FNO-CCSDTQ with custom natural orbitals

Dear developers,

I'd like to run frozen natural orbital CCSDTQ calculations using custom natural orbitals. Is it possible in mrcc?

Thanks,
Xing

Dear Xing,
What are custom natural orbitals?

Dear Professor Kallay,

I have a FCIDUMP file including the one and two body Hamiltonians represented by some orthogonal AOs.

I also have a set of molecular/natural orbitals generated by some manipulations to the CCSD 1RDM (following this paper J. Chem. Phys. 152, 034105 (2020)).

Now I want to perform UCCSDTQ calculation within an active space defined by a subset of these orbitals.

I tried to transform the AO integrals using the active orbitals to form a Hamiltonian written to fort.55 file.
Then I set
iface = cfour
to run the calculation.
But the result seemed wrong.

I also tried to transform the AO integrals using both the active orbitals plus the inactive occupied orbitals to form a Hamiltonian written to fort.55 file.
Then I set
iface=cfour
core=number of inactive occupied orbitals
to run the calculation.
But the program did not respect the core argument and all the orbitals were treated as active.

I'm not sure what the proper way is to perform such calculations, and any help would be appreciated.

Thank you in advance.
Xing

Dear Xing,
Probably, the format of your fort.55 file is not correct. I am attaching two files that explain the format for restricted and unrestricted orbitals.

Dear Professor Kallay,

Thanks for your quick reply.

I wanted to confirm if I'm doing this correctly.

Say I have a system with 40 orbitals, 30 alpha and 28 beta electrons, and I want to freeze 10 doubly occupied orbitals.

In the fort.55 file, the first line would be 30, 38.
And the integrals are computed for all the 40 alpha/beta orbitals and written to fort.55.

Finally, in the MINP file, I set
core=10

Thank you so much for your help.

Best,
Xing

That is correct, but you do not need to specify the number of core orbitals in MINP.