[CPMD-list] Trying to understand screened exchange B3LYP for PBC solids

Thu Mar 15 23:27:24 CET 2007

Hi Axel,

Thanks for your initial response.

I haven't changed the code in any way. The B3LYP calculation gives a
quite different total energy than a plain BLYP calculation. From what
I've seen, the code at least gives the impression that it is doing
Hartree Fock exchange for k-points. And, the B3LYP energy difference I
get matches an all-electron B3LYP calculation I've done previously in
another code.

It is puzzeling though, that in hfx.F there is the line of code
      IF(TKPNT) CALL STOPGM('HFX','K-POINTS NOT IMPLEMENTED')
But, the code doesn't stop, which could indicate several things I suppose.

In the B3LYP output, the functional description looks reasonable:
 EXCHANGE CORRELATION FUNCTIONALS
 HYBRID FUNCTIONAL
    LDA EXCHANGE:                      SLATER                0.80
    LDA CORRELATION:                   LYP                   1.00
    GC EXCHANGE:                       BECKE88               0.72
    GC CORRELATION:                    LYP                   0.81
    HARTREE-FOCK EXCHANGE:                                   0.20
                        SCREENED EXACT EXCHANGE USING EXP(-A*R)/R
                        SCREENING PARAMETER [BOHR^-1]:      0.010

For BLYP, the functional description is
 EXCHANGE CORRELATION FUNCTIONALS
    LDA EXCHANGE:                        SLATER (ALPHA = 0.66667)
    LDA CORRELATION:                             LEE, YANG & PARR
       [C.L. LEE, W. YANG, AND R.G. PARR, PRB 37 785 (1988)]
    GRADIENT CORRECTED FUNCTIONAL
    DENSITY THRESHOLD:                                1.00000E-08
    EXCHANGE ENERGY
       [A.D. BECKE, PHYS. REV. A 38, 3098 (1988)]
       PARAMETER BETA:                                   0.004200
    CORRELATION ENERGY
       [LYP: C.L. LEE ET AL. PHYS. REV. B 37, 785 (1988)]

So is the code actually doing Hartree Fock exchange for k-points? And,
if it's not actually running screened B3LYP, then what is it running?
Any suggestions on how to find the answers to these questions?

I know a lot of people that would be excited to run the code if it is
actually doing hybrid functionals with k-points.

Thanks for the feedback!

Kevin Driver
The Ohio State University, Physics Dept.
driver at mps.ohio-state.edu

On Thu, 1 Mar 2007, Axel Kohlmeyer wrote:

> On Thu, 1 Mar 2007 driver at mps.ohio-state.edu wrote:
>
> KD> Hello CPMD users,
>
> kevin,
>
> KD> I am using CPMD-3.11 to do some PBC, many k-point calculations with the
> KD> screened B3LYP functional for phases of silica.
>
> hmmm... did you change the code for it?
> do you see any signs in the output, that
> you get a different answer from a plain BLYP
> calculation?
>
> to the best of my knowledge, HFX hasn't been
> implemented for k-points at all.
> on top of that, i was under the impression, that
> screened exchange only makes sense, when your box is large
> enough that the screening is (almost) complete, which in
> your case is not at all the case.
>
> cheers,
>    axel.
>
> KD>
> KD> What seems bizarre is that a phase calculation will always converge to the
> KD> same total energy no matter what options I use with the SCREENED EXCHANGE
> KD> keyword. I've tried out ERFC, EXP, and SELF screening while varying the
> KD> screening parameter over several orders of magnitude (eg. 10^-4 to 10^2)
> KD> for each case. It seems like all the long range Hartree Fock exchange is
> KD> always omitted (or always included perhaps).
> KD>
> KD> Is this the behavior I should expect when using SCREENED EXCHANGE, or is
> KD> it an indication that I'm not using the code properly (input file is
> KD> below)?
> KD>
> KD> The energies I do get seem reasonable - that is, the phase energy
> KD> differences I calculate with CPMD-screened-B3LYP agree with all-electron,
> KD> unscreened-B3LYP calculations I've done in another code.
> KD>
> KD> Kevin Driver
> KD> The Ohio State University, Physics Dept.
> KD> driver at mps.ohio-state.edu
> KD>
> KD> ------------ Input file -----------------
> KD>
> KD> &INFO
> KD>  Stishovite SiO2
> KD>  6x6x6 k-point mesh; 6 atoms (primitive cell)
> KD>  400 Ryd planewave cutoff energy; B3LYP
> KD> &END
> KD>
> KD> &CPMD
> KD>   OPTIMIZE WAVEFUNCTION
> KD>   CONVERGENCE ORBITALS
> KD>    1.0e-6
> KD>   PRINT FORCES ON
> KD>   MAXSTEP
> KD>   1000
> KD> &END
> KD>
> KD> &DFT
> KD>   FUNCTIONAL B3LYP
> KD>   SCREENED EXCHANGE ERFC
> KD>       0.1
> KD> &END
> KD>
> KD> &SYSTEM
> KD>   ANGSTROM
> KD>   CELL VECTORS
> KD>    4.19311248230   0.00000000000   0.00000000000
> KD>    0.00000000000   4.19311248230   0.00000000000
> KD>    0.00000000000   0.00000000000   2.68292300360
> KD>   CUTOFF
> KD>   400.0
> KD>   POINT GROUP
> KD>   AUTO
> KD>   KPOINTS MONKHORST-PACK
> KD>   6 6 6
> KD>   TESR
> KD>   8
> KD> &END
> KD>
> KD> &ATOMS
> KD> *SI_SG_BLYP  KLEINMAN-BYLANDER
> KD>    LMAX=P
> KD>    2
> KD>    0.000000000000  0.000000000000  0.000000000000
> KD>    2.096556241150  2.096556241150  1.341461501800
> KD>
> KD> *O_SG_BLYP KLEINMAN-BYLANDER
> KD>    LMAX=P
> KD>    4
> KD>    1.283524462631  1.283524462631  0.000000000000
> KD>   -1.283524462631 -1.283524462631  0.000000000000
> KD>   -0.813031778519  0.813031778519  1.341461501800
> KD>    0.813031778519 -0.813031778519  1.341461501800
> KD> &END
> KD>
> KD>
> KD> _______________________________________________
> KD> CPMD-list mailing list
> KD> CPMD-list at cpmd.org
> KD> http://cpmd.org/mailman/listinfo/cpmd-list
> KD>
>
> --
> =======================================================================
> Axel Kohlmeyer   akohlmey at cmm.chem.upenn.edu   http://www.cmm.upenn.edu
>    Center for Molecular Modeling   --   University of Pennsylvania
> Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
> tel: 1-215-898-1582,  fax: 1-215-573-6233,  office-tel: 1-215-898-5425
> =======================================================================
> If you make something idiot-proof, the universe creates a better idiot.
>
>