From chemistry-request ":at:" server.ccl.net Mon Jan 31 17:31:37 2000 Received: from ccl.net (atlantis.ccl.net [192.148.249.4]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id RAA13525 for ; Mon, 31 Jan 2000 17:31:37 -0500 Received: from krakow.ccl.net (krakow.ccl.net [192.148.249.195]) by ccl.net (8.8.6/8.8.6/OSC 1.1) with ESMTP id QAA23131; Mon, 31 Jan 2000 16:26:48 -0500 (EST) Date: Mon, 31 Jan 2000 16:26:48 -0500 (EST) From: Jan Labanowski To: Armando Navarro cc: Jan Labanowski , CHEMISTRY ^%at%^ ccl.net Subject: Re: CCL:A question on DFT In-Reply-To: <01bf6c07$1300e5a0$eb4a90c1:~at~:qogolem.usc.es> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII On Mon, 31 Jan 2000, Armando Navarro wrote: > > Dear members: > I have a doubt about the Kohn-Sham implementation of DFT theory. > Provide we would have the "true" functional, we can in principle = > determine the "true" electronic density. But, it could be always = > represented as a single determinant of Kohn-Sham molecular orbitals? > Thanks in Advance > Armando Navarro, Ph-D student > Facultade de quimica > Departamento de quimica Organica > Universidade de Santiago de Compostela > > e-mail:qoajnv-: at :-usc.es At the same time, you do not have to do restricted calculations. You can do unrestricted UKS calculations even for singlets (why not?...). The result may be spin contaminated (from autopsy, it usually is when you have a multiconfigurational beast like a polyradical -- you have to remember that you cannot start SCF with the same alpha and beta charge densities, coefficients, etc., since there would be nothing in the SCF method to kick you out from the local solution: F(alpha) = F(beta)). The contaminated spin state is by definition a combination or pure parent and higher spin states (all of them may be formally multideterminant). Hence, the "pure" DFT function (state) derived from the unrestricted calculations need not be a single determinant but can be a multideterminant. While we do not necessarily know what is the form of these functions (states) (though there is a way to find out, since formally the highest possible multiplet is when all electrons are unpaired, which will be a single determinant, and we can back substitute), we can guess that they are multideterminant. Hence, the pure spin parent function will be most likely also a combination of determinants. And, while we are doing a single determinant DFT, and do not count twice the correlation by introducing MCSCF, we actually work on the multiconfigurational states. And by the way, purists will tell you that KS orbitals are not MOs... (I am joking here, I will not get involved in religious wars... I just do not have the ammo... I am not a DFT guru, but it is a damn interesting problem. I wish I knew this stuff better...). Jan jkl |-at-| ccl.net Jan K. Labanowski | phone: 614-292-9279, FAX: 614-292-7168 Ohio Supercomputer Center | Internet: jkl-!at!-ccl.net 1224 Kinnear Rd, | http://www.ccl.net/chemistry.html Columbus, OH 43212-1163 | http://www.ccl.net/