Alex Granovsky
gran@classic.chem.msu.su
your results look just fine with me.
Some, sometimes strong, mixing of zero-order (i.e., CAS) states
is quite natural for many systems and is OK. What should be avoided
are the situations with very large off-diagonal elements of the
effective Hamiltonian, rather than of its eigenvectors. Another test
is to look at overlap matrix of eigenvectors of the non-symmetric
effective Hamiltonian. It is better to avoid large overlaps.
All the best,
Alex
On Thu Oct 6 '11 7:19am, Solntsev Pasha wrote
---------------------------------------------
>Dear Alex.
>Thank you very much for your advice. The trick with symmetry really works. I tried and indeed immediately obtained reasonable results. Just small question. I remember you told to avoid situation then off-diagonal components of the EIGENVECTORS OF THE EFFECTIVE HAMILTONIAN matrix higher than 0.1. From my results, probably it is ok if 4th state contributes to 2nd. The same situation with another states too. How do you think is it acceptable or it is some hidden problem?
>Thank you.
>Best,
>Pavel.
>
>
>
> *** MC-XQDPT2 ENERGIES ***
> -----------------------------------------------------------------------
> STATE 1ST ORDER 2ND ORDER
> 1 E(MCSCF)= -983.4871597547 E(MP2)= -986.9133168011
> 2 E(MCSCF)= -983.3604225136 E(MP2)= -986.8498610520
> 3 E(MCSCF)= -983.3523597356 E(MP2)= -986.8325743544
> 4 E(MCSCF)= -983.2863223917 E(MP2)= -986.7965576332
> 5 E(MCSCF)= -983.2856404629 E(MP2)= -986.7955479327
> 6 E(MCSCF)= -983.2674851341 E(MP2)= -986.7720505935
> 7 E(MCSCF)= -983.2673364992 E(MP2)= -986.7719948030
> 8 E(MCSCF)= -983.2459151153 E(MP2)= -986.7636054435
>
>
> EIGENVECTORS OF THE EFFECTIVE HAMILTONIAN
>
>
>
> 1 2 3 4 5
>
>
>
> -986.913317-986.849861-986.832574-986.796558-986.795548
>
>
>
> 1 0.997444 0.000000 0.000000 0.000000 0.000000
> 2 0.000000 0.977961 0.000000 0.208789 0.000000
> 3 0.000000 0.000000 0.973468 0.000000 0.228821
> 4 0.000000 0.208789 0.000000 -0.977961 0.000000
> 5 0.000000 0.000000 0.228821 0.000000 -0.973468
> 6 -0.061205 0.000000 0.000000 0.000000 0.000000
> 7 0.000000 0.000000 0.000000 0.000000 0.000000
> 8 0.036867 0.000000 0.000000 0.000000 0.000000
>
>
>
> 6 7 8
>
>
>
> -986.772051-986.771995-986.763605
>
>
>
> 1 0.000000 -0.053977 0.046816
> 2 0.000000 0.000000 0.000000
> 3 0.000000 0.000000 0.000000
> 4 0.000000 0.000000 0.000000
> 5 0.000000 0.000000 0.000000
> 6 0.000000 -0.983535 0.170035
> 7 1.000000 0.000000 0.000000
> 8 0.000000 -0.172466 -0.984325
>
>
>
> EIGENVALUES OF THE NON-SYMMETRIC EFFECTIVE HAMILTONIAN
>
>
>
> 1 -986.91325880851070000000 + I * 0.00000000000000000000
> 2 -986.84986062440300000000 + I * 0.00000000000000000000
> 3 -986.83257411029940000000 + I * 0.00000000000000000000
> 4 -986.79655806075540000000 + I * 0.00000000000000000000
> 5 -986.79554817681560000000 + I * 0.00000000000000000000
> 6 -986.77205059351850000000 + I * 0.00000000000000000000
> 7 -986.77199810675930000000 + I * 0.00000000000000000000
> 8 -986.76366013241270000000 + I * 0.00000000000000000000
>
>
>
>
>
>
>
>
>
>On Wed Oct 5 '11 9:23pm, Alex Granovsky wrote
>---------------------------------------------
>>Dear Pasha,
>>there is a mismatch between your CASSCF and XMCQDPT job settings.
>>Exactly as I explained here, CASSCF which you have performed in C1 group
>>includes states belonging to different symmetry types when labeled
>>in the terms of D2H group's irreps.
>>At the same time, xmcqdpt (as well as mcqdpt) always applies symmetry
>>group specified in $data. Look at the energies of CASSCF states and
>>the states found by xmcqdpt's CASCI routine (which is called MQCACI in
>>the output). For most of states you'll find that the energies are not
>>the same between CASSCF and CASCI based on optimized orbitals
>>(as they should). In fact, the states you are modeling in xmcqdpt are
>>A1 states of the full D2h group.
>>
>>
>>The solution is as follows. Split this job into two parts, namely
>>CASSCF and then separate XMCQDPT. Add to your input $contrl wide=.t.
>>$end (if not already added) and perform CASSCF first. In the second
>>job, moread all these orbitals in the $guess and specify
>>$xmcqdpt inorb=2 $end. You also need to add nosym=1 to the $contrl
>>group. This will bypass CASSCF (as you already got converged orbitals)
>>and perform XMCQDPT2 in C1 group. The job will take some more time to
>>complete as it will not use symmetry. I'd also recommend to omit
>>$mcqfit group as there are only a few CSFs in the active space.
>>
>>
>>Hope this helps.
>>Alex
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>On Wed Oct 5 '11 5:56pm, Solntsev Pasha wrote
>>---------------------------------------------
>>>Dear Sania.
>>>I apologize, i didn't clarify very well. I was talking about regular unsubstituted porphyrin. This molecule indeed has D2h symmetry, you can see that very clear in electronic spectra. If you deprotonate or insert metal in it, of course you increase symmetry and due to degeneracy of LUMO (D4h) you have less number of transition in electronic spectra. But this is not my case yet. The 2nd and 3rd states lie close to each other as well as 4th and 5th. We can se that very clear from experiment. Difference not so small (~50-100 nm). I think it should be enough to separate all excitations in CASSCF job. You can correct me if i am not right.
>>>I attached file. I used different values for wstate array in $xmcqdpt - no luck.
>>>Thank you for you willing to help me.
>>>Best,
>>>Pavel.
>>>
>>>
>>>
>>>On Wed Oct 5 '11 3:33pm, sanya wrote
>>>------------------------------------
>>>>I think it's OK. Probably, your structure is close to (a sort of) D4 or C4 symmetry, and these groups do have E representations. Since the symmetry is approximate, the degeneracy is approximate, too. Any asymmetric distorsion will force your molecule out of this point and remove the degeneracy.
>>>>On Wed Oct 5 '11 1:11am, Solntsev Pasha wrote
>>>>---------------------------------------------
>>>>>Dear Alex.
>>>>>Thank you for willing to help me with my proble.
>>>>>My goal is to get excitation energies and electronic structure of the 1st and 2nd excited state of porphyrin molecule. After MCSCF i did i expected 5 states: 1st - 2200, 2nd - 2110, 3rd-2101, 4th - 1210, 5th - 1201. Or similar to that. I bu got another picture. Two configurations have almost same contribute to a state. I could expect that if we have E irrep (LUMO), but as you know in D2h we don't have E. So my question is: "Is it ok in my case?". If so how can i get reasonable data from XMCQDPT job?
>>>>>Any advice will be accepted.
>>>>>Best,
>>>>>Pavel.
>>>>>
>>>>>
>>>>>
>>>>>On Tue Oct 4 '11 7:53pm, Alex Granovsky wrote
>>>>>---------------------------------------------
>>>>>>Dear Pasha,
>>>>>>I would be glad to help. Could you please be a bit more specific formulating the problem itself?
>>>>>>Kind regards,
>>>>>>Alex Granovsky
>>>>>>
>>>>>>
>>>>>>On Tue Oct 4 '11 6:44am, Solntsev Pasha wrote
>>>>>>---------------------------------------------
>>>>>>>Dear colleagues.
>>>>>>>I am working on aromatic system related to porphyrin. So, i decided to check unsubstituted porphyrin by CASSCF/XMCQDPT(CASSCF(4,4)/6-311G(d,p)//b3lyp/6-311G(d,p)). I need only 1st and 2nd excited states. Their formation can be described by four orbital approach. Indeed, such approximation quite reasonable, because HOMO,HOMO-1 are separated from another HOMO's by big enough energy gap. All intense transitions in UV-Vis spectra rise from transitions HOMO->LUMO/LUMO+1, HOMO-1->LUMO/LUMO+1. In literature, people used CASPT2(4,4) method and showed such active space can be used for porphyrins (actually 4,4 is the smallest active space for such mycrocycles). So i did SS-CASSCF(4,4)+ntrack and caught all single electron excitations (4 for such active space). After optimization my 1st state was ground state and all single electron excited states were states from 2 to 5. Afterwards, i performed SA-CASSCF(4,4) over 5 lowest states ($drt wstate(1)=1,1,1,1,1 $end ) and got interesting picture(all configuration with zerro's coefficients were removed):
>>>>>>>STATE # 1 ENERGY = -983.487159757
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> CSF COEF OCCUPANCY (IGNORING CORE)
>>>>>>> --- ---- --------- --------- -----
>>>>>>> 1 -0.961493 2200
>>>>>>> 4 0.113581 2002
>>>>>>> 6 0.072134 0202
>>>>>>> 9 -0.185355 1111
>>>>>>> 11 0.022668 1111
>>>>>>> 15 0.096837 2020
>>>>>>> 17 0.103561 0220
>>>>>>> 20 -0.049553 0022
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> STATE # 2 ENERGY = -983.360422516
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> CSF COEF OCCUPANCY (IGNORING CORE)
>>>>>>> --- ---- --------- --------- -----
>>>>>>> 3 0.660764 1201
>>>>>>> 7 -0.735886 2110
>>>>>>> 14 0.094658 0112
>>>>>>> 18 -0.113585 1021
>>>>>>>
>>>>>>>
>>>>>>> STATE # 3 ENERGY = -983.352359738
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> CSF COEF OCCUPANCY (IGNORING CORE)
>>>>>>> --- ---- --------- --------- -----
>>>>>>> 2 0.752067 2101
>>>>>>> 8 0.646632 1210
>>>>>>> 13 0.105209 1012
>>>>>>> 19 0.072064 0121
>>>>>>>
>>>>>>>
>>>>>>> STATE # 4 ENERGY = -983.286322394
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> CSF COEF OCCUPANCY (IGNORING CORE)
>>>>>>> --- ---- --------- --------- -----
>>>>>>> 3 0.708619 1201
>>>>>>> 7 0.627084 2110
>>>>>>> 14 0.216679 0112
>>>>>>> 18 0.240156 1021
>>>>>>>
>>>>>>>
>>>>>>> STATE # 5 ENERGY = -983.285640465
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> CSF COEF OCCUPANCY (IGNORING CORE)
>>>>>>> --- ---- --------- --------- -----
>>>>>>> 2 0.611178 2101
>>>>>>> 8 -0.722176 1210
>>>>>>> 13 0.227594 1012
>>>>>>> 19 -0.230485 0121
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>As we can see for state #2
>>>>>>> 3 0.660764 1201
>>>>>>> 7 -0.735886 2110
>>>>>>>two excitations have almost same contribution. The same thing with another states too.
>>>>>>>I performed also XMCQDPT with different size of the H_ef but only first state from CAS-CI corresponds to first state in MCQDPT reference:
>>>>>>>
>>>>>>>
>>>>>>> *** MC-XQDPT2 ENERGIES ***
>>>>>>> -----------------------------------------------------------------------
>>>>>>> STATE 1ST ORDER 2ND ORDER
>>>>>>> 1 E(MCSCF)= -983.4871597547 E(MP2)= -986.9200228420
>>>>>>> 2 E(MCSCF)= -983.2674851340 E(MP2)= -986.7845079006
>>>>>>> 3 E(MCSCF)= -983.2459151152 E(MP2)= -986.7769659778
>>>>>>> 4 E(MCSCF)= -983.2049480922 E(MP2)= -986.7676948438
>>>>>>> 5 E(MCSCF)= -983.1927574918 E(MP2)= -986.7429440001
>>>>>>> 6 E(MCSCF)= -983.1590205258 E(MP2)= -986.7369602361
>>>>>>> 7 E(MCSCF)= -983.0694701169 E(MP2)= -986.6898412033
>>>>>>> 8 E(MCSCF)= -982.8072824406 E(MP2)= -986.5873781504
>>>>>>> EIGENVECTORS OF THE EFFECTIVE HAMILTONIAN
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> 1 2 3 4 5
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> -986.920023-986.784508-986.776966-986.767695-986.742944
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> 1 -0.994895 -0.067638 0.033741 0.014361 0.032678
>>>>>>> 2 -0.067492 0.708971 -0.294819 -0.530914 0.158499
>>>>>>> 3 0.038811 0.159120 0.755152 -0.237075 0.492123
>>>>>>> 4 -0.030986 0.638493 0.157087 0.438907 -0.478557
>>>>>>> 5 0.018530 0.016946 -0.563000 0.081181 0.443499
>>>>>>> 6 0.004953 -0.211512 0.001229 -0.675617 -0.548424
>>>>>>> 7 0.034252 -0.118795 -0.004363 0.076785 0.041466
>>>>>>> 8 0.040266 -0.025619 -0.004813 0.011612 0.058562
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> 6 7 8
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> -986.736960-986.689841-986.587378
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> 1 0.040499 -0.026339 -0.029372
>>>>>>> 2 -0.270110 -0.161009 0.003527
>>>>>>> 3 0.320243 0.031284 0.031555
>>>>>>> 4 0.379967 -0.019337 -0.016894
>>>>>>> 5 0.678936 0.128538 0.040542
>>>>>>> 6 0.444792 0.005797 -0.011659
>>>>>>> 7 0.129239 -0.953704 0.225448
>>>>>>> 8 0.054535 -0.214298 -0.972233
>>>>>>> EIGENVALUES OF THE NON-SYMMETRIC EFFECTIVE HAMILTONIAN
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> 1 -986.91955413898560000000 + I * 0.00000000000000000000
>>>>>>> 2 -986.78419407332210000000 + I * 0.00000000000000000000
>>>>>>> 3 -986.77622935041040000000 + I * 0.00000000000000000000
>>>>>>> 4 -986.76715735894640000000 + I * 0.00000000000000000000
>>>>>>> 5 -986.74307507596620000000 + I * 0.00000000000000000000
>>>>>>> 6 -986.73745642166250000000 + I * 0.00000000000000000000
>>>>>>> 7 -986.68459320897090000000 + I * 0.00000000000000000000
>>>>>>> 8 -986.59405552588180000000 + I * 0.00000000000000000000
>>>>>>>OVERLAP NORM MATRIX OF NON-ORTHOGONAL EIGENVECTORS
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> 1 2 3 4 5
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> 1 1.0000000 0.0033939 0.0162086 0.0031617 0.0031535
>>>>>>> 2 0.0033939 1.0000000 0.2549902 0.1426137 0.0648010
>>>>>>> 3 0.0162086 0.2549902 1.0000000 0.1192571 0.0789034
>>>>>>> 4 0.0031617 0.1426137 0.1192571 1.0000000 0.2513337
>>>>>>> 5 0.0031535 0.0648010 0.0789034 0.2513337 1.0000000
>>>>>>> 6 0.0038417 0.0901892 0.0600024 0.2513302 0.4293901
>>>>>>> 7 0.0375041 0.0049329 0.1179925 0.0041655 0.0068307
>>>>>>> 8 0.0448358 0.0165570 0.1067129 0.0007353 0.0019392
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> 6 7 8
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> 1 0.0038417 0.0375041 0.0448358
>>>>>>> 2 0.0901892 0.0049329 0.0165570
>>>>>>> 3 0.0600024 0.1179925 0.1067129
>>>>>>> 4 0.2513302 0.0041655 0.0007353
>>>>>>> 5 0.4293901 0.0068307 0.0019392
>>>>>>> 6 1.0000000 0.0078361 0.0617392
>>>>>>> 7 0.0078361 1.0000000 0.4620405
>>>>>>> 8 0.0617392 0.4620405 1.0000000
>>>>>>>
>>>>>>>
>>>>>>>I also played with $xmcqdpt wstate(1) and avecoe(1) arrays. I used same as for $drt wstate(1)=1,1,1,1,1 $end but also default value for wstate from xmcqdpt.
>>>>>>>Could you please to help me solve this problem.
>>>>>>>If you need any extra information just let me know. I have a lot of different files related to this problem.
>>>>>>>Best,
>>>>>>>Pavel.
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>