HSE functional

tkuehne tku... at gmail.com
Wed Aug 29 19:52:14 CEST 2007


There had been a bug in the 1.1.3 version in the calculation of the
ERI derivatives, in particular d1hrr_order_*.
Edward fixed that and promised a bug-fix available within this day.

I don't know if this is of interest, but a very pragmatic (quick-and-
dirty) way to circumvent the call to C_F_PROCPOINTER is to do it in an
external C++ wrapper. In my case I made all cases explicit (just 200
lines).
This way xlf accepts that.

MfG,
Thomas

On Jul 31, 11:18 pm, "Nichols A. Romero" <naro... at gmail.com> wrote:
> Manuel,
>
> Which version of the libint library are you using? Had some trouble
> compiling 1.1.3
>
> Thanks,
>
> On 7/30/07, mguidon <manuel... at gmail.com> wrote:
>
>
>
>
>
> > Hi.
>
> > The HSE functional is a hybrid functional. It consists of screened
> > exact Hartree-Fock exchange and two GGA-functionals (PBE and hole-
> > averaged screened PBE functional). The keywords are (see /tests/QS/
> > regtest-hybrid/NE-hybrid-HSE06-lda.inp):
>
> > &XC_FUNCTIONAL
> > &XWPBE
> >     SCALE_X -0.25
> >     SCALE_X0 1.0
> >     OMEGA 0.11
> >   &END
> >   &PBE
> >      SCALE_X 0.0
> >      SCALE_C 1.0
> >   &END PBE
> > &END XC_FUNCTIONAL
> > &HF
> >    EPS_SCHWARZ 1.0E-10
> >    MAX_MEMORY  10
> >    FRACTION 0.25
> >    SCREENING_TYPE SHORTRANGE
> >    OMEGA 0.11
> > &END
>
> > As in the case of B3LYP the functional is quite expensive since exact
> > exchange has to be calculated. However, the keyword OMEGA decreases
> > the cost for the Hartree-Fock part because it results in a better
> > screening. To get accurate results it is recommended to use OMEGA
> > values within the range of 0.1 and 0.2. (Higher value means better
> > screening means faster execution). In the current version, analytical
> > derivatives up to second order are implemented which means that forces/
> > stresses are available.
> > As with all calculations including Hartree-Fock exchange you need an
> > up-to-date g95 compiler together with the libint-library in order to
> > run cp2k.
>
> > Cheers
>
> > Manuel Guidon
>
> --
> Nichols A. Romero, Ph.D.
> High Performance Technologies, Inc.
> Reston, VA
> 443-567-8328 (C)
> 410-278-2692 (O)




More information about the CP2K-user mailing list