Enormous TDDFT TPA cross sections for high states.

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
Addiw7
Posts: 77
Joined: 05 Apr 2016, 17:47
First name(s): Dawid
Last name(s): Grabarek
Affiliation: Wroclaw University of Technology
Country: Poland

Enormous TDDFT TPA cross sections for high states.

Post by Addiw7 » 12 Apr 2016, 19:59

Dear Dalton Experts,

I wanted to perform TDDFT calculations of two-photo absorption cross section for green fluorescent protein (GFP) chromophore. Although the calculations finished successfully I have serious
concerns regarding the reliability of values that I get for excitations to states over 25th.
The numbers that I get after conversion to GM are ridiculous, like millions up to ca. 20 billions GM. What is more when I add nearby residues to the chromophore I obtain TPA cross section
that is ridiculously large for transition to lower states.

In this case I am wondering whether my calculations make sense at all. Should I use some special keywords or thresholds when I perform calculations for such large number of states?

I am attaching one of my output files so you could have a look. I will be very grateful for any help!
Attachments
DALTON_MOLECULE.out
TPA calculations output file (CAM-B3LYP/6-31+G*)
(422.77 KiB) Downloaded 257 times

taylor
Posts: 545
Joined: 15 Oct 2013, 05:37
First name(s): Peter
Middle name(s): Robert
Last name(s): Taylor
Affiliation: Tianjin University
Country: China

Re: Enormous TDDFT TPA cross sections for high states.

Post by taylor » 12 Apr 2016, 20:28

A calculation that asks for 30 excited states, especially with TDHF or TDDFT, is guaranteed to produce, well, brutally, garbage. I have posted on this so often (you can search through my forum postings) I refuse to do it in detail again, but I will summarize:
1) TDDFT <-> singly excitations with respect to the ground state. It is all TDDFT or TDHF is capable of describing. The notion that the lowest 30 excited states of any system are all single excitations from the ground state is fantasy.
2) Rydberg states: basis set issues, as well as what DFT functionals are capable of describing.

Best regards
Pete

Addiw7
Posts: 77
Joined: 05 Apr 2016, 17:47
First name(s): Dawid
Last name(s): Grabarek
Affiliation: Wroclaw University of Technology
Country: Poland

Re: Enormous TDDFT TPA cross sections for high states.

Post by Addiw7 » 12 Apr 2016, 20:45

I thought so. I am pretty sure that recently I have seen a post from you on that topic but I cannot find it again.

Anyway I will keep on looking.

kennethruud
Posts: 252
Joined: 27 Aug 2013, 16:42
First name(s): Kenneth
Last name(s): Ruud
Affiliation: UiT The Arctic University of Norway
Country: Norway

Re: Enormous TDDFT TPA cross sections for high states.

Post by kennethruud » 12 Apr 2016, 20:55

In addition to the problem of reliably describing highly excited states as Peter mentions, one can also add that as you enter a part of the excited-state spectrum where the density of state is high, you will have very small differences in energy between different excited states, potentially leading to divergencies in the response formalism. This latter problem can be circumvented using the complex polarization propagator approach, but I do not think such a code is yet part of the Dalton release.


Best regards,

Kenneth

Addiw7
Posts: 77
Joined: 05 Apr 2016, 17:47
First name(s): Dawid
Last name(s): Grabarek
Affiliation: Wroclaw University of Technology
Country: Poland

Re: Enormous TDDFT TPA cross sections for high states.

Post by Addiw7 » 12 Apr 2016, 21:06

I have found one of Prof. Taylor's answers to similar issues. I couldn't find it before because I didn't know it was posted by him. Anyway I am sorry for stating the same question again and
again when it had already been answered. I know it is annoying but truly, I couldn't find it.

Prof. Ruud, I am happy you entered the conversation and thank you for your comment. However in your work on YFP and impact of the distance between the chromophore and
nearby tyrozine residue you managed to calculate as much as 35 and 20 states with RI-CC2 and CAM-B3LYP, respectively! Didn't you run into similar issues as I did in course
of these calculations? Have you tried more states in either case?

Best regards,
Dawid Grabarek

taylor
Posts: 545
Joined: 15 Oct 2013, 05:37
First name(s): Peter
Middle name(s): Robert
Last name(s): Taylor
Affiliation: Tianjin University
Country: China

Re: Enormous TDDFT TPA cross sections for high states.

Post by taylor » 13 Apr 2016, 12:35

I did not mean to imply, by saying I had posted this too many times already, that you were at fault for not reading all the hundreds of messages on the forum just to find my postings on excited-state calculations, it is more that I am too lazy to just re-post things in full detail when they are already out there! But let me be clear: I was not addressing what the program can do, but what it is meaningful, or at least plausible, for a user to do. Kenneth makes the very good practical point that using Dalton (or indeed any program) is going to encounter difficulties, and thus require special treatment, when there are such things as near-singular denominators in response or perturbation theory-type expressions: resonances or near-resonances, for instance. Such issues may well arise when trying to converge, say, 40 excited states, but equally there may well be ways, for some molecules, of persuading the program to indeed converge all those roots.

This is not the issue I was concerned about in my posting (and earlier postings). My concern is with the question not of whether the program can be made to converge 40 excited states, but with how many of those states are being meaningfully described in the calculation? (Near-)resonances aside, the likelihood of any physical significance in many of these solutions is vanishingly small. Unless the basis set is suitably extended it is unlikely that the Rydberg spectrum of the system (at least assuming it's a neutral species) will be described adequately, which not only means Rydberg states will appear at the wrong energies, but, more insidiously, the mixing between valence-like and Rydberg states will be wrong. And I have already referred to the issue of multiply-excited states (which occur in the spectrum of some small aromatic systems at less than 5eV above the ground state, just as a small-molecule example) and which are simply not able to be described by single-configuration response methods.

The late Bjoern Roos used to (quite correctly) point out how difficult it was to set about calculating excited states adequately, and one could argue there have been few computational quantum chemists with as much expertise in this area as Bjoern. It is not that difficult, especially with Dalton and the effort that has gone into ensuring robust convergence of various iterative procedues, to find a lot of states, but do they mean anything? Especially given the basis set used, and the many-electron treatment? I will close with an anecdotal example from my youth...

Consider the aluminium atom (note that I use the correct-as-agreed-by-IUPAC spelling...). The Al ground state is 3s23p, and a naive lad like myself would think, comparing to what happens in B atom, there would be a low-lying excited electronic state with overall symmetry 2D from the occupation 3s3p2. And if you calculate the lowest 2D state with a small basis indeed there is. Now add a couple of diffuse d basis functions. You find that an occupation like 3s23d is now the lowest 2D state, but at least the next, or next-but-one, 2D state is derived from 3s3p2. Now add a couple more d functions. You get the start of a Rydberg series 3s24d, 3s25d, etc. And if you keep going, well, that's all you get! Rydberg states! There is no "valence-like" 2D state at all (unlike B). If we now turn to experiment, this is consistent with e.g. Charlotte Moore's tables: spectroscopists conclude that all of the 2D states are part of a Rydberg series and there is no 3s3p2 state. But if we had not tackled this at first sight simple atomic problem with appropriate concerns, and basis set issues, in mind, we would have established to our apparent satisfaction that the nth (second, third, whatever) 2D state was derived from 3s3p2, yet this result would be completely bogus. If you are calculating excited states, and are not able to investigate the possibilities of multiple excitations or of valence-Rydberg mixing, you have to accept that the results are questionable.

Best regards
Pete

Post Reply

Who is online

Users browsing this forum: No registered users and 4 guests