EOM Calculation

I am having trouble computing the 3^Sigma_u^- state of N2 using MRCC's EOM code. In Abelian symmetry, this is the second 3^A_u state. If I simply request, say, 20 triplets, I see the lowest 3^A_u state at 8.8 eV and I see many higher ones, but I never can get this one, which is at 9.7-9.8 eV. I am able to get it in CFOUR with CCSD, but ultimately I want EOM-CCSDT results.

Since simply requesting many states isn't working, I have also tried to compute this state with a CI guess vector, but I get the following error.

My input file in the second case looks like this: where the orbital indices are based on the following output.
The guess vector is based on my CFOUR computation, where the converged eigenvector looks as follows:
based on the following orbital indices.
Do you have any suggestions for how to proceed? My output file is attached.

Unfortunately I cannot reproduce this. I could get that state using the default initial guess (see attached output). If I request 20 states, I can still get that state.

Thank you very much for testing this. It turned out that there was a bug in our version of the code. Up to that point, the code was throwing an error whenever I requested ntrip=1, but that was fixed by installing the most recent version.

However, I am still experiencing similar troubles to the one I described. In particular, I am consistently missing one or both of the triplet A2 valence states for linear molecules:

N2
The converged triplet Au roots are as follows:
Excitation energy [eV]: 9.9057
Excitation energy [eV]: 9.9070
Excitation energy [eV]: 17.1230
Excitation energy [eV]: 17.1230
Excitation energy [eV]: 21.9653
Excitation energy [eV]: 25.7402
Excitation energy [eV]: 25.7401
Excitation energy [eV]: 24.7251
Excitation energy [eV]: 25.4102
Excitation energy [eV]: 33.2781

Missing is the 3Delta_u state at 9.0 eV.

CO
The converged triplet A2 roots are as follows:
Excitation energy [eV]: 9.3187
Excitation energy [eV]: 9.3187
Excitation energy [eV]: 14.8049
Excitation energy [eV]: 14.8048
Excitation energy [eV]: 15.4686
Excitation energy [eV]: 17.1859
Excitation energy [eV]: 17.1864
Excitation energy [eV]: 21.8488
Excitation energy [eV]: 21.8798
Excitation energy [eV]: 22.8377

Missing is the 3Sigma^- state at 9.9 eV.

HCN
The converged triplet A2 roots are as follows:
Excitation energy [eV]: 11.6202
Excitation energy [eV]: 11.6201
Excitation energy [eV]: 13.7495
Excitation energy [eV]: 13.7484
Excitation energy [eV]: 13.8701
Excitation energy [eV]: 13.7485
Excitation energy [eV]: 16.3399
Excitation energy [eV]: 16.4657
Excitation energy [eV]: 15.6129
Excitation energy [eV]: 18.9638

Here both valence states are missing. The Sigma^- state should be at 8.3 eV and the Delta state should be at 7.5 eV. Here I do see that the code is nearly converging to the Delta state during the fifth and sixth CI root computations, where the CI energy briefly approaches -93.091 Hartree before jumping to a higher state.

HNC
The converged triplet A2 roots are as follows:
Excitation energy [eV]: 8.1161
Excitation energy [eV]: 8.1161
Excitation energy [eV]: 11.8554
Excitation energy [eV]: 11.8565
Excitation energy [eV]: 12.0058
Excitation energy [eV]: 12.0058
Excitation energy [eV]: 14.0108
Excitation energy [eV]: 17.1046
Excitation energy [eV]: 17.1047
Excitation energy [eV]: 19.7559

Missing is the 3Sigma^- state at 8.7 eV.

HCCH
The converged triplet Au roots are as follows:
Excitation energy [eV]: 6.4887
Excitation energy [eV]: 6.4887
Excitation energy [eV]: 10.9156
Excitation energy [eV]: 10.9157
Excitation energy [eV]: -0.0002
Excitation energy [eV]: 17.8306
Excitation energy [eV]: 17.7395
Excitation energy [eV]: 17.7392
Excitation energy [eV]: 20.2386
Excitation energy [eV]: 19.3533

Missing is the 3Sigma_u^- state at 7.2 eV.


I have attached some output files. Is this just a defect of the algorithm? In that case, could you give me any pointers on what I did wrong in my guess vector for N2? If I can get the guess vectors working for N2 I think I'll be able to figure it out for the other systems.

Probably you should check the irreps you use in the calculations. E.g., for N₂, with symm=B_1u I am getting the lowest triplet state at 9.0 eV

I am aware of this. That works for N2, because the D2h projection of Delta _u is Au + B1u, but that does not help me for the other cases.

I've attached outputs for the other molecules listed in my previous post.

1) You will find an updated mrcc.f in the download area. Please recompile the program using this file.
2) You will get the desired roots if you restart the CCSD calculation from the converged CCS roots. For example, use the rest=2 option:
calc=CCSD
symm=A2
ntrip=5
rest=2
Alternatively, you can also try rest=3 to select specific roots, see manual.