Polarizability Calculation Not Converging

1

I have three radical anion molecules (Total charge = -3, Multiplicity = 4) that I am trying to calculate the polarizability for, and I am having trouble converging it. The current output is:

 ==> Requested Responses <==

    Dipole polarizabilities

   ==> Coupled-Perturbed UKS Solver <==

    Maxiter             =         100
    Convergence         =   1.000E-06
    Number of equations =           6
   -----------------------------------------------------
     Iter   Residual RMS      Max RMS  Remain  Time [s]
   -----------------------------------------------------
    Guess      1.032e+02    1.393e+02       3      1619
        1      1.664e+01    2.245e+01       3      3236
        2      5.781e+01    6.663e+01       3      5059
        3      3.228e+02    3.774e+02       3      6860
        4      1.201e+03    1.443e+03       3      8637
        5      3.459e+03    4.274e+03       3     10391
        6      8.420e+03    1.068e+04       3     12131
        7      1.818e+04    2.363e+04       3     13902
        8      3.586e+04    4.767e+04       3     15638
        9      6.591e+04    8.948e+04       3     17274
       10      1.144e+05    1.584e+05       3     19079
       11      1.896e+05    2.672e+05       3     20914
       12      3.019e+05    4.328e+05       3     22541
       13      4.650e+05    6.771e+05       3     24401
       14      6.956e+05    1.028e+06       3     26166
       15      1.015e+06    1.521e+06       3     27909
       16      1.448e+06    2.198e+06       3     29631
       17      2.025e+06    3.113e+06       3     31462
       18      2.783e+06    4.328e+06       3     33336
       19      3.766e+06    5.921e+06       3     35053
       20      5.022e+06    7.979e+06       3     36850
       21      6.611e+06    1.061e+07       3     38743
       22      8.601e+06    1.393e+07       3     40649
       23      1.107e+07    1.809e+07       3     42478
       24      1.410e+07    2.325e+07       3     44284
       25      1.780e+07    2.959e+07       3     46066
       26      2.228e+07    3.734e+07       3     47883

I had no trouble converging the geometry.

I would appreciate some advice.

2

I suspect you’ve landed on an excited SCF solution. What DFT functional are you using? I’d also try to start this computation without guess orbitals at the converged geometry and see if you converge to the same SCF solution as in this case.

3

I bet you’re right.

I’m using–> DFT; B3LYP-D3BJ; aug-cc-pVTZ.

Even though it did converge for the optimization, I was holding certain portions fixed. And, I also didn’t working slowly up to this level of theory (which I usually do). Lesson learned.

Other than calculating the Hessian, do you have a go-to calculation for checking if the system is in an excited state?

4

My normal go-to is to do a stability analysis, which detects whether you are in a local minima on the electronic state. Rather than compute the entire Hessian, we compute its lowest eigenvectors directly.

Unfortunately, it doesn’t look like Psi supports these calculations yet. Disappointing. This is something I’d like to add (we can reuse a great deal of the TD-DFT infrastrucure for this), but I’m doing some cleanup in the CC code and can’t commit to taking on that project for a few months.