optimization and evaluation of results in the GW calculation

[ErkanTetik]

Dear friends,
I run two structures with the GW approximation. The first results,
E^0_gap 0.280
E^GW_gap 0.001
DeltaE^GW_gap -0.277
in this case, can we say this structure show metallic properties with GW and this structure has semiconductor band gap without GW?
next results,
E^0_gap 0.691
E^GW_gap -0.166
DeltaE^GW_gap -0.857
GW_gap is negative, in this case, can we say this structure certainly has metallic properties?

Finally, I examine first tutorial on GW, there is an explanation in the convergence studies in the tutorial,
---can be considered converged within 0.01eV.
which the 0.01 eV value represents the difference between the parameters?

[david.waroquiers]

Hello,

For the convergence studies, you have to check that each parameter of the calculation (nband, ecuteps, ....) is high enough so that your results do not change by more than 0.01 eV if you were to run a calculation with a VERY high parameter (you will usually see an asymptotic behaviour).

For a metal, dont forget you need to increase the number of k-points a lot.


Hope it helps...

David