ABINIT Discussion Forums

Dear sir
I use atompaw generated B88 exchange and LYP correlation everything was fine, after that I use BLYP in abinit to run acell we found that
and I got warning see below
mkdenpos : WARNING -
Density went < 0 at 60441 points
and was set to 1.00E-14. Lowest was -0.10E+01.
Likely due to too low boxcut or too low ecut for pseudopotential core charge.
Functional 'Lee, Yang & Parr' does not provide an implementation of fxcFunctional 'Lee, Yang & Parr' does not provide an implementation of fxcAPPLICATION TERMINATED WITH THE EXIT STRING: Hangup (signal 1)

What does it mean
Best Regards
Abdulmutta

You are getting this message because abinit does not support hybrid xc functionals like BLYP yet. As atompaw is linkable to libxc, you can make PAW datasets based on all kinds of different xc functionals, but abinit itself does not support all the xc functionals that libxc provides. Please read carefully the input variable documentation for "ixc" for the version of abinit you are using. The latest version of abinit can use LDA, GGA, and meta-GGA xc's (though again, check "ixc" to determine which ones). Support for hybrid xc's like BLYP is I believe under development. There is the capability to use the hybrid xc PBE0 in an approximate (though in my experience, very effective) way, through use of the "useexexch" input variable.

Thank you sir
For PBE0, I must be used PBE pp files right
and then set Exact Exchange as an input like
useexexch 1
lexexch 2 -1
exchmix 0.25

Best.
Abdulmutta

Dear All,

I am afraid that there is a confusion here. As I know, being a combination of the B88 exchange and the LYP correlation, the BLYP functional is a GGA one, whereas, e.g., the B3LYP functional that contains the HF exact-exchange part and contributions coming from GGA and LDA functionals is a hybrid one. As regards the GGA BLYP functional, I would like to ask a question that is of vital importance to me. Several months ago, we performed numerous calculations with the use of abinit 6.6.2. We have obtained results on structure optimization and respective energy bands for a wide spectrum of the exchange-correlation functionals (including the GGA BLYP and AM05). As it has been recommended, we used atompaw and abinit built with libxc. We generated PAW's with, e.g., ixc=-106131 and after that with the same ixc we ran abinit. In accordance with the documentation (I mean the description of possible values for ixc), there was no Warning message telling us that this or another fxc is not implemented or supported yet. However, recently we tested the new version of abinit and surprisingly have found that ''Lee, Yang & Parr' does not provide an implementation of fxc'. We have the same for 'Armiento & Mattsson 05'. Does it mean that in the 6.6.2 version were was no, say, stopper for the respective ixc values? If it is so, can I say that our results obtained with the GGA BLYP and AM05 are not valid?

Thanks in advance.

any news on adding hybrid xc functionals into abinit?