Hello,
i have a question regarding the implememtation of vdW density functional. I was wondering which van der Waals exchange-correlation functional one can use at this moment in ABINIT, in order to have the 'vdw_xc' input variable working properly.
Thanks for your attention,
Valentina
Van der Waals implementation
Re: Van der Waals implementation
Hi,
during this time i've tried to solve this problem by generating and then using APE's pseudopotentials. If i use the 'usual' LDA (PW) pseudopotential, i will get the right acell, and the same happens for GGA (PBE). When i use the GGA composed by PBE for correlation and revPBE for exchange, then ixc= -102130, the cell parameter is just a bit over-estimated. Then, once i try the psp needed for the vdw_xc variable :
vdw_xc 1 ixc -102012 and psp= LDA_PW_correlation + revPBE_exchange
vdw_xc 2 ixc -108012 psp= LDA_PW_correlation + PW86_exchange
I do not get the right acell (i started from checking the acell as a check for the reliability of the psp), but it is in both cases really under-estimated. This scenario doesn't change if i don't use the vdW variable. I tried to set ixc = -108012 with psp= 'usual' LDA (PW) without setting vdw_xc, and what i've got is the same bad result. Then i tried with ixc -102 with psp= 'usual' LDA (PW) without setting vdw_xc and in this case the calculation went wrong, (The new primitive vectors rprimd (an evolving quantity) are too large with respect to the old rprimd and the accompanying dilatmx :this large change of unit cell parameters is not allowed by the present value of dilatmx.). My last attempt was to set ixc -102 and using as psp LDA_PW_correlation + revPBE_exchange, but again, i've got the previous error. I was using ixc -102 because it was the value that was given as an example in the share/abinit/vdwxc directory of abinit. So, now, i am wondering about what to do/try next, because, in my opinion, it is the libxc that seems having problems, and not the psp, but i might be wrong.
I am looking forward to hear an answer,
best regards,
Valentina
during this time i've tried to solve this problem by generating and then using APE's pseudopotentials. If i use the 'usual' LDA (PW) pseudopotential, i will get the right acell, and the same happens for GGA (PBE). When i use the GGA composed by PBE for correlation and revPBE for exchange, then ixc= -102130, the cell parameter is just a bit over-estimated. Then, once i try the psp needed for the vdw_xc variable :
vdw_xc 1 ixc -102012 and psp= LDA_PW_correlation + revPBE_exchange
vdw_xc 2 ixc -108012 psp= LDA_PW_correlation + PW86_exchange
I do not get the right acell (i started from checking the acell as a check for the reliability of the psp), but it is in both cases really under-estimated. This scenario doesn't change if i don't use the vdW variable. I tried to set ixc = -108012 with psp= 'usual' LDA (PW) without setting vdw_xc, and what i've got is the same bad result. Then i tried with ixc -102 with psp= 'usual' LDA (PW) without setting vdw_xc and in this case the calculation went wrong, (The new primitive vectors rprimd (an evolving quantity) are too large with respect to the old rprimd and the accompanying dilatmx :this large change of unit cell parameters is not allowed by the present value of dilatmx.). My last attempt was to set ixc -102 and using as psp LDA_PW_correlation + revPBE_exchange, but again, i've got the previous error. I was using ixc -102 because it was the value that was given as an example in the share/abinit/vdwxc directory of abinit. So, now, i am wondering about what to do/try next, because, in my opinion, it is the libxc that seems having problems, and not the psp, but i might be wrong.
I am looking forward to hear an answer,
best regards,
Valentina