1: \documentclass[aps,pra,twocolumn,showpacs]{revtex4}
2:
3: \usepackage{amssymb}
4: \usepackage{amsmath}
5: \usepackage{graphicx}
6:
7: \begin{document}
8:
9: \title{Long-range entanglement generation via frequent measurements}
10: \author{L.-A. Wu, D.A. Lidar, S. Schneider}
11: \affiliation{Chemical Physics Theory Group, Chemistry Department, University of Toronto,
12: 80 St. George Street, Toronto, Ontario M5S 3H6, Canada}
13:
14: \begin{abstract}
15: A method is introduced whereby two non-interacting quantum subsystems, that
16: each interact with a third subsystem, are entangled via repeated projective
17: measurements of the state of the third subsystem. A variety of physical
18: examples are presented. The method can be used to establish long range
19: entanglement between distant parties in one parallel measurement step, thus
20: obviating the need for entanglement swapping.
21: \end{abstract}
22:
23: \pacs{03.67.Lx, 03.65.Xp,74.20.Fg}
24: \maketitle
25:
26: \section{Introduction}
27:
28: Entanglement is at the heart of quantum information processing \cite{Gruska:book}, and is a resource that allows quantum processes to outperform
29: their classical counterparts for tasks such as computation \cite{Ekert:98},
30: communication \cite{Raz:99}, and cryptography \cite{Ekert:91}. The standard
31: way to generate entanglement between distinguishable and initially separable
32: particles is to let them interact directly for a certain amount of time:
33: any non-trivial two-body Hamiltonian is capable of generating entanglement
34: in this manner \cite{Zhang:03}. However, the direct interaction method poses
35: limitations in the context of the generation of \emph{long-range}
36: entanglement, since in many systems the interaction strength typically
37: decreases at least as fast as some power of the distance between particles.
38: Here we introduce a different paradigm for entanglement generation: we show
39: that it is possible \emph{to entangle two particles that never interact
40: directly by means of repeated measurements of a third subsystem that interacts with both}. In addition to its
41: conceptual interest, we show that this scheme offers practical advantages for long-range
42: entanglement generation.
43:
44: We remark that other alternatives to the
45: direct interaction method for entanglement generation are well known. E.g.,
46: in \cite{Knill:00} linear optics is augmented by \emph{measurements} to
47: generate entanglement and perform quantum computation. Measurements of the
48: phase of light transmitted through a cavity have also been proposed in \cite{Sorensen:03} as a method to prepare entangled states and implement quantum
49: computation in the case of atoms in optical cavities. \emph{Decoherence} can
50: also be used to create entanglement: e.g., atoms in a leaky cavity can
51: become entangled conditioned on null detection at a photo detector placed
52: outside the cavity \cite{Plenio:99,Beige:00a}. In Ref.~\cite{Cabrillo:99}
53: entanglement is generated in a similar manner conditioned instead upon
54: detection.
55:
56: Here we present a rather general theoretical framework for
57: measurement-generated entanglement. Our scheme is inspired by the recent
58: work by Nakazato, Takazawa, and Yuasa (NTY) \cite{Nakazato:03}, who
59: investigated the effects of repeated rapid projective measurements on one
60: subsystem of a bipartite system. Provided that the resulting operator on the
61: unobserved subsystem has a non-degenerate largest eigenvalue, NTY showed that \emph{the unobserved subsystem is gradually
62: projected into a pure state, independent of the initial state}. This result
63: leads to the question of whether a similar measurement scheme is capable of
64: generating entanglement when the unobserved subsystem is itself
65: multi-partite, non-interacting, and in an arbitrary initial state. We answer
66: this question in the affirmative for a wide range of model systems, and
67: establish general conditions for validity of the method. We note that an
68: additional advantage of this measurement-based method for generating
69: entanglement over the direct-interaction based method is that does not
70: depend on sensitive timing of interactions: instead, entanglement is
71: gradually purified as the number of measurements increases, and (under
72: appropriate conditions) can be made arbitrarily high. An important
73: conclusion from our study is that the method can be used to establish long
74: range entanglement between distant particles, by measuring along a chain of
75: intermediate particles. This provides an alternative to entanglement
76: swapping that does not scale with the chain length. A similar result
77: -- long range entanglement from local (but non-repeated) measurements -- was obtained in
78: Ref.~\cite{Verstraete:04} for the ground state of an antiferromagnetic spin chain. We now turn a detailed
79: description of our method.
80:
81: \section{Preliminaries}
82:
83: NTY considered the following scenario: Consider a bi-partite system composed
84: of subsystems $A,B$, initially in the separable state $\rho _{AB}(0)=|\phi
85: \rangle \langle \phi |\otimes \rho _{B}$, where $A$ is in a pure state $%
86: |\phi \rangle $ and $\rho _{B}$ is arbitrary. Subsystem $A$ is subject to
87: projective measurements $P_{A}=|\phi \rangle \langle \phi |$ applied with
88: period $\tau $. In between measurements the system evolves under the
89: Hamiltonian
90: \begin{equation}
91: H=H_{A}+H_{B}+H_{AB}
92: \end{equation}
93: (resp., the sum of free Hamiltonians and an interaction). It can be shown
94: \cite{Nakazato:03} that after $M$ such measurements the state of subsystem $B
95: $, \emph{given} that all outcomes were $|\phi \rangle $, is
96: \begin{equation}
97: \rho _{B}(M\tau )=\frac{V_{B}(\tau )^{M}\rho _{B}V_{B}^{\dagger }(\tau )^{M}%
98: }{P_{M}}, \label{eq:rhoB}
99: \end{equation}%
100: where $P_{M}$ is the survival probability, i.e., the probability of finding
101: subsystem $A$ in its initial state,%
102: \begin{equation}
103: P_{M}=\mathrm{Tr}_{B}[V_{B}(\tau )^{M}\rho _{B}V_{B}^{\dagger }(\tau )^{M}]
104: \label{successprob}
105: \end{equation}%
106: where $U=\exp (-i\tau H)$ (we use units where $\hbar =1$), and where
107: \begin{equation}
108: V_{B}(\tau )\equiv \langle \phi |U|\phi \rangle \label{eq:VB}
109: \end{equation}%
110: is an operator on subsystem $B$ that is in general \emph{not} Hermitian.
111: However, one may still find left- and right-eigenvectors with complex
112: eigenvalues, whose modulus can be shown to be bounded between $0$ and $1$.
113: The central result of NTY is that in the limit of large $M$ and small but
114: finite $\tau $, $\rho _{B}(M\tau )$ \emph{tends to a pure state independent
115: of }$B$\emph{'s initial state}. This result assumes that the largest
116: eigenvalue $\lambda _{\mathrm{\max }}$ of $V_{B}(\tau )$ is non-degenerate.
117: The final pure state $\rho _{B}(M\tau )$ is then the corresponding right
118: eigenvector $|u_{\max }\rangle $ and this outcome is found with probability
119: \begin{equation}
120: P_{M}\rightarrow |\lambda _{\mathrm{\max }}|^{2}\langle u_{\max }|u_{\max
121: }\rangle \langle v_{\max }|\rho _{B}|v_{\max }\rangle ,
122: \end{equation}
123: where $|v_{\max }\rangle $ is the corresponding left eigenvector. Note that
124: because of $\tau $'s finiteness the dynamics here is distinct from the Zeno
125: effect and \textquotedblleft quantum Zeno dynamics\textquotedblright\ \cite%
126: {Facchi:PRL02}.
127:
128: We now extend NTY's model by allowing $B$ itself to be composed of multiple
129: \emph{non-interacting} subsystems, $B=\{B_{1},B_{2},...,B_{N}\}$, and pose
130: the question: is the NTY measurement procedure capable of generating
131: entanglement amongst $B$'s subsystems, assuming that they all interact with
132: the \textquotedblleft station\textquotedblright\ $A$? The situation is
133: illustrated in Fig.~\ref{figone}. To answer this questions we consider the
134: following model:\ We set all internal Hamiltonians $H_{A}=H_{B_{i}}=0$, so
135: that $H=H_{AB}$ (equivalently, we can always transform to an interaction
136: picture rotating with the internal Hamiltonians; this will make $H_{AB}$
137: time dependent, but it will not introduce couplings between subsystems $%
138: B_{i} $). We assume that the $B$ subsystems are all qubits and the
139: interaction between $A$ and all the $B_{i}$ is identical. We can then define
140: total quasi-spin operators $\sigma _{\beta }\equiv \sum_{i=1}^{N}\sigma
141: _{\beta }^{i}$ and write the interaction as $H_{AB}=\sum_{\alpha \in
142: \{x,y,z\}}h_{\alpha }A_{\alpha }\otimes \sigma _{\alpha }$ ($h_{\alpha }$
143: real) or, more simply,
144: \begin{equation}
145: H_{AB}=\overrightarrow{A}\cdot \overrightarrow{\sigma } \label{eq3-1}
146: \end{equation}%
147: where the parameters $h_{\alpha }$ are included in the vector $%
148: \overrightarrow{A}$. We do not restrict subsystem $A$.
149:
150: \begin{figure}[tbp]
151: %\hspace{1.5cm}
152: \includegraphics[width=7cm]{fig1.eps}
153: \caption{Schematic illustration of our model: Systems $B_1$ and $B_2$ are
154: coupled to system $A$; system $A$ can be measured with a projective
155: measurement. Note that no direct coupling exists between systems $B_1$ and $B_2$.}
156: \label{figone}
157: \end{figure}
158:
159: \section{General theory}
160:
161: \subsection{A good basis}
162:
163: We construct a basis that block-diagonalizes the effective post-measurement
164: evolution operator $V_{B}(\tau )$. Let a basis for the $A$ subsystem be the
165: measurement state $\left\vert \phi \right\rangle $ and any set of states
166: orthonormal to it, denoted $\left\vert \phi _{l}^{\bot }\right\rangle $, $%
167: l=1,...,d-1$. Let $N$ be even and let a basis for the $B$ subsystem be
168: constructed out of the usual spin basis $|S,M_{S}\rangle $, where $S$ is the
169: total quasi-spin of the $N$ particles and $M_{S}$ its projection along the $%
170: z $-axis. We denote the (orthonormal) singlet states by $|s_{j}\rangle $, $%
171: j=1,...,D_{N}=N!/[(N/2+1)!(N/2)!]$, and the remaining (orthonormal) states
172: with $S>0$ by $|t_{k}\rangle $, $k=1,...,K_{N}$, $K_{N}=2^{N}-D_{N}$.
173:
174: \textit{Proposition 1}:\ Consider the ordered basis $\{\left\vert \phi
175: \right\rangle ,\left\vert \phi _{1}^{\bot }\right\rangle ,...,\left\vert
176: \phi _{d-1}^{\bot }\right\rangle \}\otimes \{|s_{1}\rangle
177: ,...,|s_{D_{N}}\rangle ,|t_{1}\rangle ,...,|t_{K_{N}}\rangle \}$. In this
178: basis we have the block-diagonal representation
179: \begin{equation}
180: V_{B}(\tau )=\left(
181: \begin{tabular}{l|l}
182: $I_{D_{N}}$ & $\mathbf{0}$ \\ \hline
183: $\mathbf{0}$ & $V_{B}^{s}$%
184: \end{tabular}%
185: \right) , \label{eq:VB-blockrep}
186: \end{equation}%
187: where $I_{D_{N}}$ is a $D_{N}\times D_{N}$-dimensional identity matrix. The
188: maximal eigenvalue of $V_{B}(\tau )$ is $1$, and is at least $D_{N}$-fold
189: degenerate.
190:
191: \textit{Proof}: The singlets states, $|S,M_{S}\rangle =|0,0\rangle $, are
192: annihilated by the $\sigma _{\beta }$ when $N$ is even (since they are
193: states with zero total quasi-spin). Therefore $H_{AB}|s_{j}\rangle =0$,
194: independent of the state of subsystem $A$. In the ordered basis $H_{AB}$ is
195: thus represented as%
196: \begin{equation}
197: H_{AB}=\left(
198: \begin{tabular}{l|l}
199: $\mathbf{0}$ & $\mathbf{0}$ \\ \hline
200: $\mathbf{0}$ & $H_{AB}^{^{\prime }}$%
201: \end{tabular}%
202: \right)
203: \end{equation}%
204: where the dimension of the upper-left block is $D_{N}\times D_{N}$ and that
205: of the lower-right block is $K_{N}\times K_{N}$. Then%
206: \begin{equation}
207: U_{AB}=e^{-i\tau H_{AB}}=\left(
208: \begin{tabular}{l|l}
209: $I_{D_{N}}$ & $\mathbf{0}$ \\ \hline
210: $\mathbf{0}$ & $U_{AB}^{\prime }$%
211: \end{tabular}%
212: \right) , \label{eq5}
213: \end{equation}%
214: where $U_{AB}^{\prime }=W^{\prime }e^{-i\tau \Lambda }W^{\prime \dag }$ is
215: unitary and $W^{\prime }$ is the unitary matrix that diagonalizes $%
216: H_{AB}^{\prime }$: $W^{\prime \dag }H_{AB}^{\prime }W^{\prime }=\Lambda =%
217: \mathrm{diag}(\lambda _{1},...,\lambda _{K_{N}})$. Taking the expectation
218: value wrt $|\phi \rangle $ we then find:%
219: \begin{equation}
220: V_{B}(\tau )=\sum_{j=1}^{D_{N}}|s_{j}\rangle \langle s_{j}|+V_{B}^{s},
221: \label{eq7}
222: \end{equation}%
223: which is the claimed result, with $V_{B}^{s}\equiv \sum_{k,k^{\prime
224: }=1}^{K_{N}}|t_{k}\rangle \left( U_{AB}^{\prime }\right) _{\phi k,\phi
225: k^{\prime }}\langle t_{k^{\prime }}|$ a $K_{N}\times K_{N}$-dimensional
226: matrix. That the maximal eigenvalue is $1$ follows from unitarity of $%
227: U_{AB}^{\prime }$ \cite{Nakazato:03}, and that it is at least $D_{N}$-fold
228: degenerate is immediate from the representation (\ref{eq:VB-blockrep}). QED
229:
230: \textit{Proposition 2 }\cite{Nakazato:03}: Non-degeneracy of $V_{B}(\tau )$
231: is a necessary condition for obtaining a pure state in the limit of large $M$%
232: .
233:
234: \textit{Proof}: In the degenerate case it follows from Eq.~(\ref{eq:rhoB})\
235: that the method yields an equally weighted sum of degenerate pure states
236: corresponding to the maximum eigenvalue of $V_{B}(\tau )$. QED
237:
238: Note that $D_{2}=1$. Hence in the totally symmetric case we have been
239: considering so far, for $N=2$ \emph{the method will generate a pure
240: (maximally entangled) singlet state}, provided $V_{B}^{s}$ has maximal
241: eigenvalue with modulus smaller than $1$. We give examples of corresponding
242: Hamiltonians below. On the other hand, $D_{4}=2$ and hence due to degeneracy
243: the method will not produce a pure entangled state for $N\geq 4$. However,
244: we can still generate pure-state entanglement in the $N=4$ case by breaking
245: the total permutation symmetry and only preserving the symmetry between $1,2$
246: and $3,4$ [e.g., having a coupling to subsystem $A$ such that $\sigma
247: _{\beta }=a_{1}(\sigma _{\beta }^{1}+\sigma _{\beta }^{2})+a_{2}(\sigma
248: _{\beta }^{3}+\sigma _{\beta }^{4})$, $a_{1}\neq a_{2}\neq 0$]. We can then
249: project out a \emph{one}-dimensional subspace $|s\rangle _{12}|s\rangle
250: _{34} $, $|s\rangle _{ij}=(|0_{i}1_{j}\rangle -|1_{i}0_{j}\rangle )/\sqrt{2}$
251: is the singlet state, which is entangled for $1,2$ and $3,4$ but separable
252: across the $12:34$ partition. Similarly, for $N=2n\geq 6$ and $\sigma
253: _{\beta }=\sum_{i=1}^{n}a_{i}(\sigma _{\beta }^{2i-1}+\sigma _{\beta }^{2i})$%
254: , $a_{i}\neq a_{j}\neq 0$ the projected state is $|s\rangle _{12}|s\rangle
255: _{34}......|s\rangle _{n-3,n-2}|s\rangle _{n-1,n}$.
256:
257: \subsection{Invariance}
258:
259: The results of the above discussion are invariant as long as the
260: Hamiltonians belong to the family $\{U_{B}H_{AB}U_{B}^{\dagger }\}$, where $%
261: U_{B}$ is an arbitrary unitary transformation of the Hamiltonian of
262: subsystem $B$. This enables the method to generate entangled states
263: equivalent under local transformations. The invariance of $%
264: H_{AB}|s_{j}\rangle =0$ is $H_{AB}^{\prime }|s_{j}^{\prime }\rangle =0$,
265: where $H_{AB}^{\prime }=U_{B}H_{AB}U_{B}^{\dagger }$ and $|s_{j}^{\prime
266: }\rangle =U_{B}|s_{j}\rangle $. Thus, e.g., in the case of $N=2$, we can
267: generate the other three Bell states by applying $U_{B}=X_{2},Y_{2},Z_{2}$
268: (where $X_{2}$ is the Pauli matrix $\sigma _{x}$ acting on subsystem $B_{2}$%
269: , etc.), which results, respectively, in $|s^{\prime }\rangle =\frac{1}{%
270: \sqrt{2}}(|11\rangle +|00\rangle ),\frac{1}{\sqrt{2}}(|00\rangle -|11\rangle
271: ),\frac{1}{\sqrt{2}}(|01\rangle +|10\rangle )$ as the outcome of the method.
272:
273: \subsection{Degenerate maximal eigenvalue}
274:
275: What happens when there are other states, either singlets or coming from $%
276: V_{B}^{s}$, that also have an eigenvalue with modulus $1$? In this case
277: NTY's result does not apply and entanglement may or may not be spoiled
278: (though it is always reduced from maximally entangled). Two examples will
279: illustrate: (i) $N=2$:\ Suppose there is a triplet state $|00\rangle $ that
280: also has $|\lambda _{\mathrm{\max }}|=1$. Then the resulting state is the
281: mixture $|s\rangle \langle s|+|00\rangle \langle 00|$. This state is
282: entangled as its partial transpose \cite{Peres:96} has a negative eigenvalue
283: of $-0.207$. We will encounter this case in the Heisenberg model below. (ii)
284: $N=4$:\ Suppose the two singlet states $|s_{1}\rangle =|s\rangle
285: _{12}|s\rangle _{34}$ and $|s_{2}\rangle =\frac{1}{\sqrt{3}}[|t_{+}\rangle
286: _{12}|t_{-}\rangle _{34}+|t_{-}\rangle _{12}|t_{+}\rangle
287: _{34}-2|t_{0}\rangle _{12}|t_{0}\rangle _{34}]$ (where $|t_{\alpha }\rangle $
288: are triplets with projection quantum number $\alpha $) appear with
289: eigenvalue one, but no other states do. With respect to the $12:34$ cut $%
290: |s_{1}\rangle $ is a product state but $|s_{2}\rangle $ is clearly
291: entangled. This state, just like in the previous example, has negative
292: partial transpose, and we have entanglement \emph{across the }$12:34$\emph{\
293: cut}. These examples illustrate that degeneracy still allows for \emph{mixed
294: state entanglement} to be generated by our method. This is useful
295: entanglement in the sense that it can be used for teleportation and all
296: other QIP primitives \cite{Horodecki:97}.
297:
298: \section{Examples}
299:
300: We now discuss examples, first limiting ourselves to $A$ being a qubit and $%
301: N=2$. Our task then reduces to calculating the eigenvalues of the $3\times 3$
302: matrix $V_{B}^{s}=\langle \phi |U_{AB}^{\prime }|\phi \rangle $. However,
303: this requires diagonalization of $U_{AB}^{\prime }$, an $6\times 6$ matrix,
304: so cannot be done analytically in complete generality. Our basis for $A$ is $%
305: \{|\phi \rangle ,|\phi ^{\bot }\rangle \}$ and for $B$ is
306: \begin{eqnarray}
307: \vec{\beta}\equiv
308: \{|s\rangle &=&\frac{1}{\sqrt{2}}(|01\rangle -|10\rangle ),|t_{-}\rangle
309: =|00\rangle ,\notag \\
310: |t_{0}\rangle &=&\frac{1}{\sqrt{2}}(|01\rangle +|10\rangle
311: ),|t_{+}\rangle =|11\rangle \}\},
312: \label{eq:beta}
313: \end{eqnarray}
314: without loss of generality (recall the
315: invariance discussion above).
316:
317: \subsection{Axial symmetry model}
318:
319: Suppose $A_{z}=0$ and $A_{x},A_{y}$ satisfy $%
320: [A_{x},A_{y}^{2}]=[A_{y},A_{x}^{2}]=0$ (or $A_{x}$ or $A_{y}=0),$ whence the
321: Hamiltonian reads $H=[A_{x}(X_{1}+X_{2})+A_{y}(Y_{1}+Y_{2})]/2$. We first
322: consider as a special case the $XY$ model: $A_{x}=JX$ and $A_{y}=JY$,
323: relevant for quantum information processing in solid state
324: \cite{Mozyrsky:01,Imamoglu:99,Quiroga:99} and atomic \cite{Zheng:00}
325: systems. It follows after some algebra
326: that $H^{3}=\left\vert A\right\vert ^{2}H$, where $\left\vert A\right\vert
327: ^{2}=A_{x}^{2}+A_{y}^{2}$. Therefore, the evolution is
328: \begin{equation}
329: U_{AB}=e^{-i\tau H}=I-2\left( \frac{H}{|A|}\right) ^{2}\sin ^{2}\frac{\tau
330: \left\vert A\right\vert }{2}-i\frac{H}{|A|}\sin \tau \left\vert A\right\vert
331: , \label{eq8}
332: \end{equation}%
333: which means that $V_{B}^{s}$ is a function only of $\langle \phi |H|\phi
334: \rangle $ and $\langle \phi |H^{2}|\phi \rangle $. In particular, for the $%
335: XY $ model we find, using $\left\vert A\right\vert =\sqrt{2}J$,
336:
337: \begin{widetext}
338: \begin{eqnarray*}
339: V_{B}(\tau ) &=&\cos ^{2}\frac{\tau J}{\sqrt{2}}-\frac{%
340: X_{1}X_{2}+Y_{1}Y_{2}-\langle Z\rangle (Z_{1}+Z_{2})}{2}
341: \sin ^{2}\frac{\tau J}{\sqrt{2}}-i\frac{\langle X\rangle
342: (X_{1}+X_{2})+\langle Y\rangle (Y_{1}+Y_{2})}{2\sqrt{2}}\sin \sqrt{2}\tau J
343: \end{eqnarray*}%
344: \end{widetext}
345: where $\langle X\rangle \equiv \langle \phi |X|\phi \rangle $, etc. It is
346: simple to check that, as required from our general result, $V_{B}|s\rangle
347: =|s\rangle $, i.e., the singlet state has eigenvalue $1$. Whether this
348: eigenvalue is degenerate is now seen to depend on the measurement of the $A$
349: subsystem: If we choose $|\phi \rangle $ to be a $\sigma _{z}$-eigenstate
350: then the additional eigenvalues of $V_{B}$ are found to be $\{1,\cos \sqrt{2}%
351: \tau J,\cos \sqrt{2}\tau J\}$, while if we choose $|\phi \rangle $ to be a $%
352: \sigma _{x}$ or $\sigma _{y}$-eigenstate then we find that the additional
353: eigenvalues are $\{\cos ^{2}q,1+3\cos 2q\pm \sqrt{\cos 4q+\sin ^{4}q-1}\}$,
354: where $q=\frac{\tau J}{\sqrt{2}}$. Thus, if we measure along $z$ then there
355: is never pure state entanglement, but at times $\tau $ other than $n\pi /%
356: \sqrt{2}J$ we have a mixed entangled state, since the eigenstate
357: corresponding to the additional eigenvalue $1$ is found to be $|11\rangle $
358: (and recall the discussion above). In the case of measurement along $x$ or $%
359: y $ the eigenvalues periodically have modulus $1$, and with the exception of
360: those times degenerary is avoided and we do obtain a pure entangled state
361: (in particular, it is simple to show that this is true for all $\tau <\sqrt{2%
362: }\arccos (1/3)/J$).
363:
364: Next we consider the performance of the scheme after a \emph{finite} number
365: of steps (Fig.~\ref{fig2}). To do so we calculate the concurrence \cite%
366: {Wootters:98} after $M$ steps of evolution/measurement for $J\tau =\pi /2<%
367: \sqrt{2}\arccos (1/3)$, in the case of $|\phi \rangle $ a $\sigma _{x}$
368: eigenstate. In this case, as shown above, the singlet state is generated by
369: the scheme. We also plot the survival probability $P_{M}$ as given by Eq.~(%
370: \ref{successprob}). After as few as three measurements the concurrence is
371: essentially unity, indicating that the system is already in the desired
372: singlet state. The scheme does, however, not work with unit probability: The
373: probability $P_{M}$ of having projected system $A$ on the desired $\sigma
374: _{x}$ eigenstate converges to $\sim 0.2$ for the values above, a value
375: reached for $M=3$.
376:
377: \begin{figure}[tbp]
378: %\hspace{1.5cm}
379: \includegraphics[width=7cm]{concetc_dip.ps}
380: \caption{Concurrence ($\triangle $) and success probability ($\blacksquare $%
381: ) $P_{M}$ for $|\protect\phi \rangle =\frac{1}{\protect\sqrt{2}}(|0\rangle
382: +|1\rangle )$ and $\protect\tau J=\protect\pi /2$.}
383: \label{fig2}
384: \end{figure}
385:
386: As another special case we consider a simplified form of the \textit{d}-wave
387: grain boundary qubit \cite{Blais:00}:$\ H=\Delta
388: (X_{1}+X_{2})+JZ(Z_{1}+Z_{2})$ ($\Delta $ is a tunneling parameter and $J$
389: is the Josephson coupling). Thus $\overrightarrow{A}=2(\Delta ,0,JZ)$, and
390: we may use the general result (\ref{eq8}), replacing $A_{y}$ by $A_{z}$. \
391: If we now choose $|\phi \rangle $ to be a $\sigma _{z}$-eigenstate then $%
392: V_{B}$ turns out to be unitary and hence all its eigenvalues have modulus
393: one, and no entanglement is generated. However, if we choose $|\phi \rangle $
394: to be a $\sigma _{x}$-eigenstate, we find
395: \begin{eqnarray*}
396: V_{B} &=&\cos ^{2}\phi -\sin ^{2}\phi (\cos ^{2}\theta X_{1}X_{2}+\sin
397: ^{2}\theta Z_{1}Z_{2}) \\
398: &&-\frac{i}{2}\sin 2\phi \cos \theta (X_{1}+X_{2})
399: \end{eqnarray*}%
400: where $\phi =\tau \sqrt{\Delta ^{2}+J^{2}}$ and $\tan \theta =J/\Delta $. It
401: is again easy to check that $V_{B}|s\rangle =|s\rangle $. One of the other three
402: eigenvalues is $1-2\sin ^{2}\phi \sin ^{2}\theta .$ The other two are
403: complex conjugates and for short times $\tau $ such that $\tan ^{2}\phi
404: <4\cos ^{2}\theta /\sin ^{4}\theta $ have the same amplitude $(1-2\sin
405: ^{2}\phi \sin ^{2}\theta )^{1/2}$. For instance, when $\theta =\pi /4,$ the
406: amplitudes of the three eigenvalues are $\cos ^{2}\phi ,\cos \phi
407: ,\cos \phi$, which vanish rapidly when raised to the power of the number of measurements.
408:
409: Finally, note that the results presented so far are not restricted to $A$
410: being a qubit: the unitary operator (\ref{eq8}) is valid even when $A$ is a
411: multi-level system, as long as the condition $%
412: [A_{x},A_{y}^{2}]=[A_{y},A_{x}^{2}]=0$ is satisfied.
413:
414: \subsection{Heisenberg model}
415:
416: We now consider the Heisenberg interaction $H=J\overrightarrow{\sigma }\cdot
417: (\overrightarrow{\sigma }_{1}+\overrightarrow{\sigma }_{2})$. After showing
418: that $(H+J\overrightarrow{\sigma }_{1}\cdot \overrightarrow{\sigma }%
419: _{2})^{2}=9J^{2}$ and $[H,J\overrightarrow{\sigma }_{1}\cdot \overrightarrow{%
420: \sigma }_{2}]=0$, it is simple to compute $e^{-i\tau (H+J\overrightarrow{%
421: \sigma }_{1}\cdot \overrightarrow{\sigma }_{2})}$, from which we directly
422: obtain
423:
424: \begin{eqnarray*}
425: U_{AB} &=&e^{-i\tau H}=e^{i\tau J\overrightarrow{\sigma }_{1}\cdot
426: \overrightarrow{\sigma }_{2}}\{\cos (3\tau J) \\
427: &&-i\sin (3\tau J)[\overrightarrow{\sigma }\cdot (\overrightarrow{\sigma }%
428: _{1}+\overrightarrow{\sigma }_{2})+\overrightarrow{\sigma }_{1}\cdot
429: \overrightarrow{\sigma }_{2}]/3\}.
430: \end{eqnarray*}%
431: $V_{B}$ then is identical to $U_{AB}$ provided one replaces with $%
432: \left\langle \phi \right\vert $ $\overrightarrow{\sigma }\left\vert \phi
433: \right\rangle $ the operator $\overrightarrow{\sigma }$ in $U_{AB}$. The
434: four eigenvalues are $\{1,e^{-2i\tau J},e^{i\tau J}\left( \cos 3\tau J\pm i%
435: \frac{1}{3}\sin 3\tau J\right) \}$, the last two having the same amplitude $%
436: 1-\frac{8}{9}\sin ^{2}3\tau J<1$. The corresponding eigenvectors are,
437: respectively, $\vec{\beta}$ [Eq.~(\ref{eq:beta})]. Since the first two eigenvalues both have
438: magnitude $1$, we have the case discussed above:\ a pure entangled state
439: cannot be projected out by our method, but we can prepare a mixed entangled
440: state that is useful for all QIP\ protocols. Note further that these
441: Heisenberg model results hold for the entire class of Hamiltonians $H=%
442: \overrightarrow{A}_{\sigma }\cdot (\overrightarrow{\sigma }_{1}+%
443: \overrightarrow{\sigma }_{2})$, when $\overrightarrow{A}_{\sigma }$ is
444: generated from an arbitrary $2$-dimensional unitary transformation $U_{A}$, $%
445: \overrightarrow{A}_{\sigma }=U_{A}J$ $\overrightarrow{\sigma }U_{A}^{\dagger
446: }$ (this is different from the invariance under a rotation of the $B$ system
447: considered above).
448:
449: \subsection{Bosonic media}
450:
451: We now consider a photon or phonon ($A$) interacting symmetrically with two
452: identical qubits ($B$).\ The Jaynes-Cummings Hamiltonian is%
453: \begin{equation*}
454: H=\epsilon b^{\dagger }b+g(\sigma _{1}^{z}+\sigma _{2}^{z})+\frac{J}{2}%
455: [b(\sigma _{1}^{+}+\sigma _{2}^{+})+b^{\dagger }(\sigma _{1}^{-}+\sigma
456: _{2}^{-})]
457: \end{equation*}%
458: where $b$ is a bosonic annihilation operator, and whence $\overrightarrow{A}%
459: =(J(b+b^{\dagger }),iJ(b-b^{\dagger }),g)$. Such a model can easily be
460: realized using microwave cavity QED \cite{Rauschenbeutel:99} (two atoms in
461: one or two cavities) or trapped ions \cite{Wineland:98}. We can exactly
462: diagonalize $H$ by writing it as a direct sum over three-dimensional matrices in the basis $%
463: \left\vert n-1,t_{+}\right\rangle ,$ $\left\vert n,t_{0}\right\rangle $ and $%
464: \left\vert n+1,t_{-}\right\rangle $, where $\left\vert n\right\rangle $ are
465: number states and $\left\vert t_{\alpha }\right\rangle $ are triplet states
466: of the two qubits. Note that $\mathcal{N}\equiv \lbrack b^{\dagger
467: }b+(\sigma _{1}^{z}+\sigma _{2}^{z})]$ is a conserved quantity and hence $H$
468: is block diagonal in its eigenvalues. For simplicity we consider the case
469: with $g=\epsilon $ and measure the single photon state \ $\left\vert \phi
470: \right\rangle =b^{\dagger }\left\vert 0\right\rangle =|1\rangle $ (this
471: projects onto a single block of the infinite matrix $U_{AB}$). In this case
472: we readily find the already diagonal
473: \begin{eqnarray*}
474: V_{B}(\tau ) &=& \mathrm{diag}\{1,e^{-2i\epsilon \tau }\frac{3+2\cos (\sqrt{10}%
475: \tau J)}{5},\\
476: &e^{-i\epsilon \tau }&\cos (\sqrt{6}\tau J),\cos (\sqrt{2}\tau J)\}
477: \end{eqnarray*}
478: in the ordered basis $\{|1\rangle |s\rangle ,\left\vert 1,t_{+}\right\rangle
479: ,\left\vert 1,t_{0}\right\rangle ,\left\vert 1,t_{-}\right\rangle \}$. Thus,
480: as long as we make sure that $\tau <2\pi /\sqrt{10}J$ the method will
481: project out a pure singlet state. It follows from our discussion of
482: invariance above that another Bell state $\frac{1}{\sqrt{2}}(|01\rangle
483: +|10\rangle )$ can be projected out if the two qubits couple to the photon
484: or phonon with opposite signs.
485:
486: \subsection{Multilevel Systems}
487:
488: Let us now consider a rather general, though abstract multilevel case. Assume that
489: subsystem $B$ consists of two particles that have $M\geq 2$ levels each, and
490: that the interaction Hamiltonian with subsystem $A$ (also an $M$-level
491: system) is of the form
492:
493: \begin{equation*}
494: H=\sum_{i,j=1}^{M}A_{ij}(O_{ij}^{1}+O_{ji}^{2}),
495: \end{equation*}%
496: where $O_{ij}=E_{ij}-E_{ji}$ and $E_{ij}=\left\vert i\right\rangle
497: \left\langle j\right\vert $ is a matrix whose elements are zero everywhere
498: except for a $1$ at position $(i,j)$. Namely, there is an $SO(M)$ symmetry
499: for odd $M$ or an $Sp(M)$ symmetry for even $M$. In this case $H$ has a
500: \emph{single} non-degenerate eigenvector with \emph{zero} eigenvalue. This
501: state is entangled. For instance, when $N=3$, we have two qutrits, which can
502: be represented in the spherical basis $\{\left\vert -1\right\rangle
503: ,\left\vert 0\right\rangle ,\left\vert +1\right\rangle \}$. The state with
504: zero eigenvalue is
505: \begin{equation*}
506: |\Psi \rangle =(\left\vert 1\right\rangle _{1}\left\vert -1\right\rangle
507: _{2}-\left\vert 0\right\rangle _{1}\left\vert 0\right\rangle _{2}+\left\vert
508: -1\right\rangle _{1}\left\vert 1\right\rangle _{2})/\sqrt{3},
509: \end{equation*}%
510: and is maximally entangled. In general, the one-dimensional subspace
511: containing the state with zero eigenvalue is the irreducible representation $%
512: (0,0,...,0)$ of $SO(M)$ or $Sp(M)$. This state is the generalization of the
513: singlet state that arose in the general theory and examples treated above
514: when the subsystems were qubits.
515:
516: \section{Perturbative treatment}
517:
518: It is apparent from the above examples that the success of our method
519: depends on keeping the period $\tau $ between two measurements short. For
520: instance, in the bosonic example, we require $\tau <2\pi /\sqrt{10}J$ in
521: order to prevent the appearance of another eigenvalue with amplitude one.
522: Let us therefore consider a short-time expansion of $V_{B}(\tau )$. To first
523: order
524: \begin{eqnarray*}
525: V_{B}(\tau )^{M} &=& (I-i\tau \left\langle \phi \right\vert
526: H_{AB}\left\vert \phi \right\rangle )^{M} \\
527: &\overset{M\rightarrow \infty }{%
528: \longrightarrow }& \exp (-it\left\langle \phi \right\vert H_{AB}\left\vert
529: \phi \right\rangle ),
530: \end{eqnarray*}
531: where $M\tau =t$ (constant), i.e., the evolution is
532: unitary. This is the quantum Zeno effect, which completely decouples the
533: interaction among all parties of subsystem $B$, so that no entanglement can
534: be generated in this limit. Thus the dominant contribution to entanglement
535: generation originates from the second-order term, which contains the \emph{%
536: self-correlation} of subsystem $B$. Consider for simplicity the case $%
537: \left\langle \phi \right\vert H_{AB}\left\vert \phi \right\rangle =0$ (as in
538: our last example). Letting $M\tau ^{2}/2=t^{2}$ (constant), we have
539: \begin{equation*}
540: V_{B}(\tau )^{M}\overset{M\rightarrow \infty }{\longrightarrow }%
541: e^{-t^{2}\left\langle \phi \right\vert H_{AB}^{2}\left\vert \phi
542: \right\rangle }
543: \end{equation*}%
544: All eigenstates of $\left\langle \phi \right\vert H_{AB}^{2}\left\vert \phi
545: \right\rangle $ with nonzero eigenvalues are rapidly suppresed as $t$ (in
546: practice $M$) increases, while those with eigenvalue zero survive. Since it
547: is much simpler to analytically calculate $\left\langle \phi \right\vert
548: H_{AB}^{2}\left\vert \phi \right\rangle $ than $V_{B}(\tau )$, this
549: perturbative method provides a relatively simple tool for estimating the
550: possibility of entanglement generation via our method, for complicated
551: systems. We consider spin-orbital coupling as another illustrative example: $%
552: H=hL_{x}(X_{1}+X_{2})+hL_{y}(Y_{1}+Y_{2})$, denoting two spin qubits that
553: couple with the same orbital angular momentum. If we measure the eigenstate $%
554: \left\vert \phi \right\rangle =\left\vert l,0\right\rangle $ of the orbital
555: component, we find $\left\langle \phi \right\vert H\left\vert \phi
556: \right\rangle =0$ and $\left\langle \phi \right\vert H^{2}\left\vert \phi
557: \right\rangle =hl(l+1)(2+X_{1}X_{2}+Y_{1}Y_{2})$. The eigenvalues are $%
558: hl(l+1)\{0,2,4,2\}$ with respective eigenvectors $\vec{\beta}$, meaning that
559: the singlet state $\left\vert s\right\rangle $ is projected out in the
560: second-order analysis. Note that this example is qualitatively different
561: from the previous one since subsystem $A$ here does not refer to a
562: physically separate particle.
563:
564: \section{Preparation of long-distance entanglement}
565:
566: Finally we come to our main result: the generation of long-distance
567: bi-partite entanglement. Since in our discussion above there was no
568: restriction on the size of the $A$ subsystem, in a sense long-distance
569: entanglement generation already follows from the results
570: above. However, it is important to specify how the measurements on $A$
571: can be carried out. Assume that the $A$ subsystem is composed of $N-2$
572: qubits $2,3,...,N-1$ and we wish to entangle the two $B$-subsystem qubits $%
573: 1,N$. We consider again as an illustrative example a simplified form of the
574: \textit{d}-wave grain boundary qubit \cite{Blais:00}:
575: \begin{widetext}
576: \begin{eqnarray*}
577: U_{AB} =\exp [-iJ\tau (X_{1}+X_{N})
578: -i\Delta \tau (Z_{1}Z_{2}+Z_{2}Z_{3}+h_{3}+Z_{N-2}Z_{N-1}+Z_{N-1}Z_{N})]
579: \end{eqnarray*}%
580: where $h_{3}=\sum_{i=3}^{N-3}Z_{i}Z_{i+1}$. We choose $|\phi \rangle $ to be
581: the state $\left\vert +\right\rangle _{2}\left\vert R\right\rangle
582: \left\vert +\right\rangle _{N-1}$, where $\left\vert R\right\rangle
583: =\left\vert +\right\rangle _{3}\left\vert +\right\rangle _{4}...\left\vert
584: +\right\rangle _{N-2}$, where $|+\rangle =(|0\rangle +|1\rangle )/\sqrt{2}$.
585: It follows after some calculations that
586: \begin{eqnarray*}
587: V_{B} &=&\frac{1}{2}\{a[\cos ^{2}\phi -\sin ^{2}\phi (\cos ^{2}\theta
588: X_{1}X_{N}+\sin ^{2}\theta Z_{1}Z_{N})-i\frac{1}{2}\sin 2\phi \cos \theta
589: (X_{1}+X_{N}) \\
590: &&+b[\cos ^{2}\phi -\sin ^{2}\phi (\cos ^{2}\theta X_{1}X_{N}-\sin
591: ^{2}\theta Z_{1}Z_{N})-i\frac{1}{2}\sin 2\phi \cos \theta (X_{1}+X_{N})\}
592: \end{eqnarray*}%
593: where
594: \begin{eqnarray*}
595: a &=&\left\langle R\right\vert e^{-i\Delta \tau
596: (Z_{3}+Z_{N-2}+h_{3})}\left\vert R\right\rangle =\left\langle R\right\vert
597: e^{-i\Delta \tau (-Z_{3}-Z_{N-2}+h_{3})}\left\vert R\right\rangle \\
598: b &=&\left\langle R\right\vert e^{-i\Delta \tau
599: (Z_{3}-Z_{N-2}+h_{3})}\left\vert R\right\rangle =\left\langle R\right\vert
600: e^{-i\Delta \tau (-Z_{3}+Z_{N-2}+h_{3})}\left\vert R\right\rangle
601: \end{eqnarray*}
602: \end{widetext}
603: (and are real numbers for odd $N$), $\phi =\tau \sqrt{J^{2}+\Delta ^{2}}$, $%
604: \tan \theta =\Delta /J$. Two of the eigenvalues and eigenstates are:
605:
606: \begin{eqnarray*}
607: E_{1} &=&\frac{1}{2}\{a+b(1-2\sin ^{2}\phi \sin ^{2}\theta )\};\frac{1}{%
608: \sqrt{2}}(\left\vert 1_{1}0_{N}\right\rangle -\left\vert
609: 0_{1}1_{N}\right\rangle ), \\
610: E_{2} &=&\frac{1}{2}\{b+a(1-2\sin ^{2}\phi \sin ^{2}\theta )\};\frac{1}{%
611: \sqrt{2}}(\left\vert 0_{1}0_{N}\right\rangle -\left\vert
612: 1_{1}1_{N}\right\rangle ).
613: \end{eqnarray*}%
614: Thus, as desired, entanglement is generated between the $B$ subsystem
615: particles $1,N$. Note that since the measurement of all the particles of the
616: $A$ subsystem is carried out in parallel \emph{this method for long-range
617: entanglement generation is independent of the distance} $N$.
618:
619: The amplitudes of the other two eigenvalues are $\sqrt{E_{1}E_{2}}$ in the
620: case of odd $N\geq 5$. Since $\max (|E_{2}|,|E_{1}|)>\sqrt{E_{1}E_{2}}$ for
621: sufficiently short time, and $E_{1}-E_{2}=2\left( \sin \phi \right)
622: ^{2}\left( \sin \theta \right) ^{2}\left( a-b\right) $ as long as $a\neq b$,
623: the procedure prepares either $\frac{1}{\sqrt{2}}(\left\vert
624: 1_{1}0_{N}\right\rangle -\left\vert 0_{1}1_{N}\right\rangle )$ or $\frac{1}{%
625: \sqrt{2}}(\left\vert 0_{1}0_{N}\right\rangle -\left\vert
626: 1_{1}1_{N}\right\rangle )$. We find, e.g, for $N=5$: $a=\cos 2\Delta \tau $,
627: $b=1$ and $\frac{1}{\sqrt{2}}(\left\vert 00\right\rangle -\left\vert
628: 11\right\rangle )$ is projected out, while for $N=7$: $a=\frac{1}{4}(3+\cos
629: 4\Delta \tau )$, $b=\cos 2\Delta \tau $ and $\frac{1}{\sqrt{2}}(\left\vert
630: 01\right\rangle -\left\vert 10\right\rangle )$ is projected out. The case of
631: even $N$ is more complicated since the amplitudes of other two eigenvalues
632: depend on the values of $a$ and $b$. E.g., in the $N=4$, $V_{B}$ has the
633: same form as above but $a=e^{-i\Delta \tau }$ and $b=e^{i\Delta \tau }$; all
634: roots have the same amplitude and no pure state will be projected out.$%
635: \allowbreak $
636:
637: \section{Conclusions}
638:
639: We have introduced a measurement-based method for entangling two systems
640: that only interact indirectly, via a third system. Applications range from
641: solid state to atomic quantum information processing. One of the advantages
642: of this measurement-based method of entanglement generation is that it does
643: not depend on precisely timed interactions, as are interactions-based
644: schemes. The method can be used to prepare arbitrarily long-distance
645: entanglement via a chain of intermediate particle in one parallel
646: measurement step. This may have useful applications in reducing the latency
647: overhead in quantum communication in quantum computing architectures. An
648: interesting open question which we leave for future research is whether the
649: same method can be used to apply quantum logic gates between non-interacting
650: particles.
651:
652: Acknowledgents.--- D.A.L. acknowledges support from the Sloan
653: Foundation, D-Wave Systems, Inc., and the
654: DARPA-QuIST program (managed by AFOSR under agreement
655: No. F49620-01-1-0468).
656:
657:
658: %\bibliographystyle{/nfs/theory/u0/dlidar/revtex/prsty}
659: %\bibliography{/nfs/theory/u0/dlidar/articles/bib}
660: %\bibliographystyle{prsty}
661: %\bibliography{bib}
662:
663: \begin{thebibliography}{10}
664:
665: \bibitem{Gruska:book}
666: J. Gruska, {\em Quantum Computing} (Mc Graw-Hill, London, 1999).
667:
668: \bibitem{Ekert:98}
669: {A. Ekert and R. Jozsa}, Phil. Trans. Roy. Soc. (Lond.) {\bf 356}, 1769
670: (1998).
671:
672: \bibitem{Raz:99}
673: {R. Raz}, in {\em {Proceedings of 31st Annual ACM Symposium on Theory of
674: Computing (STOC)}} ({ACM}, {New York, NY}, 1999), p. 358.
675:
676: \bibitem{Ekert:91}
677: {A. Ekert}, Phys. Rev. Lett. {\bf 67}, 661 (1991).
678:
679: \bibitem{Zhang:03}
680: {J. Zhang, J. Vala, S. Sastry, K.B. Whaley}, Phys. Rev. A {\bf 67}, 042313
681: (2003).
682:
683: \bibitem{Knill:00}
684: {E. Knill, R. Laflamme, and G.J. Milburn}, Nature {\bf 409}, 46 (2001).
685:
686: \bibitem{Sorensen:03}
687: {A. S$\o$rensen and K. M$\o$lmer}, Phys. Rev. Lett. {\bf 91}, 097905 (2003).
688:
689: \bibitem{Plenio:99}
690: {M.B. Plenio, S.F. Huelga, A. Beige and P.L. Knight}, Phys. Rev. A {\bf 59},
691: 2468 (1999).
692:
693: \bibitem{Beige:00a}
694: {A. Beige, S. Bose, D. Braun, S.F. Huelga, P.L. Knight, M.B. Plenio, V.Vedral}, J. Mod. Optics {\bf 47}, 2583 (2000).
695:
696: \bibitem{Cabrillo:99}
697: {C. Cabrillo, J.I. Cirac, P. Garcia-Fern\'{a}ndez, P. Zoller}, Phys. Rev. A {\bf 59}, 1025 (1999).
698:
699: \bibitem{Nakazato:03}
700: {H. Nakazato, T. Takazawa, K. Yuasa}, Phys. Rev. Lett. {\bf 90}, 060401
701: (2003).
702:
703: \bibitem{Verstraete:04}
704: {F. Verstraete, M.A. Martin-Delgado, J.I. Cirac}, Phys. Rev. Lett. {\bf 92}, 087201 (2004).
705:
706: \bibitem{Facchi:PRL02}
707: {P. Facchi and S. Pascazio}, Phys. Rev. Lett. {\bf 89}, 080401 (2002).
708:
709: \bibitem{Peres:96}
710: {A. Peres}, Phys. Rev. Lett. {\bf 77}, 1413 (1996).
711:
712: \bibitem{Horodecki:97}
713: {M. Horodecki, P. Horodecki, R. Horodecki}, Phys. Rev. Lett. {\bf 78}, 574
714: (1997).
715:
716: \bibitem{Mozyrsky:01}
717: {D. Mozyrsky, V. Privman, and M.L. Glasser}, Phys. Rev. Lett. {\bf 86}, 5112
718: (2001).
719:
720: \bibitem{Imamoglu:99}
721: {A. Imamo$\bar{\rm g}$lu, D.D. Awschalom, G. Burkard, D.P. DiVincenzo, D. Loss,
722: M. Sherwin and A. Small}, Phys. Rev. Lett. {\bf 83}, 4204 (1999).
723:
724: \bibitem{Quiroga:99}
725: {L. Quiroga and N.F. Johnson}, Phys. Rev. Lett. {\bf 83}, 2270 (1999).
726:
727: \bibitem{Zheng:00}
728: {S.-B. Zheng and G.-C Guo}, Phys. Rev. Lett. {\bf 85}, 2392 (2000).
729:
730: \bibitem{Wootters:98}
731: {W.K. Wootters}, Phys. Rev. Lett. {\bf 80}, 2245 (1998).
732:
733: \bibitem{Blais:00}
734: {A. Blais and A.M. Zagoskin}, Phys. Rev. A {\bf 61}, 042308 (2000).
735:
736: \bibitem{Rauschenbeutel:99}
737: {A. Rauschenbeutel, G. Nogues, S. Osnaghi, P. Bertet, M. Brune, J.M. Raimond,
738: and S. Haroche}, Phys. Rev. Lett. {\bf 83}, 5166 (1999).
739:
740: \bibitem{Wineland:98}
741: {D. J. Wineland, C. Monroe, W. M. Itano, D. Leibfried, B. E. King, and D. M.
742: Meekhof}, {J. of Res. of the National Inst. of Standards and Technology} {\bf
743: 103}, 259 (1998).
744:
745: \end{thebibliography}
746:
747:
748: \end{document}
749: