From chemistry-request -x- at -x- server.ccl.net  Tue Feb  1 11:02:16 2000
Received: from chemistry.mps.ohio-state.edu (chemistry.mps.ohio-state.edu [128.146.33.22])
	by server.ccl.net (8.8.7/8.8.7) with ESMTP id LAA18819
	for <chemistry(+ at +)ccl.net>; Tue, 1 Feb 2000 11:02:16 -0500
Received: (from dheister |-at-| localhost)
	by chemistry.mps.ohio-state.edu (8.9.3+Sun/8.9.3) id JAA01015
	for chemistry- at -ccl.net; Tue, 1 Feb 2000 09:57:27 -0500 (EST)
Date: Tue, 1 Feb 2000 09:57:27 -0500 (EST)
From: David Heisterberg <dheister%!at!%chemistry.ohio-state.edu>
Message-Id: <200002011457.JAA01015#* at *#chemistry.mps.ohio-state.edu>
To: chemistry: at :ccl.net
Subject: RE: CCL:A question on DFT

>> Yes, any density can be represented as a single slater determinant
>If the density should correspond to a pure
>spin state (pure <s**2>) it can in general NOT
>be represented by a single determinant and this

The proof that any reasonable density may be obtained from
a single determinant of orthonormal orbitals is pretty
straight-forward, and is by construction.  The resulting
iso-density orbitals may not look anything like typical HF
or DFT orbitals, however.  I'm sure Mel Levy discusses this
in a few papers.

It is not the case that every reasonable first order density
matrix may be obtained from a single determinant.

What constitutes reasonable is integrability (or some sort
of normalizability in the case of continuum states) and
sufficient smoothness which more or less implies finite
kinetic energy.

Dave Heisterberg