[molpro-user] Desired MCSCF orbital moving out of active space
Seth Olsen
seth.olsen at uq.edu.au
Tue Aug 21 02:41:23 BST 2012
It looks like you are averaging over many states with different symmetry, but using equal weights for all states. Have you tried playing with the weights? Since you only need the orbitals, you shouldn't be bothered by the potentially thorny conceptual issues of interpretation that arise when the state-averaged ensemble is not evenly weighted... Maybe start with a CAS-CI to figure out where your target state is, then do a CASSCF with a weighting scheme designed to target that state.
Another, probably more straightforward, way to do this is to first do a CASSCF that is rigged to target only the state of interest (by heavily constricting the active space to only the few orbitals most involved). Then, if that works, try to do the same with the other states of interest and use MERGE to construct an initial guess from the pre-optimized orbitals. This may put the initial guess in a region where it is harder to kick the state in question out.
-Seth
On 17/08/2012, at 10:06 AM, Terry Frankcombe wrote:
> I have had similar experiences when trying to do specific things with
> specific orbitals. To me this suggests that the orbital you are
> interested in does not "contribute" much correlation energy. The
> algorithm decides, rightly, that a lower energy results from rotating
> your orbital of interest out of the space.
>
> In a sense, this simply means that the optimisation algorithm in molpro
> multi is "too good". The solution I have has some success with is to
> hamstring the optimisation algorithm; choose a known inferior one. Try
> something like:
>
> ITERATIONS; DO AUGMENT 1 TO <whatever>
>
>
> On Thu, 2012-08-16 at 16:06 +0000, Little, Duncan wrote:
>> Dear Molpro users,
>>
>> I am trying to generate orbitals using MULTI for use in a separate calculation. I have previously posted about this: http://www.molpro.net/pipermail/molpro-user/2012-July/004968.html (thank you for your responses) but now feel I have refined my problem. The problem I am experiencing is that an orbital that is needed for a particular (Rydberg) state in my separate calculation is constantly pushed above the active space orbitals every time I try to increase the active space to incorporate it. I use the same active space for the separate calculation as the MOLPRO one for generating the orbitals. For example:
>>
>> 31102110 CAS, the orbital I would like to be in the active space is highlighted by ****. It is the first virtual orbital after the active orbitals:
>>
>> angstrom
>> geometry={
>> N,, 0.0000000, 0.00000000, r,; N,, 0.0000000, 0.00000000, -r,;
>> }
>>
>> {rhf
>> wf,charge=1,symmetry=1,spin=1;}
>>
>> {mcscf
>> config
>> CLOSED,1,0,0,0,1,0,0,0,
>> occ,3,1,1,0,2,1,1,0;
>> }
>> wf,charge=1,symmetry=1,spin=1;state,1;weight,1;
>> wf,charge=1,symmetry=2,spin=1;state,1;weight,1;
>> wf,charge=1,symmetry=3,spin=1;state,1;weight,1;
>> wf,charge=1,symmetry=5,spin=1;state,1;weight,1;
>> wf,charge=1,symmetry=6,spin=1;state,1;weight,1;
>> wf,charge=1,symmetry=7,spin=1;state,1;weight,1;
>> ORBPRINT,110}
>>
>>
>> NATURAL ORBITALS
>> ================
>>
>> Orb Occ Energy Coefficients
>> 1.5 2.00000 -16.127629 0.992816 0.000121 0.018420 -0.009890 0.005998 0.000584 -0.001178 -0.000138 0.002840 -0.002171
>> -0.000081 0.000261 0.000000 -0.000380 0.000000 -0.000908 0.000000 0.000344 0.000000 -0.000018
>> 0.000000 0.000087 0.000000 0.000252 0.000000 -0.000067 0.000000 0.000000 0.000057 0.000000
>> 0.000000
>>
>> 2.5 1.74152 -1.093616 0.007334 0.343864 0.283056 0.632141 0.353841 -0.036050 0.080350 0.155758 0.160042 0.002537
>> 0.006503 -0.002890 0.000000 -0.006509 0.000000 -0.003557 0.000000 0.003376 0.000000 0.000340
>> 0.000000 -0.000493 0.000000 0.000567 0.000000 -0.000049 0.000000 0.000000 -0.000722 0.000000
>> 0.000000
>>
>> **** 3.5 -0.00000 -0.082578 -0.001345 -0.059745 -0.000497 -0.499029 -4.563792-10.178761 -0.007343 -0.022340 0.460257 2.317907
>> 2.931892 0.004592 0.000000 -0.029487 -0.000000 -0.242067 -0.000000 -0.418116 -0.000000 0.000852
>> 0.000000 0.028458 0.000000 0.036191 0.000000 -0.002525 0.000000 0.000000 -0.003243 -0.000000
>> -0.000000
>>
>> 4.5 -0.00000 0.064433 -0.014736 -0.110028 -0.261324 0.916306 13.950224 33.887139 -0.006373 0.010013 -1.576450 -7.800327
>> -6.758397 -0.008343 -0.000000 0.122852 0.000000 0.860659 0.000000 1.388081 0.000000 -0.004092
>> -0.000000 -0.100116 -0.000000 -0.092591 -0.000000 0.009613 -0.000000 -0.000000 0.005993 0.000000
>> 0.000000
>>
>> 5.5 -0.00000 0.184409 0.007943 -0.020624 -0.010409 0.329126 9.566157 2.631699 0.034462 0.107873 -0.598887 -3.426771
>> -0.612603 0.000287 0.000000 0.060834 0.000000 0.428530 0.000000 2.382657 0.000000 -0.001865
>> 0.000000 -0.018177 -0.000000 -0.363481 -0.000000 -0.006317 0.000000 -0.000000 0.047744 -0.000000
>> 0.000000
>>
>> -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
>>
>> Separate calculation. Increasing the active space to 31103110, the orbital then moves to 4.5 (again highlighted by ****).
>>
>> {mcscf
>> config
>> CLOSED,1,0,0,0,1,0,0,0,
>> occ,3,1,1,0,3,1,1,0;
>> wf,charge=1,symmetry=1,spin=1;state,1;weight,1;
>> wf,charge=1,symmetry=2,spin=1;state,1;weight,1;
>> wf,charge=1,symmetry=3,spin=1;state,1;weight,1;
>> wf,charge=1,symmetry=5,spin=1;state,1;weight,1;
>> wf,charge=1,symmetry=6,spin=1;state,1;weight,1;
>> wf,charge=1,symmetry=7,spin=1;state,1;weight,1;
>> ORBPRINT,110}
>>
>> NATURAL ORBITALS
>> ================
>>
>> Orb Occ Energy Coefficients
>>
>> 1.5 2.00000 -16.116798 0.992724 0.000033 0.018523 -0.009949 0.006088 0.000623 -0.001119 -0.000248 0.002746 -0.002229
>> -0.000087 0.000264 0.000000 -0.000318 0.000000 -0.000885 0.000000 0.000350 0.000000 -0.000025
>> 0.000000 0.000058 0.000000 0.000251 0.000000 -0.000057 0.000000 0.000000 0.000066 0.000000
>> 0.000000
>>
>> 2.5 1.72734 -1.087097 0.007619 0.344272 0.282977 0.634465 0.353815 -0.035906 0.080098 0.155189 0.159046 0.002279
>> 0.006469 -0.002711 0.000000 -0.005941 0.000000 -0.002921 0.000000 0.003430 0.000000 0.000462
>> 0.000000 -0.000255 0.000000 0.000588 0.000000 -0.000068 0.000000 0.000000 -0.000748 0.000000
>> 0.000000
>>
>> 3.5 0.01910 0.678758 -0.045703 -0.347049 -0.397846 -0.973099 -0.704256 -0.193372 0.285598 0.660047 0.831059 0.300589
>> 0.028733 0.014519 0.000000 0.009487 0.000000 -0.086907 0.000000 -0.030679 0.000000 -0.001631
>> 0.000000 0.000472 0.000000 0.010586 0.000000 -0.001806 0.000000 0.000000 0.003928 0.000000
>> 0.000000
>>
>> **** 4.5 -0.00000 -0.079708 -0.000612 0.038633 -0.019481 0.455560 4.697201 10.697348 0.022228 0.053456 -0.431397 -2.373266
>> -3.050646 -0.003252 0.000000 0.036710 0.000000 0.267211 0.000000 0.458377 0.000000 -0.001136
>> 0.000000 -0.030899 -0.000000 -0.039953 -0.000000 0.002579 -0.000000 -0.000000 0.003774 0.000000
>> 0.000000
>>
>> ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
>>
>>
>> Separate calculation. Increasing the active space to 31104110 (I am aware this may not be an ideal CAS but it is solely to illustrate my point), the orbital then moves to 4.5 (again highlighted by ****).
>>
>> {rhf
>> wf,charge=1,symmetry=1,spin=1;}
>>
>> {mcscf
>> config
>> CLOSED,1,0,0,0,1,0,0,0,
>> occ,3,1,1,0,4,1,1,0;
>> wf,charge=1,symmetry=1,spin=1;state,1;weight,1;
>> wf,charge=1,symmetry=2,spin=1;state,1;weight,1;
>> wf,charge=1,symmetry=3,spin=1;state,1;weight,1;
>> wf,charge=1,symmetry=5,spin=1;state,1;weight,1;
>> wf,charge=1,symmetry=6,spin=1;state,1;weight,1;
>> wf,charge=1,symmetry=7,spin=1;state,1;weight,1;
>> ORBPRINT,110}
>>
>>
>>
>>
>> NATURAL ORBITALS
>> ================
>>
>> Orb Occ Energy Coefficients
>> 1.5 2.00000 -16.116631 0.992710 0.000036 0.018568 -0.009984 0.006086 0.000608 -0.001132 -0.000246 0.002772 -0.002235
>> -0.000085 0.000265 0.000000 -0.000325 0.000000 -0.000894 0.000000 0.000351 0.000000 -0.000024
>> 0.000000 0.000060 0.000000 0.000254 0.000000 -0.000057 0.000000 0.000000 0.000066 0.000000
>> 0.000000
>>
>> 2.5 1.72897 -1.087327 0.007169 0.341606 0.280128 0.623621 0.377514 -0.027605 0.079566 0.153830 0.158453 0.001259
>> 0.005524 -0.002652 0.000000 -0.006146 0.000000 -0.003580 0.000000 0.003723 0.000000 0.000469
>> 0.000000 -0.000277 0.000000 0.000759 0.000000 -0.000060 0.000000 0.000000 -0.000700 0.000000
>> 0.000000
>>
>> 3.5 0.01915 0.692432 -0.040626 -0.321662 -0.357695 -0.917013 -0.889834 -0.209679 0.295834 0.683983 0.808532 0.322875
>> 0.030401 0.016586 0.000000 0.010275 0.000000 -0.087661 0.000000 -0.035449 0.000000 -0.000774
>> 0.000000 0.002678 0.000000 0.009216 0.000000 -0.002156 0.000000 0.000000 0.003225 0.000000
>> 0.000000
>>
>> 4.5 0.00307 0.796155 -0.083489 -0.513488 -1.211317 0.066232 3.073560 0.258293 -0.118911 -0.311993 0.001787 -0.556182
>> -0.029572 -0.017016 0.000000 -0.025841 0.000000 0.009608 0.000000 0.086108 0.000000 -0.006459
>> 0.000000 -0.014354 0.000000 0.019845 0.000000 0.003748 0.000000 0.000000 0.003494 0.000000
>> 0.000000
>>
>> ***** 5.5 -0.00000 -0.042134 0.008871 0.032148 0.127032 -0.475218 -7.748323-19.960582 0.005297 0.000261 0.835343 3.953541
>> 4.903763 0.005937 -0.000000 -0.058749 -0.000000 -0.432178 -0.000000 -0.718211 -0.000000 0.002261
>> -0.000000 0.052030 0.000000 0.061429 0.000000 -0.004754 0.000000 0.000000 -0.005687 -0.000000
>> -0.000000
>>
>>
>>
>>
>> Does any one have any idea what may be causing this or what it is indicative of? I've tried a variety of orbital restrictions, increasing the CAS in other symmetries but the orbital constantly jumps up in everything I do.
>>
>> Any suggestions would be much appreciated.
>>
>> Duncan Little
>>
>> University College London
>> _______________________________________________
>> Molpro-user mailing list
>> Molpro-user at molpro.net
>> http://www.molpro.net/mailman/listinfo/molpro-user
>
>
> _______________________________________________
> Molpro-user mailing list
> Molpro-user at molpro.net
> http://www.molpro.net/mailman/listinfo/molpro-user
---------------------------------------------------
Seth Olsen
ARC Australian Research Fellow
6-431 Physics Annexe
School of Mathematics and Physics
The University of Queensland
Brisbane QLD 4072 Australia
seth.olsen at uq.edu.au
+61 7 3365 2816
---------------------------------------------------
Unless stated otherwise, this e-mail represents only the views of the Sender and not the views of The University of Queensland
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20120821/9d50dffd/attachment.html>
More information about the Molpro-user
mailing list