Molcas Forum

FranzOst

Member

Registered: 2025-03-15

Posts: 3

Single-determinant expansion for CI states

I am considering to use Molcas in the context of our strong-field interaction code tRecX (https://doi.org/10.1016/j.cpc.2024.109279). For that I need objects like the three particle reaction density matrices between two CI states |A> and |B>:

<A| a^+a^+a^+ a a a |B>.

I do not assume that these (and similar objects) can be directly obtained from Molcas. However, our code can work from single-determinant expansions (even if expensive). For that I need  to undo the GUGA representation into single determinants (we use that with a different quantum chemistry code already).

Is there any code that partially (or fully) does that?
An interpretation of the GUGA table without the need for digging into the original code would already help a lot.

Thanks for any hint,

Franz

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,097

Re: Single-determinant expansion for CI states

Does the PRSD keyword in RASSCF help?

FranzOst

Member

Registered: 2025-03-15

Posts: 3

Re: Single-determinant expansion for CI states

Thank you for your very rapid response!

For RASSCF I could reconstruct the wave function in determinants and translate MOs to primitives as needed. What I need is MRCI, as we compute the neutral and ions on the same MO basis (neglecting nuclear relaxation on the attosecond time scale).

I have not found (yet) the full precision CI expansion coefficients, but I believe that will be solved.

What troubles me is the correct interpretation of the CSF description, e.g.

      CONFIGURATION      1   COEFFICIENT -0.010372     VALENCE
 SYMMETRY             1  1  1  1  2  2  2  3  3  4
 ORBITALS             3  4  5  6  2  3  4  1  2  1
 OCCUPATION           0  2  2  2  2  2  2  2  2  1
 SPIN-COUPLING        0  3  3  3  3  3  3  3  3  1

(1) the last keyword in the first line, here VALENCE: I suspect this is just informative, not defining, right?
(2) importantly, the spin coupling: since my first post I have advanced to the following guess
     Spin-coupling 0,3: no change of total spin, 1 increase by 1/2, 2 decrease.

The numbers 0...3 suggest their origin in the GUGA, so there may be a rather different interpretation, unfortunately I am rather unfamiliar with the whole business.

Based on my interpretation I will implement
Jeppe Olsen J. Chem. Phys. 141, 034112 (2014)
to obtain the determinant expansion.

If I am in error, I will be grateful for enlightenment, also about how to best get the full precision coefficients.

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,097

Re: Single-determinant expansion for CI states

The MRCI program is quite old and unmaintained. You may find more information with the GUGACI program instead... I'd also guess the spin-couplings correspond to the labels used in RASSCF: 0=0 (empty orbital), 1=u, 2=d, 3=2 (doubly occupied orbital).

FranzOst

Member

Registered: 2025-03-15

Posts: 3

Re: Single-determinant expansion for CI states

Thanks again.
Does it mean MRCI use is discouraged?
I tried GUGACI, which runs OK, but does not seem to print the CI vector expansion into CSFs. I find in documentation the PRINT and THRPRINT commands, same as MRCI, but different from MRCI, where I obtained the CSF description quoted above, they do not seem to have an effect in GUGACI.

Last edited by FranzOst (2025-03-19 21:11:10)