Abnormally High Ionic Polarization

[mashton]

I'm calculating the polarization of BiFeO3 unit cells under varying amounts of epitaxial strain using LCALCPOL = .TRUE.

For the unstrained unit cell, the ionic polarization seems normal (~7.5 electrons Angst, from the OUTCAR file), but for all strained calculations, it becomes abnormally high (~27 electrons Angst).

I'm "straining" the cells by constraining their POSCARS to compressed values in the x and y directions (~ -1% strain). All calculations, including the initial relaxation, are being done with an LDA+U correction.

Can anyone please explain why the ionic contribution to the polarization is so high for these compressed unit cells? Am I doing something wrong?

[mashton]

In case anyone has this issue in the future, I discovered that the Bi atom at (0 0 0) was traversing to the other end of the box (0.999 0.999 0.999) during the relaxation. After fixing its position the issue was resolved.