Re: Offdiagonal Dipole Moment Matrix Element.

Posted by Chaehyuk Ko on September 27, 2005 at 22:46:34:

In Reply to: Re: Offdiagonal Dipole Moment Matrix Element. posted by Roland Lindh on September 26, 2005 at 20:57:02:

Thank you for the answer. I would like to ask a few more questions to make sure I understand the things correctly.

1. According to your answer, it seems that the offdigonal dipole moment maxtrix element(e.g, mu_x between S1 and S2)can not be calculated at CASPT2 level. Is this right?

2. In CASPT2 property section, MolCAS prints out 1-st cartesian moment(dipole moment). I assume this is calculated in the way you explained (i.e, FFPT is used although I didn't tell MolCAS to use it.), and the wavefunction used for the dipolemoment calculation is obtained at MSPT2 level (not single state CASPT2). Do I understand this correctly?

3. If the 1-st catesian moments are for the MSPT2 states, I think the ordering of the states for the dipole moment should be the same as in that of MSPT2 eigenvectors which is in increasing order of eigenvalues in MULTI-STATE CASPT2 SECTION. Am I right ?

The reason I am asking this is I want to obtain the dipole moment change on electronic transition (compared to S0) at MSPT2 level with the solvation model.

Thanks.

: Hi,
: the CASPT2 method is not a variational method, hence properties can not be computed as expectation values as a simple trace of the density matrix with the property integrals. One common way to get around this problem is to use a so-called finite field technique. In this one apply a small electric field to the molecule and register the change of the energy. The energy is Taylor expaned with respect to the electric field. The first derivative in this series is the dipole moment. Do as follows: apply a small electric field, both plus and minus, along each of the cartesian direction (note that if your molecule is symmetric along one direction that you have to remove this symmetry, better and simpler remove any explicit symmetry specification). Compute the dipole moment numerically from

: mu_x = (E(deltaF_x) - E(-deltaF_x))/(2*deltaF_x)

: Look at the manual for the FFPT module for further instructions.

: Take care,

: -roland

Follow Ups:

Post a Followup

Name:
E-Mail:

Subject:

if B is 1s22s22p1, what is Li?

Passfield:

Comments:
: Thank you for the answer. I would like to ask a few more questions to make sure I understand the things correctly. : 1. According to your answer, it seems that the offdigonal dipole moment maxtrix element(e.g, mu_x between S1 and S2)can not be calculated at CASPT2 level. Is this right? : 2. In CASPT2 property section, MolCAS prints out 1-st cartesian moment(dipole moment). I assume this is calculated in the way you explained (i.e, FFPT is used although I didn't tell MolCAS to use it.), and the wavefunction used for the dipolemoment calculation is obtained at MSPT2 level (not single state CASPT2). Do I understand this correctly? : 3. If the 1-st catesian moments are for the MSPT2 states, I think the ordering of the states for the dipole moment should be the same as in that of MSPT2 eigenvectors which is in increasing order of eigenvalues in MULTI-STATE CASPT2 SECTION. Am I right ? : The reason I am asking this is I want to obtain the dipole moment change on electronic transition (compared to S0) at MSPT2 level with the solvation model. : Thanks. : : : : Hi, : : the CASPT2 method is not a variational method, hence properties can not be computed as expectation values as a simple trace of the density matrix with the property integrals. One common way to get around this problem is to use a so-called finite field technique. In this one apply a small electric field to the molecule and register the change of the energy. The energy is Taylor expaned with respect to the electric field. The first derivative in this series is the dipole moment. Do as follows: apply a small electric field, both plus and minus, along each of the cartesian direction (note that if your molecule is symmetric along one direction that you have to remove this symmetry, better and simpler remove any explicit symmetry specification). Compute the dipole moment numerically from : : mu_x = (E(deltaF_x) - E(-deltaF_x))/(2*deltaF_x) : : Look at the manual for the FFPT module for further instructions. : : Take care, : : -roland : : -roland

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