<br> <a href=3D"http://www.democritos.it/pipermail/pw_forum/2008-September/01012= 3.html" target=3D"_blank">http://www.democritos.it/pipermail/pw_forum/2008-= September/010123.html</a><br> <a href=3D"http://www.democritos.it/pipermail/pw_forum/2010-February.txt"; t= arget=3D"_blank">http://www.democritos.it/pipermail/pw_forum/2010-February.= txt</a><br> <br> I understand that the third derivatives were never implemented and<br> that only the LDA part was calculated, for the Perdew Zunger<br> parameterization. Is this third derivative the third derivative that<br> is mentioned in the routine PH/raman_mat.f90 (i.e, the third<br> derivative of the energy with respect to the atomic displacements and<br> with respect to two electric fields, as described in Lazzeri and<br> Mauri, PRL 90,036401)?<br> <br> I am interested in getting a feel for what would be needed to<br> implement the third derivatives with GGA into the code, as well as<br> what sort of errors one might expect when using the third-order term<br> as currently coded with LDA pseudopotentials (comments in the forums<br> have hinted that the errors are likely small). =A0The shift in the Raman<br= > spectra for cd-Si is pretty minor when using a GGA pseudo compared to<br> that of an LDA calculation, but I wanted to be able to quantify the<br> error better for other systems where comparing the GGA raman spectra<br> to one calculated within LDA was not possible (e.g., when the system<br> doesn't exhibit a gap in LDA).<br> <br> Thanks for the help!<br> <br> Best,<br> Brad<br> UC Berkeley<br> </blockquote></div><br></div>
--000e0cd1e7da1a89630494a936f2--
