Page 1 of 1
LAPACK: Routine ZPOTRF failed!
Posted: Mon Sep 23, 2024 10:51 am
by triggin5
I am receiving the following error: LAPACK: Routine ZPOTRF failed! INFO:211 KPOINT:1 SPIN:1, when I attempt to run a job. I tried changing the ALGO as suggested from previous suggestions within the forum but to no avail. Are there any suggestions to fix this error?
Re: LAPACK: Routine ZPOTRF failed!
Posted: Mon Sep 23, 2024 12:47 pm
by martin.schlipf
Please share the input and output files of your calculation. Often this error is related to an incorrect setup. Did VASP print any warning or error message before the calculation crashed?
Re: LAPACK: Routine ZPOTRF failed!
Posted: Tue Sep 24, 2024 1:43 pm
by triggin5
It did warn of small ion distances. I have attached the input and output files.
4c1.zip
Re: LAPACK: Routine ZPOTRF failed!
Posted: Tue Sep 24, 2024 8:57 pm
by martin.schlipf
Your atoms sit on top of each other. I think you wanted to define the positions in fractional coordinates but in the POSCAR file it says Cartesian instead. It is generally a good idea to look at the structure of the calculation to spot these issues. You can also run VASP in dry run mode and look at the neighbor list. In your case, you will find the lines
Code: Select all
ion position nearest neighbor table
1 0.007 0.004 0.000- 26 0.15 2 0.20 6 0.20 7 0.28 31 0.30 27 0.34 11 0.40 3 0.40
32 0.43 8 0.45 12 0.45 36 0.49 28 0.54 13 0.57 37 0.57 53 0.59
in the OUTCAR file indicating that distance between atom 1 and 26 is only 0.15Å. Please replace Cartesian by fractional and then look at the structure to make sure all the atoms are where you expect them to be.
On an unrelated note, please do not set both NPAR and NCORE in the INCAR file. VASP will automatically determine the appropriate choice for the other parameter and overwrite your choices if you set both. We recommend to choose NCORE based on the available cores per node of your system (cores per node / NCORE should be an integer).