"Fock matrix approximation error" in spin-orbit CI

[Karl Irikura]

Dear Molpro experts,

The title error occurred many times during the calculation, but the calculation completed normally (after six days).  Unfortunately, only the lowest levels look reasonable; all higher levels lack the proper degeneracies.  The analogous calculation with the smaller TZ basis set was successful.  I tried tight thresholds but that did not help. I would be grateful for any tips. 

Thanks,

Karl

This is all-electron Br atom with a (7/8) active space, to include the low Rydberg levels.  Here is the input:

! active space (7/8)
! much tighter thresholds, seeking correct level degeneracies
memory,4000,M;

gprint,orbitals,civector;
gthresh,energy=1.d-10,coeff=1.d-8,gradient=1.d-7,step=1.d-4;twoint=1.d-13;

symmetry,xyz
geometry={Br};

basis=aug-cc-pwCVQZ-X2C

set,dkho=101

{rhf;wf,charge=-1}

NSING=3
NTRIP=8

{multi
    occ,10,12
    closed,8,6
    wf,charge=0,sym=2,spin=1;state,NSING;
    wf,charge=0,sym=2,spin=3;state,NTRIP;
    expec2,lxx,lyy,lzz
}
table, energy, ll
title, Energies and <L**2> values

! KL-shell uncorrelated
core,2,3

{ci;wf,sym=2,spin=1;state,NSING; save,5203.2}
hlsdiag = energd4
{ci;wf,sym=2,spin=3;state,NTRIP; save,5303.2}
hlsdiag(NSING+1) = energd4

table,hlsdiag

lsint
{ci;hlsmat,ls,5203.2,5303.2;print,vls=0,hls=0} !compute and diagonalize SO matrix