Impression of the A and B matrices of the Polarization Propagator in RPA

[DaniBajac]

Hello, I would like to print the matrices A and B of the Principal Propagator in the RPA aproximation, defined as eq. 3.6 and 3.7 in the work of Oddershede et. al. Computer Physics Reports 2 (1984) 33-92.
So far, I printed the whole principal propagator (REDUCED E(2)-W*S(2) MATRIX) with the .dal (for example, for a HF molecule), but I need the matrix A and B separately.

**DALTON
.DIRECT
.RUN RESPONSE
**INTEGRALS
.FC
**WAVE FUNCTIONS
.HF
**RESPONSE
.TRPFLG
*LINEAR
.PROPRT
FC F1001
.PROPRT
FC H1002
.PRINT
31
*END OF

[hjaaj]

There is a debug keyword, which will calculate the triplet A and B separately in the original orbital basis:

Code: Select all

**RESPONSE
.TRPFLG   ! singlet A and B without this keyword
*LINEAR
.ABCHK

[DaniBajac]

Sorry but I can't print the matrices A and B. Specifically, I only need the two-body components of that matrices: A(1) y B(1).
I used the keyword .ABOCHK instead of .ABCHK, because the last is used with .SINGLE RESIDUE

[DaniBajac]

Hi Mr. Hans Jensen.

It took me a while to understand that the matrices FCOCO and FCOEX represents the Coulomb and Exchange contributions to the Fock matrix.
However, the FCOCO matrix always returns the zero matrix (for the HF and H2O2 molecules). Why could this be happening?
On the other hand, in the file fone.F, I saw that FCOEX = FCOCO + FCOEX. This may indicates that the matrix FCOEX of the output represents both contribution?
Finally, can I consider that the matrices FCOCO and FCOEX are the A and B matrices of the principal propagator?

Best regards
Daniel

[hjaaj]

Note that the A and B matrices for .SINGLE RESIDUE are the same as for linear response (e.g. for FC).

The FCOCO and FCOEX contain Fock matrix type contributions to the A and B matrices. Note that the program will iteratively solve the linear response equations and not calculate the A and B matrices explicitly. The .ABCHK will calculate the A and B matrices for debugging a column at the time by using trial vectors with one and only one non-zero element (corresponding to the current column in A and B). FCOCO and FCOEX are intermediates for efficiency for trial vectors having many non-zero coefficients.

There is no easy way of getting only the two-electron contributions to A and B.

[DaniBajac]

OK Mr. Jensen, I understand. Thank you for the clarification about this themes.
However I'm going to keep looking this matrices because I need the two electron contributions to A and B for my investigation.

Thanks a lot for your answers!
Daniel