1: \documentclass[twocolumn,showpacs,superscriptaddress]{revtex4}
2: %twocolumn,
3: \usepackage{graphicx}
4: \usepackage{amsmath}
5: \usepackage{amsfonts}
6: \usepackage{amssymb}
7:
8: \newcommand{\HF}{{\mathrm{HF}}}
9: \newcommand{\Tr}{{\mathop{\rm{Tr}}\nolimits\,}}
10: \newcommand{\br}{{\bf r}}
11:
12: %\newcommand{\modified}[1]{{\bf #1}}
13: %\newcommand{\added}[1]{\textcolor{red}{#1}}
14: %\newcommand{\deleted}[1]{\textcolor{yellow}{#1}}
15: %\newenvironment{emodified}{\bfseries}{\normalfont}
16:
17: %\newcommand{\modified}[1]{\relax #1}
18: %\newcommand{\added}[1]{\relax #1}
19: %\newcommand{\deleted}[1]{\relax #1}
20: %\newenvironment{emodified}{\relax}{\relax}
21:
22:
23: \begin{document}
24: \title{Proposed definitions of the correlation energy density from a
25: Hartree-Fock starting point: The two-electron Moshinsky model atom as an exactly
26: solvable model}
27: \author{N. H. March}
28: \affiliation{Department of Physics, University of Antwerp,\\
29: Groenenborgerlaan 171, B-2020 Antwerp, Belgium}
30: \affiliation{International Centre for Theoretical Physics,\\
31: Strada Costiera, 11, Miramare, Trieste, Italy}
32: \affiliation{Oxford University, Oxford, UK}
33: \author{A. Cabo}
34: \affiliation{International Centre for Theoretical Physics,\\
35: Strada Costiera, 11, Miramare, Trieste, Italy}
36: \affiliation{Grupo de F\'{\i}sica Te\'orica, Instituto de Cibern\'etica,
37: Matem\'atica y F\'{\i}sica,\\ Calle E, No. 309, Vedado, La Habana, Cuba}
38: \author{F. Claro}
39: \affiliation{International Centre for Theoretical Physics,\\
40: Strada Costiera, 11, Miramare, Trieste, Italy}
41: \affiliation{Facultad de F\'{\i}sica, Pontificia Universidad Cat\'{o}lica de
42: Chile,\\ Campus San Joaquin, Santiago de Chile, Chile}
43: \author{G. G. N. Angilella}
44: \email[Corresponding author. E-mail: ]{giuseppe.angilella@ct.infn.it}
45: \affiliation{Dipartimento di Fisica e Astronomia, Universit\`a di
46: Catania,\\ and CNISM, UdR Catania, and INFN, Sez. Catania,\\ 64, Via S. Sofia,
47: I-95123 Catania, Italy}
48:
49: \begin{abstract}
50: In both molecular physics and condensed matter theory, deeper understanding of
51: the correlation energy density $\epsilon_{c}(\mathbf{r})$ remains a high
52: priority. By adopting L\"owdin's definition of correlation energy as the
53: difference between the exact and the Hartree-Fock values, here we propose two
54: alternative routes to define this. One of these involves both exact and
55: Hartree-Fock (HF) wavefunctions, while the second requires a coupling constant
56: integration. As an exact analytical example of the first route, we treat the
57: two-electron model atom of Moshinsky, for which both confinement potential and
58: interactions are harmonic. Though the correlation energy density
59: $\epsilon_{c}(\mathbf{r})$ is known analytically, we also investigate
60: numerically its relation to the exact ground-state density in this example.\\
61: \pacs{%
62: 31.15.Ew,% Density-functional theory
63: 31.25.-v,% Electron correlation calculations for atoms and molecules
64: 31.25.Eb% Electron correlation calculations for atoms and ions: ground state
65: }
66: \end{abstract}
67: \maketitle
68:
69:
70: \section{Introduction}
71:
72: It is true to say that one of the remaining problems in both molecular physics
73: and condensed matter theory is to gain deeper understanding of the correlation
74: energy density. We have recently approached this problem via M{\o}ller-Plesset
75: (MP) perturbation theory \cite{Cabo:06,Grassi:07}. By adopting L\"owdin's
76: \cite{Loewdin:55a} definition of the ground-state correlation energy as the
77: difference between the exact and the Hartree-Fock (HF) values, here we shall
78: propose two, formally exact, routes to the correlation energy density. The
79: first of these, as in \cite{Grassi:07}, starts out from the so-called level-shift
80: formula \cite{March:67}, but in contrast to \cite{Grassi:07}, where low-order
81: perturbation theory is invoked, our central example is exactly solvable, which
82: means that the MP series has been summed to all orders. This example is the
83: model two-electron atom introduced by Moshinsky \cite{Moshinsky:68} and it is
84: therefore natural enough that we pose the two-electron atom example formally
85: exactly in section~\ref{sec:els} immediately below. Section~\ref{sec:euls}
86: presents the exact theory for the Moshinsky model. Comparison is made in
87: section~\ref{sec:densdep} of the total kinetic energy, including correlation of
88: the Moshinsky atom with that of the (non-relativistic) He-like series of atomic
89: ions for large atomic number. An alternative route for defining the correlation
90: energy density is then proposed in section~\ref{sec:green}, which may prove to
91: come into its own in solid-state theory rather than molecular physics.
92: Section~\ref{sec:summary} constitutes a summary, plus proposals for future
93: studies which should be fruitful.
94:
95: \section{Exact level-shift theory for the ground state of atoms and molecules}
96: \label{sec:els}
97:
98: Let us consider an $N$-body system described by the Hamiltonian
99: \begin{subequations}
100: \begin{eqnarray}
101: \label{hamil}
102: H &=& H_{0}+H_{I},\\
103: H_{0} &=& \int d\mathbf{r}_{1}\,d\mathbf{r}_{2}\Psi^{\dag}(\mathbf{r}_{1})
104: h (\mathbf{r}_{1},\mathbf{r}_{2})\Psi(\mathbf{r}),\\
105: H_{I} &=& \frac{1}{2}\int d\mathbf{r}_{1}\,d\mathbf{r}_{2}
106: \Psi^{\dag}(\mathbf{r}_{1})\Psi^{\dag}(\mathbf{r}_{2}) v(\mathbf{r}_{1}
107: ,\mathbf{r}_{2})\Psi(\mathbf{r}_{2})\Psi(\mathbf{r}_{1}) ,
108: \end{eqnarray}
109: \end{subequations}
110: where the field operators in second quantization are defined as usual by their
111: expressions in terms of the creation and annihilation operators and their
112: standard commutation relations:
113: \begin{subequations}
114: \begin{eqnarray}
115: \Psi(\mathbf{r}) &=& \sum_{k}\Psi_{k}(\mathbf{r})\,a_{k},\\
116: \Psi^{\dag}(\mathbf{r}) &=& \sum_{k}\Psi_{k}^{\ast}(\mathbf{r})\,a_{k}^{\dag},\\
117: \relax[ a_{k},a_{k^{\prime}}^{\dag}] _{+} &=& \delta_{kk^{\prime}},
118: \quad\left[ a_{k},a_{k^{\prime}}\right] _{+}=0.
119: \end{eqnarray}
120: \end{subequations}
121: Here, $v(\mathbf{r}_{1},\mathbf{r}_{2})$ is the interaction potential between
122: the particles and $h(\mathbf{r}_{1},\mathbf{r}_{2})$ is the kernel of a
123: one particle operator that reduces to the kinetic energy operator in the usual
124: cases. However, it may also embody the effect of an external potential and
125: other effects. The creation (annihilation) operator $a^{\dag}_k$ ($a_k$)
126: is assumed to create (annihilate) particles in states described by the
127: wave-functions $\Psi_{k}(\mathbf{r})$. Here, as is usual, $\mathbf{r} \equiv
128: (\mathbf{x},s)$, with $\mathbf{x}$ denoting the particle
129: position, and $s$ the particle spin, and $k$ is the collective quantum
130: number associated with the basis states $\left\{ \Psi_{k}\right\}$. We
131: will assume that the free Hamiltonian is diagonal in the spin variable and
132: also that the interaction is spin-independent, \emph{i.e.}
133: \begin{subequations}
134: \begin{eqnarray}
135: \label{eq:hh1}
136: h(\mathbf{r}_{1},\mathbf{r}_{2}) &=& h({\mathbf{x}}_{1},{\mathbf{x}}_{2})
137: \delta_{s_{1}s_{s}}\\
138: v(\mathbf{r}_{1},\mathbf{r}_{2}) &=& v({\mathbf{x}}_{1},{\mathbf{x}}_{2}).
139: \end{eqnarray}
140: \end{subequations}
141:
142:
143:
144: Following \cite{Grassi:07}, let us use the level-shift formula \cite{March:67},
145: taking as the unperturbed problem the Fock Hamiltonian $H_\HF$, with ground
146: state energy $E_{0}$. Then, for two electrons, we can write explicitly for the
147: correlation energy density $\epsilon_{c}(\mathbf{r}_{1})$
148: \begin{equation}
149: \epsilon_{c}(\mathbf{r}_{1})=\frac{\int\Psi^\ast (\mathbf{r}_{1},
150: \mathbf{r}_{2})[H-H_\HF]\Phi_\HF (\mathbf{r}_{1},\mathbf{r}_{2})
151: d\mathbf{r}_{2}}{\int\Psi^\ast (\mathbf{r}_{1},\mathbf{r}_{2})\Phi_\HF
152: (\mathbf{r}_{1},\mathbf{r}_{2})d\mathbf{r}_{1} d\mathbf{r}_{2}},
153: \label{2.1}%
154: \end{equation}
155: where $\Psi$ is the exact ground-state wave function. It should be
156: noted that the definition above of the correlation energy density,
157: Eq.~(\ref{2.1}), is not unique. However, this circumstance is not
158: necessarily problematic, because any alternative definition should lead to
159: the same total integrated correlation energy. This is a similar situation, and
160: moreover also seems to be close connected, with the known lack of precise
161: definitions of the energy-momentum tensor for general physical systems.
162: Therefore, such a property should not restrict the value and utility of the
163: concept, whenever it becomes possible to construct a theoretical scheme in
164: which this energy density plays a relevant role independently of its non-unique
165: definition. From Eq.~(\ref{2.1}) we then have, in an obvious notation
166: \begin{eqnarray}
167: \langle\Psi |\Phi_\HF \rangle\epsilon_{c}(\mathbf{r}_{1})&=&
168: E\int
169: \Psi^\ast (\mathbf{r}_{1},\mathbf{r}_{2})\Phi_\HF (\mathbf{r}_{1},
170: \mathbf{r}_{2})d\mathbf{r}_{2} \nonumber\\
171: &-&
172: E_{0}\int\Psi^\ast (\mathbf{r}_{1},\mathbf{r}_{2})\Phi_\HF
173: (\mathbf{r}_{1},\mathbf{r}_{2})d\mathbf{r}_{2},
174: \label{2.2}
175: \end{eqnarray}
176: where in reaching Eq.~(\ref{2.2}) from Eq.~(\ref{2.1}), $H$ has been allowed to act
177: on $\Psi^\ast$, and $H_\HF$ on $\Phi_\HF$. Of course, by integrating
178: both sides of Eq.~(\ref{2.2}) over all $\mathbf{r}_{1}$, we obtain a trivial
179: identity for $E-E_{0}$, the latter quantity being simply the total L\"owdin
180: correlation energy $E_{c} = \int\epsilon_{c}(\mathbf{r}) d\mathbf{r}$.
181:
182: While Eq.~(\ref{2.2}) is of course, valid for the (nonrelativistic) He-like
183: series of two-electron ions with atomic number $Z$, we do not presently know
184: $\Psi^\ast$ and $E$ analytically. Therefore in section~\ref{sec:euls}
185: immediately below, we turn to illustrate Eq.~(\ref{2.2}) analytically by appeal
186: to the Moshinsky two-electron atom model \cite{Moshinsky:68}.
187:
188: \section{Exact use of the level-shift formula for the harmonic Moshinsky
189: two-electron model}
190: \label{sec:euls}
191:
192: The Moshinsky model atom has confining (external) potential
193: $\frac{1}{2}(|\mathbf{r}_{1}|^{2}+|\mathbf{r}_{2}|^{2})$ and particle interaction also of harmonic
194: form $\frac{1}{2} k |\mathbf{r}_{1}-\mathbf{r}_{2}|^{2}$. Using coordinates
195: $\mathbf{R}=(\mathbf{r}_{1}+\mathbf{r}_{2})/\sqrt{2}$ and
196: $\mathbf{r}=(\mathbf{r}_{1}-\mathbf{r}_{2})/\sqrt{2}$, the exact ground-state
197: wavefunction takes the form
198: \begin{equation}
199: \Psi(\mathbf{r},\mathbf{R})=\frac{(1+2k)^{\frac{3}{8}}}{\pi^{\frac{3}{2}}}%
200: \exp\left(-\frac{1}{2}\mathbf{R}^{2}\right)\exp\left(-\frac{1}{2}(1+2k)^{\frac{1}{2}}%
201: \mathbf{r}^{2}\right).
202: \label{3.1}%
203: \end{equation}
204: Less well known is the fact that the corresponding $\Phi_\HF$ wavefunction
205: entering the key expression (\ref{2.2}) for the correlation energy density has
206: the exact form \cite{Ballentine:98}
207: \begin{equation}
208: \Phi_\HF (\mathbf{r}_{1},\mathbf{r}_{2})=\phi(\mathbf{r}_{1})\phi
209: (\mathbf{r}_{2}),
210: \label{3.2}
211: \end{equation}
212: where
213: \begin{equation}
214: \phi(\mathbf{r})=\frac{(1+k)^{\frac{3}{8}}}{\pi^{\frac{3}{4}}}\exp\left(-\frac
215: {1}{2}(1+k)^{\frac{1}{2}}\mathbf{r}^{2}\right).
216: \label{3.3}
217: \end{equation}
218: It is understood that Eq.~(\ref{3.2}) only provides the spatial dependence of
219: the ground state, whose overall antisymmetric character is to be provided by the
220: spin dependence. Since the spin structure of the ground state problem is fixed
221: by its singlet character, we can safely employ the symbol $\mathbf{r}$ below to
222: denote only the spatial coordinates.
223:
224: Plots of the ``overlap'' $\langle\Psi|\Phi_\HF\rangle$ are already available as
225: functions of the particle-particle interaction strength $k$: \emph{e.g.} for
226: $k=1$ the overlap is 0.94. Inserting Eqs.~(\ref{3.1}) and (\ref{3.2}) into
227: the right-hand-side of Eq.~(\ref{2.2}) we obtain an exact result for
228: $\epsilon_{c}(\mathbf{r})$, now spherically symmetric in the Moshinsky atom
229: model. The correlation energy density, the HF, and exact electron densities,
230: respectively, have all the analytic form
231: \begin{subequations}
232: \begin{align}
233: \epsilon_{c}(\mathbf{r}) &= E_{c}\left(\frac{\alpha}{\pi}\right)^{\frac{3}{2}}%
234: \exp(-\alpha r^{2}), & \alpha &=\frac{b^{2}-(1-a)^{2}}{4b} ,
235: \label{3.4}\\
236: \rho(\mathbf{r}) &= 2\left(\frac{\beta}{\pi}\right)^{\frac{3}{2}}\exp(-\beta r^{2}),
237: & \beta &= \frac{2a}{1+a},
238: \label{3.5}\\
239: \rho_\HF (\mathbf{r}) &=2\left(\frac{\gamma}{\pi}\right)^{\frac{3}{2}}\exp
240: (-\gamma r^{2}), & \gamma&=\sqrt{1+k} ,
241: \label{3.6}\\
242: a &=\sqrt{1+2k}, & b&= 1+a+2\gamma.
243: \label{3.7}%
244: \end{align}
245: \end{subequations}
246:
247: \begin{figure}[t]
248: \centering
249: \includegraphics[width=0.9\columnwidth]{Fig1.eps}
250: \caption{Decay rate for the correlation energy density ($\alpha$, full line),
251: and the exact ($\beta$, short dashed) and Hartree-Fock ($\gamma$, long dashed)
252: one-particle densities, in terms of the coupling parameter $k$ (All
253: quantities are in atomic units).}
254: \label{grafico1}
255: \end{figure}
256:
257: Notice that all three functions are gaussians, albeit with a different decay
258: rate exponent. Figure~\ref{grafico1} exhibits the $k$-dependence of the latter.
259: At $k=0$ the interaction vanishes, and all three coefficients equal one. As the
260: interaction is turned on they increase, departing slowly from each other.
261: Figure~\ref{grafico2} shows the reduced correlation energy density
262: $\epsilon_{c}(\mathbf{r})/E_{c}$ (full line) together with the HF (long dashes)
263: and exact (short dashes) densities normalized to one, for $k=1$. It is apparent
264: from the functional identity and the weak parameter divergence that there is an
265: intimate connection between the correlation energy density and the exact or HF
266: electron densities. This result is compatible with the definition of
267: $\epsilon_{c}(\mathbf{r})$ as proportional to the HF one particle density, as
268: proposed in \cite{Grassi:07}. Exact relations for the Moshinsky model are presented
269: in the Appendix.
270:
271: \begin{figure}[t]
272: \centering
273: \includegraphics[width=0.9\columnwidth]{Fig2.eps}
274: \caption{Radial dependence of the reduced correlation energy density (full
275: line), exact (short dashed) and Hartree-Fock (long dashed) particle
276: densities. Here, $k=1$. All quantities are in atomic units.}
277: \label{grafico2}
278: \end{figure}
279:
280: \section{Density dependence of the total kinetic energies in the Moshinsky
281: model atom and the H\lowercase{e}-like atomic ions at large atomic number $Z$}
282: \label{sec:densdep}
283:
284: Let us illustrate in this section two examples of physical systems in which the
285: total kinetic energy can be expressed as a functional of the density. We note
286: first that the total kinetic energy of the Moshinsky model treated in
287: section~\ref{sec:euls} above can be expressed exactly in terms solely of $\rho
288: (\mathbf{r})$. From \cite{Holas:03} one knows that the kinetic energy density $t$
289: now defined from the wavefunction form $(\nabla\Psi)^{2}$ is given by
290: \begin{equation}
291: t_{\mathrm{Mosh}}(\mathbf{r})=\frac{1}{2}\rho(\mathbf{r})\left[ \frac{3}{2}
292: \frac{(d-1)^{2}}{d}-\frac{2d-1}{d} \log \frac{\rho(\mathbf{r})}{\rho
293: (\mathbf{0})}\right] ,
294: \end{equation}
295: where
296: \begin{equation}
297: d^{-1}=2-\pi\left[\frac{\rho(\mathbf{0})}{2}\right]^{\frac{2}{3}} .
298: \label{alfa}
299: \end{equation}
300: As noted earlier, in the Moshinsky atom the ground state $\rho(\mathbf{r})$ is
301: purely Gaussian. Notice that, in contrast to the He-ions at large $Z$, to be
302: discussed below, there is a $\log\rho(\mathbf{r})$ term. Because everything is
303: harmonic we expect that the potential energy $U$ will coincide with the kinetic
304: term $T$ and that both will be a half of the total energy $E$, that is
305: \begin{equation}
306: T=U=\frac{E}{2}.
307: \end{equation}
308: Thus the correlation energy density can be obtained as
309: \begin{equation}
310: E_{\mathrm{corr}}\simeq 2T_{\mathrm{corr}}=2\int
311: t_{\mathrm{corr}}(\mathbf{r})d\mathbf{r}.
312: \end{equation}
313: Here, $t_{\mathrm{corr}} (\mathbf{r}) = t_{\mathrm{Mosh}} (\mathbf{r}) -
314: \epsilon_{\mathrm{HF}} (\mathbf{r})/2$, and $\epsilon_{\mathrm{HF}}
315: (\mathbf{r})/2$ is defined as a density of HF energy constructed from one of
316: its expressions as spatial integrals. In such a way, the correlation energy
317: density is defined as $\epsilon_c (\mathbf{r}) =2 t_{\mathrm{corr}}
318: (\mathbf{r})$.
319:
320:
321: \subsubsection{Kinetic energy, including correlation, of the He-like atomic
322: ions with large atomic number}
323:
324: Following the work of Schwartz \cite{Schwartz:59} on the He-like atomic
325: ions with large $Z$ (however, still non-relativistic) the total kinetic
326: energy $T$ has been obtained by Gal, March and Nagy \cite{Gal:99} as
327: \begin{equation}
328: T=-\frac{1}{2}\left[ \frac{\rho^{\prime}(\mathbf{r})}{\rho(\mathbf{r})}
329: \right]_{\mathbf{r=0}}\int\frac{\rho(\mathbf{r})}{|\mathbf{r|}}
330: d\mathbf{r-}\frac{1}{8}\left[ \frac{\rho^{\prime}(\mathbf{r})}
331: {\rho(\mathbf{r})}\right]_{\mathbf{r}=0}^{2}\int\rho(\mathbf{r})d\mathbf{r}.
332: \end{equation}
333: This expression for the total kinetic energy, again including kinetic
334: correlation energy is evidently determined therefore solely by $\rho
335: (\mathbf{r})$ and the derivative of $\log\rho(\mathbf{r})$. Thus, the
336: total kinetic energy density $t(\mathbf{r})$ takes the form
337: \begin{equation}
338: t(\mathbf{r})=-\frac{1}{2}\left[ \frac{\rho^{\prime}(\mathbf{r})}
339: {\rho(\mathbf{r})}\right]_{\mathbf{r=0}}\frac{\rho(\mathbf{r})}
340: {|\mathbf{r|}}\mathbf{-}\frac{1}{8}\left[ \frac{\rho^{\prime}(\mathbf{r}
341: )}{\rho(\mathbf{r})}\right]_{\mathbf{r}=0}^{2}\rho(\mathbf{r}) .
342: \end{equation}
343: In closing this section we emphasize that in contrast to the Moshinsky atom,
344: the total kinetic energy including correlation now depends on both
345: $\rho(\mathbf{r})$ and $\nabla\rho(\mathbf{r})$.
346:
347: \section{Differential second-order density matrix formula for correlation energy
348: density via a coupling constant integration}
349: \label{sec:green}
350:
351: The differential level shift formula (\ref{2.1}) for the correlation energy
352: density $\epsilon_{c}(\mathbf{r})$ is entirely appropriate for two-electron
353: systems like the Moshinsky atom. But repeated volume integrations make it
354: unwieldy for $N$-electron problems, with $N>2$. Therefore, in this section we
355: derive an alternative route via a coupling constant integration. This leads to a
356: formula for the L\"owdin correlation energy density $\epsilon_{c}(\mathbf{r})$
357: characterized by second-order density matrices. Since the
358: development of efficient methods for the evaluation of second-order density
359: matrices is in rapid progess, the reduced number of integrals to be evaluated
360: can represent a helpful technical advancement, once the expressions for the
361: density matrices are already at hand.
362:
363: \subsection{Introductory example of uniform electron liquid}
364:
365: To point the way, let us consider the homogeneous electron liquid (HEL). As
366: emphasized in early work, one of us \cite{March:58,March:59a} has used as the
367: `coupling constant' the mean interelectronic separation $r_s$. A variant of
368: Hellmann-Feynman theorem enables to express the ground-state energy for the HEL
369: (see \emph{e.g.} \cite{Giuliani:05} for a review). In particular, the
370: ground-state energy per electron, $E(r_s )$, satisfies the virial theorem
371: \cite{March:58}
372: \begin{equation}
373: 2T(r_s ) + U(r_s ) = -r_s \frac{dE}{dr_s}
374: \label{eq:virial}
375: \end{equation}
376: where $T$ and $U$ represent kinetic and potential contributions, respectively. As
377: shown by March and Young \cite{March:59a}, removing $T=E-U$ from
378: Eq.~(\ref{eq:virial}) allows the resulting first-order differential equation for
379: $E(r_s )$ to be integrated to yield
380: \begin{equation}
381: E(\lambda) = -\lambda \int^\lambda \frac{U(\lambda)}{\lambda^2} d\lambda ,
382: \end{equation}
383: where $\lambda = r_s^2$ plays the role of a coupling constant adiabatically
384: connecting the unperturbed and the exact Hamiltonians, as in Hellmann-Feynman
385: theorem. But it is well known that $E_\HF (r_s ) = (A/r_s^2 ) - (B/r_s )$, with
386: $A=\frac{3}{5}$ and $B=\frac{3}{2\pi} \left( \frac{9\pi}{4} \right)^{1/3}$.
387: Hence $E(\lambda) - E_\HF$ is known via a coupling constant
388: integration. In HEL, $U$ is determined by the diagonal element of the
389: second-order density matrix, and hence we seek next a generalization of such a
390: formula for the L\"owdin correlation energy for an $N$-electron system without
391: the translational invariance of the HEL.
392:
393: \subsection{Coupling constant formula}
394:
395: Motivated by the above HEL example, we have gone back to the treatment by
396: Stanton \cite{Stanton:62} of the L\"owdin correlation energy $E-E_\HF$ in terms
397: of such a coupling constant integration. We write the Hamiltonian as
398: \begin{equation}
399: H(\lambda) = H_\HF + \lambda H_1 ,
400: \end{equation}
401: where $0\leq \lambda \leq 1$ is a coupling constant adiabatically connecting the
402: HF Hamiltonian to the exact Hamiltonian $H\equiv H(\lambda=1)$. Stanton
403: \cite{Stanton:62} then generalizes Hellmann-Feynman's theorem to embrace the
404: case of an $N$-particle Hamiltonian. Beginning with the elementary identity
405: $E-E_\HF \equiv E(1) -E(0) = \int_0^1 (dE(\lambda)/d\lambda) d\lambda$, Eq.~(12)
406: in Ref.~\cite{Stanton:62} reads, in obvious notation,
407: \begin{equation}
408: E-E_\HF = \int_0^1
409: [\langle \Psi(\lambda) | \frac{\partial H}{\partial\lambda} | \Psi(\lambda)\rangle -
410: \langle \Phi_\HF | \frac{\partial H}{\partial\lambda} |
411: \Phi_\HF \rangle ]
412: d\lambda ,
413: \label{eq:Stanton}
414: \end{equation}
415: where $\Psi(\lambda)$ is the ground-state eigenvector of the exact Hamiltonian
416: $H(\lambda)$ with eigenvalue $E(\lambda)$, and $\Phi_\HF$ is the HF ground-state
417: with energy $E_\HF$. The scalar products $\langle\ldots\rangle$ obviously imply
418: an integration over the coordinates of $N$ particles. The important point to
419: stress here is that instead of the quantum-mechanical average
420: $\langle\Psi|H|\Phi_\HF \rangle$ entering the level shift formula,
421: Eq.~(\ref{2.1}), Eq.~(\ref{eq:Stanton}) involves `symmetric' averages like
422: $\langle \Psi | H | \Psi\rangle$ and $\langle \Phi_\HF | H | \Phi_\HF \rangle$.
423: These, of course, are achieved at the cost of the coupling constant integration.
424: But since $H(\lambda=1)$ involves only one- and two-body operators, all the
425: volume integrations but two for an $N$-electron atom, molecule of cluster can be
426: achieved by use of the second-order density matrix $\Gamma(\br_1 ,\br_1^\prime ;
427: \br_2 , \br_2^\prime )$ defined by \cite{Loewdin:55a}
428: \begin{widetext}
429: \begin{equation}
430: \Gamma(\br_1 ,\br_1^\prime ; \br_2 , \br_2^\prime ) =
431: \frac{N(N-1)}{2} \int \Psi^\ast (\br_1 , \br_2 ,\br_3 ,\ldots \br_N )
432: \Psi(\br_1^\prime , \br_2^\prime , \br_3 , \ldots \br_N ) d\br_3 \ldots d\br_N .
433: \label{eq:Gamma}
434: \end{equation}
435: Inserting Eq.~(\ref{eq:Gamma}) into Eq.~(\ref{eq:Stanton}) we hence find
436: \begin{equation}
437: E-E_\HF = \frac{2}{N(N-1)} \int_0^1 d\lambda \int d\br_1 d\br_2
438: \left[ h(\br^\prime_1 , \br^\prime_2 ) - h_\HF (\br^\prime_1 , \br^\prime_2 )
439: \right] \left.\Gamma(\br_1 ,\br_1^\prime ; \br_2 , \br_2^\prime
440: ;\lambda)\right|_{\br_1^\prime = \br_1 , \br_2^\prime = \br_2} ,
441: \label{eq:hh}
442: \end{equation}
443: where use has been made of the following results:
444: \begin{subequations}
445: \begin{eqnarray}
446: \frac{\partial H(\lambda)}{\partial\lambda} &=& H_1 = H - H_\HF \\
447: \langle \Phi_\HF | \frac{\partial H(\lambda)}{\partial\lambda} | \Phi_\HF
448: \rangle &=& \langle \Phi_\HF | H-H_\HF | \Phi_\HF \rangle = 0.
449: \end{eqnarray}
450: \end{subequations}
451: In Eq.~(\ref{eq:hh}), the kernel $h(\br^\prime_1 , \br^\prime_2 )$
452: associated with the `free' Hamiltonian has been defined in Eq.~(\ref{eq:hh1}),
453: whereas the expression of the Hartree-Fock kernel $h_\HF (\br^\prime_1 ,
454: \br^\prime_2$ will be given by Eq.~(\ref{eq:hh2}) below.
455:
456:
457: The final step, as with the level shift formula Eq.~(\ref{2.1}), is to drop the
458: volume integration over $\br_2$ and hence to achieve the desired result for the
459: L\"owdin correlation energy density $\epsilon_c (\br_1 )$ as the coupling
460: constant integration over second-order density matrices as
461: \begin{equation}
462: \epsilon_c (\br_1 ) = \frac{2}{N(N-1)} \int_0^1 d\lambda \int d\br_2
463: \left[ h(\br^\prime_1 , \br^\prime_2 ) - h_\HF (\br^\prime_1 , \br^\prime_2 )
464: \right] \left.\Gamma(\br_1 ,\br_1^\prime ; \br_2 , \br_2^\prime
465: ;\lambda)\right|_{\br_1^\prime = \br_1 , \br_2^\prime = \br_2} ,
466: \label{4.6}
467: \end{equation}
468: \end{widetext}
469: The explicit form of the HF kernel $h_{\mathrm{HF}}
470: (\mathbf{r}_{1},\mathbf{r}_{2})$
471: in Eq.~(\ref{4.6}) can be written as
472: \begin{eqnarray}
473: h_{\mathrm{HF}}(\mathbf{r}_{1},\mathbf{r}_{2})&=&
474: h(\mathbf{r}_{1},\mathbf{r}_{2}) \nonumber\\
475: &&+\delta(\mathbf{r}_{1}-\mathbf{r}_{2})
476: \int d\mathbf{r}_{3} v(\mathbf{r}_{1},\mathbf{r}_{3})\,
477: \sum_{k} \Psi_{k}^{\ast}(\mathbf{r}_{3})\Psi_{k}
478: (\mathbf{r}_{3})
479: \nonumber\\
480: &&-
481: v(\mathbf{r}_{1},\mathbf{r}_{2})\sum_{k} \Psi_{k}
482: (\mathbf{r}_{1})\Psi_{k}^{\ast}(\mathbf{r}_{2}) ,
483: \end{eqnarray}
484: where the sum over $k$ runs over the filled orbitals of the mean field
485: problem. After explicitly writing the spin and spatial dependence, this
486: expression takes the form
487: \begin{widetext}
488: \begin{eqnarray}
489: h_{\mathrm{HF}}(\mathbf{x}_{1},s_{1};\mathbf{x}_{2},s_{1})&=&
490: h(\mathbf{x}_{1},\mathbf{x}_{2})\delta_{s_{1}s_{2}}
491: \nonumber\\
492: &&+
493: \delta_{s_{1}s_{2}}
494: \delta(\mathbf{x}_{1}-\mathbf{x}_{2})
495: \sum_{s_{3}=\pm1}\int d
496: \mathbf{x}_{3} v(\mathbf{x}_{1},\mathbf{x}_{3})\,\sum_{k}
497: \Psi_{k}^{\ast}(\mathbf{x}_{3},s_{3})
498: \Psi_{k}(\mathbf{x}_{3},s_{3}) \nonumber\\
499: && -v(\mathbf{x}_{1},\mathbf{x}_{2})
500: \sum_{k} \Psi_{k}(\mathbf{x}_{1},s_{1})
501: \Psi_{k}^{\ast}(\mathbf{x}_{2} ,s_{2}) ,
502: \label{eq:hh2}
503: \end{eqnarray}
504: where the single particle orbitals $\Psi_{k}$ satisfy the HF equations
505: \begin{equation}
506: \int d\mathbf{r}_{2}h_{\mathrm{HF}}
507: (\mathbf{r}_{1},\mathbf{r}_{2})\Psi_{k}(\mathbf{r}_2 )=
508: \epsilon_{k}\Psi_{k}(\mathbf{r}_1 ).
509: \end{equation}
510: Now, the second-quantized HF Hamiltonian can be expressed as
511: \begin{subequations}
512: \begin{eqnarray}
513: H_{\mathrm{HF}} &=&
514: \int d\mathbf{r}_{1}\,d\mathbf{r}_{2}\Psi^{\dag}(\mathbf{r}_{1})
515: h_{\mathrm{HF}}
516: (\mathbf{r}_{1},\mathbf{r}_{2})\Psi(\mathbf{r}_{2})-\sum_{k}
517: \frac{1}{2} v_{k} \nonumber\\
518: &=& \sum_{k} \epsilon_{k} a_{k^{^{\prime}}}^{\dag} a_{k}-\sum
519: _{k}\frac{1}{2} v_{k}\\
520: \epsilon_{k} &=& h_{k}+ v_{k}\\
521: h_{k} &=& \int d\mathbf{r}_{1}\,d\mathbf{r}_{2}
522: \Psi_{k}^{\ast}(\mathbf{r}_{1})
523: h(\mathbf{r}_{1},\mathbf{r}_{2})\Psi_{k}(\mathbf{r}_{2}),\\
524: v_{k} &=& \sum_{k^\prime}\int d\mathbf{r}_{1}\,d\mathbf{r}_{2}
525: \Psi_{k}^{\ast}(\mathbf{r}_{1})\Psi_{k^\prime}^{\ast}(\mathbf{r}_{2})
526: v(\mathbf{r}_{1},\mathbf{r}_{2})
527: \left(\Psi_{k}(\mathbf{r}_{1})\Psi_{k^\prime} (\mathbf{r}_{2})
528: -\Psi_{k^\prime}(\mathbf{r}_{1})\Psi_{k}(\mathbf{r}_{2}) \right).
529: \end{eqnarray}
530: \end{subequations}
531: \end{widetext}
532: It should be noted that the field operator $\Psi(\mathbf{r})$ in the above
533: expressions are now constructed in terms of the single particle mean field
534: orbitals.
535: It should also be recalled that in general the kernel $h_\HF (\br_1^\prime
536: ,\br_2^\prime)$ is not simply given by the Fock
537: operator. It should also incorporate an additive constant which implements the
538: property
539: \begin{equation}
540: \langle\Phi_\HF| H_\HF |\Phi_\HF \rangle = \langle\Phi_\HF |H|\Phi_\HF \rangle
541: =E_\HF .
542: \end{equation}
543: A possible optional form for $H_\HF$ could be the one introduced in
544: Ref.~\cite{Cabo:06} in order to propose an improvement of the M\o{}ller-Plesset
545: perturbative expansion. In this approach the above mentioned additive
546: constant is not required and the second quantized version of $H_\HF$
547: becomes a pure bilinear form in the creation and anihilation operators.
548:
549: We can, so far, only see a way to evaluate Eq.~(\ref{4.6}) wholly analytically
550: for the two-electron Moshinsky atom. But we have not pressed the details of that
551: ourselves, since for this simple two-electron model the level shift formula has
552: overwhelming advantages over the coupling constant integration formula. But with
553: recent progress in evaluating correlated two-body density matrices for systems
554: with $N>2$, we expect Eq.~(\ref{4.6}) to rapidly become the advantageous route
555: to employ \cite{Mazziotti:06}. Therefore, in App.~\ref{app:green}, we present an
556: alternative method based on the Green function.
557:
558:
559: \section{Summary and future directions}
560: \label{sec:summary}
561:
562: The two proposals made here for the correlation energy density
563: $\epsilon_{c}(\mathbf{r}_{1})$\ are embodied in Eqs.~(\ref{2.2}) and
564: (\ref{4.6}). The first one involves knowledge of the exact many-electron
565: ground-state function $\Psi$, which of course is generally not available. For
566: the Moshinsky two-electron atom model, however, both $\Psi$ and its HF
567: counterpart are known. Our results show that the functional form of the
568: correlation energy density, HF and exact density is the same, in support of a
569: simple proportionality expression.
570:
571: The second proposal has a more general character. An integration over a coupling
572: constant $\lambda$ involving the second order density matrix must be
573: accomplished. This alternative seems promising for solid state applications.
574: However, we delay its analysis for further extensions of this work. The
575: homogenous electron liquid case of the so-called Sawada Hamiltonian (see
576: \cite{Sawada:57}) would appear then to afford a promising starting point.
577:
578: Finally, returning to the level shift, the early analytical work of Schwartz
579: \cite{Schwartz:59} on He-like atomic ions with large $Z$ referred to in section
580: 4, may provide further insight into the use of the differential level shift
581: formula (\ref{2.1}). This formula, of course, is readily generalized beyond this
582: two-electron example, but applications then are likely to involve considerable
583: computational effort.
584:
585:
586: \begin{acknowledgments}
587: The authors gratefully acknowledge support from the ICTP through the Condensed
588: Matter Section. AC received partial support from the Network on \emph{Quantum
589: Mechanics Particles and Fields} (Net-35) of the OEA at the ICTP. FC was funded
590: in part by Fondecyt, Grant 1060650.
591: \end{acknowledgments}
592:
593: \appendix
594:
595: \begin{figure}[t]
596: \centering
597: \includegraphics[width=0.7\columnwidth]{Fig3.eps}
598: \caption{Reduced correlation energy density for the Moshinsky model as a
599: function of the Hartree-Fock density, for a range of values of the latter
600: evaluated at the origin. All quantities are in atomic units.}
601: \label{grafico3}
602: \end{figure}
603:
604:
605:
606: \section{Dependence of reduced correlation energy on the electron density}
607:
608: This Appendix is motivated by the simple observation that, for the Moshinsky
609: model, where we know both the exact electron density $\rho(r)$ and the HF
610: counterpart $\rho_\HF (r)$, we expect the difference $\rho(r)-\rho_\HF (r)$ to
611: depend on the reduced correlation energy density
612: $\epsilon_{c}(\mathbf{r})/E_{c}$.
613:
614: We have from Eqs.~(\ref{3.4}) and (\ref{3.6}) in the main text, that
615: \begin{eqnarray}
616: \rho(\mathbf{r}) &=& \rho(0)\exp(-\beta r^{2}), \label{A1}\\
617: \rho_\HF (\mathbf{r}) &=& \rho_\HF (0)\exp(-\gamma r^{2}).
618: \label{A2}
619: \end{eqnarray}
620: Hence,
621: \begin{equation}
622: \rho(\mathbf{r})-\rho_\HF (\mathbf{r})=\rho(0)(\exp(- r^{2}))^{\beta}
623: -\rho_\HF (0)(\exp(-r^{2}))^{\gamma}.
624: \label{A3}
625: \end{equation}
626: But we also have
627: \begin{equation}
628: \frac{\epsilon_{c}(\mathbf{r})}{E_{c}}=\frac{\epsilon_{c}(0)}{E_{c}}
629: (\exp(-r^{2}))^{\alpha}.
630: \label{A4}%
631: \end{equation}
632: Hence from Eq.~(\ref{A4}) we can write
633: \begin{equation}
634: \exp(- r^{2})=\left(\frac{\epsilon_{c}(\mathbf{r})}{\epsilon_{c}
635: (\mathbf{0})}\right)^{\frac{1}{\alpha}}.
636: \label{A5}
637: \end{equation}
638: Substituting this equation into Eq.~(\ref{A3}) we find,
639: \begin{equation}
640: \rho(\mathbf{r})-\rho_\HF (\mathbf{r})=\rho(0)
641: \left(\frac{\epsilon_{c}(\mathbf{r})}{\epsilon_{c}
642: (\mathbf{0})}\right)^{\frac{\beta}{\alpha}}-\rho_\HF (0)
643: \left(\frac{\epsilon_{c}(\mathbf{r})}{\epsilon_{c}
644: (\mathbf{0})}\right)^{\frac{\gamma}{\alpha}}.
645: \label{A6}
646: \end{equation}
647: This appears to be a functional relation between
648: $\epsilon_{c}(\mathbf{r}) / \epsilon_{c}(\mathbf{0})$
649: and $\rho(\mathbf{r})-\rho_\HF (\mathbf{r})$.
650: But of course, on the RHS, the four quantities $\rho(0)$,
651: $\rho_\HF (0)$,
652: $\gamma/\alpha$ and $\beta/\alpha$ all
653: depend on the interaction strength $k$. However, one of these four quantities
654: is sufficient to carry the fingerprints of $k$ for the others: we single out
655: therefore $\rho_\HF (0)$. But $\rho(0)$ and $\rho_\HF (0)$ are known from
656: normalization of the densities to two. Hence we can write,
657: \begin{equation}
658: \frac{\rho(0)}{\rho_\HF (0)}=\left(\frac{\beta}{\gamma}\right)^{\frac{3}{2}}.
659: \label{A7}
660: \end{equation}
661: and also
662: \begin{equation}
663: \rho(\mathbf{r})=\rho(0)(\exp(-r^{2}))^{\beta}
664: \label{A8}
665: \end{equation}
666: plus
667: \begin{equation}
668: \rho_\HF (\mathbf{r})=\rho_\HF (0)(\exp(- r^{2}))^{\gamma}
669: \label{A9}
670: \end{equation}
671: From the latter equations it follows that
672: \begin{equation}
673: \rho(\mathbf{r})=\rho_\HF (\mathbf{0})\left(\frac{\beta}{\gamma}\right)^{\frac{3}{2}}
674: \left(\frac{\rho_\HF(\mathbf{r})}{\rho_\HF(\mathbf{0})}
675: \right)^{\frac{\beta}{\gamma}}
676: \end{equation}
677: Substituting this equation in Eq.~(\ref{A6}) yields
678: \begin{eqnarray}
679: \left(\frac{\beta}{\gamma}\right)^{\frac{3}{2}}
680: \left(\frac{\rho_\HF (\mathbf{r})}{\rho_\HF
681: (\mathbf{0})}\right)^{\frac{\beta}{\gamma}}-\frac{\rho_\HF (\mathbf{r)}}{\rho_\HF
682: (\mathbf{0})} &=&\nonumber \\
683: && \hspace{-3.5truecm}
684: \left(\frac{\beta}{\gamma}\right)^{\frac{3}{2}}
685: \left(\frac{\epsilon_{c}(\mathbf{r})}{\epsilon_{c}(\mathbf{0})}
686: \right)^{\frac{\beta}{\alpha}}
687: -\left(\frac{\epsilon_{c}(\mathbf{r})}{\epsilon_{c}(\mathbf{0})}
688: \right)^{\frac{\gamma}{\alpha}}.
689: \label{A11}
690: \end{eqnarray}
691: Notice that the strength of the interaction does not appear explicitly in this
692: formula. Figure~\ref{grafico3} shows a three dimensional plot of
693: $\epsilon_{c}(\mathbf{r}) / \epsilon_{c}(\mathbf{0})$, $\rho_\HF (\mathbf{r)}$
694: and $\rho_\HF (\mathbf{0})$, where the latter plays the role of the interaction
695: strength $k$.
696:
697: \section{Coupling constant integration in a Green function formula}
698: \label{app:green}
699:
700: Following the discussion of Fetter and Walecka \cite{Fetter:04}, we here
701: attempt to solve the time-independent Schr\"odinger equation for an arbitrary
702: value of $\lambda$, namely
703: \begin{equation}
704: H(\lambda)\Psi(\lambda)=E(\lambda)\Psi(\lambda),
705: \label{4.2}
706: \end{equation}
707: where the wave function $\Psi$ is assumed to be normalized: that is
708: $\langle\Psi(\lambda)|\Psi(\lambda)\rangle=1$. One immediately finds then from
709: Eq.~(\ref{4.2}) that
710: \begin{equation}
711: E(\lambda)=\langle\Psi(\lambda)|H(\lambda)|\Psi(\lambda)\rangle.
712: \label{4.3}
713: \end{equation}
714: Differentiating with respect to the parameter $\lambda$ yields
715: \begin{equation}
716: \frac{d}{d\lambda}E(\lambda)=\langle\Psi(\lambda)|
717: H_{1}|\Psi(\lambda)\rangle,
718: \end{equation}
719: where $H_{1}=H(\lambda=1)-H_\HF$, in the present study. Integrating this
720: expression produces
721: \begin{equation}
722: E-E_{0}=\int_{0}^{1}\frac{d\lambda}{\lambda}\langle\Psi(\lambda)|\lambda H_{1}(\lambda
723: )|\Psi(\lambda)\rangle.
724: \label{4.4}
725: \end{equation}
726: Hence again one is led to a level-shift formula, with $E_{0}$ denoting the
727: HF ground state energy but now via the coupling constant integration in
728: Eq.~(\ref{4.4}).
729:
730: Fetter and Walecka then display the way in which Eq.~(\ref{4.4}), for $N$
731: electrons, can be reduced to integrations over just two vectors $\mathbf{r}_{1}$ and $\mathbf{r}_{2}$, by means of a time-dependent Green function,
732: denoted now by $G^{\lambda}(\mathbf{r}_{1} , t;\mathbf{r}_{2}, t^{\prime})$.
733: This result, shown in their equation (7.31), then
734: reads
735: \begin{widetext}
736: \begin{equation}
737: E-E_\HF =\pm\frac{i}{2}\int_{0}^{1}\frac{d\lambda}{\lambda}\int d\mathbf{r}_{1}
738: \lim_{t^{\prime}\to t^{+}}
739: \lim_{\mathbf{r}_{2}\to\mathbf{r}_{1}}
740: \left[ \int d\mathbf{y}
741: \delta(\mathbf{r}_{1} - \mathbf{y})
742: \left(i \hbar
743: \frac{\partial}{\partial t}-h_\HF
744: (\mathbf{r}_{1},\mathbf{y})\right)
745: \Tr G^{\lambda}(\mathbf{y} , t;\mathbf{r}_{2} ,
746: t^{\prime})\right] ,
747: \label{4.5}
748: \end{equation}
749: where in place of the kinetic energy operator, in our case the Fock operator
750: $h_\HF$ appears. As with the level-shift formula in section~\ref{sec:els}, we
751: now take the differential form of Eq.~(\ref{4.5}) to find the second result
752: proposed in the present study for the correlation energy density
753: $\epsilon_{c}(\mathbf{r})$, namely
754: \begin{equation}
755: \epsilon_{c}(\mathbf{r}_{1})=\pm\frac{i}{2}\int_{0}^{1}\frac{d\lambda}{\lambda}
756: \lim_{t^{\prime}\to t^{+}}
757: \lim_{\mathbf{r}^{\prime}\to\mathbf{r}}
758: \left[ \int d\mathbf{y}
759: \delta(\mathbf{r}_{1} - \mathbf{y})
760: \left( i\hbar
761: \frac{\partial}{\partial t}-h_\HF
762: (\mathbf{r}_{1},\mathbf{y})\right)
763: \Tr G^{\lambda}(\mathbf{y} ,
764: t;\mathbf{r}_{2} , t^{\prime})\right] .
765: \label{4.6a}%
766: \end{equation}
767: \end{widetext}
768: While Eq.~(\ref{4.6a}) is able to furnish a local definition of correlation
769: energy density, the explicit determination of the Green function $G^{\lambda
770: }(\mathbf{r}_{1} , t;\mathbf{r}_{2} , t^{\prime})$ represents a
771: considerable challenge.
772:
773: It is worth noticing that when the Green function is taken in the HF
774: approximation, since the operator $\left(i \hbar\frac{\partial}{\partial
775: t}-h_\HF \right)$ furnishes the inverse of this HF Green function, it follows
776: that the correlation energy vanishes as it should do, since its definition is
777: the difference between the exact energy $E$ and the HF value $E_\HF$. This
778: property has an interesting implication. Let us assume that the correlation
779: energy density defined in Eq.~(\ref{2.1}) has a dependence on $\mathbf{r}$ which
780: closely follows the one associated of the HF density (or the exact one )
781: \cite{Grassi:07}. Then, from definition (\ref{4.6a}) it follows that the HF
782: correction to the propagator does not contribute at all to the correlation
783: energy. Therefore, the approximate validity of the correlation energy density
784: formula proposed in \cite{Grassi:07} means that the contributions to the
785: correlation energy density coming from the higher order corrections to the Green
786: function, should approximately follow the behavior of either the total or the HF
787: density. Therefore, the validity of the proposals of \cite{Grassi:07} for the
788: correlation energy density, directly implies that the exact (or the HF) density
789: should have a close relation with the exact or the HF densities. This conclusion
790: arises because the higher corrections to the HF density turn to be approximately
791: proportional to the same exact (or HF) densities as implied by the correctness
792: of the definitions given in \cite{Grassi:07}. This property is supported by the
793: analytical results of section~\ref{sec:euls} in which a close similarity between
794: the HF electron density and the correlation energy density emerged. These
795: general issues are expected to be analyzed in future extensions of the work.
796:
797: Returning briefly to the theme of the correlation energy density $\epsilon_c
798: (\br_1 )$, we have written in Eq.~(\ref{4.6a}) a formula involving the Green
799: function $G^\lambda$ which can be brought into contact with a similar result for
800: the level shift formula. Let us note that Eq.~(\ref{4.6a}), as well
801: as Eq.~(\ref{2.1}), can always be rewritten in the form:
802: \begin{equation}
803: \epsilon_c (\br_1 ) = \int C(\br_1 ,\br_2 ) d\br_2
804: \end{equation}
805: following a similar generalization for kinetic and exchange energy densities,
806: but now in HF theory, proposed by March and Santamaria
807: \cite{March:91b,March:06c}. Then the level shift (LS) formula, Eq.~(\ref{2.1})
808: reads
809: \begin{equation}
810: C^{\mathrm{LS}}
811: (\br_{1} ,\br_2)=\frac{\int\Psi^\ast [H-H_\HF]\Phi_\HF
812: d\br_3 \ldots \br_N}{\int\Psi^\ast \Phi_\HF
813: d\br_1 \ldots d\br_N}.
814: \label{eq:appast}
815: \end{equation}
816: The conventional exact wavefunction (WF) theory reads
817: \begin{equation}
818: C^{\mathrm{WF}} (\br_1 , \br_2 ) =
819: \frac{\int \Psi^\ast H \Psi d\br_3 \ldots d\br_N}{\int\Psi^\ast \Psi
820: d\br_1 \ldots d\br_N }
821: -
822: \frac{\int \Phi_\HF^\ast H \Phi_\HF d\br_3 \ldots d\br_N}{\int\Phi_\HF^\ast \Phi_\HF
823: d\br_1 \ldots d\br_N } .
824: \label{eq:appdag}
825: \end{equation}
826: While, in general, these forms of $C(\br_1 ,\br_2 )$ are different, they must
827: all integrate to the same $\int\epsilon_c (\br_1 ) d\br_1$. This is true also for the two
828: coupling constant integration formulae derivable from Eqs.~(\ref{4.6}) and
829: (\ref{4.6a}).
830:
831: What we stress in this Appendix is the $N$-particle character of
832: Eqs.~(\ref{eq:appast}) and (\ref{eq:appdag}), and the Green function
833: generalization of $C(\br_1 ,\br_2 )$ following from Eq.~(\ref{4.5}). To date, of
834: course, analytical progress is restricted to the Moshinsky atom. But since we
835: have treated this fully in the body of the text, we shall omit further details.
836:
837:
838: \bibliographystyle{mprsty}
839: \bibliography{a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p,q,r,s,t,u,v,w,x,y,z,zzproceedings,Angilella}
840:
841:
842:
843: \end{document}
844: