSCAR file due to research purposes. But I could say it is a surface of 20 slabs, having transition metal ions.
Maybe the cell size is related to the problem. Is there any general reason for this error?

Code: Select all

POSCAR1                              
1.0
        5.5554594994         0.0000000000         0.0000000000
       -0.1009658000         5.5665397091         0.0000000000
        0.4677609768         1.3795390124        59.1420671357

[fabien_tran1]

Ok. Could you at least grep for TOTEN in the OUTCAR ("grep TOTEN OUTCAR") and show the values.

[duccoi]

Please help me find the data here. https://drive.google.com/drive/folders/ ... sp=sharing

[fabien_tran1]

The TOTEN file shows 22 steps. Are these the last 22 steps of the 200 that you mentioned in your 1st post?

Actually, is this calculation your 1st attempt or have you already done several attempts with different parameters? For instance, the last sentence at this page: https://www.vasp.at/wiki/index.php/Slow-growth_approach mentions some potential problems and it is suggested to use a less strict value for SHAKETOL.

At the moment it is difficult to make more suggestions, since not much information is provided.

[duccoi]

Thank you.
There are about 20 steps, not 200. Maybe the number of constraints is too much for a large system (I set 8 constraints in ICONST). I am going to reduce the number of constraints.