Talk:Density functional theory

From Wikipedia, the free encyclopedia

WikiProject Physics This article is within the scope of WikiProject Physics, which collaborates on articles related to physics.
Start This article has been rated as Start-Class on the assessment scale.
High This article is on a subject of High importance within physics.

Help with this template This article has been rated but has no comments. If appropriate, please review the article and leave comments here to identify the strengths and weaknesses of the article and what work it will need.
WikiProject Chemistry
Chemistry Portal
This article is within the scope of WikiProject Chemistry, which collaborates on Chemistry and related subjects on Wikipedia. To participate, help improve this article or visit the project page for details on the project.

Article Grading: The article has not been rated for quality and/or importance yet. Please rate the article and then leave comments here to explain the ratings and/or to identify the strengths and weaknesses of the article..
To-do list for Density functional theory: edit · history · watch · refresh
  • Separate HK theorem and KS into separate sections
  • Make clear the notion that the HK theorem is an exact theorem for the ground state and the functionals are where the approximations come in
  • Introduce LDA
  • Write about time-dependent dft
  • Write about hybrid functionals, especially B3LYP.

[edit] Comments

When discussing DFT and dispersion forces I would be more careful in saying that"Despite the improvements in DFT, there are still difficulties in using density functional theory to properly describe intermolecular interactions, especially van der Waals forces (dispersion)". In literature there are some examples of how to address this issue. In particular, it has been shown that the BHandH functional is able of reproducing with great accuracy correlated ab initio performances, see for instance "Hybrid density functional theory for pi-stacking interactions: application to benzenes, pyridines, and DNA bases", Robertazzi et al., Journal of Computional Chemistry, 2006 Mar;27(4):491-504.

Thank you for the reference! It seems however that they do not simply treat dispersion interactions, but the special case that is pi-pi stacking. A very interesting problem, of course, but still a special case. -- Mipmip 06:58, 15 November 2006 (UTC)
There are also some other derived or ad-hoc (see Grimme's B2PLYP) dispersion-cabable methods that include DFT. Nevertheless, a semi-local KS-DFT functional can, in principle, never describe dispersion interactions properly due to the simple fact that dispersion is intrinsically a non-local correlation effect.129.69.55.52 (talk) 13:35, 4 April 2008 (UTC)

I would explain the different opinions about DFT vs ab initio. While DFT is able to treat bigger, more complex systems, many scientists view it just as a good semiempyrical technique, not to be used if a "proper" ab initio calculation is possible.

True, also should be mentioned are techniques that are less accurate than DFT but can be used for bigger systems still.

DFT is not in itself semi-empirical (Hohenberg-Kohn is in itself just a reformulation of the Schrodinger equation.) LDA and LSDA are ab initio. Many popular functionals are however semi-empirical.

But the debate on true ab initio vs. hybrid DFT may be a bit too specific for a general article such as this though. Although DFT looks somewhat empirical in comparison to wave-based methods, by comparison to the very empirical methods such as MM, DFT looks very 'ab initio' indeed.

I made some changes to the introduction of the article. I felt some statements were incorrect, or at least very vague. For example, the sentence beginning with "Within the framework of DFT, the ..." actually describes Kohn-Sham theory. DFT, of course, can be done without Kohn-Sham (see the original Hohenberg-Kohn paper).

The same sentence says that DFT reduces the many body problem to a "solvable problem of a single electron moving in an averaged effective force field." Being picky and precise, I don't think this is true. DFT does not reduce the N-body problem to a single electron problem, Kohn-Sham reduces it to an effective non-interacting electron problem. Moreover, this is not always solvable: if you have 1023 electrons you still cannot solve the equivalent non-interacting problem. Where, by solve, I mean to numerically solve the problem.

I will probably make more modifications, further down the article, as time goes on. I know that some of the material in the introduction is now repeated later, but that shouldn't be a big deal. Please comment. WikipAcct 07:58, 2 Jan 2005 (UTC)

Also, I think the article needs to emphasize that KS-DFT is a groundstate theory. And it should mention the time-dependent generalization. I will add these to the to-do list.

The statement: DFT is an exact theory only for the free electron gas, while for the treatment of extended atomic systems various approximations have to be made, is slightly misleading.

The paradigm of DFT development, starting with Thomas-Fermi and Kohn-Sham has been to work as a series of corrections to the free electron gas model. All functionals I know of use that approach, and can thus be said to be 'exact for the free electron gas'. But this does not mean DFT by necessity is like that. DFT can in principle be exact for any potential fullfilling the criterium of the Hohenberg-Kohn proof.

Yes, I agree. To be precise, the sentence in the article refered to
by this person is incorrect. Feel free to fix it! I just stumbled upon
the article yesterday. I thought about quickly fixing it to say
that "LDA is exact for theuniform gas", but that doesn't help
the article much. In fact, in my not so humble opinion, the article
needs some work. I'll try to add some improvements. WikipAcct 05:52, 4 Jan 2005 (UTC)

Some of the statements seemed to be still odd, and so i cleaned up some sections, and expanded some. I think it still needs a discussion of time-dependent DFT, as well as some more applications, and maybe some of the work on including dispersion into functionals. Salsb 18:30, 26 July 2005 (UTC)

If I may add something: The Spartan package distributed by Wavefunction, Inc. also supports DFT computations. The version we are running in our laboratory is properly cited as: Spartan'04 for Macintosh; Wavefunction, Inc.; Irvine, CA. Iamthealchemist 22:03, 6 December 2005 (UTC)

I think there needs to be more on the various exchange and correlation functions that are widely used, particularly in programs such as Gaussian. There is a mention of B3LYP without stating that it is a hybrid method and this needs defining as noted on the TODO list above. There is no mention of non-hybrid methods such as BLYP. Elsewhere there is an empty link to B3LYP, so perhaps we need a set of sections on different widely used methods, so they can be used as links for B3LYP and so on. I am rather busy at present with other pages and I am not a real expert on DFT. Anyone else want to do it? Bduke 01:02, 19 December 2005 (UTC)

[edit] Kohn-Sham functional

Which part of Kohn-Sham equations is the "Kohn-Sham functional"? --HappyCamper 14:14, 4 October 2006 (UTC)

E[ρ] WilliamDParker 15:15, 4 October 2006 (UTC)
Thanks. --HappyCamper 20:27, 4 October 2006 (UTC)

[edit] Software

I think it might be helpful to separate the list of software that supports DFT into programs available for free (including GPL'd software and proprietary software released gratis) and programs for which a paid site license is required. Does this make sense? --WilliamJenkins09 (talk) 14:02, 4 January 2008 (UTC)

It might be better to make it into a table, with columns for license (free etc as you suggest), methods (basis set KS or pure DFT, only DFT or DFT + HF etc) and perhaps other things. This would make it rather large and it might then have to become a separate article like Quantum chemistry computer programs. --Bduke (talk) 00:40, 5 January 2008 (UTC)