The ABINIT Users Mailing List ( CLOSED )

[z.wu@gl.ciw.edu]

Dear developers,

Recently I implemented a new xc functional (PRB 73, 235116). Basically I 
implemented fx, dfxdss and d2fxdss2 in xcpbe.F90 for the new functional, also 
other programs driving xcpbe. I think it should be enough for even linear 
response, and I checked dfxdss and d2fxdss2 by finite difference. The 
calculated force agrees well with energy finite difference. However the 
second energy derivative doesn't mach: the direct computation gives 177.17, 
whereas the finite difference of gradients gives 189.49.

Could you give me any hints that any other subroutines (such as those in 
16response directory) need to be modified also? If you want to implement a 
new xc functional, what you would be doing? By the way I am working on the 
version 5.3.4.

Many thanks!

Zhigang