I have compiled abinit-7.0.5 against libxc-2.0.1, and I am using the following input file to calculate the energy binding of the water dimer using the meta-GGA VS98 XC-functional:
acell 15 15 15 Angstrom
ntypat 2
znucl 8 1
natom 6
typat 1 1 2 2 2 2
usekden 1
xcart
-1.470106000 -0.118885000 -0.038009000
1.427463000 0.016250000 -0.106185000
-0.512677000 -0.061567000 -0.142015000
-1.810870000 -0.005544000 -0.924779000
1.765972000 -0.792385000 0.282511000
1.704352000 0.706974000 0.498617000 Angstrom
chksymbreak=0
ecut 60.0
ixc=-204232
kptopt 0
nkpt 1
nstep 10
toldfe 1.0d-6
diemac 2.0
optforces 1
where the pw basis is combined with the O and H PBE-Goedecker pseudopotentials found in http://cvs.berlios.de/cgi-bin/viewvc.cgi/cp2k/potentials/Goedecker/abinit/pbe/.
The calculation finishes giving a positive Etotal energy, and I would like to ask you the reason why. Do you see something wrong or missed in the input file? Are there any pseudopotentials tuned for this functional?
It is worth to mention that I have used the same input and pseudos to make a PBE calculation (i.e. just changing ixc to 11), and the result perfectly matches the one coming from an all-electron calculation (4.96 kcal/mol vs 4.95 kcal/mol for PBE/Aug-CC-pV5Z). Besides, I have also tried ixc=-204 (only X-VS98) and the corresponding binding energy is not bad (4.59 vs 4.31 kcal/mol).
Thanks in advance,
Pablo