VASP Forum

Dear VASP admin,

When using LDIPOL for molecules or slabs in VASP, does the molecule or slab need to be placed at a specific position inside the simulation cell?

We are confused by the following statement in the VASP wiki https://www.vasp.at/wiki/Electrostatic_corrections: "Note that the system has plenty of vacuum on either side. This empty space is important for the potential corrections needed for the LDIPOL tag."

Many thanks in advance,
Ibrahim

VASP has a built-in mechanism to determine the position where the center of the cell is. We recommend that you set it with the tag DIPOL manually if you want to be more precise. Otherwise you need to make sure that VASP got it correct

Thanks for your reply.

- That means that for LDIPOL tag, there is no specific position inside the simulation cell, and we should instead take care of the cell center defined by the DIPOL tag, right?

- How can we make sure that VASP interpreted it correctly when the default value is used? When we checked the XML files, we found <v name="DIPOL"> -100.00000000 -100.00000000 -100.00000000 </v>. Also, the SCF procedure was stable and the calculations converged.

Many thanks in advance,
Ibrahim

If you do not set the DIPOL flag, VASP will report something like this

 DIPCOR: dipole corrections for dipol
 direction  3 min pos    58,

in the OUTCAR file. The position (here 58) tells you where VASP put the dipole. If you do set DIPOL, VASP will use it to compute position from that. The position is given as a grid coordinate.

Thanks again for your reply.

In fact, we tested one of our systems with and without DIPOL and with the cluster is centered and when it is shifted away from the center.

We observed the following:

- When the cluster is centered and DIPOL is set to (0.5, 0.5, 0.5), the calculation converges.

- When the cluster is centered and DIPOL is not specified in the INCAR file, the calculation does not converge.

- When the cluster is shifted away from the center and DIPOL is not specified in the INCAR file, the calculation converges.

So, if the calculation converges, does this mean that VASP has correctly interpreted the DIPOL tag and that the dipole correction has been correctly applied?

Many thanks in advance,
Ibrahim

If you set DIPOL in the INCAR file, VASP knowns where the center of the cluster is and will always use the correct position. Otherwise it will guess where it is by looking at the density. If can make a mistake and that then leads to the calculation not converging. Think of it this way: You should typically tell VASP where the center of the cell is, but if you forget, VASP will try to converge the calculation anyway using reasonable defaults.

You can see that in the output that you produced by looking at the line that I had suggested echo $file; grep 'direction.*min pos' $file | tail -n 5. If you set DIPOL in the INCAR file, VASP does not need to guess where the center is and always uses the same position. In the case that does not converge, VASP guesses very different positions in every iteration. In contrast, when VASP finds the correct position like in the last case, the calculation converges.

DIPOL-position-centered/OUTCAR
 direction  1 min pos     1, direction  2 min pos     1, direction  3 min pos     1,
 direction  1 min pos     1, direction  2 min pos     1, direction  3 min pos     1,
 direction  1 min pos     1, direction  2 min pos     1, direction  3 min pos     1,
 direction  1 min pos     1, direction  2 min pos     1, direction  3 min pos     1,
 direction  1 min pos     1, direction  2 min pos     1, direction  3 min pos     1,

No-DIPOL-position-centered/OUTCAR
 direction  1 min pos   120, direction  2 min pos   193, direction  3 min pos    98,
 direction  1 min pos   183, direction  2 min pos   224, direction  3 min pos   243,
 direction  1 min pos   256, direction  2 min pos   130, direction  3 min pos    67,
 direction  1 min pos   168, direction  2 min pos    86, direction  3 min pos    45,
 direction  1 min pos   159, direction  2 min pos    82, direction  3 min pos    42,

No-DIPOL-position-shifted/OUTCAR
 direction  1 min pos    97, direction  2 min pos    86, direction  3 min pos    94,
 direction  1 min pos    96, direction  2 min pos    86, direction  3 min pos    93,
 direction  1 min pos    95, direction  2 min pos    86, direction  3 min pos    92,
 direction  1 min pos    94, direction  2 min pos    87, direction  3 min pos    92,
 direction  1 min pos    93, direction  2 min pos    87, direction  3 min pos    92,

Thank you very much for the detailed reply.

martin.schlipf wrote: Wed May 13, 2026 6:33 am

If you set DIPOL in the INCAR file, VASP does not need to guess where the center is and always uses the same position.

Does this have an effect in the case of ionic optimization, where the center of the cluster can change relative to its initial position?

Many thanks in advance,
Ibrahim

VASP will not included a force contribution from the DIPOL moving in the cell while in principle it should. But if that would lead to a force then the dipole correction is not a good approximation.

martin.schlipf wrote: Thu May 14, 2026 4:44 pm

VASP will not included a force contribution from the DIPOL moving in the cell while in principle it should. But if that would lead to a force then the dipole correction is not a good approximation.

Thank you.

Could you please clarify this point further?
We would set the tag DIPOL manually in INCAR to be more precise, as you recommended. Should we be concerned about this during ionic optimization?

Many thanks in advance,
Ibrahim

Also, I f we restarted the calculation with CONTCAR, should every time DIPOL is modified in INCAR.

Many thanks in advance,
Ibrahim

In principle you should change DIPOL but that should almost never necessary in practice. It is only required if the atoms move by such a large amount that was the center of the cluster is not anymore. It does not need to be the exact center, approximately in the center is fine.