1:
2: \documentclass[aps,prl,twocolumn]{revtex4}
3: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4: \usepackage{amssymb}
5: \usepackage{amsmath}
6: \usepackage{graphicx}
7: \usepackage{epsfig}
8:
9:
10: \begin{document}
11:
12: \title{Quantum information storage and state transfer based on spin systems}
13: \author{Z. Song$^{1,a}$ and C. P. Sun $^{1,2,a,b}$}
14: \affiliation{$^{1}$Department of Physics, Nankai University, Tianjin 300071, China}
15: \affiliation{$^{2}$ Institute of Theoretical Physics, Chinese Academy of Sciences,
16: Beijing, 100080, China}
17:
18: \begin{abstract}
19: The idea of quantum state storage is generalized to describe the coherent
20: transfer of quantum information through a coherent data bus. In this
21: universal framework, we comprehensively review our recent systematical
22: investigations to explore the possibility of implementing the physical
23: processes of quantum information storage and state transfer by using quantum
24: spin systems, which may be an isotropic antiferromagnetic spin ladder system
25: or a ferromagnetic Heisenberg spin chain. Our studies emphasize the physical
26: mechanisms and the fundamental problems behind the various protocols for the
27: storage and transfer of quantum information in solid state systems.
28: \end{abstract}
29:
30: \pacs{PACS number:03.67.-a, 03.67.Lx, 03.65.Ud, 75.10.Jm}
31: \maketitle
32:
33: \section{I. Introduction}
34:
35: The current development of quantum information science and technology
36: demands optimal systems serving as long-lived quantum memories, through
37: which the quantum information carried by a quantum system with short
38: decoherence time can be coherently transferred \cite{q-inf}. In this sense a
39: quantum channel or a quantum data bus is needed for perfect transmission of
40: quantum states. In this article, we will demonstrate that both the quantum
41: information storage and the quantum state transfer can be uniquely described
42: in a universal framework.
43:
44: There exist some schemes \cite{lukin,Flei,sun-prl} concerning about quantum
45: storage of photon states, while there are also some efforts devoted to the
46: universal quantum storage for a qubit (a basic two-level system) state,
47: which is necessary in quantum computation. For example, most recently an
48: interesting protocol \cite{lukin1,zoller,exp} was presented to reversibly
49: map the electronic spin state onto the collective spin state of the
50: surrounding nuclei. Because of the long decoherence time of the nuclear
51: spins, the information stored in them can be robustly preserved. It was
52: found that \cite{szs}, only under two homogeneous conditions with low
53: excitations, such many-nuclei system approximately behaves as a single mode
54: boson to serve as an efficient quantum memory.
55:
56: The low excitation condition requires a ground state with all spins
57: orientated, which can be prepared by applying a magnetic field polarizing
58: all spins along the same direction. With the concept of spontaneous symmetry
59: breaking (SSB), one can recognize that a ferromagnetic Heisenberg spin chain
60: usually has a spontaneous magnetization, which naturally offers such a kind
61: of ground state. In happen of SSB, the intrinsic interaction between spins
62: will strongly correlate with the nuclei to form the magnon, a collective
63: mode of spin wave, even without any external magnetic field. With these
64: considerations, Wang, Li, Song, and Sun \cite{wlss} explored the possibility
65: of using a ferromagnetic quantum spin system, instead of the free nuclear
66: ensemble, to serve as a robust quantum memory. A protocol was present to
67: implement a quantum storage element for the electronic spin state in a ring
68: array of interacting nuclei. Under appropriate control of both the electron
69: and the external magnetic field, an arbitrary quantum state of the
70: electronic spin qubit, either pure or mixed state, can be coherently stored
71: in the nuclear spin wave and then read out in reverse process.
72:
73: On the other hand, designed for a more realistic quantum computing, a
74: scalable architecture of quantum network should be based on the solid state
75: system \cite{Div,Bose}. However, the intrinsic feature of solid state based
76: channels, such as the finiteness of the correlation length \cite%
77: {White,Dagotto} and the environment induced noise (especially the low
78: frequency noise) may block this scalability. Fortunately, analytical study
79: shows that a spin system possessing a commensurate structure of energy
80: spectrum matched with the corresponding parity can ensure the perfect state
81: transfer \cite{Matt1,songz1,Matt2}. Based on this fact, an isotropic
82: antiferromagnetic spin ladder system can be pre-engineered as a novel robust
83: kind of quantum data bus \cite{songz2}. Because the effective coupling
84: strength between the two spins connected to a spin ladder is inversely
85: proportional to the distance of the two spins, the quantum information can
86: be transferred between the two spins separated by a longer distance. Another
87: example of the near-perfect transfer of quantum information was given to
88: illustrate an application of the theorem. The protocol of such near-perfect
89: quantum state transfer is proposed by using a ferromagnetic Heisenberg chain
90: with uniform coupling constant, but an external parabolic magnetic field
91: \cite{songz1}.
92:
93: The present paper will give a broad overview of the present situation of the
94: our investigations mentioned above on quantum state storage and quantum
95: information coherent transfer based on quantum spin systems. We will
96: understand the physical mechanisms and the fundamental problems behind these
97: protocols in the view of a unified conception, the generalized quantum
98: information storage.
99:
100: \section{II. Generalized Quantum Storage as a Dynamic Process}
101:
102: For the dynamic process recording and reading quantum information carried by
103: quantum states, we first describe the idea of generalized quantum storage,
104: which was also introduced in association with the Berry's phase factor\cite%
105: {geo}. Let $M$ be a quantum memory possessing a subspace spanned by $%
106: |M_{n}\rangle ,(n=1,2,...,d$, $\left\langle M_{n}\right. \left\vert
107: M_{m}\right\rangle =\delta _{nm}),$ which can store the quantum information
108: of a system $S$ with basis vectors $|S_{n}\rangle ,n=1,2,...,d$. If there
109: exists a controlled time evolution interpolating between the initial state $%
110: |S_{n}\rangle $ $\otimes |M\rangle $ and the final state $|S\rangle \otimes
111: |M_{n}\rangle $ for each index $n$ and arbitrarily given states $|S\rangle $
112: and $|M\rangle $, we define the usual quantum storage by using a factorized
113: evolution of time $T_{m}$
114: \begin{equation}
115: |\Phi (T_{m})\rangle =U(T_{m})|\Phi (0)\rangle =|S\rangle \otimes
116: |M_{n}\rangle ,
117: \end{equation}%
118: starting from the initial state $|\Phi (0)\rangle =|S_{n}\rangle \otimes
119: |M\rangle $. The corresponding readout process is an inverse evolution of \
120: time $T_{f}(>T_{m})$
121: \begin{equation}
122: |\Phi (T_{f})\rangle =U(T_{f})|\Phi (0)\rangle =|S_{n}\rangle \otimes
123: |M\rangle.
124: \end{equation}
125: In this sense, writing an arbitrary state $|S(0)\rangle
126: =\sum_{n}c_{n}\,|S_{n}\rangle $ of $S$ into $M$ with the initial state $%
127: |M\rangle $ of quantum memory can be realized as a controlled evolution from
128: time $t=0$ to $t=T_{m}$
129: \begin{equation}
130: \sum_{n}c_{n}|S_{n}\rangle \otimes |M\rangle \rightarrow |S\rangle \otimes
131: \sum_{n}c_{n}|M_{n}\rangle.
132: \end{equation}%
133: The readout process from $M$ is another controlled evolution from time $%
134: t=T_{m}$ to $t=T_{f}$
135: \begin{equation}
136: |S\rangle \otimes \sum_{n}c_{n}|M_{n}\rangle \rightarrow
137: \sum_{n}c_{n}|S_{n}\rangle \otimes |M\rangle.
138: \end{equation}
139: Obviously, the combination of these two processes forms a cyclic evolution
140: that a state totally returns to the initial one.
141:
142: However, in the view of the decoding approach, one need not the
143: \textquotedblleft totally returning" to revival the information of initial
144: state and a difference is allowed by $n-independent$ unitary transformation $%
145: W=W_{S}$ $\otimes 1$, namely,
146: \begin{equation}
147: |S\rangle \otimes W_{M}\sum_{n}c_{n}|M_{n}\rangle \rightarrow
148: (W_{S}\sum_{n}c_{n}|S_{n}\rangle )\otimes |M\rangle .
149: \end{equation}%
150: This is a quantum dynamic process for recording and reading, which defines
151: a quantum storage. Because the factor $W_{S}$ is known to be independent of
152: the initially state, it can be easily decoded from $W_{S}\sum_{n}c_{n}|S_{n}%
153: \rangle $ by the inverse transformation of $W_{S}$.
154: \begin{figure}[tbp]
155: \includegraphics[bb=90 280 520 580, width=7 cm, clip]{fq1.eps}
156: \caption{Demonstration of quantum state transfer as a process of generalized
157: quantum information storage by grouping the data bus $D$ and the target
158: subsystem $S^{B}$as a generalized quantum memory.}
159: \end{figure}
160: We notice that the quantum storage usually relates to two quantum
161: subsystems.
162:
163: \bigskip
164:
165: We will show as follows that the quantum state transfer can be understood as
166: a generalized quantum storage with three subsystems, the input one with the
167: Hilbert space $S^{A}$, the data bus with $D$ and output one with $S^{B}$. As
168: illustrated in Fig. 1, the two subsystems $S^{A}$ and $S^{B}$ located at two
169: distant locations $A$ and $B$ respectively. Then the Hilbert space of the
170: total system can be written as
171: \begin{equation}
172: S_{T}=S^{A}\otimes D\otimes S^{B}\equiv S^{A}\otimes M,
173: \end{equation}%
174: where $M=D\otimes S^{B}$ can be regarded as the generalized quantum memory
175: with the memory space spanned by $|M_{n}\rangle =\left\vert D\right\rangle
176: \otimes U_{B}\left\vert S_{n}^{B}\right\rangle $. Here $\left\vert
177: D\right\rangle $ is a robust state of the data bus and $U_{B}$ represents
178: some local unitary transformations with respect to $B$, which are
179: independent of the initial state. With this notation, the quantum state
180: transfer indeed can be regarded as a generalized QDR.
181:
182: In fact, if one input a state of $\left\vert S^{A}\right\rangle
183: =\sum\nolimits_{n}c_{n}\left\vert S_{n}^{A}\right\rangle $ localized at $A$
184: at $t=0$, the initial state of whole system can be written as
185: \begin{equation}
186: \left\vert \psi (0)\right\rangle =\sum\nolimits_{n}c_{n}\left\vert
187: S_{n}^{A}\right\rangle \otimes |M\rangle
188: \end{equation}%
189: where $|M\rangle =\left\vert D\right\rangle \otimes \left\vert
190: S^{B}\right\rangle $. The quantum state transfer can be usually described as
191: a factorized time evolution at time $t=T_{f}$
192: \begin{eqnarray}
193: \left\vert \psi (T_{f})\right\rangle &=&\left\vert S\right\rangle \otimes
194: \left\vert D\right\rangle \otimes \sum\nolimits_{n}c_{n}U_{B}\left\vert
195: S_{n}^{B}\right\rangle \notag \\
196: &=&|S\rangle \otimes \sum\nolimits_{n}c_{n}|M_{n}\rangle
197: \end{eqnarray}%
198: with $|M_{n}\rangle =\left\vert D\right\rangle \otimes U_{B}\left\vert
199: S_{n}^{B}\right\rangle $. The above equations just demonstrate that the
200: quantum state transfer is essentially a generalized quantum memory with $%
201: W_{M}=(1\otimes U_{B})$. In this sense the revisable quantum state transfer
202: can be regarded as a general readout process.
203:
204: Now we would like to remark on the differences between generalized quantum
205: state storage and other two types of quantum processes, quantum
206: teleportation and quantum copy. In fact, quantum teleportation is
207: theoretically perfect, yielding an output state which revival the input with
208: a fidelity $F=1$. Actually one of necessary procedure in teleportation is to
209: measure the Bell state at location $A$, which will induce the collapse of
210: wavepacket. On the other way around, the quantum state storage process is
211: always on time evolution without any measurement. As for quantum copy the
212: initial state remains unchanged during its copy can be generated in a
213: dynamic process.
214:
215: \section{III. Quantum state transfer in spin systems}
216:
217: A robust quantum information processing based on solid state system is
218: usually implemented in a working spaces panned by the lowest states, which
219: are well separated from other dense spectrum of high excitations. In this
220: sense the energy gap of the solid state system is an important factor we
221: should take into account. The decoherence induced by the environmental noise
222: can also destroy the robustness of quantum information processing, such as
223: the low frequency (e.g, $1/f$ ) noise dominating in the solid state devices.
224: People believe that the gap of the data bus can suppress the stay of
225: transferred state in the middle way in order to enhance the fidelity, but
226: the large gap may result in a shorter correlation length. The relationship
227: between correlation length and the energy gap is usually established in the
228: system with translational symmetry. So we need to consider some
229: modulated-coupling systems or artificially engineered irregular quantum spin
230: systems where the strong correlation between two distant site can be
231: realized.
232:
233: \subsection{1. Theorem for the perfect quantum state transfer}
234:
235: Quantum mechanics shows that perfect state transfer is possible. To sketch
236: our central idea, let us first consider a single particle system with the
237: usual spatial refection symmetry (SRS) in the Hamiltonian $H$. Let $P$ be
238: the spatial refection operator. The SRS is implied by $[H,P]=0$. Now we
239: prove that at time $\pi /E_{0}$ any state $\psi (\mathbf{r})$ can evolve
240: into the reflected state $\pm \psi (-\mathbf{r})$ if the eigenvalues $%
241: \varepsilon _{n}$ match the parities $p_{n}$ in the following way
242: \begin{equation}
243: \varepsilon _{n}=N_{n}E_{0},p_{n}=\pm (-1)^{N_{n}} \label{spmc}
244: \end{equation}%
245: for arbitrary positive integer $N_{n}$ and
246: \begin{equation}
247: H\phi _{n}(\mathbf{r})=\varepsilon _{n}\phi _{n}(\mathbf{r}),P\phi _{n}(%
248: \mathbf{r})=p_{n}\phi _{n}(\mathbf{r}).
249: \end{equation}%
250: Here, $\phi _{n}(\mathbf{r})$ is the common eigen wave function of $H$ and $P
251: $, $\mathbf{r}$ is the position of the particle. We call Eq. (\ref{spmc})
252: the spectrum-parity matching condition (SPMC). The proof of the above
253: rigorous conclusion is a simple, but heuristic exercise in basic quantum
254: mechanics. In fact, for the spatial refection operator, $P\psi (\mathbf{r}%
255: )=\pm \psi (-\mathbf{r})$. For an arbitrarily given state at $t=0,$ $\psi (%
256: \mathbf{r},t)\left\vert _{t=0}\right. =\psi (\mathbf{r})$, it evolves to
257: \begin{equation}
258: \psi (\mathbf{r},t)=e^{-iHt}\psi (\mathbf{r})=\sum_{n}C_{n}e^{-iN_{n}E_{0}t}%
259: \phi _{n}(\mathbf{r})
260: \end{equation}%
261: at time $t$, where $C_{n}=\left\langle \phi _{n}\right\vert \psi \rangle $.
262: Then at time $t=\pi /E_{0}$, we have
263: \begin{equation}
264: \psi (\mathbf{r},\frac{\pi }{E_{0}})=\sum_{n}C_{n}(-1)^{N_{n}}\phi _{n}(%
265: \mathbf{r})=\pm P\psi (\mathbf{r})
266: \end{equation}%
267: that is $\ \psi (\mathbf{r},\pi /E_{0})=\pm \psi (-\mathbf{r}).$This is just
268: the central result \cite{MIS} discovered for quantum spin system that the
269: evolution operator becomes a parity operators $\pm P$ at some instant $%
270: t=(2n+1)\pi /E_{0}$, that is $\exp [-iH\pi /E_{0}]=\pm P.$From the above
271: arguments we have a consequence that if the eigenvalues $\varepsilon
272: _{n}=N_{n}E_{0}$ of a 1-D Hamiltonian $H$ with spatial refection symmetry
273: are odd-number spaced, i.e. $N_{n}-N_{n-1}$ are always odd, any initial
274: state $\psi (x)$ can evolve into $\pm \psi (-x)$ at time $t=\pi /E_{0}$. In
275: fact, for such 1-D systems, the discrete states alternate between even and
276: odd parities. Consider the eigenvalues $\varepsilon _{n}=N_{n}E_{0}$ with
277: odd-number spaced. The next nearest level must be even-number spaced, then
278: the SPMC is satisfied. Obviously, the 1-D SPMC is more realizable for the
279: construction of the model Hamiltonian to perform perfect state transfer.
280:
281: Now, we can directly generalize the above analysis to many particle systems.
282: \ For the quantum spin chain, one can identify the above SRS as the MIS with
283: respect to the center of the quantum spin chain. As the discussion in Ref.
284: \cite{MIS}, we write MIS operation
285: \begin{equation}
286: P\Psi (s_{1,}s_{2,}...,s_{N-1,}s_{N})=\Psi (s_{N},s_{N-1,}...,s_{2},s_{1,})
287: \end{equation}%
288: for the wave function $\Psi (s_{1,}s_{2,}...,s_{N-1,}s_{N})$ of spin chain.
289: Here, $s_{n}=0,1$ denotes the spin values of the n-th qubit.
290:
291: \subsection{2. Perfect state transfer in modulated coupling system}
292:
293: Based on the above analysis, in principle, perfect quantum state transfer is
294: possible in the framework of quantum mechanics. According to SPMC, many spin
295: systems can be pre-engineered for perfect quantum states transfer. For
296: instance, two-site spin-$\frac{1}{2}$ Heisenberg system is the simplest
297: example which meets the SPMC. Recently, M. Christandl et al \cite%
298: {Matt1,Matt2} proposed a $N$-site $XY$ chain with an elaborately designed
299: modulated coupling constants between two nearest neighbor sites, which
300: ensures a perfect state transfer. It is easy to find that this model
301: corresponds the SPMC for the simplest case $N_{n}=n$. A natural extension of
302: the application of the theorem leads to discover other models with $%
303: N_{n}\neq n$. Following this idea, a new class of different models whose
304: spectrum structures obey the SPMC exactly were proposed for perfect state
305: transfer. Consider an $N$-site spin$-\frac{1}{2}$ $XY$ chain with the
306: Hamiltonian
307: \begin{equation}
308: H=2\sum_{i=1}^{N-1}J_{i}[S_{i}^{x}S_{i+1}^{x}+S_{i}^{y}S_{i+1}^{y}]
309: \end{equation}%
310: where $S_{i}^{x},S_{i}^{y}$ and $S_{i}^{z}$ are Pauli matrices for the $i-$%
311: th site, $J_{i}$ is the coupling strength for nearest neighbor interaction.
312: For the open boundary condition, this model is equivalent to the spin-less
313: fermion model. The equivalent Hamiltonian can be written as
314: \begin{equation}
315: H=\sum\limits_{i=1}^{N-1}J_{i}^{[k]}a_{i}^{\dag }a_{i+1}+h.c,
316: \end{equation}%
317: where $a_{i}^{\dag },a_{i}$ are the fermion operators. This describes a
318: simple hopping process in the lattice. According to the SPMC, we can present
319: different \ models (labelled by different positive integer $k\in
320: \{0,1,2,...\}$) through pre-engineering of the coupling strength as $%
321: J_{i}=J_{i}^{[k]}=\sqrt{i\left( N-i\right) }$for even $i$ \ and $%
322: J_{i}=J_{i}^{[k]}=\sqrt{\left( i+2k\right) \left( N-i+2k\right) }$for odd $i$%
323: . By a straightforward calculation, one can find the k-dependent spectrum$%
324: \varepsilon _{n}=-N+2(n-k)-1$ for $n=1,2,...,N/2,$and $\varepsilon _{n}=-N+$
325: $2(n+k)-1$ for $n=N/2+1,...,N$. The corresponding k-dependent eigenstates
326: are
327: \begin{equation}
328: \left\vert \phi _{n}\right\rangle =\sum_{i=1}^{N}c_{ni}\left\vert
329: i\right\rangle =\sum_{i=1}^{N}c_{ni}a_{i}^{\dag }\left\vert 0\right\rangle
330: \end{equation}%
331: where the coefficients $c_{ni}$ can be explicitly determined by the
332: recurrence relation presented in Ref. \cite{songz2}.
333:
334: It is obvious that the model proposed in Ref. \cite{Matt1} is just the
335: special case of our general model in $k=0$. For arbitrary $k$, one can
336: easily check that it meets the our SPMC by a straightforward calculation.
337: Thus we can conclude that these spin systems with a set of pre-engineered
338: couplings $J_{i}^{[k]}$ can serve as the perfect quantum channels that allow
339: the qubit information transfer.
340:
341: \subsection{3. Near-perfect state transfer}
342:
343: In real many-body systems, the dimension of Hilbert space increase with the
344: size $N$ exponentially. For example, $N$-site spin-$\frac{1}{2}$ system, the
345: dimension is $D=2^{N}$, and the symmetry of the Hamiltonian can not help so
346: much. So it is almost impossible to obtain a model to be exactly engineered.
347: In the above arguments we just show the possibility to implement the perfect
348: state transfer of any quantum state over arbitrary long distances in a
349: quantum spin chain. It sheds light into the investigation of near-perfect
350: quantum state transfer. There is a naive way that one select some special
351: states to be transported, which is a coherent superposition of commensurate
352: part of the whole set of eigenstates. For example, we consider a truncated
353: Gaussian wavepacket for an anharmonic oscillator with lower eigenstates to
354: be harmonic. It is obvious that such system allows some special states to
355: transfer with high fidelity. We can implement such approximate harmonic
356: system in a natural spin chain without the pre-engineering of couplings, but
357: the present of a modulated external field. Another way to realize near
358: perfect state transfer is\ to achieve the entangled states and fast quantum
359: states transfer of two spin qubits by connecting two spins to a medium which
360: possesses a spin gap. A perturbation method, the Fr\H{o}hlich
361: transformation, shows that the interaction between the two spins can be
362: mapped to the Heisenberg type coupling.
363:
364: \subsubsection{3.1 Spin ladder}
365:
366: We sketch our idea with the model illustrated in Fig. 2. The whole quantum
367: system we consider here consists of two qubits (A and B) and a $2\times N$%
368: -site two-leg spin ladder. In practice, this system can be realized by the
369: engineered array of quantum dots \cite{QD array}. The total Hamiltonian $%
370: H=H_{M}+H_{q}$
371: \begin{figure}[h]
372: \includegraphics[bb=40 240 550 660, width=7 cm, clip]{l_fig1.eps}
373: \caption{Two qubits $A$ and $B$ connect to a $2\times N$-site spin ladder.
374: The ground state of $H$ with a-type connection (a) is singlet (triplet) when
375: $N$ is even (odd), while for b-type connection (b), one should have opposite
376: result.}
377: \end{figure}
378: contains two parts, the medium Hamiltonian
379: \begin{equation}
380: H_{M}=J\sum_{\left\langle ij\right\rangle \perp }\mathbf{S}_{i}\cdot \mathbf{%
381: S}_{j}+J\sum_{\left\langle ij\right\rangle \parallel }\mathbf{S}_{i}\cdot
382: \mathbf{S}_{j} \label{2}
383: \end{equation}%
384: describing the spin-1/2 Heisenberg spin ladder consisting of two coupled
385: chains and the coupling Hamiltonian
386: \begin{equation}
387: H_{q}=J_{0}\mathbf{S}_{A}\cdot \mathbf{S}_{L}+J_{0}\mathbf{S}_{B}\cdot
388: \mathbf{S}_{R} \label{3}
389: \end{equation}%
390: describing the connections between qubits $A$, $B$ and the ladder. In the
391: term $H_{M},$ $i$ denotes a lattice site on which one electron sits, $%
392: \left\langle ij\right\rangle $ $\perp $ denotes nearest neighbor sites on
393: the same rung, $\left\langle ij\right\rangle $ $\parallel $ denotes nearest
394: neighbors on either leg of the ladder. In term $H_{q}$, $L$ and $R$ denote
395: the sites connecting to the qubits $A$ and $B$ at the ends of the ladder.
396: There are two types of the connection between $\mathbf{S}_{A}(\mathbf{S}_{B})
397: $ and the ladder, which are illustrated in Fig. 2. According to the Lieb's
398: theorem \cite{Lieb}, the spin of the ground state of $H$ with the connection
399: of type a is zero (one) when $N$ is even (odd), while for the connection of
400: type b, one should have an opposite result. For the two-leg spin ladder $%
401: H_{M},$ analytical analysis and numerical results have shown that the ground
402: state and the first excited state of the spin ladder have spin $0$ and $1$
403: respectively \cite{Dagotto,White}. It is also shown that there exists a
404: finite spin gap $\triangle =E_{1}^{M}-E_{g}^{M}\sim J/2.$between the ground
405: state and the first excited state (see the Fig. 3). This fact has been
406: verified by experiments \cite{Dagotto} and is very crucial for our present
407: investigation.
408: \begin{figure}[h]
409: \includegraphics[bb=45 320 540 650, width=7 cm, clip]{fq3.eps}
410: \caption{Schematic illustration of the energy levels of the system. (a) When
411: the connections between two qubits and the medium switch off ($J_{0}=0$) the
412: ground states are degenerate. (b), (c) When $J_{0}$ switches on, the ground
413: state(s) and the first excited state(s) are either singlet or triplet. This
414: is approximately equivalent to that of two coupled spins. }
415: \end{figure}
416: Thus, it can be concluded that the medium can be robustly frozen to its
417: ground state to induce the effective Hamiltonian $H_{eff}=J_{eff}\mathbf{S}%
418: _{A}\cdot \mathbf{S}_{B}$ between the two end qubits. With the effective
419: coupling constant $J_{eff}$ to be calculated in the following, this
420: Hamiltonian depicts the direct exchange coupling between two separated
421: qubits. As the famous Bell states, $H_{eff}$ has singlets and triplets
422: eigenstates $\left\vert j,m\right\rangle _{AB}$: $\left\vert
423: 0,0\right\rangle =\frac{1}{\sqrt{2}}\left( \left\vert \uparrow \right\rangle
424: _{A}\left\vert \downarrow \right\rangle _{B}-\left\vert \downarrow
425: \right\rangle _{A}\left\vert \uparrow \right\rangle _{B}\right) $ and $%
426: \left\vert 1,1\right\rangle =\left\vert \uparrow \right\rangle
427: _{A}\left\vert \uparrow \right\rangle _{B}$, $\left\vert 1,-1\right\rangle
428: =\left\vert \downarrow \right\rangle _{A}\left\vert \downarrow \right\rangle
429: _{B}$, $\left\vert 1,0\right\rangle =\frac{1}{\sqrt{2}}\left( \left\vert
430: \uparrow \right\rangle _{A}\left\vert \downarrow \right\rangle
431: _{B}+\left\vert \downarrow \right\rangle _{A}\left\vert \uparrow
432: \right\rangle _{B}\right) $, which can be used as a channel to share
433: entanglement for a perfect quantum communication in a longer distance.
434:
435: The above central conclusion can be proved with both analytical and
436: numerical methods as follows. To deduce the above effective Hamiltonian we
437: use $|\psi _{g}\rangle _{M}$ ($|\psi _{\alpha }\rangle _{M}$) and $E_{g}$ ($%
438: E_{\alpha }$) to denote ground (excited) states of $H_{\mathrm{M}}$ and the
439: corresponding eigen-values. The zero order eigenstates $\left\vert
440: m\right\rangle $ can then be written as in a joint way
441: \begin{eqnarray}
442: \left\vert j,m\right\rangle _{g} &=&\left\vert j,m\right\rangle _{AB}\otimes
443: |\psi _{g}\rangle _{M}, \notag \\
444: \left\vert \psi _{\alpha }^{jm}(s^{z})\right\rangle &=&\left\vert
445: j,m\right\rangle _{AB}\otimes |\psi _{\alpha }\rangle _{M}
446: \end{eqnarray}%
447: Here, we have considered that $z$-component $%
448: S^{z}=S_{M}^{z}+S_{A}^{z}+S_{B}^{z}$ of total spin is conserved with respect
449: to the connection Hamiltonian $H_{q}$. Since $S_{M}^{z}$ and $S_{M}^{2}$
450: commute with $H_{M}$, we can label $|\psi _{g}\rangle _{M}$ as $|\psi
451: _{g}(s_{M},s_{M}^{z},)\rangle _{M}$ and then $s^{z}=m+s_{M}^{z}$ can
452: characterize the non-coupling spin state $\left\vert \psi _{\alpha
453: }^{jm}(s^{z})\right\rangle $.
454:
455: When the connections between the two qubits and the medium are switched off,
456: i.e., $J_{0}=0,$ the degenerate ground states of $H$ are just $\left\vert
457: j,m\right\rangle _{g}$ with the degenerate energy $E_{g}$ and spin $0,1$
458: respectively, which is illustrated in Fig. 3(a). When the connections
459: between the two qubits and the medium are switched on, the degenerate states
460: with spin $0,1$ \cite{Song} should split as illustrated in Fig. 3(b) and
461: (c). In the case with $J_{0}\ll J$ at lower temperature $kT<J/2$, the medium
462: can be frozen to its ground state and then we have the effective Hamiltonian
463: \begin{eqnarray}
464: H_{\mathrm{eff}} &\cong &\sum_{j^{\prime },m^{\prime },j,m,s^{z}}\frac{%
465: |_{g}\left\langle j,m\right\vert H_{\mathrm{q}}\left\vert \psi _{\alpha
466: }^{j^{\prime }m^{\prime }}(s^{z})\right\rangle |^{2}}{E_{g}-E_{\mathrm{%
467: \alpha }}}\left\vert j,m\right\rangle _{gg}\left\langle j,m\right\vert
468: \notag \\
469: &=&J_{eff}.\mathit{Diag}.(\frac{1}{4},\frac{1}{4},\frac{1}{4},-\frac{3}{4}%
470: )+\varepsilon
471: \end{eqnarray}
472: where%
473: \begin{eqnarray}
474: J_{eff} &=&\sum\limits_{\alpha }\frac{J_{0}^{2}[L(\alpha )R^{\ast }(\alpha
475: )+R(\alpha )L^{\ast }(\alpha )]}{E_{g}-E_{\alpha }}, \label{10} \\
476: \varepsilon &=&\sum\limits_{\alpha }\frac{3J_{0}^{2}\left[ \left\vert
477: L(\alpha )\right\vert ^{2}+\left\vert R(\alpha )\right\vert ^{2}\right] }{%
478: 4\left( E_{g}-E_{\alpha }\right) }. \notag
479: \end{eqnarray}%
480: This just proves the above effective Heisenberg Hamiltonian
481: $H_{eff}$. Here, the matrix elements of interaction $K(\alpha
482: )=_{M}\left\langle \psi _{g}\right\vert S_{K}^{z}\left\vert \psi
483: _{\alpha }\left( 1,0\right) \right\rangle _{M}$ ($K=S,L)$ can be
484: calculated only for the variables of data bus medium. We also
485: remark that, because $S^{z}$ and $S^{2}$ are conserved for
486: $H_{q},$ off-diagonal elements in the above effective Hamiltonian
487: vanish.
488:
489: In temporal summary, we have shown that at lower temperature $kT<J/2$, $H$
490: can be mapped to the effective Hamiltonian $H_{eff}$, which seemingly
491: depicts the direct exchange coupling between two separated qubits. Notice
492: that the coupling strength has the form $J_{eff}\sim g(L)J_{0}^{2}/J$, where
493: $g(L)$ is a function of $L=N+1$, the distance between the two qubits we
494: concerned. Here we take the $N=2$ case as an example. According to Eq. (\ref%
495: {10}) one can get $J_{eff}=-(1/4)J_{0}^{2}/J$ and $(1/3)J_{0}^{2}/J$ when $A$
496: and $B$ connect the plaquette diagonally and adjacently, respectively. This
497: result is in agreement with the theorem \cite{Lieb} about the ground state
498: and the numerical result when $J_{0}\ll J$. In general cases, the behavior $%
499: g(L)$ vs $L$ is very crucial for quantum information since $L/\left\vert
500: J_{eff}\right\vert $ determines the characteristic time of quantum state
501: transfer between the two qubits $A$ and $B$. In order to investigate the
502: profile of $g(L)$, a numerical calculation is performed for the systems $%
503: L=4,5,6,7,8,$ and $10$, with $J=10,20,40,$ and $J_{0}=1$. The spin gap
504: between the ground state(s) and first excite state(s) are calculated, which
505: corresponds to the magnitude of $J_{eff}$. The numerical result is plotted
506: in Fig. 4, which indicates that $J_{eff}\sim 1/(LJ)$. It implies that the
507: characteristic time of quantum state transfer linearly depends on the
508: distance and then guarantees the possibility to realize the entanglement of
509: two separated qubits in practice.
510:
511: In order to verify the validity of the effective Hamiltonian $H_{eff}$, we
512: need to compare the eigenstates of $H_{eff}$ with those reduced states from
513: the eigenstates of the whole system. In general the eigenstates of $H$ can
514: be written formally as
515: \begin{equation}
516: \left\vert \psi \right\rangle =\sum_{jm}c_{jm}\left\vert j,m\right\rangle
517: _{AB}\otimes |\beta _{jm}\rangle _{M}
518: \end{equation}%
519: where \{$|\beta _{jm}\rangle _{M}$ \} is a set of vectors of the data bus,
520: which is not necessarily orthogonal. Then we have the condition $%
521: \sum_{jm}|c_{jm}|_{M}^{2}\langle \beta _{jm}|\beta _{jm}\rangle _{M}=1$ for
522: normalization of $\left\vert \psi \right\rangle $. In this sense the
523: practical description of the A-B subsystem of two quits can only be given by
524: the reduced density matrix
525: \begin{figure}[h]
526: \includegraphics[bb=30 300 510 760, width=7 cm, clip]{l_fig3.eps}
527: \caption{The spin gaps obtained by numerical method for the systems $%
528: L=4,5,6,7,8,$ and $10$, with $J=10,20,40,$ and $J_{0}=1$ are potted, which
529: is corresponding to the magnitude of $J_{eff}$. It indicates that $%
530: J_{eff}\sim 1/(LJ)$.}
531: \end{figure}
532: \begin{eqnarray}
533: \rho _{AB} &=&Tr_{M}(\left\vert \psi \right\rangle \left\langle \psi
534: \right\vert )=\sum_{jm}|c_{jm}|^{2}\left\vert j,m\right\rangle _{AB}\langle
535: j,m| \\
536: &&+\sum_{j^{\prime }m^{\prime }\neq jm}c_{j^{\prime }m^{\prime }}^{\ast
537: }c_{jmM}\langle \beta _{j^{\prime }m^{\prime }}|\beta _{jm}\rangle
538: _{M}\left\vert j,m\right\rangle _{AB}\langle j^{\prime },m^{\prime }| \notag
539: \end{eqnarray}%
540: where $Tr_{M}$ means the trace-over of the variables of the medium. By a
541: straightforward calculation we have
542: \begin{eqnarray}
543: \left\vert c_{11}\right\vert ^{2} &=&\left\vert c_{1-1}\right\vert
544: ^{2}=\left\langle \psi \right\vert \left( \frac{1}{4}+S_{A}^{z}\cdot
545: S_{B}^{z}\right) \left\vert \psi \right\rangle , \notag \\
546: \left\vert c_{00}\right\vert ^{2} &=&\left\langle \psi \right\vert \left(
547: \frac{1}{4}-\mathbf{S}_{A}\cdot \mathbf{S}_{B}\right) \left\vert \psi
548: \right\rangle , \\
549: \left\vert c_{10}\right\vert ^{2} &=&1-2\left\vert c_{11}\right\vert
550: ^{2}-\left\vert c_{00}\right\vert ^{2}. \notag
551: \end{eqnarray}
552: Now we need a criteria to judge how close the practical reduced eigenstate
553: is to the pure state for the effective two sites coupling $H_{eff}$. As we
554: noticed, it has the singlet and triplet eigenstates $\left\vert
555: j,m\right\rangle _{AB}$ in the subspace spanned by $\left\vert
556: 0,0\right\rangle _{AB}$ with $S^{z}=S_{A}^{z}+S_{B}^{z}=0$, we have $%
557: \left\vert c_{11}\right\vert ^{2}=\left\vert c_{10}\right\vert
558: ^{2}=\left\vert c_{1-1}\right\vert ^{2}=0,$ $\left\vert c_{00}\right\vert
559: ^{2}=1;$ for triplet eigenstate $\left\vert 1,0\right\rangle _{AB}$, we have
560: $\left\vert c_{11}\right\vert ^{2}=\left\vert c_{1-1}\right\vert
561: ^{2}=\left\vert c_{00}\right\vert ^{2}=0,$ $\left\vert c_{10}\right\vert
562: ^{2}=1$. With the practical Hamiltonian $H,$ the values of $\left\vert
563: c_{jm}\right\vert ^{2}, i=1,2,3,4$ are numerically calculated for the ground
564: state $\left\vert \psi _{g}\right\rangle $ and first excited state $%
565: \left\vert \psi _{1}\right\rangle $ of finite system systems $L=4,5,6,7,8$
566: and $10$ with $J=10,20,$ and $40,$ $(J_{0}=1)$ in $S^{z}=0$ subspace, which
567: are listed in the Table 1(a,b,c) of Ref. \cite{songz1}. It shows that, at
568: lower temperature, the realistic interaction leads to the results about $%
569: \left\vert c_{jm}\right\vert ^{2}$, which are very close to that described
570: by $H_{eff},$ even if $J$ is not so large in comparison with $J_{0}$.
571:
572: We address that the above tables reflect all the facts distinguishing the
573: difference between the results about the entanglement of two end qubit
574: generated by $H_{eff}$ and $H.$ Though we have ignored the off-diagonal
575: terms in the reduced density matrix, the calculation of the fidelity $%
576: F(|j,m\rangle )\equiv ._{M}\langle j,m|\rho _{AB}|j,m\rangle
577: _{M}=|c_{jm}|^{2}$ further confirms our observation, that the effective
578: Heisenberg type interaction of two end qubits can approximates the realistic
579: Hamiltonian very well. Then the quantum information can be transferred
580: between the two ends of the $2\times N$-site two-leg spin ladder, that can
581: be regarded as the channel to share entanglement with separated Alice and
582: Bob. Physically, this is just due to a large spin gap existing in such a
583: perfect medium, whose ground state can induce a maximal entanglement of the
584: two end qubits. We also pointed out that our analysis is applicable for
585: other types of medium systems as data buses, which possess a finite spin
586: gap. Since $L/\left\vert J_{eff}\right\vert $ determines the characteristic
587: time of quantum state transfer between the two qubits, the dependence of $%
588: J_{eff}$ upon $L$ becomes important and relies on the appropriate choice of
589: the medium.
590:
591: In conclusion, we have presented and studied in detail a protocol to quantum
592: state transfer. Numerical results show that the isotropic antiferromagnetic
593: spin ladder system is a perfect medium through which the interaction between
594: two separated spins is very close to the Heisenberg type coupling with a
595: coupling constant inversely proportional to the distance even if the spin
596: gap is not so large comparing to the couplings between the input and output
597: spins with the medium.
598:
599: \subsubsection{3.2 Spin chain in modulated external magnetic field}
600:
601: Let us consider the Hamiltonian of $(2N+1)$-site spin-$\frac{1}{2}$
602: ferromagnetic Heisenberg chain
603: \begin{equation}
604: H=-J\sum_{i=1}^{2N}\mathbf{S}_{i}\cdot \mathbf{S}_{i+1}+%
605: \sum_{i=1}^{2N+1}B(i)S_{i}^{z}
606: \end{equation}%
607: with the uniform coupling strength $-J<0,$ but in the parabolic magnetic
608: field
609: \begin{equation}
610: B(i)=2B_{0}(i-N-1)^{2}
611: \end{equation}%
612: where $B_{0}$ is a constant. In single-excitation invariant subspace with
613: the fixed z-component of total spin $S^{z}=N-1/2$, this model is equivalent
614: to the spin-less fermion hopping model with the Hamiltonian
615: \begin{equation}
616: H=-\frac{J}{2}\sum\limits_{i=1}^{2N}(a_{i}^{\dag }a_{i+1}+h.c)+\frac{1}{2}%
617: \sum_{i=1}^{2N+1}B(i)a_{i}^{\dag }a_{i}
618: \end{equation}%
619: where we have neglected a constant in the Hamiltonian for simplicity. For
620: the single-particle case with the set basis$\{\overset{\qquad \ \ \ \ \ \ \
621: n^{\prime }th}{\left\vert n\right\rangle =\left\vert
622: 0,0,...,1,0...\right\rangle }|n=1,2,..\}$
623: \begin{figure}[h]
624: \includegraphics[bb=75 350 525 770, width=7 cm, clip]{s_fig1.eps}
625: \caption{Schematic illustration of the time evolution of a Gaussian
626: wavepacket. It shows that the near-perfect state transfer over a long
627: distance is possible in the quasi-harmonic system.}
628: \end{figure}
629: which is just the same as that of the Hamiltonian of Josephson junction in
630: the Cooper-pair number basis Ref. \cite{Shn} for $E_{J}=J,E_{c}=2B_{0}$.
631: Analytical analysis and numerical results have shown that the lower energy
632: spectrum is indeed quasi-harmonic in the case $E_{J}\gg E_{c}$ \cite{Shi}.
633: Although the eigenstates of the Hamiltonian (36) does not satisfy the SPMC
634: precisely, especially for high energy range, there must exist some Gaussian
635: wavepacket states expanded by the lower eigenstates. Such kind of state can
636: be transferred with high fidelity.
637:
638: We consider a Gaussian wavepacket at $t=0$, $x=N_{A}$ as the initial state
639: \begin{equation}
640: \left\vert \psi (N_{A},0)\right\rangle =C\sum_{i=1}^{2N+1}e^{-\frac{1}{2}%
641: \alpha ^{2}(i-N_{A}-1)^{2}}\left\vert i\right\rangle
642: \end{equation}%
643: where $\left\vert i\right\rangle $ denotes the state with $2N$ spins in down
644: state and only the $i$th spin in up state, $C$ is the normalization factor.
645: The coefficient $\alpha ^{2}=4\ln 2/\Delta ^{2}$ is determined by the width
646: of the Gaussian wavepacket $\Delta $. The state $\left\vert \psi
647: (0)\right\rangle $ evolves to $\left\vert \psi (t)\right\rangle
648: =e^{-iHt}\left\vert \psi (N_{A},0)\right\rangle $ at time $t$ and the
649: fidelity for the state $\left\vert \psi (0)\right\rangle $ transferring to
650: the position $N_{B}$ is defined as
651: \begin{equation}
652: F(t)=\left\vert \left\langle \psi (N_{B},0)\right\vert e^{-iHt}\left\vert
653: \psi (N_{A},0)\right\rangle \right\vert .
654: \end{equation}%
655: In Fig. 5 the evolution of the state $\left\vert \psi (0)\right\rangle $ is
656: illustrated schematically. From the investigation of Ref. \cite{Shi}, we
657: know that for small $N_{A}=-N_{B}=-x_{0}$, where $N_{B}$ is the mirror
658: counterpart of $N_{A}$, but in large $\Delta $ limit, if we take $%
659: B_{0}=8\left( \ln 2/\Delta ^{2}\right) ^{2}$, $F(t)$ has the form
660: \begin{equation}
661: F(t)=\exp [-\frac{1}{2}\alpha ^{2}N_{A}^{2}(1+cos\frac{2t}{\alpha ^{2}})]
662: \end{equation}%
663: which is a periodic function of $t$ with the period $T=\alpha ^{2}\pi $ and
664: has maximum of 1. This is in agreement with our above analysis. However, in
665: quantum communication, what we concern is the behavior of $F(t)$ in the case
666: of the transfer distance $L\gg \Delta $, where $L=2\left\vert
667: N_{A}\right\vert =2\left\vert N_{B}\right\vert $. For this purpose the
668: numerical method is performed for the case $L=500,\Delta =2,4,6$ and $%
669: B_{0}=8\left( \ln 2/\Delta ^{2}\right) ^{2}\lambda $. The factor $\lambda $
670: determines the maximum fidelity and then the optimal field distribution can
671: be obtained numerically. In the Ref. \cite{songz2}, Fig. 2(a), (b) and (c)
672: the functions $F(t)$ are plotted for different values of $\lambda $. It
673: shows that for the given wavepackets with $\Delta =2,4$ and $6$, there
674: exists a range of $\lambda $, during which the fidelities $F(t)$ are up to $%
675: 0.748,0.958$ and $0.992$ respectively . For finite distance, the maximum
676: fidelity decreases as the width of Gaussian wavepacket increases. On the
677: other hand, the strength of the external field also determines the value of
678: the optimal fidelity for a given wavepacket. There exists an optimal
679: external field to obtain maximal fidelity, meanwhile the period of $F(t)$
680: close to $T=\alpha ^{2}\pi $. This shows a difference from the ideal system,
681: i.e. continuous harmonic systems, in which the fidelity is independent of
682: the strength of the external field. Numerical results indicate that it is
683: possible to realize near-perfect quantum state transfer over a longer
684: distance in a practical ferromagnetic spin chain system.
685:
686: In summary, we have shown that a perfect quantum transmission can be
687: realized through a universal quantum channel provided by a quantum spin
688: system with spectrum structure, in which each eigenenergy is commensurate
689: and matches with the corresponding parity. According to this SPMC for the a
690: mirror inversion symmetry \cite{MIS}, we can implement the perfect quantum
691: information transmission with several novel pre-engineered quantum spin
692: chains. For more practical purpose, we prove that an approximately
693: commensurate spin system can also realize near-perfect quantum state
694: transfer in a ferromagnetic Heisenberg chain with uniform coupling constant
695: in an external field. Numerical method has performed to study the fidelity
696: for the system in a parabolic magnetic field. The external field plays a
697: crucial role in the scheme. It induces a lower quasi-harmonic spectrum,
698: which can drive a Gaussian wavepacket from the initial position to its
699: mirror counterpart. The fidelity depends on the initial position (or
700: distance $L$), the width of the wavepacket $\Delta$ and the magnetic field
701: distribution $B(i)$ via the factor $\lambda $. Thus for given $L$ and $%
702: \Delta $, proper selection of the factor $\lambda $ can achieve the optimal
703: fidelity. Finally, we conclude that it is possible to implement near-perfect
704: Gaussian wavepacket transmission over a longer distance in many-body system.
705:
706: \section{V. Quantum storage based on the spin chain}
707:
708: Recently a universal quantum storage protocol \cite{lukin1,zoller,exp} was
709: presented to reversibly map the electronic spin state onto the collective
710: spin state of the surrounding nuclei ensemble in a quantum well (see the
711: Fig. 6). Because of the long decoherence time of the nuclear spins, the
712: information stored in them can be robustly preserved.
713: \begin{figure}[tbp]
714: \includegraphics[bb=80 350 500 600, width=7 cm, clip]{fq4.eps}
715: \caption{the electronic spin state onto the collective spin state of the
716: surrounding nuclei ensemble in a quantum well }
717: \end{figure}
718: When all nuclei (with spin operators $I_{x}^{(i)},I_{y}^{(i)},I_{z}^{(i)}$)
719: of spin $I_{0}$ are coupled with a single electron spin with strength $g_{i}$%
720: , a pair of collective operators\cite{szs}
721: \begin{equation}
722: B=\frac{\sum_{i=1}^{N}g_{i}I_{-}^{(i)}}{\sqrt{2I_{0}\sum g_{j}^{2}}}
723: \end{equation}%
724: and its conjugate $B^{+}$ are introduced to depict the collective
725: excitations in ensemble of nuclei with spin $I_{0}$ from its polarized
726: initial state$\left\vert G\right\rangle =\left\vert -NI_{0}\right\rangle
727: =\prod\limits_{i=1}^{N}\left\vert -I_{0}\right\rangle _{i}$which denotes the
728: saturated ferromagnetic state of nuclei ensemble. There is an intuitive
729: argument that if the $g_{i}$s have different values, while the distribution
730: is "quasi-homogeneous", $B$ and $B^{\dagger }$ can also be considered as
731: boson operators satisfying $[B,B^{+}]\rightarrow 1$ approximately.
732:
733: Song, Zhang and Sun analyzed the universal applicability of this protocol in
734: practice \cite{szs}. It was found that only under two homogeneous conditions
735: with low excitations, the many-nuclei system approximately behaves as a
736: single mode boson and its excitation that can serve as an efficient quantum
737: memory. The low excitation condition requires a ground state with all spins
738: orientated, which can be prepared by applying a magnetic field polarizing
739: all spins along a single direction. With the consideration of spontaneous
740: symmetry breaking for all spins orientated, a protocol of quantum storage
741: element was proposed to use a ferromagnetic quantum spin system, instead of
742: the free nuclear ensemble, to serve as a robust quantum memory.
743:
744: \begin{figure}[h]
745: \includegraphics[width=4cm,height=4cm]{w_fig1.eps}
746: \caption{The configuration geometry of the nuclei-electron system. The
747: nuclei are arranged in a circle within a quantum dot to form a ring array.
748: To turn on the interaction one can push a single electron towards the center
749: of the circle along the axis that perpendicular to the plane.}
750: \end{figure}
751:
752: The configuration of the quantum storage element is illustrated in Fig. 7.
753: The nuclei are arranged in a circle within a quantum dot to form a spin ring
754: array. A single electron is just localized in the center of the ring array
755: surrounded by the nuclei. The interaction of the nuclear spins is assumed to
756: exist only between the nearest neighbors while the external magnetic field $%
757: B_{0}$ threads through the spin array. Then the electron-nuclei system can
758: be modelled by a Hamiltonian $H=H_{e}+H_{n}+H_{en}$. It contains the
759: electronic spin Hamiltonian $H_{e}=g_{e}\mu _{B}B_{0}\sigma ^{z},$the
760: nuclear spin Hamiltonian
761: \begin{equation}
762: H_{n}=g_{n}\mu _{n}B_{0}\sum_{l=1}^{N}S_{l}^{z}-J\sum_{l=1}^{N}\mathbf{S}_{l}%
763: \mathbf{\cdot S}_{l+1} \label{**}
764: \end{equation}%
765: with the Zeeman split and the ferromagnetic interaction $J>0$, and the
766: interaction between the nuclear spins and the electronic spin
767: \begin{equation}
768: H_{en}=\frac{\lambda }{2N}\sigma ^{+}\sum_{l=1}^{N}S_{l}^{-}+h.c. \label{*}
769: \end{equation}%
770: Here, $g_{e}$ ($g_{n}$) is the Lande $g$ factor of electron (nuclei), and $%
771: \mu _{B}$ ($\mu _{N}$) the Bohr magneton (nuclear magneton). The Pauli
772: matrices $S_{l}^{-}$ and $\sigma ^{+}$ represent the nuclear spin of the $l$%
773: -th site and the electronic spin respectively. The denominator $N$ in Eq. (%
774: \ref{*}) originates from the envelope normalization of the localized
775: electron wave-function \cite{lukin1,zoller,exp}. The hyperfine interactions
776: between nuclei and electron are proportional to the envelope function of
777: localized electron. The electronic wave function is supposed to be
778: cylindrical symmetric, e.g., the $s$-wave component. Thus the coupling
779: coefficient $\lambda \propto |\psi \left( \mathbf{r}\right) |^{2}$ is
780: homogenous for all the $N$ nuclei in the ring array.
781:
782: To consider the low spin wave excitations, the discrete Fourier
783: transformation defines the bosonic operators
784: \begin{equation}
785: b_{k}=\frac{1}{\sqrt{N}}\sum_{l=1}^{N}e^{i\frac{2\pi kl}{N}}S^{-}_{l},
786: \end{equation}%
787: in the large $N$ limit. Then one can approximately diagonalize the
788: Hamiltonian (\ref{**}) as
789: \begin{equation*}
790: H_{T}=H_{N}+\sum_{k=1}^{N-1}\omega _{k}b_{k}^{\dag }b_{k}
791: \end{equation*}%
792: where $H_{N}$ is a Jaynes-Cummings (JC) type Hamiltonian
793: \begin{equation}
794: H_{N}=\omega _{N}b_{N}^{\dag }b_{N}+\frac{\Omega }{2}\sigma ^{z}+\lambda
795: \sqrt{\frac{s}{2N}}\left( \sigma ^{+}b_{N}+\sigma ^{-}b_{N}^{\dag }\right)
796: \end{equation}%
797: Then we obtain the dispersion relation for magnon or the spin wave
798: excitation
799: \begin{equation}
800: \omega _{k}=g_{n}\mu _{n}B_{0}+2Js-2Js\cos \frac{2\pi k}{N}.
801: \end{equation}%
802: The above results show that $H_{T}$ only contains the interaction of the $N$%
803: -th magnon with the electronic spin and the other $N-1$ magnons decouples
804: with it. Here, the frequency of the boson $\omega _{N}=g_{n}\mu _{n}B_{0}$
805: and the two level spacing $\Omega =2g^{\ast }\mu _{B}B_{0}$ can be modulated
806: by the external field $B_{0}$ simultaneously.
807:
808: The process of quantum information storage can be implemented in the
809: invariant subspace of the electronic spin and the $N$-th magnon. Now we can
810: describe the quantum storage protocol based on the above spin-boson model.
811: Suppose the initial state of the total system is prepared so that there is
812: no excitation in the $N$ nuclei at all while the electron is in an arbitrary
813: state $\rho _{e}\left( 0\right) =\sum_{n,m=\pm }\rho _{nm}\left\vert
814: n\right\rangle \left\langle m\right\vert $ \ where $\left\vert
815: +\right\rangle $ ($\left\vert -\right\rangle $) denotes the electronic spin
816: up\ (down) state. The initial state of the total system can then be written
817: as
818: \begin{equation}
819: \rho \left( 0\right) =\rho _{b}\left( 0\right) \otimes \left\vert
820: 0_{N}\right\rangle \left\langle 0_{N}\right\vert \otimes \rho _{e}\left(
821: 0\right)
822: \end{equation}%
823: in terms of $\rho _{b}\left( 0\right) =\left\vert \{0\}\right\rangle
824: _{N-1}\left\langle \{0\}\right\vert $ where $\left\vert n_{1},n_{2},\cdots
825: ,n_{N-1}\right\rangle \equiv \left\vert \left\{ n_{k}\right\} \right\rangle
826: _{N-1}$ ( $k=1$, $2$, $\cdots $, $N-1$) denotes the Fock state of the other $%
827: N-1$ magnons. If we set $B_{0}=0$, at $t=T\equiv (\pi /\lambda )\sqrt{N/2s}$%
828: , the time evolution from $\rho \left( 0\right) $ is just described as a
829: factorized state
830: \begin{equation}
831: \rho \left( T\right) =\rho _{b}\left( 0\right) \otimes w_{F}\otimes
832: \left\vert -\right\rangle \left\langle -\right\vert ,
833: \end{equation}%
834: where $w_{F}=\sum_{n,m=0,1}w_{nm}\left\vert n_{N}\right\rangle \left\langle
835: m_{N}\right\vert $ \ is the storing state of the $N$-th magnon with
836: \begin{equation}
837: w_{nm}=\rho _{nm}\exp \left[ \frac{i}{2}\left( m-n\right) \pi \right] .
838: \end{equation}%
839: Here, to simplify our expression, we have denoted $\rho _{++}\equiv \rho
840: _{00}$, $\rho _{+-}\equiv \rho _{01}$, $\rho _{-+}\equiv \rho _{10}$, $\rho
841: _{--}\equiv \rho _{11}$. The difference between $w_{F}$ and $\rho _{e}\left(
842: 0\right) $ is only an unitary transformation independent of the stored
843: initial state $\rho _{e}\left( 0\right) $.
844:
845: So far we have discussed the ideal case with homogeneous coupling between
846: the electron and the nuclei, that is, the coupling coefficients are the same
847: constant $\lambda $ for all the nuclear spins. However, the inhomogeneous
848: effect of coupling coefficients has to be taken into account if what we
849: concern is beyond the s-wave component, in which the wave function is not
850: strictly cylindrical symmetric. In this case, the quantum decoherence
851: induced by the so-called quantum leakage has been extensively investigated
852: for the atomic ensemble based quantum memory \cite{sun-you}. We now discuss
853: the similar problems for the magnon based quantum memory.
854:
855: For general case, $\lambda _{l}\propto |\psi \left( \mathbf{r}_{l}\right)
856: |^{2}$ vary with the positions of the nuclear spins where $\psi \left(
857: \mathbf{r}_{l}\right) $ is the envelope function of the electron at site $%
858: \mathbf{r}_{l}$. In this case, the Hamiltonian contains terms other than the
859: interaction between the spin and $N$-th mode boson, that is, the
860: inhomogeneity induced interaction
861: \begin{equation}
862: V=\lambda \sqrt{\frac{s}{2N}}(\sigma _{+}\sum_{k=1}^{N-1}\chi _{k}b_{k}+h.c.)
863: \end{equation}%
864: should be added in our model Hamiltonian $H_{T}$ where $\chi
865: _{k}=\sum_{l=1}^{N}\frac{\lambda _{l}}{\lambda N}\exp [i2\pi kl/N]$.
866:
867: For a Gaussian distribution of $\lambda _{l}$, e.g. $\lambda _{l}=(\lambda /%
868: \sqrt{2\pi }\sigma )\exp (-(l-1)^{2}/(2\sigma ^{2}))$with width $\sigma $
869: and $\lambda _{1}=\lambda $, the corresponding inhomogeneous coupling is
870: depicted by
871: \begin{equation}
872: \chi _{k}=\frac{1}{N}\sum_{l-1=0}^{N-1}\frac{1}{\sqrt{2\pi }\sigma }e^{\frac{%
873: -(l-1)^{2}}{2\sigma ^{2}}+i\frac{2\pi kl}{N}}
874: \end{equation}%
875: Fig. 8 shows the magnitude of $\chi _{k}$ for different Gaussian
876: distributions of $\lambda _{l}$ with different widths $\sigma $. It
877: indicates that the modes near $1$ and $N-1$ have a stronger coupling with
878: the electron. When the interaction gets more homogenous (with larger $\sigma
879: $) the coupling coefficients $\chi _{k}$ for all the modes from $1$ to $N-1$
880: become smaller. When the distribution is completely homogeneous, all the
881: couplings with the $N-1$ magnon modes vanish and then we obtain the
882: Hamiltonian $H_{T}$.
883:
884: In the following we will adopt a rather direct method to analyze the
885: decoherence problem of our protocol resulting from dissipation. If $N$ is so
886: large that the spectrum of the quantum memory is quasi-continuous, this
887: model is similar to the "standard model" of quantum dissipation for the
888: vacuum induced spontaneous emission \cite{Louisell}. The $N-1$ magnons will
889: induce the quantum dissipation of the electronic spin with a decay rate%
890: \begin{equation}
891: \gamma =2\pi \sum_{k=1}^{N}\frac{\lambda ^{2}s|\chi _{k}|^{2}}{2N}\delta
892: \left( \omega _{k}-2\lambda \sqrt{\frac{s}{2N}}\right) .
893: \end{equation}%
894: Let $\left\vert \Psi \right\rangle $ be the ideal evolution governed by the
895: expected Hamiltonian $H_{T}$ without dissipation while the realistic
896: evolution $\left\vert \Psi ^{\prime }\right\rangle $ governed by the
897: Hamiltonian with dissipation. Suppose the initial state of the electron is $%
898: (\left\vert +\right\rangle +\left\vert -\right\rangle )/\sqrt{2}$, we can
899: analytically calculate the fidelity
900: \begin{eqnarray}
901: &&F(t)=|\langle \Psi \left\vert \Psi ^{\prime }\right\rangle |=\frac{1}{2}%
902: (1+e^{-\frac{\gamma }{2}t})\times \notag \\
903: &&\sec \varphi (\cos gt\cos \left( \Delta _{1}^{\prime }t+\varphi \right)
904: +\sin gt\sin \Delta _{1}^{\prime }t),
905: \end{eqnarray}%
906: where $\varphi =\arcsin \sqrt{2N\gamma ^{2}/\lambda ^{2}s},g=\lambda \sqrt{%
907: s/2N}$ and $\Delta _{1}^{\prime }=\sqrt{g^{2}-\gamma ^{2}}$.
908:
909: Fig. 8 shows the curve of the fidelity $F(t)$ changing with time $t$. We
910: can see that the fidelity exhibits a exponential decay behavior with a
911: sinusoidal oscillation. At the instance when we have just implemented the
912: quantum storage process, the fidelity is about $1-\pi \gamma /8$. Therefore,
913: the deviation from the ideal case with homogeneous couplings is very small
914: for $\gamma /g<<1$. Since the ring-shape spin array with inhomogeneous
915: coupling is just equivalent to an arbitrary Heisenberg spin chain in the
916: large $N$ limit, the above arguments means that an arbitrary Heisenberg
917: chain can be used for quantum storage following the same strategy addressed
918: above if $\gamma /g$ is small, i.e., the inhomogeneous effect is not very
919: strong.
920: \begin{figure}[h]
921: \includegraphics[width=5cm,height=3cm]{w_fig3.eps}
922: \caption{The fidelity $F\left( t\right) $ in the large $N$ limit. The
923: vertical line indicates the instant\ $\frac{\protect\pi }{2g}$, at which the
924: quantum storage is just implemented. Here $\frac{\protect\gamma }{g}=\frac{1%
925: }{50}$. The inset shows the decaying oscillation with details of $F\left(
926: t\right) $ in a small region near the instant $\frac{\protect\pi }{2g}$. }
927: \end{figure}
928: On the other hand, if $N$ is small, the spectrum of the quantum memory is
929: discrete enough to guarantee the adiabatic elimination of the $N-1$ magnon
930: modes, i.e., $\lambda \sqrt{s/2N}\chi _{k}/|\omega _{k}|<<1$ for the $N-1$
931: magnon modes. As a consequence of this adiabatic elimination, the quantum
932: decoherence or de-phasing can result from the mixing of different magnon
933: modes.
934:
935: \textbf{Acknowledgement:} \textit{We acknowledge\ the \
936: collaborations \ with P. Zhang, Yong Li, Y.D Wang, B.Chen, \
937: X.F.Qian, T.Shi, Ying Li and R. Xin, which resulting in our
938: systematical researches on the quantum spin based quantum
939: information processing. SZ's work is supported by the Innovation
940: Foundation of Nankai university. CPS acknowledge the support of
941: the CNSF (grant No. 90203018), the Knowledge Innovation Program
942: (KIP) of the Chinese Academy of Sciences, the National Fundamental
943: Research Program of China (No. 001GB309310).}
944:
945: \begin{thebibliography}{a}
946: \bibitem[a]{email} Electronic addresses: songtc@nankai.edu.cn\newline
947: suncp@itp.ac.cn
948:
949: \bibitem[b]{www} Internet www site: http://www.itp.ac.cn/\symbol{126}suncp
950:
951: \bibitem{q-inf} D. P. DiVincenzo and C. Bennet, Nature \textbf{404}, 247
952: (2000) and references therein
953:
954: \bibitem{lukin} M. D. Lukin, Rev. Mod. Phys. \textbf{75}, 457 (2003).
955:
956: \bibitem{Flei} M. Fleischhauer and M. D. Lukin, Phys. Rev. Lett, \textbf{84}%
957: , 5094 (2000); Phys. Rev. A, \textbf{65}, 022314 (2002)
958:
959: \bibitem{sun-prl} C. P. Sun, Y. Li, and X. F. Liu, Phys. Rev. Lett. \textbf{%
960: 91}, 147903 (2003).
961:
962: \bibitem{za-store} E. Pazy, I. D'Amico, P. Zanardi, and F. Rossi, Phys. Rev.
963: B \textbf{64}, 195320 (2001)
964:
965: \bibitem{lukin1} J. M. Taylor, C. M. Marcus and M. D. Lukin, Phys. Rev.
966: Lett, \textbf{90}, 206803 (2003).
967:
968: \bibitem{zoller} A. Imamoglu, E. Knill, L. Tian and P. Zoller, Phys. Rev.
969: Lett. \textbf{91}, 017402 (2003)
970:
971: \bibitem{exp} M. Poggio et al., Phys. Rev. Lett. \textbf{91}, 207602 (2003)
972:
973: \bibitem{szs} Z. Song, P. Zhang and C. P. Sun, quant-ph/0409120,Effective
974: boson-spin model for nuclei ensemble based universal quantum memory,
975: submitted to Phys. Rev. B.
976:
977: \bibitem{wlss} Y. D. Wang, Y. Li, Z. Song and C. P. Sun, cond-mat/0409120
978: Magnon based quantum storage of electronic spin with a ring array of nuclear
979: spins, submitted to Phys. Rev. A (2004).
980:
981: \bibitem{Div} D. P. DiVincenzo, D. Bacon, J. Kempe, G. Burkard, and K. B.
982: Whaley, Nature 408, 339 (2000).
983:
984: \bibitem{Bose} S. Bose, Phys. Rev. Lett. \textbf{91}, 207901 (2003).
985:
986: \bibitem{Dagotto} E. Dagotto and T. M. Rice, Science \textbf{271}, 618
987: (1996).
988:
989: \bibitem{White} S. White, R. Noack, and D. Scalapino, Phys. Rev. Lett.
990: \textbf{73}, 886 (1994); R. Noack, S. White, and D. Scalapino, Phys. Rev.
991: Lett. \textbf{73}, 882 (19940).
992:
993: \bibitem{Matt1} M. Christandl, N. Datta, and J. Landahl, Phys. Rev. Lett.
994: \textbf{92}, 187902 (2004).
995:
996: \bibitem{Matt2} M. Christandl, N Datta, T. C. Dorlas, A. Ekert, A. Kay and
997: A. J. Landahl quant-ph/0411020.
998:
999: \bibitem{songz1} T. Shi, Ying Li, Z. Song, and C. P. Sun, cond-mat/0408152,
1000: Quantum state transfer via the ferromagnetic chain in a spatially modulated
1001: field, submitted to Phys. Rev. A (2004).
1002:
1003: \bibitem{songz2} Ying Li, T.Shi, B. Chen, Z. Song, C.P.Sun,
1004: quant-ph/0406159, Quantum state transmission via a spin ladder as a robust
1005: data bus, Phys. Rev. A (2004) in press.
1006:
1007: \bibitem{geo} C. P. Sun, P. Zhang and Y. Li,quant-ph/0311052, Geometric
1008: Quantum Information Storage Based on Atomic Ensemble ; Y. Li, P. Zhang, P.
1009: Zanardi, and C. P. Sun Phys. Rev. A 70, 032330 (2004)
1010:
1011: \bibitem{MIS} Claudio Albanese, Matthias Christandl, Nilanjana Datta, Artur
1012: Ekert, quant-ph/0405029.
1013:
1014: \bibitem{QD array} D. Loss and D. P. DiVincenzo, Phys. Rev. A \textbf{57},
1015: 120 (1998); B. E. Kane, Nature (London) \textbf{393}, 133 (1998).
1016:
1017: \bibitem{Lieb} E. Lieb, Phys. Rev. Lett. \textbf{62}, 1201 (1989); E. Lieb
1018: and D. Mattis, J. Math. Phys. \textbf{3}, 749 (1962).
1019:
1020: \bibitem{Song} Z. Song, Phys. Lett. A \textbf{231}, 135 (1997); Z. Song,
1021: Phys. Lett. A \textbf{233}, 135 (1997). (2004)
1022:
1023: \bibitem{Shn} Yu. Makhlin, G. Schon, and A. Shnirman, Rev. Mod. Phys. 73,
1024: 357 (2001)
1025:
1026: \bibitem{Shi} T. Shi, B. Chen, Z. Song and C.P. Sun, On the harmonic
1027: approximation for large Josephson junction coupling charge qubits, Comm.
1028: Theor. Phys. (2004),in press.
1029:
1030: \bibitem{sun-you} C. P. Sun, S. Yi, L. You, Phys. Rev. A \textbf{67}, 063815.
1031:
1032: \bibitem{Louisell} W.H. Louisell, "Quantum Statistical Properties of
1033: Radiation", John Wiley and Son's, New York, (1990).
1034: \end{thebibliography}
1035:
1036: \end{document}
1037: