I have checked, but didn't find this problem, there is no sign convention issue in the current implementation. Perhaps the equation you used to evaluate transition dipole moment based on TrESP is incorrect. The correct code should like as follows:

    xdip=0
    ydip=0
    zdip=0
    do iatm=1,ncenter
        xdip=xdip+a(iatm)%x*a(iatm)%charge
        ydip=ydip+a(iatm)%y*a(iatm)%charge
        zdip=zdip+a(iatm)%z*a(iatm)%charge
    end do
    write(*,*)
    write(*,"(' Dipole moment (a.u.): ',3f14.6)") xdip,ydip,zdip

where a%charge directly corresponds to the TrESP charges given by Multiwfn, and no additional negative sign should be included.

Best regards,

Tian

Thank you very much for developing and maintaining such a great program. Multiwfn’s ability to work with many different quantum chemistry packages is really impressive.

I noticed something unusual when using TrESP charges from quantum chemistry software like Gaussian or ORCA, followed by transition density analysis in Multiwfn. The transition dipole calculated by projecting the charges onto atomic positions seems to point in a different direction compared to the transition dipole moment vector reported by the quantum chemistry software.

I understand the direction of the transition dipole can be arbitrary due to phase conventions, but since both results are based on the same wavefunction, I expected consistency. It seems there may be a sign mismatch in how Multiwfn interprets the transition density compared to the original quantum chemistry software.

Thanks.

Best regards,
Sayan