User Tools

Site Tools


howto:troubleshooting

Troubleshooting

Something going wrong ? Better to realize this than to think everything is OK …

In this page, the user will find tips to fix several problems with ABINIT.

Many other tips might be found on the forum, or in the standard ABINIT documentation.

ABINIT does not run

(to be filled …)

ABINIT stops without finishing its tasks

Make sure you get and read the error messages. They are usually written at the end of the log file, and look like:

--- !ERROR
src_file: m_kg.F90
src_line: 179
mpi_rank: 1
message:   (here follows the error message)

If you do not see such error message at the end of the log file, the error message might be found alternatively in an “error” file, depending on the command that you used to launch ABINIT. Also, if you are running in batch, the behaviour might be different, and error messages might end up in another file than the log file. If you run in parallel, the error messages might have been written in still another, separate file, entitled

__ABI_MPIABORTFILE__

If ABINIT stops quickly, consider to run in sequential, interactively (not in parallel neither in batch) to ease the understanding of what is happening.

Sometimes, the error message is prepared by another message giving preliminary information. Please, be sure to have identified the relevant information in the log file.

If you think to have obtained all information that ABINIT was supposed to give you, try to identify whether ABINIT stops because of geometry optimization problem or SCF problem. In the first case, your input geometry is perhaps crazy. Anyhow, see the next items in this troubleshooting page.

Incorrect initial geometry

Many mistakes done by beginners are related to incorrect starting geometry.

Here is a check list :

  • Check that the units are correct for your cell parameters and atomic positions. ABINIT uses atomic unit by default, but can use several other units if specified by the user, see ABINIT input parameters.
  • Check that the typat atom types are correct with respect to xred or xcart
  • Check that the number of atoms natom is coherent with your list of coordinates, xred or xcart. ABINIT reads only the coordinates of natom nuclei, and ignore others.
  • Relax first the atomic positions at fixed primitive vectors before optimizing the cell. Explicitly, use a first datadet with optcell = 0, then a second dataset with non-zero optcell, in which you tell ABINIT to read optimized atomic positions using getxred or getxcart. In this second dataset, do not forget to usedilatmx bigger than 1 if you expect the volume of the cell to increase during the optimization. See the additional suggestions in the documentation of optcell .
  • ABINIT can read VASP POSCAR external files containing unit cell parameters and atomic positions. See the input variable structure. This might help in setting the geometry correctly.
  • Try to visualize your primitive cell. There are numerous possibilities, using Abipy, VESTA, XCrysDen (to be documented).

SCF cycle does not converge

(to be filled)

Geometry optimization does not converge

(to be filled)

ABINIT delivers strange results

(to be filled)

howto/troubleshooting.txt · Last modified: 2020/08/05 16:05 by Jean-Michel Beuken