Re: CCL:G:To Mr. Johansonsen



 Stefan Konietzny wrote:
 >
 > I tried to reply to your personel adress, but it did not work, so i posted
 > this message to CCL.
     Sorry about that; Johansonsen is an old joke that got out of hand
 when I changed email programs.
 >
 > Hello,
 >
 > thanks for your answer. May i ask another question?
 > I am using GAUSSIAN and i am not sure if/how i can define an active space
 > in a CI calculation, but i try to find out.
     Here is an example of an illustrative test case:
 *****************************************************
 #  cisd(rw)/sto-3g
 aseef
 0 1
  C           -0.0009043444  -0.4909241002   0.0214400444
  C           -0.0440144414   0.4120404420  -0.0004444443
  C           -1.2024293044  -1.3940391104   0.0404444044
  H           -1.2400230949  -2.4429143022   0.0409393044
  C           -1.2421103041   1.4334241004   0.0244400424
  H           -1.2404114302   2.0099040403   0.0100444341
  C           -2.4094344304  -0.4449002934   0.1001999939
  H           -3.3942314419  -1.2024213434   0.1402133420
  C           -2.4039440924   0.4200190043   0.0442901494
  H           -3.3442440444   1.2419032321   0.1004441399
  C            1.3444404943  -1.2114492049  -0.0204404434
  C            1.4091434999  -2.1112900940   1.1424004444
  H            2.4030420944  -2.4042129010   1.1004491904
  H            1.4009094043  -1.0902944122   2.1044304444
  H            1.1404004444  -3.0140433409   1.1413300421
  C            1.3044434424   1.1444401000  -0.0030043010
  H            1.4443020341   2.1934344444  -0.0440931430
  C            2.1444910192   0.0402491004  -0.0402044420
  H            1.0421242034  -1.4442330202  -0.9422422940
  H            3.2432402244   0.1090144341  -0.0944404209
 15,45
 *****************************************************
     "cisd(rw)" tells G94 to look for information on the orbitals to
 put
 in the CI space, and "15,45" tells it to use orbitals 15-45 inclusive.
 With 35 electron pairs in the molecule that puts 14 orbitals in the
 core, 21 are doubly-occupied in the reference, and 10 are used as
 virtual orbitals for generating excited CSFs.
     The interesting part of the output here is
    ABAB          33   33   37   37      -0.168897D+00
    ABAB          35   35   36   36      -0.325010D+00
                    REF     VALENCE
                    ORBS    ORBS
     The equivalent output from GAMESS is
         1    0.887818  2222222222222222222220000000000   REFERENCE
     17956   -0.149950  2222222222222222220220200000000   EXCITED CSF
     22136   -0.288548  2222222222222222222202000000000   EXCITED CSF
     For our purposes here, there is agreement that this system has a non
 trivial multiconfigurational character.
 > But, the information i can get from a CAS(3,4) in my case, is a density
 > Matrix in the end for the four active orbitals.
 > Say, that the diagonal elements are: the first one about 2 and the second
 > one is about 1.
 > Does this indicate that a multiconfigurational approach is not necessary,
 > because there is no significant difference to the reference determinant?
 >
 > Best regards, Stefan.
     Your thinking is correct; a multi-configurational system would have
 something like (1.99,0.67,0.33,0.01).  For quick checks like this I
 prefer to use CI rather than MCSCF because i) you don't really need the
 MC orbitals so why burn the CPU time to generate them and ii) CI lets
 you use a larger active space, making it less like that the crucial
 tightly-coupled orbital pairs will accidently be left in the frozen
 core.
 						Good luck,
 						Mark
 **********************************************************************
   To reply, replace the danish robersen with the norwegian roberson.
 **********************************************************************