Excited states permanent dipole moment

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
carlos.diaz
Posts: 1
Joined: 11 Nov 2013, 12:45
First name(s): Carlos
Last name(s): Diaz
Affiliation: University of Central Florida - Chemistry Department
Country: United States

Excited states permanent dipole moment

Post by carlos.diaz » 07 Feb 2014, 16:17

Hello,

I am interested in calculating the permanent dipole moment of excited states calculated via TD-DFT, I looked up in the manual but could find only a way to do it for CC calculations.

Thanks for your help,
Carlos.

olav
Posts: 143
Joined: 28 Aug 2013, 06:20
First name(s): Olav
Last name(s): Vahtras

Re: Excited states permanent dipole moment

Post by olav » 07 Feb 2014, 16:39

You have an example of this in the test case library: look at DALTON/test/rsp_dresqr and replace the wave-function part with DFT.
The double residue matrix of the quadratic response function has diagonal elements which are permenent dipole moments
of the excited states relative to the ground state e.g

<k|z|k> - <0|z|0>

Regards,
Olav

Addiw7
Posts: 76
Joined: 05 Apr 2016, 17:47
First name(s): Dawid
Last name(s): Grabarek
Affiliation: Wroclaw University of Technology
Country: Poland

Re: Excited states permanent dipole moment

Post by Addiw7 » 09 May 2016, 20:08

Dear Olav,

Although my supercomputing centre provides Dalton, there are no test files where they should be. Could you or anyone else be so kind to post proper input here, please?

Best regards,
Dawid

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: Excited states permanent dipole moment

Post by taylor » 10 May 2016, 07:19

First, you should open a new topic for something since what you are asking has nothing obvious to do with excited-state dipole moments.

Second, if you are not sure whether your centre's Dalton installation is working properly, you should ask them whether they ran the tests and if they were successful.

Third, if you are unhappy with your centre's arrangements you can apply for your own Dalton licence and build and install it yourself.

Fourth, do you really expect someone to send you the 400+ test jobs and their outputs? For what purpose when you can do it all yourself by getting a licence?!

Best regards
Pete

Addiw7
Posts: 76
Joined: 05 Apr 2016, 17:47
First name(s): Dawid
Last name(s): Grabarek
Affiliation: Wroclaw University of Technology
Country: Poland

Re: Excited states permanent dipole moment

Post by Addiw7 » 10 May 2016, 07:40

First, apparently you did not understand my question as it has everything to do with excited-state dipole moments.

Second, my Dalton installation works absolutely fine. Administrators just did not put the example input files.

Third, I asked only about posting one input file or just a part requesting the excited-state dipole moments.

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: Excited states permanent dipole moment

Post by taylor » 10 May 2016, 07:51

So "Although my supercomputing centre provides Dalton, there are no test files where they should be. Could you or anyone else be so kind to post proper input here, please?" clearly indicates some issue with respect to excited-state dipole moments? I must admit my powers of telepathy are so limited that this escaped me!

Further, is the manual so inadequate that you cannot find what you need? If so, it is important that we improve the documentation, so please let us know what should be done.

Finally, I repeat, if you need sample jobs, either obtain your own licence and download the source, which includes all the test jobs, or tell your centre to make them available. I do not see how we can respond, on an individual basis, to people asking for specific test jobs. We want to provide general solutions!

Best regards
Pete

bast
Posts: 1206
Joined: 26 Aug 2013, 13:22
First name(s): Radovan
Last name(s): Bast
Affiliation: none
Country: Germany

Re: Excited states permanent dipole moment

Post by bast » 10 May 2016, 08:14

Addiw7 wrote:Dear Olav,

Although my supercomputing centre provides Dalton, there are no test files where they should be. Could you or anyone else be so kind to post proper input here, please?

Best regards,
Dawid

dear Dawid,

here is the test case posted unedited:

Code: Select all

**DALTON INPUT
.RUN RESPONS
**WAVE FUNCTIONS
.HF
.MP2
.MCSCF
*SCF INPUT
.DOUBLY OCCUPIED
 2 0 0 0
*CONFIGURATION INPUT
.INACTIVE
 1 0 0 0
.ELECTRONS
 2
.CAS SPACE
 2 0 0 0
.SYMMET
 1
.SPIN MULT
 1
**RESPONS
*QUADRATIC
.DIPLEN
.DOUBLE
.ROOTS
 2 1 0 0
**END OF DALTON INPUT

Code: Select all

BASIS
4-31G
Double residue of the quadratic response function

    2    2  X  Y    1 1.00D-12
        1.0   1
H           .0000000000             .0000000000            2.0969699107
        3.0   1
Li          .0000000000             .0000000000            -.9969699107
I was a bit hesitant to post it because the tests
are designed to have short run time and maximize code test
coverage, they are not necessarily designed to produce physically/chemically
meaningful results. So please carefully inspect and edit (also according to what
Olav commented).

But hopefully it is useful as a starting point.

Good luck!
radovan

Addiw7
Posts: 76
Joined: 05 Apr 2016, 17:47
First name(s): Dawid
Last name(s): Grabarek
Affiliation: Wroclaw University of Technology
Country: Poland

Re: Excited states permanent dipole moment

Post by Addiw7 » 11 May 2016, 17:09

Hello again,

Even after getting this test input I have troubles to get the right results. The issue is that when I order calculations for 12 states, the calculations do not converge for 67 states, which is obvious.

In that case, is there any other way to get those excited-state permanent dipole moments with TDDFT?

I went through the manual and tried to use either **PROPERTIES or **RESPONSE with linear response but it does not provide anything what would look like dipole moments.

Best regards,
Dawid

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: Excited states permanent dipole moment

Post by taylor » 11 May 2016, 17:45

CAN YOU PLEASE POST YOUR OUTPUT!

Your posting is difficult to understand (especially without the output...). If you are requesting 12 states, the program will (try to) solve for 12 states. No more, no less. How does it somehow not converge for 67 states? Where does the 67 come from? What states are these, and where/how did you request them?

I can tell you that the chance of getting 67 excited states converged is close to zero, and the results for about 60 of them will be garbage anyway (I have posted so often about this I refuse to do it again).

Best regards
Pete

Addiw7
Posts: 76
Joined: 05 Apr 2016, 17:47
First name(s): Dawid
Last name(s): Grabarek
Affiliation: Wroclaw University of Technology
Country: Poland

Re: Excited states permanent dipole moment

Post by Addiw7 » 11 May 2016, 17:52

Yes, of course, I am attaching the output. I do not understand myself where those states come from.
Attachments
TDDFT-dipole-mom.tar.gz
(62.12 KiB) Downloaded 239 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: Excited states permanent dipole moment

Post by taylor » 11 May 2016, 20:15

So now I can see what you are trying to do. The program needs to calculate response "between" 12 excited states. These occur in the bra and in the ket, but the operator is Hermitian so we need only the lower triangle of 12*13/2 elements, which is 78 calculations for which the response equations must be solved. In many cases the solutions to these equations are not converged and in some cases there appear to be instabilities, hence the error messages.

An immediate problem, in my opinion, is that your proposed computational approach (DFT/B3LYP in a 6-31+G* basis) has less chance of describing 12 excited states in any sensible way than I have of being elected US president (and as a non-US citizen I am not even eligible...). B3LYP was parametrized to give reasonable energetics and related quantities such as geometries and vibrational frequencies, using a 6-31G** basis, or thereabouts, for the ground states of small organic molecules. At no point were excited states considered, for example. Since DFT is a semiempirical method, at least until someone comes up with the exact functional, treating things like excited states (let alone electric properties of excited states!) is extrapolation. Semiempirical methods work pretty well when we interpolate, but not when we extrapolate. There is no reason, for example, to expect that B3LYP will even give good ground-state dipole moments. It may be that other functionals are more appropriate, and many would at this point jump to CAM-B3LYP. I do not run enough DFT calculations to make a recommendation. But I am sceptical...

A second issue is the basis set. I am very fortunate to have counted the late John Pople as a friend and mentor for many years, but if I can fault him for anything it is the various basis sets he and colleagues introduced. They are rubbish. They are usually not even split-valence quality and even for ground states the results will be poor. For excited states the results will in general be worthless. The odd cancellation of errors may occasionally delude the observer into thinking they provide useful answers, but relying on cancellation of errors is a very risky way of doing science! While I could not, above, suggest which functional to use at the DFT level, I would suggest that the smallest basis set for which reliable results for excited states might be obtained is aug-cc-pVDZ, or perhaps (I haven't tried them for this) some of Frank Jensen's pc-n sets.

Best regards
Pete

Addiw7
Posts: 76
Joined: 05 Apr 2016, 17:47
First name(s): Dawid
Last name(s): Grabarek
Affiliation: Wroclaw University of Technology
Country: Poland

Re: Excited states permanent dipole moment

Post by Addiw7 » 11 May 2016, 20:38

Right, I use various basis sets and various DFT functionals to check the performance.

Can anyone simply answer a question about calculating excited-state dipole moments with TDDFT in other manner than the one proposed at the very beginning of this topic by Olav?
I mean, is it possible at all? If not, then fine, I will just not calculate it.

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: Excited states permanent dipole moment

Post by taylor » 11 May 2016, 20:57

I'm confused again. What other method could there be for calculating excited state dipole moments (with Dalton) other than quadratic response, at least using DFT methods? You could certainly use the MCSCF code to calculate CASSCF or RASSCF wave functions for individual excited states and evaluate their individual dipole moments, but I'm not sure that's what you want to do.

Best regards
Pete

hjaaj
Posts: 317
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: Excited states permanent dipole moment

Post by hjaaj » 11 May 2016, 22:13

Addiw7 wrote:Yes, of course, I am attaching the output. I do not understand myself where those states come from.
Your output helped to understand what you want to do, and to suggest some answers.

1. Yes, it is possible to calculate excited state dipole moments, and your input is basically correct.

2. Convergence problem: because you need to solve linear response equations at 67 different frequencies (12*11/2 = 66 difference frequencies + 1 zero frequency [i.e. static]) and one generally needs about 10 iterations to converge. The needed dimension of the reduced equations is approximately 10*67*2 - the *2 because of the paired structure of response equations, each trial vector adds two entries to the reduced equations. The default max dim of 600 is not sufficient for 12 roots, I would recommend to ask for something like .MAXRM 1500.

3. Serious problem: You get a lot of warnings for negative eigenvalues from the solutions of the 67 response equations. That ought not to be possible, because the highest difference frequency of approx. 0.05 a.u. is well below the lowest excitation energy of approx. 0.11 a.u. My guess is you run into a numerical noise problem. You have set .AO DELETE to 1.d-10, that is calling for trouble. You should let Dalton remove the numerical linearly dependent basis functions. Some of the basis functions you chose not to be deleted will be extremely diffuse, with some MO coefficients of approx. +1000 and -1000. Two-electron integrals from such orbitals will be close to garbage, and the DFT grid is definitely not OK for such diffuse functions.

-- Hans Jørgen.

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: Excited states permanent dipole moment

Post by taylor » 12 May 2016, 08:30

Sorry, by the way, my "78" should have been 66 in my previous posting since it's the lower sub-triangle, as HJJ points out. Apologies for any confusion there.

Best regards
Pete

Post Reply

Who is online

Users browsing this forum: No registered users and 3 guests