Re: Sorry, but once again about SDCI

Posted by perake on November 22, 2001 at 09:18:13:

In Reply to: Sorry, but once again about SDCI posted by Evgeniy on November 21, 2001 at 14:35:39:

Dear Evgeniy,
I may have misunderstood your question. First about using MRCI:

: Thanks a lot for your detail answers.
: If I get You right there are certain "difficulties" to carry out single-reference
: SDCI (what is rather surprisingly) using MRCI program.

There should be no such difficulties for any ordinary single-reference SDCI. But your
request was for getting also excited states from that SDCI, moreover, states of
other symmetry. For the first case, MRCI was not built for experiments but for
application calculation. It was not then assumed that anyone would want to get also
such excited states that are not properly approximated already by the reference CI.
The calculation you suggest would usually be done by having more than a single reference
function, i.e. MR-SDCI, which is what the program is built to do. As for different symmetry,
the natural thing would be to do several CI calculations, with different symmetries; but
again, the calculation you ask for would require the states of other symmetries to appear
as the excited states of the SDCI, which means that they would have to be treated in the
same calculation.

Perhaps if you explained why you need to do this particular type of calculation, I may
give you a better answer.

: But somehow You did not mention about LUCITA module.

Well, I do not know much about LUCITA, which is the reason I do not write to a billboard
about it when this was not even asked. I suggest you write MOLCAS and ask, or directly to
the authors of that program.

But I have another suggestion: If the problem does not involve very many correlated and
virtual orbitals, it can be done by the RASSCF module. The keyword CIONLY will prevent
orbital update, the correlated orbitals that were occupied in the reference are in RAS1,
the correlating orbitals that were unoccupied in the reference are in RAS3. RAS2 is
empty. You allow at most two holes in RAS1, and two electrons in RAS3.

Good luck,

Per Ake

Follow Ups:

Post a Followup

Name:
E-Mail:

Subject:

Comments:
: Dear Evgeniy, : I may have misunderstood your question. First about using MRCI: : : Thanks a lot for your detail answers. : : If I get You right there are certain "difficulties" to carry out single-reference : : SDCI (what is rather surprisingly) using MRCI program. : There should be no such difficulties for any ordinary single-reference SDCI. But your : request was for getting also excited states from that SDCI, moreover, states of : other symmetry. For the first case, MRCI was not built for experiments but for : application calculation. It was not then assumed that anyone would want to get also : such excited states that are not properly approximated already by the reference CI. : The calculation you suggest would usually be done by having more than a single reference : function, i.e. MR-SDCI, which is what the program is built to do. As for different symmetry, : the natural thing would be to do several CI calculations, with different symmetries; but : again, the calculation you ask for would require the states of other symmetries to appear : as the excited states of the SDCI, which means that they would have to be treated in the : same calculation. : Perhaps if you explained why you need to do this particular type of calculation, I may : give you a better answer. : : But somehow You did not mention about LUCITA module. : Well, I do not know much about LUCITA, which is the reason I do not write to a billboard : about it when this was not even asked. I suggest you write MOLCAS and ask, or directly to : the authors of that program. : But I have another suggestion: If the problem does not involve very many correlated and : virtual orbitals, it can be done by the RASSCF module. The keyword CIONLY will prevent : orbital update, the correlated orbitals that were occupied in the reference are in RAS1, : the correlating orbitals that were unoccupied in the reference are in RAS3. RAS2 is : empty. You allow at most two holes in RAS1, and two electrons in RAS3. : Good luck, : Per Ake

[ Follow Ups ] [ Post Followup ] [ Molcas user's WWWBoard ]