Firefly and PC GAMESS-related discussion club


 
Learn how to ask questions correctly  
 
 
We are NATO-free zone
 



Re^3: Poor SCF convergence with re-read orbitals

Slawomir Janicki
slawomir.janicki@comcast.net


To close this discussion thread, the fix worked. The SCF converged within 4 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  -434.359186149  -434.359186149   0.000428083   0.000000014     0.000000000     1.000000000             -1 230828006
   2  1  0  -434.359186149     0.000000000   0.000200000   0.000000013     0.000000000     1.000000000             -1 230828015
   3  2  0  -434.359186149     0.000000000   0.000013786   0.000000004     0.000000000     1.000000000             -1 230828011
   4  3  0  -434.359186149     0.000000000   0.000023126   0.000000017     0.000000000     1.000000000             -1 230828011


$GUESS ASIS=.T. $END
should be used when $VEC section contains orbitals saved with
WIDE=.T. 

Slawomir


On Mon Dec 9 '13 3:02am, Alex Granovsky wrote
---------------------------------------------
>Dear Slawomir,

>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
>>>


[ Previous ] [ Next ] [ Index ]           Wed Dec 11 '13 1:11am
[ Reply ] [ Edit ] [ Delete ]           This message read 607 times