LEPSILON vs LOPTICS

[dekoning]

I'm trying to determine both the static and frequency-dependent dielectrical matrix of a sample of water molecules using, respectively, a calculation with LEPSILON=.TRUE. and another with LOPTICS=.TRUE. In the latter, I set NEDOS=2000 and use a variety of value of NBANDS, ranging between 2 and 3 times the VASP default, as recommended in the VASP manual. All the other parameters are the same for both the LEPSILON and LOPTICS calculations.

The manual also recommends verifying that the zero-frequency value obtained using the LOPTICS calculation should exactly match the static value obtained from the LEPSILON calculation. Here is the problem. They are NOT the same. the LOPTICS calculation gives around 1.5 (with variations only in the second digit by increasing NBANDS), whereas the LEPSILON calculation gives 1.95, which is significantly higher.

Any idea as to the origin of this discrepancy?

[wahyu]

The f=0 real dielectric from LOPTICS should match with the MICROSCOPIC dielectric from LEPSILON. In all my calculations, the micro > macro in the LEPSILON case, and by incresing NBANDS I get smaller and smaller value pf f=0 real dielectric in the LEPSILON case, so I am not sure about your case cause your LOPTIC is already smaller than the LEPSILON case....