1: %%%%Sample article, book, LateX
2: % Revised An 14.02
3: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4: \documentclass[a4paper,aps,pra,%twocolumn,
5: %showpacs,
6: preprintnumbers]{revtex4}
7: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
8:
9: %\documentclass[12pt,fleqn]{article}
10: \usepackage{graphics,graphicx,epsfig}
11:
12:
13: %\documentclass[]{article}
14: %\documentstyle[aps,pra,twocolumn,epsfig]{revtex}
15: %\documentclass[fleqn,twocolumn]{article}
16: %\documentclass{book}
17: %\documentstyle[12pt,fleqn,russcorr]{article}% ýòî äëÿ ÒåÕà îò
18: %ÐÔÔÈ.
19:
20: % Ýòî ÷òîáû ïèñàòü ïî ðóññêè.
21: %\usepackage[cp1251]{inputenc}
22: %\usepackage[russianb]{babel}
23: %\usepackage{mathtext}
24: % Ýòî ÷òîáû ïèñàòü ôîðìóëû áåç ïðîáëåì
25: \usepackage[centertags]{amsmath}
26: \usepackage{amsfonts}
27: \usepackage{amssymb}
28: %%%%Sample article, book, LateX
29: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
30: %\documentclass[twocolumn,fleqn]{article}
31: %\documentclass[fleqn]{article}
32: %\documentclass{book}
33: %\documentstyle[12pt,fleqn,russcorr]{article}% ýòî äëÿ ÒåÕà îò ÐÔÔÈ.
34: %\documentstyle[aps,epsf,draft,12pt]{revtex}
35: %\pagestyle{empty} Ýòî ÷òîáû íå áûëî íóìåðàöèè ñòàíèö
36:
37: % Ýòî ÷òîáû ïèñàòü ïî ðóññêè.
38: %\usepackage[cp1251]{inputenc}
39: %\usepackage[russianb]{babel}
40:
41: % Ýòî ÷òîáû ïèñàòü ôîðìóëû áåç ïðîáëåì
42:
43: \usepackage[centertags]{amsmath}
44: \usepackage{amsfonts}
45: \usepackage{amssymb}
46: \usepackage{amsthm}
47: \newcommand{\bbbone}{{\mathchoice {\rm 1\mskip-4mu l} {\rm 1\mskip-4mu l}
48: {\rm 1\mskip-4.5mu l} {\rm 1\mskip-5mu l}}}%Ýòà êîìàíäà ïèøåò êðàñèâî åäèíè÷íûé îïåðàòîð 1.
49: \newcommand{\beq}{\begin{equation}}
50: \newcommand{\eeq}{\end{equation}}
51: \newcommand{\beqa}{\begin{eqnarray}}
52: \newcommand{\eeqa}{\end{eqnarray}}
53:
54: \def\tr{\mathrm{tr}}
55: \newcommand{\ket} [1] {\vert #1 \rangle}
56: \newcommand{\bra} [1] {\langle #1 \vert}
57: \newcommand{\braket}[2]{\langle #1 | #2 \rangle}
58: \newcommand{\proj}[1]{\ket{#1}\bra{#1}}
59: \newcommand{\mean}[1]{\langle #1 \rangle}
60: %\newcommand{\proj}[1]{\ket{#1}\bra{#1}}
61: \newcommand{\kb}[2]{\ket{#1}\bra{#2}}
62:
63: \renewcommand{\Re}{\mathrm{Re}}
64: \renewcommand{\Im}{\mathrm{Im}}
65: \newcommand{\Eqref}[1]{Eq.~(\ref{#1})}
66: \newcommand{\Eqsref}[1]{Eqs.~(\ref{#1})}
67: \newcommand{\Sref}[1]{Sec.~\ref{#1}}
68: \newcommand{\Aref}[1]{Appendix \ref{#1}}
69: \newcommand{\Fref}[1]{Fig.~\ref{#1}}
70:
71: \newcommand{\one}{\mbox{$1 \hspace{-1.0mm} {\bf l}$}}
72:
73: \tolerance = 10000
74:
75: %\numberwithin{equation}{section}
76: %\renewcommand{\theequation}{\thesection.\arabic{equation}}
77: %\numberwithin{equation}{subsection}
78: %\textwidth=17cm \textheight=21cm
79: %\oddsidemargin=-20pt %ëåâîå ïîëå=1äþéì
80: %\headsep=-10mm %ïðîáåë îò êîëîí.òèò äî òåêñòà
81:
82: \begin{document}
83:
84: %\date{}
85: \sloppy
86: \begin{abstract}
87: Multiparticle entangled states generated via interaction between
88: narrow-band light and an ensemble of identical two-level atoms are
89: considered. Depending on the initial photon statistics, correlation
90: between atoms and photons can give rise to entangled states of these
91: systems. It is found that the state of any pair of atoms interacting
92: with weak single-mode squeezed light is inseparable and robust
93: against decay. Optical schemes for preparing entangled states of
94: atomic ensembles by projective measurement are described.
95: \\
96: \\
97: \\
98: PACS numbers: 03.67.Mn
99: \end{abstract}
100: %\pacs{PACS numbers: 03.67.Mn}
101: \title{On Multiparticle Entanglement via Resonant Interaction
102: between Light and Atomic Ensembles}
103: \author{V.N. Gorbachev, A.I. Trubilko}.
104: \affiliation {Laboratory for quantum information $\&$
105: computation,
106: University of AeroSpase Instrumentation,\\
107: St-Petersburg, 190000, Bolshaia Morskaia 67, Russia.}
108:
109: %\begin{document}
110: \maketitle
111: %\tableofcontents
112: \section{Introduction}
113: Now properties of multiparticle entangled states, their
114: preparation and application are the subject of extensive discussion.
115: The desired state of a physical system can be prepared either by
116: projective measurement or as a result of evolution. For atomic
117: systems, both methods have already been implemented in experiments.
118: In particular, two atomic ensembles were used in \cite{Pol} to
119: create an EPR pair by projective measurement. The latter method was
120: demonstrated in several studies: entangled states of alkali ions
121: were generated via Coulomb interaction \cite{M}, neutral Rydberg
122: atom were used to create an EPR pair in a micromaser setup
123: \cite{HMN}, and resonant dipole - dipole interaction was used for
124: entangling neutral atoms in an optical lattice \cite{Lat}. The most
125: popular methods for preparing entangled photon states are still
126: mostly based on parametric down-conversion. For example, an
127: entangled state equivalent to a three-state quantum system (qutrit)
128: was prepared and examined by using quantum state tomography in
129: \cite{MS}. These examples suggest that an entangled state of two
130: systems can be prepared experimentally by using a certain
131: interaction. Systems of this kind are well studied. With regard to
132: applications, it is important to know how entanglement can be
133: utilized and its robustness against decoherence. In this respect, of
134: special interest are multiparticle systems, whose entangled states
135: are characterized by much more complicated and diverse behavior.
136:
137: Previous efforts were mainly focused on analysis of entanglement
138: between several particles. In particular, the W class of tripartite
139: entanglement defined in \cite{6} includes the symmetric three-photon
140: polarization entangled state implemented in the experiment reported
141: in \cite{7}. An extension to four qubits was proposed in \cite{8},
142: where nine inequivalent classes were distinguished that cannot be
143: connected by local operations and quantum communication. Studies of
144: multiparticle systems are relatively few, being focused on
145: entanglement criteria and application to problems in quantum
146: information theory. Whereas the Peres - Horodecki criterion for
147: bipartite entanglement found in \cite{9} was applied to a real
148: physical system in \cite{10}, no operational criterion is known for
149: entanglement in the general case, and various approaches are often
150: used. In \cite{11}, the concept of entanglement molecules \cite{12}
151: was introduced to propose a classification using graphs, in which
152: particles and classical or quantum correlations represented,
153: respectively, by vertices and edges connecting pairs of vertices.
154: Graphs of this kind can be used to describe both pure and mixed
155: entangled states and distinguish several classes differing by
156: topological properties of the graphs. In \cite{13}, symmetric states
157: (including Dicke states) were studied by using several entanglement
158: measures (entanglement entropy, negativity, and entanglement of
159: formation) defined by the eigenvalues of a partial transpose of the
160: density matrix. A numerical analysis was performed to find that
161: symmetric states are robust to particle loss even if the number of
162: particles is large (up to $10^{3}$ ). Note that the calculation of
163: eigenvalues is a difficult task, because the dimension of an
164: ensemble’s Hilbert space exponentially increases with the number of
165: constituent particles. Owing to their robustness, symmetric states
166: can be used in such applications as cloning and telecloning
167: protocols for quantum information transmission \cite{14}, quantum
168: key distribution \cite{15}, and quantum teleportation or dense
169: coding \cite{16}. The formulation of two models of a one-way quantum
170: computer using measurements on multiparticle entangled states
171: \cite{17}, \cite{18} has strongly stimulated studies of the
172: properties of multiparticle systems, in particular, Ising- and Bose
173: - Hubbard-like models.
174:
175: The present study focuses on the Dicke states arising as a result of
176: collective interaction of many atoms with electromagnetic field
177: \cite{19}, which has been analyzed in numerous studies (e.g., see
178: \cite{20}). This system exhibits many physical properties of
179: interest for quantum information processing. Photon trapping in
180: chain configurations of atoms was considered in \cite{21}. When the
181: system is placed in a cavity, this effect reduces the photon escape
182: rate and increases the decoherence time of the cavity mode. In
183: \cite{22}, this effect was used for generating W states and
184: anticloning \cite{23}, which can be implemented with high fidelity
185: by means of photon trapping. In those studies, only single-photon
186: traps and single-photon initial states were analyzed. Here, we
187: consider the more general case of multiphoton processes, assuming
188: that the photon statistics is arbitrary.
189:
190: The main questions addressed below are the following: what types of
191: entangled states are produced by interaction between atoms and
192: field? What states can be prepared from independent atomic ensembles
193: entangled with a photon? How can these states be utilized? We
194: consider resonant interaction between narrow-band light and an
195: ensemble of identical two-level atoms coupled to a common heat bath.
196: The analysis is restricted to a simple model of radiative decay.
197: Multiphoton processes, such as Raman scattering, are described in
198: terms of effective Hamiltonians, which can be obtained by unitary
199: transformation \cite{24}. The behavior of an atomic system
200: interacting with light characterized by arbitrary photon statistics
201: is analyzed by using perturbation theory in the interaction strength
202: for arbitrary statistics oh light particularly for Gaussian,
203: coherent, and squeezed states. We find that weak single- mode
204: squeezed light is required to create multiparticle entanglement
205: between atoms. As distinct to the case considered in \cite{25}, the
206: steady state discussed here is robust against atomic decay. When
207: decay is neglected and analysis is restricted to a single-photon
208: initial state, simple exact solutions describing exchange of
209: excitation between the field mode and atoms can be obtained
210: \cite{26}. These solutions can be used for generating and
211: transforming symmetric Dicke states and for processing and storing
212: quantum information. The optical schemes for projective measurement
213: considered here can be used to generate entangled states of atomic
214: ensembles. An EPR entangled pair of macroscopic ensembles was
215: created in an experiment \cite{Pol}. The new states produced in our
216: schemes have hierarchical structure, thus differing from the cluster
217: states introduced in \cite{27} as a resource for one-way computing.
218:
219: The paper is organized as follows. First, we formulate a basic model
220: and write out the second-order perturbation solutions obtained by
221: taking into account radiative decay. These solutions are then used
222: to analyze the states of the atomic system corresponding to various
223: photon statistics. Exact solutions obtained under certain initial
224: conditions by neglecting radiative decay are used to describe
225: generation and transformation of symmetric Dicke states. Finally, we
226: consider optical schemes for preparing entangled states of atomic
227: ensembles by projective measurement.
228:
229: \section{Basic equations}
230: In the dipole approximation, the ensemble of $N$ identical, but
231: distinguishable, two-level atoms interacting with electromagnetic
232: field is described by the Hamiltonian
233: \begin{eqnarray}
234: \nonumber
235: H=i\hbar\vartheta, &&\\
236: \nonumber \vartheta=\sum_{k}
237: %\epsilon_{m}
238: g_{k}a_{k}S^{\dagger}_{k}-h.c.,&&
239: \end{eqnarray}
240: where
241: $g_{k}=(\hbar\omega_{k}/2\varepsilon_{0}L^{3})^{1/2}(\mu,e_{k})$ is
242: the coupling constant, $\mu$ is the dipole transition matrix
243: element, $e_{k}$ is the polarization vector for the mode with wave
244: vector $k$, $a_{k}$ and $a_{k}^{\dagger}$ are photon creation and
245: annihilation operators,
246: $S^{\dagger}_{k}=\sum_{a}s_{10}(a)\exp(ikr_{a})$ is the atomic
247: operator for the atom located at a point $r_{a}$ (x,y = 0, 1, where
248: 0 and 1 denote the ground and excited states, respectively). When
249: analysis is restricted to interaction with a single resonant mode,
250: $S_{k}$ can be replaced with $S_{k=0}$, which makes it possible to
251: treat an atomic ensemble occupying a spatial region as a point like
252: object. Then
253: \begin{eqnarray}
254: \label{001}
255: \vartheta=S_{10}B-S_{01}B^{\dagger},
256: \end{eqnarray}
257: where $S_{10}=\sum_{a}\ket{1}_{a}\bra{0}$, $B=ga$. Effective
258: Hamiltonian (\ref{001}) is used here to describe not only
259: interaction with a single resonant mode, but also multiphoton
260: processes, such as Raman scattering. In the latter case, we set
261: $B=fa_{A}a^{\dagger}_{S}$, and assume that the photon frequencies
262: $\omega_{A}$ and $\omega_{S}$ satisfy the relation
263: $\omega=\omega_{A}-\omega_{S}$, where $\omega$ is the atomic
264: transition frequency. Hamiltonians of this form can be obtained by
265: unitary transformation \cite{24}.
266: \\
267: The density matrix $\rho$ of $N$ atoms interacting with field obeys
268: the master equation
269: \begin{eqnarray}
270: \label{002} \frac {\partial} {\partial t}
271: \rho=[\vartheta,\rho]+\mathcal{L}\rho,
272: \end{eqnarray}
273: where relaxation is represented by the Lindblad superoperator
274: \begin{eqnarray}
275: \nonumber
276: \mathcal{L}=\sum_{a}\mathcal{L}_{a},
277: \end{eqnarray}
278: \begin{eqnarray}
279: \mathcal{L}_{a}=-\frac{\gamma_{\uparrow}}{2}[s_{01}(a)s_{10}(a)\rho
280: -s_{10}(a)\rho
281: s_{01}(a)]-\frac{\gamma_{\downarrow}}{2}[s_{10}(a)s_{01}(a)\rho-
282: s_{01}(a)\rho s_{10}(a)]+h.c.&&
283: \end{eqnarray}
284: This representation corresponds to the model of purely radiative
285: decay with longitudinal and transverse decay rates
286: $\gamma=\gamma_{\downarrow}+\gamma_{\uparrow}$ and $\gamma_{\bot}$ ,
287: which satisfy the relation $\gamma_{\bot}=\gamma/2$. In general case
288: $\gamma_{\bot}>\gamma/2$ since $\gamma_{\bot}$ should be replaced by
289: $\gamma_{\bot}+\kappa$, where $\kappa$ is a dephasing collision
290: rate.
291:
292: Effective Hamiltonian (\ref{001}) may involve many field modes with
293: $\omega_{k}$ differing from the atomic transition frequency
294: by $\delta \omega_{k}$ and
295: occupying a frequency band of width $\Delta \omega$. If $\Delta
296: \omega, \delta \omega_{k}\ll \gamma_{\bot}$, then we can consider a
297: narrow- band radiation field and make use of resonance
298: approximation. Otherwise, the field must be described in terms of
299: multiple-time correlation functions. Solution of Eq. (\ref{002}) is
300: a difficult task. To describe the interaction between single atom
301: and field, the following equation for the density matrix
302: $\rho_{a}=Tr'_{a}\rho$ is derived from (\ref{002}) by tracing over
303: all atoms except for one:
304: \begin{eqnarray}
305: \label{003}
306: \frac {\partial} {\partial t}
307: \rho_{a}=[\vartheta_{a},\rho_{a}]+\mathcal{L}_{a}\rho_{a}+
308: N(N-1)Sp_{a'}[\vartheta_{a'},\rho_{aa'}],
309: \end{eqnarray}
310: where $\vartheta_{a}=s_{10}(a)B-h.c.$ and $\rho_{aa'}=Sp'_{aa'}\rho$
311: is a two-particle density matrix. The right-hand side of (\ref{003})
312: contains a multiparticle contribution proportional to $N(N-1)$,
313: because the density matrix $\rho_{aa'}$ does not commute with the
314: field operators. This leads to the Bogolyubov-
315: Born-Green-Kirkwood-Yvon chain of equations for the multiparticle
316: density matrices $\rho_{a},\rho_{aa'},\rho_{aa'a''},\dots$. In
317: physical terms, this means that fluctuations of quantized
318: electromagnetic field induce correlation between atoms. If the field
319: is assumed to be classical and noise-free, for example, a coherent
320: state is considered, then the interaction will not give rise to any
321: correlation, and the initially uncorrelated atoms will remain
322: mutually independent. In what follows, we use (\ref{002}) to analyze
323: interactions that can be used to generate symmetric Dicke states.
324:
325: \section{Dicke states}
326:
327: First, we define symmetric Dicke states and introduce a
328: representation of symmetric Dicke states that demonstrates their
329: relation to the collective interaction processes. The Dicke states
330: are eigenstates of the operators $J_{z}$ and
331: $J^{2}=J_{x}^{2}+J_{y}^{2}+J_{z}^{2}$
332: \begin{eqnarray}
333: % \nonumber to remove numbering (before each equation)
334: \nonumber
335: J_{z}\ket{jma}=m\ket{jma}&&\\
336: J^{2}\ket{jma}=j(j+1)\ket{jma},&&
337: \end{eqnarray}
338: where $[J_{s},J_{p}]=i\epsilon_{spd}J_{d}$. For example, operators
339: $J_{s}, s=x,y,z$ can be represented by Pauli matrices
340: $J_{s}=(1/2)\sum_{k}\sigma_{sk}$, $\sigma_{sk}, s=x,y,z$. Indexes
341: $j$ and $m$ are integer or half-integer numbers $|m|\leq j, \max
342: j=N/2$. If $j=N/2$, then the states are symmetric, and the quantum
343: number “a” introduced to lift degeneracy can be omitted. For h
344: excited atoms $h=m+N/2$, the states can be represented as
345: \begin{eqnarray} \label{0021}
346: \ket{j=N/2,m} \equiv \ket{h;N}=
347: \sum_{z}P_{z}\ket{1_{1},1_{2},\dots,1_{h},0_{h+1},\dots,0},&&
348: \end{eqnarray}
349: where $P_{z}$ is one of the $C_{h}^{N}=N!/(h!(N-h)!)$
350: distinguishable permutations of particles.
351:
352:
353: The vector $\ket{h;N}$ describes an atomic of $h$ excited atoms and
354: it is normalized as $\mean{h;N|h;N}=C_{h}^{N}$. Symmetric states of
355: a multiparticle system arise when interaction is described by
356: collective operators of the form
357: $S_{10}=\sum_{a}^{N}\ket{1}_{a}\bra{0}$.
358: \begin{equation}
359: \label{007} \ket{h;N}=(1/h!)S_{10}^{h}\ket{0;N}.
360: \end{equation}
361: If $h=1$, then one finds that
362: \begin{eqnarray}
363: \ket{1;N} =\ket{10\dots0}+\dots +\ket{00\dots1}.
364: \end{eqnarray}
365: Since the wavefunction $\ket{h; N}$ is not factorizable, it
366: represents an entangled state. In terms of correlation between
367: particles, it is substantially different from other entangled
368: states. For example, in the Greenberger– Horne–Zeilinger (GHZ) state
369: $GHZ=(1/2)(\ket{0}^{\otimes N}+\ket{1}^{\otimes N})$, the
370: correlation of any $M$ particles $(M < N)$ is classical. In
371: particular, the density matrix corresponding to the state $\proj{1;
372: N}$ of a group of $M$ particles is $\rho(M\leq
373: N)=N^{-1}\proj{1;M}+(N-M)N^{-1}\proj{0;N}$. The corresponding von
374: Neumann entropy depends on the relative particle number $p = M/N$:
375: $S(\rho(M\leq N))=-p\log p-(1-p)\log (1-p)$. When $p=1/2$ the
376: entropy achieves its maximum 1. If $M=2$ we can apply the necessary
377: and sufficient separability criterion proposed in \cite{9}.
378: According to this criterion, the state is inseparable (entangled) if
379: the density matrix partially transposed over the one of the atoms
380: has at least one negative eigenvalue. In the case considered here,
381: one of the four eigenvalues $\{1/N; 1/N; (N-2)(2N)^{-1}[1 \pm
382: \sqrt{1+4/(N-2)^{2}}]\}$ is negative. Note that the behavior of
383: correlation between $M$ particles depends on $p = M/N$. As the total
384: particle number $N$ increases, $p\to 0$ and the correlation
385: vanishes, since their state becomes pure as $\rho(M\leq
386: N)\to\proj{0;N}$ In what follows, we make use of the following
387: equalities:
388: \begin{eqnarray}
389: \label{DD} \nonumber
390: S_{01}\ket{0;N}=0,&&\\
391: \nonumber
392: S_{10}\ket{h;N}=(h+1)\ket{h+1;N},&&\\
393: \nonumber
394: S_{01}\ket{h;N}=(N-h+1)\ket{h-1;N},&&\\
395: \nonumber
396: S_{01}S_{10}\ket{h;N}=(h+1)(N-h)\ket{h;N},&&\\
397: S_{10}S_{01}\ket{h;N}=h(N-h+1)\ket{h;N}.&&
398: \end{eqnarray}
399: \section{Second-order perturbation theory}
400:
401: To solve Eq. (\ref{002}), we use perturbation theory in the
402: interaction strength:
403: \begin{eqnarray}
404: \rho=\rho^{(0)}+\rho^{(1)}+\rho^{(2)}+\dots,
405: \end{eqnarray}
406: Here, the zeroth-order approximation $\rho^{(0)}$ is the
407: steady-state solution of (\ref{002}) with $\vartheta=0$:
408: $\rho^{(0)}=\proj{0}\otimes\rho_{f}$, where the density matrix
409: $\rho_{f}$ represents the modes and $\ket{0}=\ket{0}^{\otimes N}$
410: corresponds to the ground state of all atoms. The operators
411: $\rho^{(k)}$, $k=1,\dots$ satisfy the equations
412: \begin{eqnarray}
413: \frac {\partial} {\partial t}
414: \rho^{(k)}=[\vartheta,\rho^{(k-1)}]+\mathcal{L}\rho^{(k)},
415: \end{eqnarray}
416: subject to the initial conditions $\rho^{(k)}(0)=0$.
417:
418: The analysis that follows is restricted to second-order perturbation
419: theory, which is sufficient to obtain statistical characteristics of
420: the excitation field. The matrix equation for $\rho^{(2)}$ is
421: \begin{eqnarray}
422: \nonumber \bra{1_{k},1_{m};N} \frac {\partial} {\partial t}
423: \rho^{(2)}\ket{0;N}=
424: -2\gamma_{\bot}\bra{1_{k},1_{m};N}\rho^{(2)}\ket{0}+
425: \bra{1_{k},1_{m};N}R\ket{0}&&\\
426: \nonumber \bra{1_{k};N} \frac {\partial} {\partial t}
427: \rho^{(2)}\ket{1_{m};N}=
428: -2\gamma_{\bot}\bra{1_{k};N}\rho^{(2)}\ket{1_{m};N}+
429: \bra{1_{k};N}R\ket{1_{m};N}, ~~~k\neq m&&\\
430: \nonumber \bra{1_{k};N}\frac {\partial} {\partial t}
431: \rho^{(2)}\ket{1_{k};N}=
432: -\gamma\bra{1_{k};N}\rho^{(2)}\ket{1_{k};N}+
433: \bra{1_{k};N}R\ket{1_{k};N}&&\\
434: \bra{0;N}\frac {\partial} {\partial t} \rho^{(2)}\ket{0;N}=
435: \gamma\sum_{k}\bra{1_{k};N}\rho^{(2)}\ket{1_{k};N}+
436: \bra{0;N}R\ket{0;N}.&&
437: \end{eqnarray}
438: where $s_{10}(k)\ket{0;N}=\ket{1_{k};N}$ and
439: $s_{10}(k)s_{10}(p)\ket{0;N}=\ket{1_{k},1_{p};N}$ represent the
440: states in which only the kth atom is excited and only the kth and
441: pth atoms are excited, respectively. The nonzero matrix elements of
442: the operator $R=[\vartheta,\rho^{(1)}]$ are
443: \begin{eqnarray}
444: \nonumber
445: \bra{1_{k},1_{m};N}R\ket{0;N}=2\kappa(t)B^{2}\rho_{f},&&\\
446: \nonumber
447: \bra{0;N}R\ket{0;N}=-\kappa(t)N(B^{\dagger}B\rho_{f}+\rho_{f}B^{\dagger}B),&&\\
448: \bra{1_{k};N}R\ket{1_{m};N}=2\kappa(t)B\rho_{sf}B^{\dagger},&&
449: \end{eqnarray}
450: where $\kappa(t)=(1/\gamma_{\bot})(1-\exp(-\gamma_{\bot}t)).$ For
451: purely radiative decay, $\gamma_{\bot}=\gamma/2$ and the
452: second-order perturbation theory yields
453: \begin{eqnarray}\label{0025}
454: \nonumber \rho=\proj{0}\otimes
455: \rho_{f}+\kappa[\ket{1;N}\bra{0;N}\otimes
456: B\rho_{sf}+h.c.]+\kappa^{2}[\ket{2;N}\bra{0;N}\otimes
457: B^{2}\rho_{f}+h.c.]
458: &&\\
459: \nonumber -N\gamma
460: \mathcal{K}\proj{0;N}\otimes[B^{\dagger}B\rho_{f}-
461: B\rho_{f}B^{\dagger}+h.c.]-(1/2)N\kappa^{2}\proj{0;N}\otimes
462: [B^{\dagger}B\rho_{f}+h.c.]&&\\
463: +\kappa^{2}\proj{1;N}\otimes B\rho_{f}B^{\dagger},&&
464: \end{eqnarray}
465: where
466: $$ \mathcal{K}= \gamma_{\bot}^{-1} \left\{\gamma^{-2}[\gamma t+1-\exp(-\gamma t)]
467: -\frac{\kappa^{2}}{2} \right\}.$$ This expression is valid to second
468: order if the field is relatively weak:
469: \begin{equation}\label{00250}
470: N\kappa^{2}\mean{B^{\dagger}B}\ll 1.
471: \end{equation}
472: In the case of interaction with a single resonant cavity mode, we
473: have $B=ga$ and $\kappa^{2}\mean{B^{\dagger}B}=n/n_{s}$, where
474: $n_{s}=(\gamma_{\bot}/g)^{2}$ is a saturation parameter and
475: $n=\mean{a^{\dagger}a}$ is the mean photon number. Then,
476: (\ref{00250}) reduces to the standard condition imposed in the case
477: of resonant coupling between the field and two-level atoms: $
478: Nn/n_{s}\ll 1$. Solution (\ref{0025}) describes the joint evolution
479: of the atomic ensemble and field starting from an ensemble of
480: ground-state atoms and an arbitrary state of the field.
481:
482: \section{Mixed nonseparable atomic states}
483:
484: Second-order perturbation theory predicts correlation between atoms
485: depending on photon statistics, i.e., provides a framework for
486: describing entangled (inseparable) atomic states. To analyze the
487: properties of the atomic system, we use second-order perturbation
488: theory to find the density matrix for a group of $M\leq N$ atoms,
489: $\rho_{A}(M\leq N)$, obtained by taking the trace of (\ref{0025})
490: over the field states represented by $\rho_{f}$ and over $N-M$
491: particles. The result has the form
492: \begin{eqnarray}
493: \label{320A} \nonumber \rho_{A}(M\leq N)
494: \nonumber
495: =\ket{0}\bra{0}[1-M\kappa^{2}\mean{B^{\dagger}B}]
496: +\kappa[\mean{B}\ket{1;M}\bra{0}+h.c.]
497: \nonumber
498: +\kappa^{2}[
499: \mean{B^{2}}\ket{2;M}\bra{0}+h.c.]&&\\
500: +\kappa^{2}\mean{B^{\dagger}B} \ket{1;M}\bra{1;M}.&&
501: \end{eqnarray}
502: Note that the density matrix $\rho_{A}(M\leq N)$ describes a mixed
503: state of the atomic ensemble. Unlike the density matrices for
504: symmetric Dicke states (\ref{0021}), $\rho_{A}(M\leq N)$ is
505: independent of both $N$ and $p = M/N$. Therefore, the correlations
506: between $M < N$ atoms are identical and are independent of the total
507: particle number. This implies that the state is robust to particle
508: loss.
509:
510: The atomic density matrix cannot be factorized because of the
511: correlation depending on photon statistics. Consider two atoms
512: described in terms of their respective observables $c_1$ and $c_2$
513: such that $[c_{1}; c_{2}] = 0$. Setting $M = 2$ in (\ref{320A}), we
514: have the two-atom density matrix
515: \begin{eqnarray}\label{322}
516: \nonumber
517: \rho_{A}(2)=\proj{00}(1-2\kappa^{2}\mean{B^{\dagger}B})+\kappa\mean{B}
518: \Big(\ket{10}\bra{00}+\ket{01}\bra{00}+h.c.\Big)+\kappa^{2}\mean{B^{2}}
519: \Big(\ket{11}\bra{00}+h.c.\Big)&&\\
520: %\nonumber
521: +\kappa^{2}\mean{B^{\dagger}B}\Big(\ket{10}+\ket{01}\Big)\Big(
522: \bra{10}+\bra{01} \Big).&&
523: \end{eqnarray}
524: Using (\ref{322}) we find that the covariance of the operators
525: $c_{1},c_{2}$ is determined by the electromagnetic field variance:
526: \begin{eqnarray} \label{Cov}
527: %\nonumber
528: \mean{c_{1}c_{2}}-\mean{c_{1}}\mean{c_{2}}=\kappa^{2}[(\mean{B^{2}}-\mean{B}^{2})
529: \bra{0}c_{1}\ket{1}\bra{0}c_{2}\ket{1}%&&\\
530: %\nonumber
531: +(\mean{B^{\dagger}B} -\mean{B^{\dagger}}\mean{B})
532: \bra{1}c_{1}\ket{0}
533: \bra{0}c_{2}\ket{1}%&&\\
534: %\nonumber
535: +c.c].&&
536: \end{eqnarray}
537: If the field is not fluctuating in the sense that its variances are
538: zero, i.e., $\mean{B^{2}}-\mean{B}^{2}=0$ etc. (which is true in the
539: present case, e.g., for a coherent state), then there is no
540: correlation between atoms. Suppose that $c_{k} (k = 1, 2)$ are
541: dipole operators: $c_{k}=d_{k}=\mu(s_{01}(k)+s_{10}(k))$, where the
542: matrix element $\mu$ is real. Then the correlation between two
543: dipole moments depends on photon statistics. We define the
544: quadrature operator $X_{f}=B^{\dagger}\exp(i\theta)+h.c.$. Then
545: (\ref{Cov}) implies that the covariance of the dipole moments is
546: determined by the variance of the quadrature operator normally
547: ordered with respect to the field operators $B$ and $B^{\dagger}$ at
548: $\theta= 0$: $ \mean{d_{1}d_{2}}-\mean{d_{1}}\mean{d_{2}}=
549: \mu^{2}\kappa^{2}D_{N} $, where
550: $D_{N}=\mean{X_{f}^{2}}-\mean{X_{f}}^{2}- \mean{[B,B^{\dagger}]}$.
551: For coherent states, the variance is $D_{N}=0$. The dipole moments
552: are correlated both for a squeezed-state field (with $D_{N}<0$) and
553: for field in a classical state (with $D_{N}>0$).
554:
555:
556: The necessary and sufficient condition for inseparability of a mixed
557: state is provided by the Peres – Horodecki criterion \cite{9}, which
558: is valid for systems with Hilbert spaces of dimension $2\times 2$
559: and $2\times 3$. In the case considered here, the state of a
560: two-atom system described by $\rho_{A}(2)$ is inseparable
561: (entangled) if at least one eigenvalue of the density matrix
562: partially transposed over the variable of atom 1
563: $\rho_{A}(2)^{T_{1}}$ is negative. As example, we consider light in
564: Gaussian and squeezed states.
565:
566:
567: For a Gaussian field ($\mean{B}=\mean{B^{2}}=0$, expression
568: (\ref{322}) reduces to the density matrix describing a superposition
569: of the ground and mixed states:
570: $\rho_{A}(2)=a\proj{00}+b[(\ket{01}+\ket{10})(\bra{01}+\bra{10})]$,
571: where $a+2b=1$ and $a=1-2\kappa^{2}\mean{B^{\dagger}B}$. The
572: eigenvalues of $\rho_{A}(2)^{T_{1}}$ are
573: \begin{eqnarray*}
574: \lambda=\Big\{b,b,
575: \frac{a}{2}\pm\sqrt{\frac{a^{2}}{4}+b^{2}}\Big\}.&&
576: \end{eqnarray*}
577: Since $\sqrt{a^{2}/4+b^{2}}\approx a/2$, in the approximation
578: considered here, we have the eigenvalues: $\{b,b,a,0\}$ i.e., a
579: separable state.
580:
581: Consider the case of resonant interaction with single- mode squeezed
582: light $(B = ga)$ generated, for example, by a parametric oscillator.
583: A simple model of the oscillator is defined by the effective
584: Hamiltonian $H=i\hbar(f/2)(a^{\dagger 2}-h.c.)$. The solution is
585: $a=a_{0}\cosh r+a_{0}^{\dagger}\sinh r$, where $r=f\tau$ is the
586: squeezing parameter, $\tau$ is the normalized length of the
587: nonlinear medium, and $a_{0}, a_{0}^{\dagger}$ and denote the input
588: field operators. For the initial vacuum state, $\mean{a}=0$,
589: $\mean{a^{2}}=\mean{a^{\dagger 2}}=\cosh r\sinh r$,
590: $\mean{a^{\dagger}a}=\sinh^{2} r$. In this case (\ref{322}) reduces
591: to the following the two-atom density matrix
592: \begin{eqnarray}\label{334}
593: \nonumber \rho_{A}(2)=\proj{00}[1-2\kappa^{2}\mean{B^{\dagger}B}] +
594: \kappa^{2}\Big[\mean{B^{2}}
595: \Big(\ket{11}\bra{00}+\ket{00}\bra{11}\Big)+h.c.\Big]&&\\
596: +\kappa^{2}\mean{B^{\dagger}B}\Big(\ket{10}\bra{10}+\ket{01}\bra{10}+
597: \ket{10}\bra{01}+\ket{01}\bra{01}\Big).&&
598: \end{eqnarray}
599: The four eigenvalues of $\rho_{A}^{T_{1}}(2)$ are
600: \begin{equation}\label{336}
601: \lambda=\left\{0;~~ 1-\frac{2}{n_{s}}\sinh^{2}r;~~ \pm
602: \frac{1}{n_{s}}\exp(\pm r)\sinh r\right\}.
603: \end{equation}
604: To be specific, we set $r > 0$, i.e., consider the state squeezed
605: with respect to canonical momentum or phase. In this case,
606: $(-1/n_{s})\sinh r\exp(-2r)<0$. However, it is clear that the degree
607: of squeezing is low, because the approximations used here imply that
608: \begin{equation}\label{337}
609: \frac{\sinh^{2}r}{n_{s}}\ll 1.
610: \end{equation}
611: Thus, the state of the atomic system is inseparable. This behavior
612: is explained as follows. Fluctuations of light give rise to
613: correlation between atoms, which leads to two-atom coherence. When
614: condition (\ref{337}) holds, this coherence plays the key role.
615: Since absorption is weak, the system is almost entirely in the
616: ground state. As distinct to the case of Gaussian statistics, the
617: density matrix has the form
618: $\rho_{A}(2)\approx\ket{00}\bra{00}+\kappa^{2}
619: [\mean{B^{2}}\ket{11}\bra{00}+h.c.].$
620:
621: Note that the following two observations can be inferred from this
622: example. First, a steady entangled atomic state can be created by
623: using weak squeezed light, which looks promising from an
624: experimental perspective. Second, the entire ensemble cannot be
625: interpreted as separable, because any pair in a group of $M\leq N$
626: atoms is entangled, i.e., the quantum correlation of the ensemble as
627: a whole is robust to particle loss
628:
629: Since no reliable universally applicable criterion is known for
630: multiparticle entanglement, we apply the Peres – Horodecki criterion
631: to two two-level subsystems and found that any pair of atoms in the
632: ensemble can be inseparable, which gives reason to interpret the
633: state of the entire system as inseparable.
634:
635: Note also that spurious entanglement may be predicted by
636: perturbation theory \cite{28}. In that study, an example of
637: expansion of the product of two wave functions in terms of a common
638: classical parameter was considered in which individual summands
639: represent entangled states. However, if entanglement entropy is used
640: as a measure, then we have initially independent systems, because
641: the entropy is either quadratic in the small parameter or zero in
642: arbitrary-order perturbation theory. Note that physical
643: implementation of such entangled states, i.e., preparation of an
644: independent state of a pair of entangled particles, requires
645: projective measurement in an entangled basis. The present analysis
646: also relies on perturbation theory, but we deal with a different
647: situation in both physical and formal sense, in which interaction
648: between particles gives rise to correlation. The wavefunction
649: obtained in first-order perturbation theory is not factorizable, and
650: the corresponding entanglement entropy is zero to the corresponding
651: accuracy. This result is physically plausible, because there is no
652: correlation in the first-order perturbation theory. In our analysis,
653: entanglement is predicted by second-order perturbation theory, which
654: describes real emission and absorption processes result in
655: correlation. In this order of perturbation theory, the existence of
656: quantum correlation is substantiated by entanglement criteria
657: consistent with approximation accuracy.
658:
659: \section{Exact solutions}
660:
661: Radiative decay can be neglected in (\ref{002}) when evolution over
662: a time $t\ll\gamma^{-1}$ is considered, and the behavior of the
663: entire system is described by the wavefunction $
664: \phi(t)=\exp(-i\hbar^{-1}H t)(\phi_{A}\otimes\phi_{f})$, where the
665: initial states of the atoms and field are assumed to be
666: uncorrelated. Then, simple solutions can be obtained under certain
667: initial conditions.
668:
669: Consider the mixing of modes $a$ and $b$ described by
670: \begin{eqnarray}
671: \label{00210} H=i\hbar f(a^{\dagger}bS-ab^{\dagger}S^{\dagger}),
672: \end{eqnarray}
673: where $S=S_{10}, S^{\dagger}=S_{01}$. If analysis is restricted to
674: single-photon Fock states of the modes
675: $\phi_{f}=c\ket{01}_{ab}+e\ket{10}_{ab}$ exact solutions can be
676: written as
677: \begin{eqnarray}
678: \label{0022} \nonumber
679: \exp\{-i\hbar^{-1}Ht\}(c\ket{01}_{ab}+e\ket{10}_{ab})\otimes\phi_{A}=c
680: \Big\{ \ket{01}\cos[tf\sqrt{SS^{\dagger}}]+
681: \ket{10}S^{\dagger}\frac{1}{\sqrt{SS^{\dagger}}}
682: \sin[tf\sqrt{SS^{\dagger}}] \Big\}\otimes\phi_{A}&&\\
683: +e\Big\{-\ket{01}S\frac{1}{\sqrt{ S^{\dagger} S}}
684: \sin[tf\sqrt{S^{\dagger} S}]+\ket{10}\cos[tf\sqrt{S^{\dagger}
685: S}]\Big\}\otimes\phi_{A}.&&
686: \end{eqnarray}
687: In the case of a single-photon process described by the Hamiltonian
688: \begin{eqnarray}
689: \label{00222}
690: H=i\hbar g(aS-a^{\dagger}S^{\dagger})
691: \end{eqnarray}
692: there also exist simple solutions. For example,
693: \begin{eqnarray}
694: \label{0023} \nonumber
695: \exp\{-i\hbar^{-1}Ht\}(c\ket{1}\otimes\ket{0;N}+e\ket{0}\otimes
696: \ket{1;N})&&\\
697: = c\Big\{\cos[gf\sqrt{N}]\ket{1}\otimes\ket{0;N}
698: +\frac{1}{\sqrt{N}}\sin[gf\sqrt{N}]\ket{0}\otimes
699: \ket{1;N}\Big\}&&\\ \nonumber + e\Big\{
700: -\sqrt{N}\sin[gf\sqrt{N}]\ket{1}\otimes\ket{0;N}+
701: \cos[gf\sqrt{N}]\ket{0}\otimes \ket{1;N}\Big\}, &&
702: \end{eqnarray}
703: where $\ket{h;N}=\ket{0}^{\otimes N}, h=0,1$ represents the ground
704: state of the atomic ensemble and a symmetric Dicke state defined in
705: accordance with (\ref{0021}). These solutions are valid only under
706: the restrictions imposed above on the initial states. They describe
707: exchange of excitation between the cavity mode and the atoms.
708:
709: \section{Generation and transformation of symmetric states }
710:
711: Now, we use the exact solutions written out above to analyze the
712: evolution of symmetric Dicke states $\ket{h;N}$ in single- photon
713: and wave-mixing processes.
714:
715: First, consider the case when the spatial inhomogeneity of the field
716: within the region occupied by the atomic ensemble can be neglected.
717: Setting (\ref{0023}) $\phi_{A}=\ket{h;N}$, we use (\ref{DD}) to
718: obtain
719: \begin{eqnarray}
720: \label{005}
721: \nonumber
722: \Big(\alpha\ket{01}+\beta\ket{10}\Big)\otimes\ket{h;N}\to%&&\\
723: \nonumber
724: \alpha\Big\{\cos\theta_{h}\ket{01}\otimes\ket{h;N}+\sqrt{\frac{h+1}{N-h}}
725: \sin\theta_{h}\ket{10}\otimes\ket{h+1;N}\Big\}&&\\
726: +\beta\Big\{-\sqrt{\frac{N-h+1}{h}}\sin\theta'_{h}\ket{01}\otimes
727: \ket{h-1;N}+\cos\theta'_{h}\ket{10}\otimes\ket{h;N}\Big\}&&
728: \end{eqnarray}
729: where $\theta_{h}=tf\sqrt{(h+1)(N-h)}$,
730: $\theta'_{h}=tf\sqrt{h(N-h+1)}$. Relation (\ref{005}) entails
731: possibilities of preparation of an entangled from ground-state atoms
732: $ \ket{0;N}\to\ket{1;N},$ and transformation of entangled states by
733: changing the number of excited atoms $ \ket{h;N}\to\ket{h\pm1;N},$
734: including disentanglement: $ \ket{h;N}\to\ket{h-1;N}\to\dots
735: \ket{0;N}.$
736:
737: Note that exact solutions (\ref{0023}) and (\ref{005}) describe
738: state swapping, which can be used to map the state of light onto
739: atoms in order to store it in a long-lived atomic ensemble, i.e., to
740: implement quantum memory. In particular, an unknown superposition of
741: photons can be transferred to atoms and back by using the following
742: transformation entailed by (\ref{0023})
743: \begin{equation}\label{5210}
744: \Big(\alpha\ket{1}+\beta\ket{0}\Big)\otimes\ket{0;N}\leftrightarrows
745: \ket{0}\otimes\Big(\alpha\frac{1}{\sqrt{N}}\ket{1;N}+\beta\ket{0;N}\Big).
746: \end{equation}
747: Solutions (\ref{0023}) and (\ref{005}) make it possible to take into
748: account the spatial configuration of atoms in the ensemble. For
749: example, consider the interaction between a one-dimensional array of
750: atoms located at points $x_{1},\dots,x_{N}$ and a single photon
751: described by Hamiltonian (\ref{00222}) with
752: $S=\sum_{p}s_{10}(p)\exp[ikx_{p}]$, where
753: $s_{10}(p)=\ket{1}_{p}\bra{0}$ corresponds to the atom located at
754: $x_{p}$, $p=1,\dots,N$. Using (\ref{0023}), we can show that
755: \begin{eqnarray}
756: % \nonumber to remove numbering (before each equation)
757: \ket{1}\otimes\ket{0;N}\to\cos\theta \ket{1}\otimes\ket{0;N}+
758: \sin\theta\ket{0}\otimes\eta_{N}(1),
759: \end{eqnarray}
760: ãäå $\theta=tg\sqrt{N}$,
761: \begin{eqnarray}
762: \label{0055}
763: % \nonumber to remove numbering (before each equation)
764: \eta_{N}=(1/\sqrt{N})\Big[e^{ikx_{1}}\ket{10\dots0}+\dots
765: e^{ikx_{N}}\ket{0\dots 01}\Big].
766: \end{eqnarray}
767: Expression (\ref{0055}) implies that an array of entangled atoms is
768: created when $\theta=\pi/2$. Note that $\eta_{N}$ is the Dicke state
769: with $j=m=N/2-1$ only if $\sum_{p}\exp[ikx_{p}]=0$.
770:
771: \section{Entangled atomic ensembles}
772:
773: Solutions (\ref{0022}), (\ref{0023})) imply that a photon and an
774: atomic ensemble are entangled via interaction. If photons are
775: entangled (e.g., by projective measurement) in a combination of such
776: independent systems, then the atomic ensembles will become
777: entangled. We consider optical measurement schemes based on this
778: method, known as entanglement swapping. The key resources used in
779: these schemes are set of atomic ensembles correlated with respective
780: photons, beamsplitters, and single-photon detectors. The analysis
781: that follows is restricted to schemes in which only specific
782: single-photon output is recorded.
783:
784: As an initial state, we use the EPR pair
785: \begin{eqnarray}
786: \label{004}
787: Z(W)=a\ket{0}_{f}\otimes\ket{0}+b\ket{1}_{f}\otimes\ket{W},
788: \end{eqnarray}
789: where Fock states are denoted by the subscript “f,”
790: $\ket{W}=\ket{1;N}/\sqrt{N}, \ket{0}=\ket{0;N}$. It is generated by
791: the mode mixing described by (\ref{00210}), where the mode $b$ is a
792: classical wave. The state of $n$ independent identical ensembles
793: entangled with respective photons is represented by the product
794: \begin{eqnarray} %\nonumber
795: \label{0041} Z_{n}(W)=Z(W)^{\otimes
796: n}%&&\\
797: =a^{n-1}b\Big[\ket{10\dots0}_{f}\otimes\ket{W0\dots 0}+\dots
798: \ket{00\dots 1}_{f}\otimes\ket{00\dots W}\Big]+\dots&&
799: \end{eqnarray}
800: As illustrated by the figure, the photons associated with atomic
801: ensembles are injected into a system of $n-1$ beamsplitters with $n$
802: input ports and n output ports.
803: \begin{figure}[h]
804: \label{Bs}
805: \centering
806: \epsfxsize=9cm \epsfbox{Ahh.eps}
807: \caption{(a) Scheme for generating entangled states of atomic
808: ensembles. (b) Preparation of entangled states by correlation of
809: photocounts recorded by two schemes.}
810: \end{figure}
811: Each beamsplitter performs the transformation $\ket{01}_{f}\to
812: c_{k}\ket{01}_{f}+s_{k}\ket{10}_{f}$,
813: $\ket{10}_{f}\to-s_{k}\ket{01}_{f}+c_{k}\ket{10}_{f}$, where
814: $c_{k}^{2}+s_{k}^{2}=1$, $k=1\dots n-1$. The scheme is described by
815: a unitary operator $U_{nf}$ and characterized by the following
816: property. There exist an input port optically coupled to every
817: output port and an output port optically coupled to every input
818: port. In Fig. 1a, the latter is output port 1. We call it the
819: optical output port, and the corresponding detector is called the
820: output detector. The scheme performs the transformation
821: \begin{eqnarray}
822: \nonumber
823: U_{nf}\ket{1\dots 0}_{f}=t_{1}\ket{1\dots 0}_{f}+\dots
824: +t_{n}\ket{0\dots 1}_{f},&&\\
825: \label{0042}
826: U_{nf}^{-1}\ket{1\dots 0}_{f}=\tau_{1}\ket{1\dots
827: 0}_{f}+\dots +\tau_{n}\ket{0\dots 1}_{f},&&
828: \end{eqnarray}
829: where the coefficients $t_{k},\tau_{k}$, $k=1,\dots n$ are
830: determined by the transmittances and reflectances of the
831: beamsplitters, and $\sum_{k}t^{2}_{k}= \sum_{k}\tau^{2}_{k}=1$.
832:
833: If output detector detects a photon which corresponds to the state
834: $\ket{1_{f}}=\ket{1\dots 0}$, then with the probability
835: \begin{eqnarray}
836: \label{000ver}
837: Prob(1)=|a^{n-1}b|^{2}
838: \end{eqnarray}
839: entangled state of atomic ensembles will be prepared
840: \begin{eqnarray}
841: \nonumber
842: \label{0043}
843: \bra{1_{f}}U_{nf}Z_{n}(W)/\sqrt{Prob(1)}=\eta_{n}(W),&&\\
844: \eta_{n}(W)=q_{1}\ket{W\dots 0}+q_{n}\ket{0\dots W},&&
845: \end{eqnarray}
846: This scheme has the following property. Since the coefficients
847: $q_{1},..q_{n}$ are completely determined by the transmittances and
848: reflectances of the beamsplitters, weakly entangled states $Z(W)$
849: can be used to prepare highly entangled states atomic ensembles.
850:
851: Let us consider several particular cases. If $n = 2$, then
852: $q_{1}=c_{1},q_{2}=s_{1}$, and we have an EPR pair of the form $
853: \eta_{2}(W)=EPR(W)=c_{1}\ket{W0}+s_{1}\ket{0W}$. When
854: $c_{1}=s_{1}=1/\sqrt{2}$ it is maximally entangled. If $n = 3$ and
855: $q_{1}=c_{1}c_{2},q_{2}=-s_{1}c_{2},q_{3}=s_{2}$, then we have a $W$
856: state. If $c_{1}=-s_{1}=1/\sqrt{2}$, $c_{2}=\sqrt{2/3}$ and
857: $c_{2}=\sqrt{2/3}$, then
858: \begin{eqnarray}
859: \label{0047}
860: \eta_{3}(W)=W(W)=(1/\sqrt{3})(\ket{W00}+\ket{0W0}+\ket{00W}).
861: \end{eqnarray}
862: In particular, one can prepare the asymmetric state
863: $\widetilde{W}(W)=(1/\sqrt{2})\ket{W00}+(1/2)\ket{0W0}+(1/2)\ket{00W}$.
864: When $N = 1$, it is unitary equivalent to the GHZ state and can be
865: used as a quantum channel for teleportation or dense coding
866: \cite{29}.
867:
868: Using correlation between photocounts in a combination of schemes
869: considered above, mixed states of atomic ensembles can be prepared,
870: including inseparable ones. For example, consider two independent
871: identical schemes $S_{2}(X)$ combined as shown in Fig. 1b, with
872: three single-photon detectors in each scheme. If a photon is
873: detected by either scheme, then we have the pair of states $
874: \bra{1_{f}}S_{2}(X)\otimes\bra{0_{f}}S_{2}(X)w=\ket{\eta_{2}(X),0}$
875: and $
876: \bra{0_{f}}S_{2}(X)\otimes\bra{1_{f}}S_{2}(X)w=\ket{0,\eta_{2}(X)}
877: $. Suppose that the detector outputs are connected so that a single
878: photon produced by either scheme is counted. This measurement is
879: described by the projector $\proj{1_{f}0_{f}}+\proj{0_{f}1_{f}}$.
880: The resulting mixed state is represented by a density matrix of the
881: form
882: \begin{eqnarray}
883: \label{0057}
884: \rho(X)=(1/2)\Big[\proj{\eta_{2}(X),0}+\proj{0,\eta_{2}(X)}\Big].
885: \end{eqnarray}
886: Its separability is an open question, because a necessary and
887: sufficient condition is known only for mixed systems of dimension
888: $2\times 2, 2\times 3$. However, if we assume that $N = 1$, i.e.,
889: consider a combination of four atoms instead of ensembles, then
890: $\eta_{2}(X)=\Psi^{+}=(1/\sqrt{2})(\ket{01}+\ket{10})$ and density
891: matrix (\ref{0057}) describes a four-particle state:
892: \begin{eqnarray}
893: \rho(4)=(1/2)(\proj{\Psi^{+}00}+\proj{00\Psi^{+}}).
894: \end{eqnarray}
895: Taking the state of the pair of atoms in the first scheme defined by
896: the two-particle reduced density matrix
897: $\rho(2)=(1/2)\Big[\proj{\Psi^{+}}+\proj{00}\Big]$, we can apply the
898: separability criterion. The density matrix partially transposed over
899: the variables of the one atom has four eigenvalues one of which is
900: negative $1/4,1/4, (1\pm\sqrt{2})/4$. Therefore, the density matrix
901: $\rho(4)$ is inseparable.
902:
903: \section{Hierarchic structure of states}
904:
905: Note that expression (\ref{0043}) is hierarchically structured. To
906: illustrate this property, we consider a combination of schemes
907: generating states of this type. As distinct to schemes using
908: correlation of photocounts, we consider optically connected schemes.
909: If an elementary scheme that performs the transformation $S_{n}(X) =
910: U_{nf}Z_{n}(W)$ with $X = W$ (see Fig. 1a) records single-photon
911: output, then the resulting state has the form of (\ref{0043}):
912: \begin{eqnarray}
913: \label{591}
914: \bra{1_{f}}S_{n}(X)w=\eta_{n}(X)=\tau^{`}_{1}\ket{X0\dots
915: 0}+\dots+\tau^{`}_{n}\ket{00\dots X},
916: \end{eqnarray}
917: where $w=1/\sqrt{Prob(1)}$. We define the optical output port of the
918: scheme $S_{n}(X)$ as the one optically coupled to every input port.
919: In Fig. 1a, it is output port 1. The input port of the scheme
920: $S_{n}(X)$ is defined as the optical input port of the system of
921: beamsplitters. Then, we can take, for example, p independent schemes
922: represented as $(S_{n}(X))^{p}$ and use their optical outputs as the
923: input of the scheme $S_{p}$. As a result, we have a new scheme
924: $S_{p}((S_{n}(X))^{p})$. If it records single-photon output, we have
925: an entangled state that consists of lower level entangled states:
926: \begin{eqnarray}
927: \bra{1_{f}}S_{p}((S_{n}(X))^{p})w=\eta_{p}(\eta_{n}(X))
928: =t_{1}\ket{\eta_{n}(X),0\dots
929: 0}+\dots+t_{p}\ket{0,0,\dots\eta_{n}(X)}.
930: \end{eqnarray}
931: By virtue (\ref{591}) it takes the forme:
932: \begin{eqnarray}
933: \eta_{p}(\eta_{n}(X)) =\eta_{pn}(X).
934: \end{eqnarray}
935: \\
936: Thus, we can formulate the following property. The state
937: $\eta_{n}(X)$ defined by (\ref{0047}) with $n=n_{1}n_{2}\dots n_{p}$
938: can be represented as
939: \begin{eqnarray}
940: \eta_{n}(X)=\eta_{n_{1}}(\eta_{n_{2}}(\dots(\eta_{n_{p}})).
941: \end{eqnarray}
942: This implies that the vector $\eta_{n}(X)$ has the structure of an
943: entangled state with respect to any group of $s$ particles, where s
944: is such that $n/s$ is a natural number greater than unity.
945:
946: When the wavefunction $\eta_{n}(X)$ is symmetric, a hierarchically
947: structured representation can be obtained by using the permanent
948: expansion defined as a determinant with a summation rule for
949: permutations depending on symmetry \cite{30}. In particular,
950: successive decomposition of a determinant with respect to rows or
951: columns and subsequent association of summands can be used to
952: represent a permanent in terms of permanents of lower dimension,
953: which reflects hierarchical structure.
954:
955: For example, when $n = 6$, it holds that
956: \begin{eqnarray}
957: \eta_{6}(X)=\eta_{3}(\eta_{2}(X))=\eta_{2}(\eta_{3}(X))).
958: \end{eqnarray}
959: This state has the structure of an EPR pair or a W state:
960: \begin{eqnarray}
961: \nonumber \eta_{3}(\eta_{2}(X))=W(EPR)=EPR(W).&&
962: \end{eqnarray}
963: This example demonstrates that the same state exhibits structure
964: characteristic of entangled states of two different types. This
965: property can be used in different applications: the EPR pair can
966: serve as a quantum channel for teleportation or dense coding, while
967: the symmetric W state can be used for cloning.
968:
969: To choose a particular structure defined by the dimension of the
970: Hilbert space of its element, appropriate basis vectors and
971: observables should be introduced. In physical terms, this is
972: equivalent to a two-level approximation. Indeed, any group of $s$
973: particles, where $s$ is such that $n/s$ is a natural number greater
974: than unity, is represented in $\eta_{n}(X)$ by two states,
975: $\ket{0}=0_{s}$ è $\eta_{s}(X)=1_{s}$. The group can be treated as a
976: two-level particle (qubit) with basis vectors $0_{s}$ and $1_{s}$.
977: Such qubits and hierarchically structured states $\eta_{n}(X)$ can
978: be used in quantum information processing. By analogy with
979: (\ref{0055}), the vector $\eta_{n}(X)$ represents a Dicke state only
980: if $\eta_{n}(X)$.
981:
982: \section{Conclusions}
983:
984: A model describing resonant interaction of identical two-level atoms
985: with a narrow-band radiation field is used to analyze multiparticle
986: entanglement. The interaction is described by an effective
987: Hamiltonian that allows for various multiphoton processes. The
988: statistics of radiation and atoms are characterized by a density
989: matrix, for which solutions are calculated in secondorder
990: perturbation theory in the interaction strength and exact solutions
991: are found in the case of negligible decay. It is shown that the
992: state of any pair of atoms interacting with weak single-mode
993: squeezed light is inseparable and robust against decay. It is
994: demonstrated that symmetric entangled multiparticle states can be
995: generated by using optical schemes based on singlephoton projection.
996: An optical scheme is described that can be used to prepare highly
997: states of entangled atomic ensembles from weakly entangled states by
998: projective measurement.
999:
1000: This work was supported, in part, by Delzell Foundation, Inc.
1001:
1002: \begin{thebibliography}{50}
1003: \bibitem{Pol}
1004: B. Julsgaard, A. Kozhekin, and E. S. Polzik, Nature 413, 400 (2001).
1005: \bibitem{M}
1006: C. Monroe, D. M. Meekhof, B. E. King, et al., Phys. Rev. Lett. 75,
1007: 4714 (1995).
1008: \bibitem{HMN}
1009: E. Hagley, X. Maitre, G. Nogues, et al., Phys. Rev. Lett. 79, 1
1010: (1997).
1011: \bibitem{Lat}
1012: G. K. Brennen, C. M. Caves, P. S. Jessen, and I. H. Deutsch, Phys.
1013: Rev. Lett. 82, 1060 (1999).
1014: \bibitem{MS}
1015: A. V. Burlakov, M. V. Chekhova, and O. A. Karabutova, Phys. Rev. A
1016: 60, 4209 (1999); A. V. Burlakov and M. V. Chekhova, Pis’ma Zh. Eksp.
1017: Teor. Fiz. 75, 505 (2002) [JETP Lett. 75, 432 (2002)]; A. V.
1018: Burlakov, L. A. Krivitskioe, S. P. Kulik, et al., Opt. Spektrosk.
1019: 94, 743 (2003) [Opt. Spectrosc. 94, 684 (2003)]; Yu. I. Bogdanov, L.
1020: A. Krivitskioe, and S. P. Kulik, Pis’ma Zh. Eksp. Teor. Fiz. 78, 804
1021: (2003) [JETP Lett. 78, 352 (2003)]; L. A. Krivitskioe, S. P. Kulik,
1022: A. N. Penin, and M. V. Chekhova, Zh. Eksp. Teor. Fiz. 124, 943
1023: (2003) [JETP 97, 846 (2003)]
1024: \bibitem{6}
1025: W. Dur, G. Vidal, and J. I. Cirac, Phys. Rev. A 62, 062314 (2000).
1026: \bibitem{7}
1027: M. Eibl, N. Kiesel, M. Bourennane, et al., Phys. Rev. Lett. 92,
1028: 077901 (2004).
1029: \bibitem{8}
1030: F. Verstraete, J. Dehaene, B. De Moore, and H. Verscheld, Phys. Rev.
1031: A 65, 052112 (2002).
1032: \bibitem{9}
1033: A. Peres, Phys. Rev. Lett. 77, 1413 (1996); M. Horodecki, P.
1034: Horodecki, and R. Horodecki, Phys. Lett. A 223, 1 (1996).
1035: \bibitem{10}
1036: A. M. Basharov, Pis’ma Zh. Eksp. Teor. Fiz. 75, 151 (2002) [JETP
1037: Lett. 75, 123 (2002)]; Zh. Eksp. Teor. Fiz. 121, 1249 (2002) [JETP
1038: 94, 1070 (2002)].
1039: \bibitem{11}
1040: M. Plesch, J. Novotny, Z. Dzurakova, and V. Buzek, J. Phys. A: Math.
1041: Gen. 37, 1843 (2004).
1042: \bibitem{12}
1043: W. Dur, Phys. Rev. A 63, R020303 (2001).
1044: \bibitem{13}
1045: J. K. Stockton, G. M. Geremia, A. C. Doherty, and H. Mabuchi, Phys.
1046: Rev. A 67, 022112 (2003).
1047: \bibitem{14}
1048: V. Buzek and M. Hillery, Phys. Rev. A 54, 1844 (1998); M. Mirao, D.
1049: Jonathan, M. B. Plenio, and V. Vedral, Phys. Rev. A 59, 156 (1999).
1050: \bibitem{15}
1051: J. Joo, Y.-J. Park, J. Lee, and J. Jang, quant-ph/0204003.
1052: \bibitem{16}
1053: V. N. Gorbachev, A. I. Trubilko, A. A. Rodichkina, and A. I.
1054: Zhiliba, Phys. Lett. A 314, 267 (2003).
1055: \bibitem{17}
1056: E. Knill, R. Laflamme, and G. J. Milburn, Nature 409, 46 (2001).
1057: \bibitem{18}
1058: H. J. Briegel and R. Raussendorf, Phys. Rev. Lett. 86, 5188 (2001).
1059: \bibitem{19}
1060: R. Dicke, Phys. Rev. 93, 99 (1954).
1061: \bibitem{20}
1062: A. V. Andreev, V. I. Emel’yanov, and Yu. A. Il’inskioe, Cooperative
1063: Phenomena in Optics (Nauka, Moscow, 1988) [in Russian]
1064: \bibitem{21}
1065: H. Hammer, quant-ph/0407094.
1066: \bibitem{22}
1067: A. Olaya-Castro, N. F. Johnson, and L. Quiroga, Phys. Rev. A 70,
1068: R020301 (2004).
1069: \bibitem{23}
1070: N. Gisin and S. Popescu, Phys. Rev. Lett. 83, 432 (1999).
1071: \bibitem{24}
1072: A. M. Basharov, Method of Unitary Transformation in Nonlinear Optics
1073: (Mosk. Inzh.–Fiz. Inst., Moscow, 1990) [in Russian]; Zh. Eksp. Teor.
1074: Fiz. 102, 1126 (1992) [Sov. Phys. JETP 75, 611 (1992)].
1075: \bibitem{25}
1076: D. V. Kupriyanov, I. M. Sokolov, and A. V. Slavgorodskii, Phys. Rev.
1077: A 63, 063811 (2001); Phys. Rev. A 68, 043815 (2003).
1078: \bibitem{26}
1079: V. N. Gorbachev, A. I. Zhiliba, A. A. Rodichkina, and A. I.
1080: Trubilko, Phys. Lett. A. 323, 339 (2004).
1081: \bibitem{27}
1082: H. J. Briegel and R. Raussendorf, Phys. Rev. Lett. 86, 910 (2001);
1083: R. Raussendorf, D. E. Browne, and H. J. Briegel, quant-ph/0301052.
1084: \bibitem{28}
1085: A. M. Basharov and E. A. Manykin, Opt. Spektrosk. 96, 91 (2004)
1086: [Opt. Spectrosc. 96, 81 (2004)].
1087: \bibitem{29}
1088: V. N. Gorbachev, A. I. Trubilko, A. A. Rodichkina, and A. I.
1089: Zhiliba, Phys. Lett. A 314, 267 (2003).
1090: \bibitem{30}
1091: S. S. Schweber, An Introduction to Relativistic Quantum Field Theory
1092: (Row, Peterson, Evanston, Ill, 1961; Inostrannaya Literatura,
1093: Moscow, 1963).
1094:
1095: \end{thebibliography}
1096:
1097: \end{document}
1098: %
1099: