problem converging molecule in doublet state

[garold]

I do not see a way to upload files.  I will share the lengthy output
file with whoever wishes to see it.

Garold

On Apr 7, 8:03 pm, garold <g.murd... at gmail.com> wrote:
> Dear CP2K Community,
>
> I am attempting to calculate the energy of a molecule (dicarboyxlate
> anion with a lithium cation; net charge is zero) in its doublet
> state.  I have tried using both restricted and unrestricted formalisms
> (LSD and ROKS, respectively).  However, I cannot get the SCF to
> converge with either.  I have been using both B3LYP and BLYP and
> various basis sets.  Ideally, I would prefer to use B3LYP/GAPW/6-31+G*
> but have been unable to obtain convergence even with BLYP/GAPW/6-31G*
> (no diffuse functions) or BLYP/GPW/DZVP-MOLOPT-SR-GTH (my original
> assumption was that diffuse functions were source of the problem).
> The input for my preferred method/basis set is below and the output is
> attached.  Please let me know if you have any suggestions.  Thank you
> in advance.
>
> Note that I am using a recent tested version of cp2k compiled with
> libint.  Both BLYP/ and B3LYP/GAPW/6-31+G* work well for the closed-
> shell, singlet version of this system (e.g., with one electron added,
> net charge of -1, same geometry).  However, a preconditioner (OT/
> OUTER_SCF option) does not seem to work well for this system so I am
> not using it.
>
> Best regards,
> Garold Murdachaew
>
> ps: Input file (gzipped output file is attached):
>
> &FORCE_EVAL
>   METHOD Quickstep
>   &DFT
>     LSD
>     CHARGE 0
>     MULTIPLICITY 2
>     BASIS_SET_FILE_NAME ../EMSL_BASIS_SETS_mine
>     POTENTIAL_FILE_NAME ../POTENTIAL
>     &MGRID
>       CUTOFF 280
>       NGRIDS 5
>     &END MGRID
>     &QS
>       METHOD GAPW
>     &END QS
>     &POISSON
>       PERIODIC NONE
>       PSOLVER MT
>     &END
>     &SCF
>       SCF_GUESS ATOMIC
>       EPS_SCF 1.0E-7
>       MAX_SCF 20
> !     &OUTER_SCF
> !       EPS_SCF 1.0E-7
> !       MAX_SCF 20
> !     &END
> !     &OT
> !       PRECONDITIONER FULL_ALL
> !     &END
>     &END SCF
>     &XC
>       &XC_FUNCTIONAL
>        &LYP
>          SCALE_C 0.81
>        &END
>        &BECKE88
>          SCALE_X 0.72
>        &END
>        &VWN
>          FUNCTIONAL_TYPE VWN3
>          SCALE_C 0.19
>        &END
>        &XALPHA
>          SCALE_X 0.08
>        &END
>       &END XC_FUNCTIONAL
>       &HF
>         &SCREENING
>           EPS_SCHWARZ 1.0E-10
>         &END
>         &MEMORY
>           MAX_MEMORY 1200
>         &END
>         FRACTION 0.20
>       &END
>     &END XC
>   &END DFT
>   &SUBSYS
>     &CELL
>       ABC 20.0 20.0 20.0
>       PERIODIC NONE
>     &END CELL
>     &COORD
>  C    -0.793023    -0.703538     0.527367
>  C    -0.046767    -0.838845    -0.834379
>  C    -0.109334     0.381305    -1.793731
>  O    -0.533605     0.203716    -2.940264
>  C     0.080655    -0.434112     1.783207
>  O    -0.014088    -1.209796     2.739982
>  O     0.835715     0.609771     1.736225
>  O     0.313892     1.501426    -1.316006
>  H    -1.529788     0.107036     0.458189
>  H    -1.351096    -1.620668     0.723529
>  H     1.012459    -1.057227    -0.647331
>  H    -0.452088    -1.690968    -1.382698
> Li     0.909030     1.669426     0.332287
>     &END COORD
>     &KIND C
>       BASIS_SET 6-31+Gx
>       POTENTIAL ALL
>     &END KIND
>     &KIND O
>       BASIS_SET 6-31+Gx
>       POTENTIAL ALL
>     &END KIND
>     &KIND H
>       BASIS_SET 6-31Gx
>       POTENTIAL ALL
>     &END KIND
>     &KIND Li
>       BASIS_SET 6-31+Gx
>       POTENTIAL ALL
>     &END KIND
>   &END SUBSYS
> &END FORCE_EVAL
> &GLOBAL
>   PROJECT d_m
>   PRINT_LEVEL MEDIUM
> &END GLOBAL
>
> pps: I constructed the 6-31+G* basis set simply by adding the diffuse
> functions to C, O, and Li, the basis set file is attached.