Computing static electro-optic coefficients by GGA+NC

Hello everyone,
The flexoelectricity is now available within the GGA+NC in Abinit v9.6, however, GGA+NC psp does work for static electro-optic coefficients (for example: tnlo_2.abi and tnlo_6.abi) and gives some unreasonable energy values.

Please give me some suggestions. Thank you.

Qiang Li

Hi Qiang,

The implementations are quite independent. The terms needed for the flexoelectricity are somehow also 3rd derivatives but the q derivative is different from the R and E field ones needed for Raman and NLO. I thought Raman had been generalized to GGA as well (no SOC, right?) - perhaps only with PAW? The strange thing is that it should complain and stop, not give random results, if the things is not implemented.

You are simply copying tnlo but changing the psp8 files?

Dear mverstra,
I run the example (tnlo_2.abi and tnlo_6.abi) with GGA+NC, and the abinit shows the message: Third-order xc kernel can only be computed for ixc = 0, 3, 7 or 8.
As you said, there are differences in calculation methods of flexoelectricity and NLO .
Using LDA+NC is good for computing static electro-optic coefficients.
Alternative, frequency-dependent electro-optical susceptibility can be calculated using LDA+NC and GGA+NC.

Thank you.
Q Li