Alex Granovsky
gran@classic.chem.msu.su
add
$guess asis=.t. $end
to your second input file. Note, as the condition number
of overlap matrix is terribly large (ca. 10), one can expect
troubles with computations.
all the best,
Alex
On Fri Dec 6 '13 10:37pm, Alex Granovsky wrote
----------------------------------------------
>Dear Slawomir,
>could you send me your input and output files?
>All the best,
>Alex
>
>
>On Fri Dec 6 '13 4:08pm, Slawomir Janicki wrote
>-----------------------------------------------
>>I run into SCF convergence trouble with re-read orbitals. Is there a way to increase the accuracy of punched orbitals beyond WIDE=.T. ?
>>In the first pass I construct the guess orbitals by:
>>- $GUESS GUESS=MOREAD NORB=904 KDIAG=0 EXTRA=.T. $END from a previous calculation with QZVPD on all atoms except one H with QZVP;
>>- adding the p diffuse functions from QZVPD to that unique H with $EXTRAF NEXTRA(1)=0,0,0,0,0,0,0,0,0,0,0,0,0,3 $END
>>- declaring the basis set in $DATA
>>This SCF calculation converges fine in 12 iterations.
>> NONZERO BLOCKS
>> ITER EX DEM TOTAL ENERGY E CHANGE DENSITY CHANGE DIIS ERROR VIR. SHIFT DAMPING INTEGRALS SKIPPED
>> * * * INITIATING DIIS PROCEDURE * * *
>> 1 0 0 -434.359184038 -434.359184038 26.000472601 0.000765616 0.000000000 1.000000000 -1 231093005
>> 2 1 0 -434.359184889 -0.000000851 21.033149847 0.000696271 0.000000000 1.000000000 -1 230685217
>> 3 2 0 -434.359185951 -0.000001062 2.264576385 0.000341393 0.000000000 1.000000000 -1 230830209
>> 4 3 0 -434.359186134 -0.000000183 1.741446876 0.000100257 0.000000000 1.000000000 -1 230858906
>> 5 4 0 -434.359186149 -0.000000015 0.110725963 0.000003935 0.000000000 1.000000000 -1 230829709
>> 6 5 0 -434.359186149 0.000000000 0.000703852 0.000000093 0.000000000 1.000000000 -1 230828030
>> 7 6 0 -434.359186149 0.000000000 0.000949385 0.000000032 0.000000000 1.000000000 -1 230828021
>> 8 7 0 -434.359186149 0.000000000 0.000186930 0.000000010 0.000000000 1.000000000 -1 230828001
>> 9 8 0 -434.359186149 0.000000000 0.000155360 0.000000006 0.000000000 1.000000000 -1 230828007
>> 10 9 0 -434.359186149 0.000000000 0.000225597 0.000000011 0.000000000 1.000000000 -1 230828011
>> 11 10 0 -434.359186149 0.000000000 0.000018039 0.000000012 0.000000000 1.000000000 -1 230828006
>> 12 11 0 -434.359186149 0.000000000 0.000035272 0.000000021 0.000000000 1.000000000 -1 230828006
>>
>>
>>Next, I use the orbitals from the first dat file and construct the guess orbitals for the second input file by:
>>- $GUESS GUESS=MOREAD NORB=907 KDIAG=0 EXTRA=.F. $END
>>- declaring the basis set in $DATA (the $DATA is the same as in the first input file)
>>The second SCF fails to converge and the energy and density oscillate wildly.
>> NONZERO BLOCKS
>> ITER EX DEM TOTAL ENERGY E CHANGE DENSITY CHANGE DIIS ERROR VIR. SHIFT DAMPING INTEGRALS SKIPPED
>> * * * INITIATING DIIS PROCEDURE * * *
>> 1 0 0 -434.359184142 -434.359184142************** 0.001427558 0.000000000 1.000000000 -1 230828006
>> 2 1 0 -220.595295982 213.763888161 2.036093324 112.110529823 0.191843150 1.000000000 -1 219960281
>> 3 2 0 -561.039293414 -340.443997432 0.609648347 34.077380819 0.091843148 1.000000000 -1 219960281
>> 4 3 0 -586.729178088 -25.689884675 1.225957379 68.635216291 0.091843148 1.000000000 -1 219960281
>> 5 4 0 -506.720698955 80.008479134 0.264003476 50.804093873 0.091843148 1.000000000 -1 219960281
>> 6 5 0 -436.096265988 70.624432967 0.493343875 33.901854332 0.091843148 1.000000000 -1 219960281
>>etc.
>>The SCF convergence options are the same in both cases:
>> $SCF DIRSCF=.T. $END
>> $CONTRL MAXIT=100 $END
>> $SCF FDIFF=.F. $END
>> $SCF NCONV=5 $END
>> $SCF EXTRAP=.F. $END
>> $SCF SODIIS=0 DIONCE=.F. $END
>> $SCF SOSCF=.F. $END
>> $SCF DIIS=.T. ETHRSH=2500 MAXDII=10 $END
>> $SCF DEM=.F. $END
>> $SCF DAMP=.T. DMPCUT=0.0 $END
>> $SCF SHIFT=.T. $END
>> $SCF RSTRCT=.F. $END
>>Any idea how to fix this?
>>Slawomir
>>