Convergence issues when running RESPONSE CUBIC

Find answers or ask questions regarding Dalton calculations.
Please upload an output file showing the problem, if applicable.
(It is not necessary to upload input files, they can be found in the output file.)

Post Reply
rlt92
Posts: 11
Joined: 20 Nov 2018, 12:29
First name(s): Richard
Last name(s): Thurston
Affiliation: UC Berkeley
Country: United States

Convergence issues when running RESPONSE CUBIC

Post by rlt92 » 07 Dec 2019, 07:12

Hi this maybe as simple as increasing the iteration number but I would like some help debugging this issue. I've been running MCSCF calculations to compute the frequency dependence of the third order response for nitrobenzene and some of those calculations haven't been converging. The confusing thing for me is that only some of the calculations failed to converge where as others converged just fine when the only thing I changed between runs is the frequency. What's more, while those calculations that converged give the shape of a curve that I expect, I'm also confused because when I compare the results to experimental results in the literature the poorly converged values give better agreement with experiment than converged values.

At this point I'm distrusting of a lot of my process. I'm no longer sure if the type and size of the basis set I've chosen is correct for this calculation, if my active space is too small, or if the root of the problem is that I just need to increase the .MATITO to a larger value.

Attached is a pair of output files from a calculation that did converge and one that did not converge. Any help diagnosing this issue would be greatly appreciated.
Attachments
freq-0.033_cubic_response_NBopt_dunningZ-2_not-converged.out
(255.67 KiB) Downloaded 85 times
freq-0.027_cubic_response_NBopt_dunningZ-2_converged.out
(264.86 KiB) Downloaded 91 times

hjaaj
Posts: 379
Joined: 27 Jun 2013, 18:44
First name(s): Hans Jørgen
Middle name(s): Aagaard
Last name(s): Jensen
Affiliation: Universith of Southern Denmark
Country: Denmark

Re: Convergence issues when running RESPONSE CUBIC

Post by hjaaj » 17 Dec 2019, 16:39

I have looked at your output files. Your choice of active space is very unbalanced and that is probably what causes convergence problems in the response part. This unbalanced active space also means I would not put much confidence in the calculated gamma values. It will not help to increase .MAXITO in this case, because it does not exit because max number of iterations exceeded.

The predicted natural orbital occupation numbers from MP2 show that the following should constitute a balanced active space choice:

Code: Select all

.INACTIVE ORBITALS
16 11 2 0
.CAS SPACE
0 0 3 3
.ELECTRONS
 6

Post Reply

Who is online

Users browsing this forum: Bing [Bot] and 7 guests