Dear All,
I have been trying to compute the electric transition dipole moments from the ground state to the excited states and the magnetic transition dipole moments between the excited states (the later by restarting the quadratic response from the linear response). I have met rather unexpected results when I used different point groups of symmetry  the magnetic transition dipole moment values I obtained were extremely different.
I have carried out the following experiment:
1. For the planar geometry of pyromellitic diimide (D2h point group) I have changed a little the orientation of two protons on the opposite site of the molecule as I get the geometries which belong to three different point groups: C2h, C2v and C2.
2. For each case I have calculated the electric transition dipole moments and the magnetic transition dipole moments between the excited states.
Everything seems good in the C2h symmetry case. However, there is a problem with solving the response equations for two remaining ones (giving an example of the error messages from the output file:
WARNING: Solution vector on RSPVEC converged to 1.88D+00
WARNING: which is less than desired convergence 1.00D03).
I should add that no error message was given in the linear response calculations.
I have tried to increase the maximum number of iterations for linear equations in the *QUADRATIC section (.MAXITL), but it didn’t solve the problem.
How is it possible that so little change in the orientation of two protons is causing such a discrepancy? Could it be connected with the presence of the inversion centre (I mean for the C2h point group where the inverse centre is present the computations finished with no error message opposite to the C2 and C2v ones)?
Please, help me to deal with this problem.
Best regards,
Mercedes Kukułka
P.S. I am using Dalton2015.1.
Magnetic transition dipole moments vs. symmetry of molecule

 Posts: 3
 Joined: 10 Feb 2015, 10:19
 First name(s): Mercedes
 Last name(s): Kukułka
 Affiliation: Jagiellonian University
 Country: Poland
Magnetic transition dipole moments vs. symmetry of molecule
 Attachments

 qrcalc_c2v.out
 (437.91 KiB) Downloaded 325 times

 qrcalc_c2h.out
 (124.25 KiB) Downloaded 367 times

 Posts: 590
 Joined: 15 Oct 2013, 05:37
 First name(s): Peter
 Middle name(s): Robert
 Last name(s): Taylor
 Affiliation: Tianjin University
 Country: China
Re: Magnetic transition dipole moments vs. symmetry of molec
A couple of points. Your C2v and C2h calculations are not strictly comparable: in one you ask for eight roots, all in one symmetry (unlikely to be all physically meaningful, by the way), and in the other you ask for six each in two symmetries. The chosen irrep in the first case is symmetric with respect to the molecular plane, as are the two you select in the latter, so one might well expect the eight roots sought in the former to turn up among the twelve roots total you ask for in the latter, although I see no guarantee of this. So there is a difference between the two calculations already.
Second, and possibly related to this, is that it is quite clear that in the C2v calculation the first attempt to solve the "LINEAR EQUATIONS FOR LINEAR RESPONSE PROPERTIES" encounters a singular matrix for the lefthand side of the equations. It is true that the program simply identifies a matrix with at least one negative eigenvalue, but I have not looked at the code to discover what is considered to be "zero" for this purpose. In any event the determinant of the reduced matrix is ludicrously large, the equations are basically indeterminate, and the iterative process does not converge at all (the residual is essentially unchanged).
I think the first thing to do is to verify whether it is necessary or desirable to take out twelve roots total to span the solutions of your C2h calculation. It may be that one of the twelve is problematic and is causing this illconditioned behaviour (perhaps some sort of resonance or nearsingular denominator). But at present I think the fact that the one calculation "works" and the other doesn't is not particularly meaningful because they are not equivalent calculations.
Best regards
Pete
Second, and possibly related to this, is that it is quite clear that in the C2v calculation the first attempt to solve the "LINEAR EQUATIONS FOR LINEAR RESPONSE PROPERTIES" encounters a singular matrix for the lefthand side of the equations. It is true that the program simply identifies a matrix with at least one negative eigenvalue, but I have not looked at the code to discover what is considered to be "zero" for this purpose. In any event the determinant of the reduced matrix is ludicrously large, the equations are basically indeterminate, and the iterative process does not converge at all (the residual is essentially unchanged).
I think the first thing to do is to verify whether it is necessary or desirable to take out twelve roots total to span the solutions of your C2h calculation. It may be that one of the twelve is problematic and is causing this illconditioned behaviour (perhaps some sort of resonance or nearsingular denominator). But at present I think the fact that the one calculation "works" and the other doesn't is not particularly meaningful because they are not equivalent calculations.
Best regards
Pete

 Posts: 590
 Joined: 15 Oct 2013, 05:37
 First name(s): Peter
 Middle name(s): Robert
 Last name(s): Taylor
 Affiliation: Tianjin University
 Country: China
Re: Magnetic transition dipole moments vs. symmetry of molec
By the way, there is no need to perturb the geometry if you want to study the effect of reducing the symmetry. If you specify the symmetry explicitly yourself you can specify any symmetry from the full group (apparently D2h for your parent molecule) to C1, simply by choosing how many and which generators you specify. This is documented in the manual and in various postings.
Best regards
Pete
Best regards
Pete
Who is online
Users browsing this forum: xiongyan21 and 9 guests