Project small basis set on larger one

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: 59
Joined: 05 Apr 2016, 17:47
First name(s): Dawid
Last name(s): Grabarek
Affiliation: Wroclaw University of Technology
Country: Poland

Project small basis set on larger one

Post by Addiw7 » 03 Aug 2018, 11:12

Dear Dalton Users,

I would like to perform sequentially wavefunction optimization using DFT/CAM-B3LYP-D3/6-31G* and use the result to obtain DFT/CAM-B3LYP-D3/aug-cc-pVDZ density, so I need to project density matrix from smaller into larger basis set. How can I do it?

If I simply use

Code: Select all

**WAVE FUNCTIONS
.RESTART
I get error from SIRIUS related to unequal number of contracted orbitals:
Input ERROR in connection with ".MOSTART"
NSYM, NBAS from input : 1 687
NSYM, NBAS from MOSTART file: 1 366

hjaaj
Posts: 284
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: Project small basis set on larger one

Post by hjaaj » 03 Aug 2018, 21:37

You cannot restart like that with a new basis set. Take your final geometry obtained with the smaller basis set and start a new geometry optimization with aug-cc-pVDZ.

This is the fastest when you already have done the geometry optimization with the smaller basis set. However, there is an option ".PREOPT" for preoptimizing with a smaller basis set, which you might want to use next time.

-- Hans Jørgen.

Addiw7
Posts: 59
Joined: 05 Apr 2016, 17:47
First name(s): Dawid
Last name(s): Grabarek
Affiliation: Wroclaw University of Technology
Country: Poland

Re: Project small basis set on larger one

Post by Addiw7 » 04 Aug 2018, 08:28

But I don't do geometry optimization. I just want to obtain CAM-B3LYP density in single-point calculations and it's usually faster to converge
SCF equations in smaller basis set and then do the projection on larger and recalculate with larger one.

Such a solution is available in many other computational packages, so I thought it is also the case in Dalton.

kennethruud
Posts: 243
Joined: 27 Aug 2013, 16:42
First name(s): Kenneth
Last name(s): Ruud
Affiliation: UiT The Arctic University of Norway
Country: Norway

Re: Project small basis set on larger one

Post by kennethruud » 04 Aug 2018, 16:22

That functionality is, as Hans Jørgen points out, currently not available. It is not clear how much of a computational saving you would get from this. If you struggle with convergence with the larger (and possibly diffuse) basis sets, it could be just as efficient to try to first optimize the +2 cation in the full basis, and then restart from this solution.

However, if you are interested in a bit of programming, it should not be too difficult to add this functionality to the program.

The file abacus/huckel.F contains our Huckel calculation routine, including at the end of the subroutine HUCDRV_1 the projection of the Huckel orbitals onto the full basis set (which is what you want to do, but now using an arbitrary basis set instead of the Huckel basis. The part which will be most difficult to set up is probably to get Dalton to calculate the overlap matrix for both the basis of your choice and the basis you want to project from. You should here be able to follow the structure of the program by grep'ing for the "Huckel" in the *F files in the abacus directory.

It is probably not going to be a priority to introduce this functionality from the point of the developers, but as noted, we would welcome a contribution to the code of this functionality if you would have to time and interest to add this functionality.


Best regards,

Kenneth Ruud

Addiw7
Posts: 59
Joined: 05 Apr 2016, 17:47
First name(s): Dawid
Last name(s): Grabarek
Affiliation: Wroclaw University of Technology
Country: Poland

Re: Project small basis set on larger one

Post by Addiw7 » 05 Aug 2018, 10:40

Dear Professor Ruud,

Thank you for your answer. I would love to help, however I don't have enough experience in programming to get involved in development.

Best wishes,
Dawid Grabarek

hjaaj
Posts: 284
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: Project small basis set on larger one

Post by hjaaj » 10 Aug 2018, 16:01

Another option to save some computer time would be to converge LDA or e.g. PBE first, and then restart the CAM-B3LYP with these orbitals. (LDA and PBE have no exact exchange.) That will work in Dalton.

Another comment is that the DFT code in Dalton is more than 10 years old. I would suggest that you try LSDALTON for your project, LSDALTON has more efficient Coulomb and exchange algorithms and also more robust DFT convergence algorithms.

-- Hans Jørgen.

Addiw7
Posts: 59
Joined: 05 Apr 2016, 17:47
First name(s): Dawid
Last name(s): Grabarek
Affiliation: Wroclaw University of Technology
Country: Poland

Re: Project small basis set on larger one

Post by Addiw7 » 10 Aug 2018, 16:13

Yes, I am aware that LSDALTON is more efficient however it lacks some functionalities, e.g. electrostatic and polarizable embedding that I will
require for further calculations.

Addiw7
Posts: 59
Joined: 05 Apr 2016, 17:47
First name(s): Dawid
Last name(s): Grabarek
Affiliation: Wroclaw University of Technology
Country: Poland

Re: Project small basis set on larger one

Post by Addiw7 » 11 Aug 2018, 12:48

Dear Hans Jorgen,

Is it possible to converge CAM-B3LYP density and KS orbitals in LSDALTON and use it as start for calculations in DALTON?

Best wishes,
DG

Post Reply

Who is online

Users browsing this forum: No registered users and 4 guests