Posted by Lori Burns on April 25, 2005 at 08:41:38:
In Reply to: Re: basis set limitations posted by Valera Veryazov on April 18, 2005 at 21:14:27:
Hi,
Thank you for your responses. Minding the linear dependence output in scf and using the deleted keyword in rasscf did indeed solve the problem I described. Unfortunately, it introduced another. I had been running state-averaged casscf optimizations with the scheme {seward rasscf mclr alaska slapaf }; however, when orbitals were deleted, the optimization sequence produced exceedingly large forces that did not converge.
DISPLACEMENTS FORCES
RMS + 0.9157E-01 0.1200E-02 No + 0.4983E+04 0.3000E-03 No +
RMS + 0.7840E-01 0.1200E-02 No + 0.4345E+04 0.3000E-03 No +
RMS + 0.8417E-01 0.1200E-02 No + 0.3159E+04 0.3000E-03 No +
RMS + 0.6179E-01 0.1200E-02 No + 0.3942E+04 0.3000E-03 No +
RMS + 0.8430E-01 0.1200E-02 No + 0.3567E+04 0.3000E-03 No +
RMS + 0.7517E-01 0.1200E-02 No + 0.7700E+04 0.3000E-03 No +
RMS + 0.7223E-01 0.1200E-02 No + 0.3632E+04 0.3000E-03 No +
RMS + 0.6404E-01 0.1200E-02 No + 0.4879E+04 0.3000E-03 No +
RMS + 0.5976E-01 0.1200E-02 No + 0.3047E+04 0.3000E-03 No +
RMS + 0.8984E-01 0.1200E-02 No + 0.2945E+04 0.3000E-03 No +
RMS + 0.7994E-01 0.1200E-02 No + 0.2417E+04 0.3000E-03 No +
RMS + 0.8777E-01 0.1200E-02 No + 0.4099E+04 0.3000E-03 No +
RMS + 0.5878E-01 0.1200E-02 No + 0.6803E+04 0.3000E-03 No +
Max + 0.1546E+00 0.1800E-02 No + 0.1189E+05 0.4500E-03 No +
Max + 0.1211E+00 0.1800E-02 No + 0.6799E+04 0.4500E-03 No +
Max + 0.1080E+00 0.1800E-02 No + 0.6496E+04 0.4500E-03 No +
Max + 0.8474E-01 0.1800E-02 No + 0.4107E+04 0.4500E-03 No +
Max + 0.1400E+00 0.1800E-02 No + 0.5745E+04 0.4500E-03 No +
Max + 0.1118E+00 0.1800E-02 No + 0.9538E+04 0.4500E-03 No +
Max + 0.1270E+00 0.1800E-02 No + 0.4383E+04 0.4500E-03 No +
Max + 0.8454E-01 0.1800E-02 No + 0.5964E+04 0.4500E-03 No +
Max + 0.8546E-01 0.1800E-02 No + 0.3354E+04 0.4500E-03 No +
Max + 0.1420E+00 0.1800E-02 No + 0.6478E+04 0.4500E-03 No +
Max + 0.9107E-01 0.1800E-02 No + 0.4043E+04 0.4500E-03 No +
Max + 0.1456E+00 0.1800E-02 No + 0.7582E+04 0.4500E-03 No +
Max + 0.9305E-01 0.1800E-02 No + 0.7831E+04 0.4500E-03 No +
Another symptom is seen in slapaf where this is in the output of one of the later optimization passes:
Internal forces
Iter. 1
nrc001 *********
nrc002 *********
nrc003 *********
nrc004 *********
nrc005 *********
nrc006 *********
nrc007 28.50323
nrc008 *********
nrc009 *********
nrc010 *********
nrc011 *********
nrc012 *********
nrc013 *********
nrc014 *********
I've tried manually removing the offending orbital from the scf output and a few different means of exchanging orbitals between modules, but the problem persists as long as mclr is present. Given how the problem appeared, I would expect some notification to mclr comparable to the deleted keyword to be required, but none of the mclr keywords looked likely. Of course, non-state-averaged calculations (without mclr) run nicely, but I am concerned by this seeming introduction of faulty data communication. A sample input file follows. Thanks for any further help.
Sincerely,
Lori Burns
*
* TIGHT OPT of (4,4) ACTIVE C2V S2 TROP with aug-cc-pvdz
*
*
>>> Set MaxIter 100 <<<
>>>>>> Do while <<<<<<<
&SEWARD &END
Title
Tropolone molecule Seward S2 S2 S2 S2
Symmetry
XY Y
Basis set
O.aug-cc-pVDZ.Dunning.10s5p2d.4s3p2d..
*
O1 0.0000000000 2.1265526996 3.7193825488 *from TR_N
End of basis
Basis set
C.aug-cc-pVDZ.Dunning.10s5p2d.4s3p2d..
*
C1 0.0000000000 1.4072601580 1.4468644620 *from TR_N
C7 0.0000000000 3.0093536770 -0.6236105960 *from TR_N
C6 0.0000000000 2.3521482927 -3.1620811364 *from TR_N
C5 0.0000000000 0.0000000000 -4.3098289930 *from TR_N
End of basis
Basis set
H.aug-cc-pVDZ.Dunning.5s2p.3s2p..
*
H0 0.0000000000 0.0000000000 4.5238998111 *from TR_N
H7 0.0000000000 4.9897334009 -0.1708538998 *from TR_N
H6 0.0000000000 3.9339121091 -4.4431755140 *from TR_N
H5 0.0000000000 0.0000000000 -6.3419807805 *from TR_N
End of basis
End of Input
>>> IF ( ITER = 1 ) <<<
&SCF &END
Title
Tropolone molecule SCF S2 S2 S2 S2
Occupied
15 3 2 12 *****
Iterations
100
End of Input
&RASSCF &END
Lumorb
Title
Tropolone molecule RASSCF S2 S2 S2 S2
Deleted
0 0 0 1
Symmetry
1 *****
Spin
1
Nactel
4 0 0 *****
Inactive
15 1 2 12 *****
Ras2
0 2 2 0 *****
Iter
100,25
End of Input
!cp $Project.JobIph $Project.JobOld
>>> ENDIF <<<<<<<<<<<<<<
&RASSCF &END
Jobiph
Cirestart
Title
Tropolone molecule outif RASSCF S2 S2 S2 S2
Deleted
0 0 0 1
Symmetry
1 *****
Spin
1
Nactel
4 0 0 *****
Inactive
15 1 2 12 *****
Ras2
0 2 2 0 *****
Ciroot
*1 1
*1
*****
*1 2
*2
*****
2 2
1 2
1 1
Levshift
*1.0
0.75
Iter
100,25
Rlxroot
2 *****
End of Input
&MCLR &END
Title
Tropolone molecule MCLR S2 S2 S2 S2
Sala
2 *****
End of Input
&ALASKA &END
End of input
&SLAPAF &END
Thrs
* 1.0D-06 3.0D-04 *gaussian opt
1.6D-09 1.0D-05 *gaussian opt = tight
* 1.6D-11 1.0D-06 *gaussian opt = verytight
End of Input
>>> ENDDO <<<<<<<<<<<<<<<<
&CASPT2 &END
Title
Tropolone molecule CASPT2 S0 S0 S0 S0
Maxit
20
Lroot
2 *****
End of Input