Manual:RELADC:Sample inputs

From Diracwiki

Example 1:

A bent triatomic molecule in C_s symmetry. We have only one fermionic IRREP and complex MOs. This is indicated in the ADC output reporting complex ADC matrices. We want to do single ionizations (DOSIPS=T) at the ADC(3) level (ADCLEVEL=3) including constant diagrams (DOCONST=default) and compute all final states in symmetry 1, the only available one (XREPS=1). If the basis set is large, we restrict the iterations for the constant diagrams by releasing convergence a bit (VCONV=1.0E-05) beacause this step takes longest. We request 1000 Lanczos iterations and consider ionization energies up to 100 eV on screen. The complete spectrum is written to SSPEC.1. You get as many entries as you have Lanczos iterations. A characteristic of the iterative diagonalizer is that the eigenvalues at the edge of the spectrum converge very fast and are reproduced for high iteration numbers. In order to get higher eigenvalues equally tightly converged the number of iterations has to be increased accordingly. In this case the spurious (already converged) eigenvalues are projected out. For the C_s molecule the ADC input would then look like this:

$RELADC DOSIPS=T, ADCLEVEL=3, XREPS=1, VCONV=1.0D-05 $END
$LANINP SIPITER=1000, SIPPRNT=100.0 $END

Since we do not activate DIPs nothing else needs to be specified in the &LANINP Namelist.

Example 2:

The cadmium atom in D_2h with .SPINFREE Hamiltonian. The D_2h double group has two fermionic IRREPS (g and u) leading to XREPS=1,2. By using the .SPINFREE option a real calculation in the ordinary D_2h point group is performed in DIRAC leading to eight symmetries as known from the character table. Therefore we set XREPS=1,2,3,4,5,6,7,8 if we need all symmetries. The relation of the irrep name to the number can be obtained from the Dirac output and is reproduced in the ADC code as well. After the calculation the A_1g final states are the in SSPEC.01, the B_2g in SSPEC.02 asf. The complete spectrum is then the merge of SSPEC.01 ... SSPEC.08. Remember that for the plot we only need the IP and the corresponding pole strength. We therefore do a grep '@' on the merged SSPEC.X files (cat SSPEC.* > SPEC.all) and obtain all lines in each symmetry. If we need only a specific range we do "sort -n SPEC.all > SPEC.range" and edit according to our needs. Then we can use gnuplot with "plot "SPEC.range" u 1:2 w i". For ADC(3) with constant diagrams and 600 Lanczos iterations we would have an input like this:

**DIRAC
.TITLE
 input for Cd atom
.WAVE F
.4INDEX
**GENERAL
.DIRECT
 1 1 1
**INTEGRALS
*READIN
*TWOINT
.SOFOCK
.SCREEN
1.E-16
**HAMILTONIAN
.SPINFREE
**WAVE FUNCTIONS
.DHF
.RELADC
*DHFCAL
.CLOSED
 30 18
.FCKCNV
5.0E-09
.INTFLG
1 1 1
**MOLTRA
.INTFLG
1 1 1
.CORE
1..8
1..6
.ACTIVE
9..25
7..22
*END OF
$RELADC DOSIPS=T, ADCLEVEL=3, XREPS=1,2,3,4,5,6,7,8, $END
$LANINP SIPITER=600 $END

Back to RELADC manual.