1: \documentclass[aps,showpacs,preprint]{revtex4}
2: \usepackage{graphicx}
3: %\usepackage{endfloat}
4: %\leftskip1in
5: %\renewcommand{\baselinestretch}{1.2}
6: \begin{document}
7: \title{Speedup of quantum state transfer by three- qubit interactions:
8: Implementation by nuclear magnetic resonance
9: \footnote{
10: Jingfu Zhang: zhangjfu2000@yahoo.com, Jingfu@e3.physik.uni-dortmund.de\\
11: Xinhua Peng: xinhua@e3.physik.uni-dortmund.de\\
12: Dieter Suter: Dieter.Suter@uni-dortmund.de}}
13: \author{Jingfu Zhang $^{1,2}$, Xinhua Peng $^{1}$, and Dieter Suter $^{1}$}
14: \address{$^{1}$Fachbereich Physik, Universit$\ddot{a}$t Dortmund, 44221 Dortmund, Germany\\
15: $^{2}$Department of Physics, Tsinghua University, Beijing, 100084,
16: P R China}
17: \date{\today}
18: \begin{abstract}
19: Universal quantum information processing requires single-qubit
20: rotations and two-qubit interactions as minimal resources. A
21: possible step beyond this minimal scheme is the use of three-qubit
22: interactions. We consider such three-qubit interactions and show
23: how they can reduce the time required for a quantum state transfer
24: in an $XY$ spin chain. For the experimental implementation, we use
25: liquid-state nuclear magnetic resonance (NMR), where three-qubit
26: interactions can be implemented by sequences of radio-frequency
27: pulses.
28: \end{abstract}
29: \pacs{03.67.Lx}
30:
31: \maketitle
32: \section{Introduction}
33: Quantum computers are capable of solving some computational problems
34: efficiently for which no efficient classical algorithms are known.
35: Examples include the factorization of large numbers
36: \cite{Shor94}, searching unsorted databases \cite{Grover97}, and
37: simulating quantum systems \cite{Feynman82,Somma02}.
38: While this advantage originates from a different scaling behavior
39: compared to classical computers, rather than a higher clock speed,
40: the time required for a single gate operation remains a critical issue:
41: Reliable quantum computation becomes possible only if
42: a sufficiently large number of gate operations can be completed
43: within the decoherence time of the system.
44:
45: An important element of many quantum information processing
46: operations is the transfer of a quantum state
47: $\alpha|0\rangle+\beta|1\rangle$ from one qubit to another
48: \cite{Bose03}.
49: We will refer to this process as quantum state transfer (QST).
50: We thus discuss a system that is initially in state
51: $|\Psi\rangle_{i}=(\alpha|0\rangle+\beta|1\rangle)_{A}|\psi\rangle_{i}$,
52: where the qubit A is in state $\alpha|0\rangle+\beta|1\rangle$ and
53: the other qubits in state $|\psi\rangle_{i}$.
54: If we denote the QST
55: operation as $T$, the state transfer from A to B can be
56: represented as $T:|\Psi\rangle_{i}\rightarrow|\Psi\rangle_{f}
57: =|\psi\rangle_{f}(\alpha|0\rangle+e^{i\phi}\beta|1\rangle)_{B}$:
58: The final state corresponds to qubit B in state
59: $\alpha|0\rangle+e^{i\phi}\beta|1\rangle$ and the other qubits in
60: state $|\psi\rangle_{f}$. A quantum state transfer must thus
61: correctly transfer the amplitudes but not necessarily the phases
62: of the state \cite{Christandl}. No condition is imposed on the
63: state of the other qubits in the system.
64:
65: Currently there are three methods that can implement the QST. The first one is
66: quantum teleportation proposed by Bennett et al \cite{Bennett},
67: and has been experimentally realized in optical and liquid-state
68: nuclear magnetic resonance (NMR) systems \cite{Pan,Nielsen98}.
69: This method is based on quantum entanglement and requires quantum
70: measurements.
71: Classical communication is also needed if
72: one wants to determine the phase factor in the final state of
73: qubit B.
74: The second method is based on swap operations, where $T$ can
75: be represented as $T=\Pi S_{jl}$. $S_{jl}$ denotes a SWAP gate
76: that exchanges the states of qubits $j$ and $l$. To realize
77: $S_{jl}$, one needs external operations to control the qubits
78: other than qubits $j$ and $l$, such as switching on and off the
79: couplings between qubits $j$ or $l$ and the other qubits.
80:
81: The third method uses a static spin-network
82: \cite{Christandl,Karbach}: the qubits are linearly connected by
83: Heisenberg interactions. Qubit A is initialized into state
84: $\alpha|0\rangle+\beta|1\rangle$ and the other qubits each into
85: state $|0\rangle$. Under the influence of a suitable static
86: coupling network, the system evolves such that qubit B ends up in
87: state $\alpha|0\rangle+e^{i\phi}\beta|1\rangle$. Unlike the second
88: method, the third method does not require spin couplings to be
89: switched on and off, so that it is one kind of quantum
90: computations with the 'always on' interactions \cite{Bose} that
91: avoids single-qubit operations. Hence it is easy to implement in
92: some solid-state systems \cite{DiVincenzo}. In this article we
93: concentrate on the third method. For example, the QST can be
94: implemented in a three- spin linear chain with the $XY$-
95: interactions
96: $\sigma_{x}^{1}\sigma_{x}^{2}+\sigma_{y}^{1}\sigma_{y}^{2}
97: +\sigma_{x}^{2}\sigma_{x}^{3}+\sigma_{y}^{2}\sigma_{y}^{3}$.
98:
99: The initial state is chosen as
100: $(\alpha|0\rangle+\beta|1\rangle)_{A}|00\rangle$ by setting spin
101: $1$ at the location A into state $\alpha|0\rangle+\beta|1\rangle$
102: and the other spins into state $|00\rangle$. Waiting for a period
103: of time $t_{0}=\frac{\pi}{2\sqrt{2}}$, one obtains the state
104: $|00\rangle(\alpha|0\rangle-\beta|1\rangle)_{B}$, which means that
105: the spin $3$ at location B is now in state
106: $\alpha|0\rangle-\beta|1\rangle$.
107: Both the initial state and the
108: final state are product state. However the middle state can be an
109: entangled states.
110: The relation between quantum entanglement and the
111: QST in the spin-network has been well discussed in Ref.
112: \cite{Sun}. In the three spin chain the maximum transfer distance
113: is 2. If one transfers a state over longer distance, one needs to
114: design and generate the couplings between spins in a linear chain,
115: or expand the chain into a spin network through introducing the
116: additional spins. The details can be found in Refs.
117: \cite{Christandl}.
118:
119: Like other quantum information processing tasks, QST can be
120: effected with a minimum set of gates \cite{physica}, typically
121: consisting of single- qubit rotations and CNOT gates that can be
122: implemented through two- qubit interactions
123: \cite{Deutsch85,Somma02}. An additional possible resource are
124: three- spin interactions \cite{Baxter}. Effective three-particle
125: interactions exist in some real physical systems, for example in
126: optical lattices constructed of equilateral triangles
127: \cite{PachosPRL}. The spin Heisenberg chain with three-spin
128: interactions can exhibit interesting phase transition phenomena,
129: such as incommensurate phases \cite{Tsvelik,Frahm}, chiral phase
130: transitions \cite{Cruz}, or a quantum entanglement phase
131: transition\cite{Lou,Yang}.
132:
133: The three- spin interaction that we consider corresponds to a
134: coupling between next- nearest- neighbors controlled by the middle
135: spin \cite{PachosPRL}. It is a rare resource in some quantum
136: systems. In this article, we use the three- spin interactions in
137: the spin $XY$- chain to increase the speed of the QST, and
138: quantitatively describe the advantages obtained by using such a
139: resource.
140:
141: While nature does not provide three-spin interactions between nuclear spins,
142: they can be simulated quite readily in liquid-state NMR \cite{Tseng99}.
143: For this purpose, one combines the natural two-spin interactions of the type
144: $J_{mn}\sigma_{z}^{m}\sigma_{z}^{n}$, where $\sigma_{z}^{m}$
145: denotes the $z$- component of the Pauli matrix for spin $m$, and
146: $J_{mn}$ denotes the coupling constant between spins $m$ and $n$.
147: In this work, we use this approach
148: to generate an effective Hamiltonian with variable three-qubit
149: coupling strength to assess the speed-up of the QST due to
150: three-qubit interactions.
151:
152: \section{$XY$ spin chain with three-spin interactions}\label{sect2}
153:
154: \subsection {System and hamiltonian}
155:
156: To test the speed-up of a state-transfer operation by three-spin
157: interactions, we consider a three spin $XY$ chain, which is
158: described by the Hamiltonian
159: \begin{equation}\label{xy}
160: H_{XY3}=(\sigma_{x}^{1}\sigma_{x}^{2}+\sigma_{y}^{1}\sigma_{y}^{2}
161: +\sigma_{x}^{2}\sigma_{x}^{3}+\sigma_{y}^{2}\sigma_{y}^{3})+
162: \frac{\lambda}{2}(\sigma_{x}^{1}\sigma_{z}^{2}\sigma_{y}^{3}-
163: \sigma_{y}^{1}\sigma_{z}^{2}\sigma_{x}^{3}) .
164: \end{equation}
165: Here, $\sigma_{x/y/z}^{j}(j=1,2,3)$ are the Pauli matrices
166: and we have set $\hbar$ and the coupling constant
167: for the two-spin terms to one.
168: To find an analytical expression for the time evolution of this system
169: and determine the conditions for state transfer, we write the Hamiltonian (1)
170: as a sum of two commuting parts, $H_{XY3}= C + D$, where
171: \begin{equation}
172: C=\sigma_{x}^{1}\sigma_{x}^{2}+\sigma_{y}^{2}\sigma_{y}^{3}+
173: \frac{\lambda}{2}\sigma_{x}^{1}\sigma_{z}^{2}\sigma_{y}^{3},
174: \hspace{1.0 cm}
175: D=\sigma_{y}^{1}\sigma_{y}^{2}+\sigma_{x}^{2}\sigma_{x}^{3}-
176: \frac{\lambda}{2}\sigma_{y}^{1}\sigma_{z}^{2}\sigma_{x}^{3}.
177: \end{equation}
178:
179: \subsection {Propagator and transfer speed}\label{2b}
180:
181: This decomposition shows directly that this Hamiltonian generates a periodic time-evolution:
182: defining $k=\sqrt{2+\frac{\lambda^{2}}{4}}$, we find $C^2 = D^2 = k^2 I$
183: and therefore
184: \begin{equation}\label{Uxy}
185: U(t)=e^{-iH_{XY3} t}
186: = e^{-i t C } e^{-i t D }
187: = [\cos(kt)I-i\frac{\sin(kt)}{k}C] [\cos(kt)I-i\frac{\sin(kt)}{k}D] .
188: \end{equation}
189: For times $t = n\tau = n\pi/k $ ($n$ integer), the propagator
190: returns to unity, $U(\tau) = I$.
191:
192: The matrix representation of the propagator is
193: \begin{equation}\label{Um}
194: U(t)=\left (
195: \begin{array}{cccccccc}
196: 1& 0 & 0 & 0 & 0 & 0 & 0 & 0 \\
197: 0 & \frac{(2kc)^{2}-(\lambda s)^{2}}{4k^{2}} & -i\frac{2kcs+\lambda s^{2}}{k^{2}} & 0 &
198: \frac{k\lambda cs-2s^{2}}{k^{2}} & 0 & 0 & 0 \\
199: 0 & -i\frac{2kcs-\lambda s^{2}}{k^{2}} & \frac{k^{2}-4s^{2}}{k^{2}} & 0
200: & -i\frac{2kcs+\lambda s^{2}}{k^{2}} & 0 & 0 & 0 \\
201: 0 & 0 & 0 & \frac{(2kc)^{2}-(\lambda s)^{2}}{4k^{2}} & 0 & -i\frac{2kcs-\lambda s^{2}}{k^{2}}
202: & -\frac{2s^{2}+k\lambda cs}{k^{2}} & 0 \\
203: 0 & -\frac{2s^{2}+k\lambda cs}{k^{2}} & -i\frac{2kcs-\lambda s^{2}}{k^{2}} & 0
204: & \frac{(2kc)^{2}-(\lambda s)^{2}}{4k^{2}} & 0 & 0 & 0 \\
205: 0 & 0 & 0 & -i\frac{2kcs+\lambda s^{2}}{k^{2}}
206: & 0 & \frac{k^{2}-4s^{2}}{k^{2}} & -i\frac{2kcs-\lambda s^{2}}{k^{2}} & 0 \\
207: 0 & 0 & 0 & \frac{k\lambda cs-2s^{2}}{k^{2}}& 0 & -i\frac{2kcs+\lambda s^{2}}{k^{2}}
208: & \frac{(2kc)^{2}-(\lambda s)^{2}}{4k^{2}} & 0 \\
209: 0 & 0 & 0 & 0 & 0 & 0 & 0 & 1 \\
210: \end{array}\right ),
211: \end{equation}
212: where $c\equiv\cos(kt)$, and $s\equiv\sin(kt)$.
213:
214: The propagator generates a state transfer at times $t_{QST} =
215: \arcsin\sqrt{\frac{8+\lambda^{2}}{8+2\lambda^{2}}}/k$: If $\lambda
216: \geq 0$, it effects a transfer from qubit 3 to qubit 1, for
217: negative 3-qubit coupling constant in the opposite direction. In
218: both cases, the periodicity of the overall evolution implies that
219: the reverse transfer occurs at time $t = \pi/k - t_{QST}$. The
220: corresponding propagators are
221: \begin{equation}\label{UQST13}
222: U(t_{QST_{1\rightarrow 3}})=\left (
223: \begin{array}{cccccccc}
224: 1& 0 & 0 & 0 & 0 & 0 & 0 & 0 \\
225: 0 & 0 & 0 & 0 & -1 & 0 & 0 & 0 \\
226: 0 & i\frac{4\lambda}{\lambda^{2}+4}
227: & \frac{\lambda^{2}-4}{\lambda^{2}+4} & 0 & 0 & 0 & 0 & 0 \\
228: 0 & 0 & 0 & 0 & 0 & i\frac{4\lambda}{\lambda^{2}+4} & \frac{\lambda^{2}-4}{\lambda^{2}+4} & 0 \\
229: 0 & \frac{\lambda^{2}-4}{\lambda^{2}+4} & i\frac{4\lambda}{\lambda^{2}+4} & 0 & 0 & 0 & 0 & 0 \\
230: 0 & 0 & 0 & 0 & 0 & \frac{\lambda^{2}-4}{\lambda^{2}+4} & i\frac{4\lambda}{\lambda^{2}+4} & 0 \\
231: 0 & 0 & 0 & -1 & 0 & 0 & 0 & 0 \\
232: 0 & 0 & 0 & 0 & 0 & 0 & 0 & 1 \\
233: \end{array}\right ).
234: \end{equation}
235: for the transfer $1 \rightarrow 3$ and
236: \begin{equation}\label{UQST31}
237: U(t_{QST_{3\rightarrow 1}})=\left (
238: \begin{array}{cccccccc}
239: 1& 0 & 0 & 0 & 0 & 0 & 0 & 0 \\
240: 0 & 0 & -i\frac{4\lambda}{\lambda^{2}+4} & 0 & \frac{\lambda^{2}-4}{\lambda^{2}+4} & 0 & 0 & 0 \\
241: 0 & 0 & \frac{\lambda^{2}-4}{\lambda^{2}+4} & 0 & -i\frac{4\lambda}{\lambda^{2}+4} & 0 & 0 & 0 \\
242: 0 & 0 & 0 & 0 & 0 & 0 & -1 & 0 \\
243: 0 & -1 & 0 & 0 & 0 & 0 & 0 & 0 \\
244: 0 & 0 & 0 & -i\frac{4\lambda}{\lambda^{2}+4} & 0 & \frac{\lambda^{2}-4}{\lambda^{2}+4} & 0 & 0 \\
245: 0 & 0 & 0 & \frac{\lambda^{2}-4}{\lambda^{2}+4} & 0 & -i\frac{4\lambda}{\lambda^{2}+4} & 0 & 0 \\
246: 0 & 0 & 0 & 0 & 0 & 0 & 0 & 1 \\
247: \end{array}\right )
248: \end{equation}
249: for the transfer $3\rightarrow 1$.
250:
251: Figure \ref{tQST13} shows the dependence of the QST time on the
252: strength $\lambda$ of the three-qubit interaction. The overall
253: cycle time $\tau$ decreases monotonically when a three-body
254: coupling is added to the Hamiltonian. However, for $\lambda \neq
255: 0$, the state transfer is no longer a simple SWAP operation, which
256: exchanges the states of qubits 1 and 3, but the transfer becomes
257: asymmetric, requiring different durations for the two directions.
258: While the overall cycle time $1 \rightarrow 3 \rightarrow 1$
259: decreases monotonically with increasing $|\lambda|$, the slower of
260: the two state transfers only gets faster than for $\lambda = 0$
261: when $|\lambda| > 2.71199$.
262:
263: \subsection{State transfer}\label{2c}
264:
265: To demonstrate the state transfer, we set qubit 1 into a superposition state $\alpha|0\rangle+\beta|1\rangle$,
266: with the other two qubits in state $|00\rangle$.
267: Applying the forward state transfer Eq. (5) to this state gives
268: \begin{equation}
269: U(t_{QST_{1\rightarrow3}})(\alpha|0\rangle+\beta|1\rangle)|00\rangle
270: =|00\rangle(\alpha|0\rangle-\beta|1\rangle)
271: \end{equation}
272: and similar for the reverse transfer.
273:
274: If we write this transfer in density operator notation, it reads
275: \begin{equation}\label{px1}
276: \left
277: (\begin{array}{cc}
278: |\alpha|^{2} & \alpha\beta^{*} \\
279: \alpha^{*}\beta & |\beta|^{2} \\
280: \end{array}\right)\bigotimes \left(
281: \begin{array}{cccc}
282: 1 & 0 & 0 & 0 \\
283: 0 &0 & 0 & 0 \\
284: 0 & 0 & 0 & 0 \\
285: 0 & 0 & 0 & 0 \\
286: \end{array}
287: \right)
288: \rightarrow
289: \left(
290: \begin{array}{cccc}
291: 1 & 0 & 0 & 0 \\
292: 0 &0 & 0 & 0 \\
293: 0 & 0 & 0 & 0 \\
294: 0 & 0 & 0 & 0 \\
295: \end{array}
296: \right)
297: \bigotimes \left (
298: \begin{array}{cc}
299: |\alpha|^{2} & -\alpha\beta^{*} \\
300: -\alpha^{*}\beta & |\beta|^{2} \\
301: \end{array}
302: \right) .
303: \end{equation}
304: This result differs when the second and third qubits are initially
305: in different states:
306: %-------------------------------------------------
307:
308: \begin{eqnarray}\label{px2}
309: \left
310: (\begin{array}{cc}
311: |\alpha|^{2} & \alpha\beta^{*} \\
312: \alpha^{*}\beta & |\beta|^{2} \\
313: \end{array}\right)\bigotimes\left(\begin{array}{cccc}
314: 0 & 0 & 0 & 0 \\
315: 0 &1 & 0 & 0 \\
316: 0 & 0 & 0 & 0 \\
317: 0 & 0 & 0 & 0 \\
318: \end{array}
319: \right )\rightarrow\left(\begin{array}{cccc}
320: 0 & 0 & 0 & 0 \\
321: 0 &\frac{16\lambda^{2}}{(\lambda^{2}+4)^{2}} & i\frac{4\lambda(\lambda^{2}-4)}{(\lambda^{2}+4)^{2}} & 0 \\
322: 0 & -i\frac{4\lambda(\lambda^{2}-4)}{(\lambda^{2}+4)^{2}} & \frac{(\lambda^{2}-4)^{2}}{(\lambda^{2}+4)^{2}} & 0 \\
323: 0 & 0 & 0 & 0 \\
324: \end{array}
325: \right )\bigotimes\left (\begin{array}{cc}
326: |\alpha|^{2} & \alpha\beta^{*} \\
327: \alpha^{*}\beta & |\beta|^{2} \\
328: \end{array}\right)
329: \end{eqnarray}
330: %--------------------------------------------------
331: \begin{eqnarray}\label{px3}
332: \left
333: (\begin{array}{cc}
334: |\alpha|^{2} & \alpha\beta^{*} \\
335: \alpha^{*}\beta & |\beta|^{2} \\
336: \end{array}\right)\bigotimes\left(\begin{array}{cccc}
337: 0 & 0 & 0 & 0 \\
338: 0 &0 & 0 & 0 \\
339: 0 & 0 & 1 & 0 \\
340: 0 & 0 & 0 & 0 \\
341: \end{array}
342: \right )\rightarrow\left(\begin{array}{cccc}
343: 0 & 0 & 0 & 0 \\
344: 0 &\frac{(\lambda^{2}-4)^{2}}{(\lambda^{2}+4)^{2}}& -i\frac{4\lambda(\lambda^{2}-4)}{(\lambda^{2}+4)^{2}} & 0 \\
345: 0 & i\frac{4\lambda(\lambda^{2}-4)}{(\lambda^{2}+4)^{2}} & \frac{16\lambda^{2}}{(\lambda^{2}+4)^{2}} & 0 \\
346: 0 & 0 & 0 & 0 \\
347: \end{array}
348: \right )\bigotimes\left (\begin{array}{cc}
349: |\alpha|^{2} & \alpha\beta^{*} \\
350: \alpha^{*}\beta & |\beta|^{2} \\
351: \end{array}\right)
352: \end{eqnarray}
353: %-------------------------------------------------
354: \begin{equation}\label{px4}
355: \left
356: (\begin{array}{cc}
357: |\alpha|^{2} & \alpha\beta^{*} \\
358: \alpha^{*}\beta & |\beta|^{2} \\
359: \end{array}\right)\bigotimes\left(\begin{array}{cccc}
360: 0 & 0 & 0 & 0 \\
361: 0 &0 & 0 & 0 \\
362: 0 & 0 & 0 & 0 \\
363: 0 & 0 & 0 & 1 \\
364: \end{array}
365: \right )\rightarrow\left(\begin{array}{cccc}
366: 0 & 0 & 0 & 0 \\
367: 0 &0 & 0 & 0 \\
368: 0 & 0 & 0 & 0 \\
369: 0 & 0 & 0 & 1 \\
370: \end{array}
371: \right)\bigotimes\left
372: (\begin{array}{cc}
373: |\alpha|^{2} & -\alpha\beta^{*} \\
374: -\alpha^{*}\beta & |\beta|^{2} \\
375: \end{array}\right).
376: \end{equation}
377: The phase of the superposition in the transferred state contains thus
378: information on the state of the other qubits.
379:
380: %--------------------------------------------------
381:
382: \subsection{Mixed states and parallel implementation}\label{2d}
383:
384: These different cases can be implemented in parallel by using a
385: mixed initial state \cite{Datta05}. By choosing
386: $\alpha=\beta=1/\sqrt2$ and adding the four initial states in Eqs.
387: (\ref{px1}-\ref{px4}), we obtain
388: $$ \frac{1}{2}\left
389: (\begin{array}{cc}
390: 1 & 1 \\
391: 1 & 1 \\
392: \end{array}\right)\bigotimes \left(
393: \begin{array}{cccc}
394: 1 & 0 & 0 & 0 \\
395: 0 &1 & 0 & 0 \\
396: 0 & 0 & 1 & 0 \\
397: 0 & 0 & 0 & 1 \\
398: \end{array}
399: \right)
400: = \frac{1}{2}(\sigma_{x}^{1}+I^{1}) \otimes I^2 \otimes I^3 .
401: $$
402: In the following, we will ignore the unit operator on qubits that
403: are in a superposition state.
404: The state transfer acting on this initial state generates then
405: \begin{equation}\label{mix13x}
406: U(t_{QST_{1\rightarrow 3}})\sigma_{x}^{1}I^{2}I^{3}
407: U(t_{QST_{1\rightarrow 3}})^{\ddag}=-\sigma_{z}^{1}
408: \sigma_{z}^{2}\sigma_{x}^{3}
409: \end{equation}
410: For related initial conditions, we find
411: \begin{equation}\label{mix13y}
412: U(t_{QST_{1\rightarrow 3}})\sigma_{y}^{1}I^{2}I^{3}U(t_{QST_{1\rightarrow
413: 3}})^{\ddag}=-\sigma_{z}^{1}\sigma_{z}^{2}\sigma_{y}^{3}
414: \end{equation}
415: \begin{equation}\label{mix13z}
416: U(t_{QST_{1\rightarrow 3}})\sigma_{z}^{1}I^{2}I^{3}U(t_{QST_{1\rightarrow
417: 3}})^{\ddag}=I^1I^2\sigma_{z}^{3}.
418: \end{equation}
419: Obviously the different phases that we found in the state transfer
420: for the pure initial states result in the introduction of
421: correlations when a mixed initial state is used. Only if the
422: initial state is not a superposition state (Eq. \ref{mix13z}), do
423: we find a state transfer that does not entangle the transferred
424: state with the other states.
425:
426: \section{Implementation in an NMR quantum computer}\label{real}
427:
428: The nuclear spin system that we use to implement the stepped-up
429: QST has the natural Hamiltonian
430: \begin{equation}\label{nmr}
431: H=-\pi\sum_{i=1}^3 \nu_{i}\sigma_{z}^{i}
432: +\frac{\pi}{2} J_{12}\sigma_{z}^{1}\sigma_{z}^{2}
433: +\frac{\pi}{2} J_{23}\sigma_{z}^{2}\sigma_{z}^{3}
434: \end{equation}
435: where $\nu_{i}$ denotes the resonance frequency of spin $i$.
436: %\begin{equation}
437: % H_n= \sum_{i=1}^3 \omega_i I^i_z + \sum_{i<j} J_{ij} I^i_z I^j_z .
438: %\end{equation}
439: Considering this system as a quantum simulator of the Heisenberg spin chain
440: described by the Hamiltonian (1), we generate an effective evolution (3)
441: by an appropriate sequence of radio frequency pulses.
442: While it is relatively straightforward to generate each of the terms of the Hamiltonian (1),
443: they do not commute with each other.
444: A sequential generation of the different terms therefore does not produce the correct
445: overall evolution.
446: Two different approaches allow one to generate such an evolution:
447: \begin{itemize}
448: \item Each term is implemented for a very short duration. In this
449: limit, the corresponding propagators are close to the unit
450: operator and the noncommuting terms appear only in second order
451: \cite{Vandersypen}.
452: \item The evolutions $U_{C}(t)=e^{-itC}$ and
453: $U_{D}(t)=e^{-itD}$ are written as a product such that each factor can be
454: implemented directly.
455: \end{itemize}
456: For the purpose of this paper, we have chosen the second approach.
457:
458: \subsection{Decomposing $U(t)$}
459:
460: A suitable decomposition of $U_{C}(t)$ uses the three operators
461: $L_{x}^{C}\equiv
462: \sigma_{x}^{1}\sigma_{x}^{2}/2$,
463: $L_{y}^{C}\equiv \sigma_{y}^{2}\sigma_{y}^{3}/2$, and
464: $L_{z}^{C}\equiv \sigma_{x}^{1}\sigma_{z}^{2}\sigma_{y}^{3}/2$.
465: These operators can be viewed as the three components of an
466: angular momentum vector $\bf{L^{C}}$, because they
467: satisfy the cyclic commutation relations
468: $[L_{x}^{C},L_{y}^{C}]=iL_{z}^{C}$ and cycl. \cite{Zhang05}.
469: In terms of these operators, $U_{C}(t)$ becomes
470: \begin{equation}\label{UAr}
471: U_{C}(t)=e^{-i(\frac{2\sqrt{2}}{\sin \theta_{c}}t)(\bf{L^{C}\cdot n_{c})}}
472: =e^{-i(\frac{2\sqrt{2}}{\sin \theta_{c}}t)(\frac{\sin \theta_{c}}{\sqrt{2}}L_{x}^{C}
473: +\frac{\sin \theta_{c}}{\sqrt{2}}L_{y}^{C}+\cos\theta_{c}L_{z}^{C})}
474: \end{equation}
475: where $\tan\theta_{c}=\frac{2\sqrt{2}}{\lambda}$, and the vector
476: ${\bf n_{c}}=(\frac{\sin\theta_{c}}{\sqrt{2}},\frac{\sin\theta_{c}}{\sqrt{2}},\cos\theta_{c})$
477: gives the direction of the rotation axis for $U_{C}(t)$, as
478: shown in Figure \ref{rotation}.
479: Using angular momentum theory, we rewrite this as
480: \begin{eqnarray}\label{UCf}
481: U_{C}(t)&=&e^{-i \frac{\pi}{4}L_{z}^{C}}
482: e^{i(\frac{\pi}{2}-\theta_{c})L_{y}^{C}}
483: e^{-i(\frac{2\sqrt{2}}{\sin \theta_{c}}t)L_{x}^{C}}
484: e^{-i(\frac{\pi}{2}-\theta_{c})L_{y}^{C}}
485: e^{i\frac{\pi}{4}L_{z}^{C}}.
486: \end{eqnarray}
487:
488: In a completely analogous way, we define
489: $L_{x}^{D}\equiv
490: \sigma_{x}^{2}\sigma_{x}^{3}/2$,
491: $L_{y}^{D}\equiv \sigma_{y}^{1}\sigma_{y}^{2}/2$, and
492: $L_{z}^{D}\equiv \sigma_{y}^{1}\sigma_{z}^{2}\sigma_{x}^{3}/2$ as
493: the three components of the angular momentum vector $\bf{L^{D}}$.
494: In terms of these operators, $U_{D}$ becomes
495: \begin{eqnarray}\label{UDf}
496: U_{D}(t)=e^{-itD}=e^{-i \frac{\pi}{4}L_{z}^{D}}
497: e^{-i(\theta_{d}-\frac{\pi}{2})L_{y}^{D}}
498: e^{-i(\frac{2\sqrt{2}}{\sin \theta_{d}}t)L_{x}^{D}}
499: e^{i(\theta_{d}-\frac{\pi}{2})L_{y}^{D}}
500: e^{i\frac{\pi}{4}L_{z}^{D}},
501: \end{eqnarray}
502: where $\theta_{d}=\pi-\theta_{c}$.
503:
504: While the two-spin terms $L_x$ and $L_y$ in Eqs. (\ref{UCf}) and
505: (\ref{UDf}) are relatively easy to implement, the three-spin terms
506: $L_z$ are less straightforward. We re-write them as
507: \begin{equation}\label{xzy67}
508: e^{i\eta L_{z}}=e^{i\frac{\pi}{2}L_{y}}e^{i\eta L_{x}} e^{-i\frac{\pi}{2}L_{y}} ,
509: \label{e.xyz}
510: \end{equation}
511: where $\eta$ is an arbitrary real number.
512: Alternatively, we may transform the propagators
513: $e^{i\eta L_{z}^{C}}$ and $e^{i\eta L_{z}^{D}}$ as
514: \begin{equation}\label{xzy}
515: e^{i\eta L_{z}^{C}}
516: = e^{i\frac{\eta}{2}\sigma_{x}^{1}\sigma_{z}^{2}\sigma_{y}^{3}}
517: =e^{\mp i\frac{\pi}{4}\sigma_{y}^{1}}e^{\pm i\frac{\pi}{4}\sigma_{x}^{3}}
518: e^{i\frac{\eta}{2}\sigma_{z}^{1}\sigma_{z}^{2}\sigma_{z}^{3}}
519: e^{\pm i\frac{\pi}{4}\sigma_{y}^{1}}e^{\mp i\frac{\pi}{4}\sigma_{x}^{3}}
520: \end{equation}
521: and
522: \begin{equation}\label{yzx}
523: e^{i\eta L_{z}^{D}}
524: = e^{i\frac{\eta}{2}\sigma_{y}^{1}\sigma_{z}^{2}\sigma_{x}^{3}}
525: =e^{\pm i\frac{\pi}{4}\sigma_{x}^{1}}e^{\mp i\frac{\pi}{4}\sigma_{y}^{3}}
526: e^{i\frac{\eta}{2}\sigma_{z}^{1}\sigma_{z}^{2}\sigma_{z}^{3}}
527: e^{\mp i\frac{\pi}{4}\sigma_{x}^{1}}e^{\pm i\frac{\pi}{4}\sigma_{y}^{3}} .
528: \end{equation}
529: and use the decomposition of
530: $\sigma_{z}^{1}\sigma_{z}^{2}\sigma_{z}^{3}$ into one- and
531: two-qubit operators \cite{Tseng99}
532: \begin{equation}\label{zzz67}
533: e^{i\frac{\eta}{2}\sigma_{z}^{1}\sigma_{z}^{2}\sigma_{z}^{3}}
534: =e^{i\frac{\pi}{4}\sigma_{x}^{2}}e^{-i\frac{\pi}{4}\sigma_{z}^{1}\sigma_{z}^{2}}
535: e^{i\frac{\pi}{4}\sigma_{y}^{2}}e^{i\frac{\eta}{2}\sigma_{z}^{2}\sigma_{z}^{3}}
536: e^{i\frac{\pi}{4}\sigma_{y}^{2}}e^{-i\frac{\pi}{4}\sigma_{z}^{1}\sigma_{z}^{2}}
537: e^{-i\frac{\pi}{2}\sigma_{y}^{2}}e^{-i\frac{\pi}{4}\sigma_{x}^{2}} .
538: \end{equation}
539: The expressions (\ref{xzy}-\ref{zzz67}) are identical to the
540: explicit forms of (\ref{e.xyz}). They use only single-qubit
541: operations $e^{i\phi\sigma_\alpha}$ and precessions under pairwise
542: couplings, $e^{i\xi\sigma_z^i\sigma_z^k}$, which are easy to
543: implement experimentally.
544:
545: Without loss of generality, we discuss here only the case $\lambda
546: \geq0$. After some simplifications \cite{Zhang05,Somma02,DUPRA03},
547: Eqs. (\ref{UCf}) and (\ref{UDf}) can be represented as
548: \begin{eqnarray}\label{UCfTs}
549: U_{C}(t)&=&e^{\mp i\frac{\pi}{4}\sigma_{y}^{1}}
550: e^{\pm i\frac{\pi}{4}\sigma_{x}^{3}}
551: e^{-i \frac{\pi}{8}\sigma_{z}^{1}\sigma_{z}^{2}\sigma_{z}^{3}}\nonumber\\
552: &\times&e^{\mp i\frac{\pi}{4}\sigma_{x}^{2}}
553: e^{-i\frac{1}{2}[\frac{\pi}{2}-\arctan(\frac{2\sqrt{2}}{\lambda})]\sigma_{z}^{2}\sigma_{z}^{3}}
554: e^{\pm i\frac{\pi}{4}\sigma_{x}^{2}}\nonumber\\
555: &\times& e^{\mp i\frac{\pi}{4}\sigma_{y}^{2}}
556: e^{-it\sqrt{2+\frac{\lambda^{2}}{4}}\sigma_{z}^{1}\sigma_{z}^{2}}
557: e^{\pm i\frac{\pi}{4}\sigma_{y}^{2}}\nonumber\\
558: &\times&e^{\pm i\frac{\pi}{4}\sigma_{x}^{2}}
559: e^{-i\frac{1}{2}[\frac{\pi}{2}-\arctan(\frac{2\sqrt{2}}{\lambda})]\sigma_{z}^{2}\sigma_{z}^{3}}
560: e^{\mp i\frac{\pi}{4}\sigma_{x}^{2}}\nonumber\\
561: &\times&
562: e^{i\frac{\pi}{8}\sigma_{z}^{1}\sigma_{z}^{2}\sigma_{z}^{3}}
563: e^{\pm i\frac{\pi}{4}\sigma_{y}^{1}} e^{\mp
564: i\frac{\pi}{4}\sigma_{x}^{3}},
565: \end{eqnarray}
566: and
567: \begin{eqnarray}\label{UDfTs}
568: U_{D}(t)&=&e^{\mp i\frac{\pi}{4}\sigma_{x}^{1}}
569: e^{\pm i\frac{\pi}{4}\sigma_{y}^{3}}
570: e^{-i \frac{\pi}{8}\sigma_{z}^{1}\sigma_{z}^{2}\sigma_{z}^{3}}\nonumber\\
571: &\times&e^{\mp i\frac{\pi}{4}\sigma_{x}^{2}}
572: e^{-i\frac{1}{2}[\frac{\pi}{2}-\arctan(\frac{2\sqrt{2}}{\lambda})]\sigma_{z}^{1}\sigma_{z}^{2}}
573: e^{\pm i\frac{\pi}{4}\sigma_{x}^{2}}\nonumber\\
574: &\times& e^{\pm i\frac{\pi}{4}\sigma_{y}^{2}}
575: e^{-it\sqrt{2+\frac{\lambda^{2}}{4}}\sigma_{z}^{2}\sigma_{z}^{3}}
576: e^{\mp i\frac{\pi}{4}\sigma_{y}^{2}}\nonumber\\
577: &\times&e^{\pm i\frac{\pi}{4}\sigma_{x}^{2}}
578: e^{-i\frac{1}{2}[\frac{\pi}{2}-\arctan(\frac{2\sqrt{2}}{\lambda})]\sigma_{z}^{1}\sigma_{z}^{2}}
579: e^{\mp i\frac{\pi}{4}\sigma_{x}^{2}}\nonumber\\
580: &\times&
581: e^{i\frac{\pi}{8}\sigma_{z}^{1}\sigma_{z}^{2}\sigma_{z}^{3}}
582: e^{\pm i\frac{\pi}{4}\sigma_{x}^{1}} e^{\mp
583: i\frac{\pi}{4}\sigma_{y}^{3}},
584: \end{eqnarray}
585: respectively, and the three-spin terms are implemented according
586: to Eq. (\ref{zzz67}).
587:
588: \subsection{System and pulse sequence}
589:
590: For the experimental implementation, we used a sample of Carbon-13
591: labelled trichloroethylene (TCE), dissolved in d-chloroform. Data
592: were taken with a Bruker DRX 500 MHz spectrometer. We denote the
593: $^{1}$H nuclear spin as qubit 2 (H2), the $^{13}$C directly
594: connected to $^{1}$H is denoted as qubit 1 (C1), and the other
595: $^{13}$C as qubit 3 (C3). The parameters of the system and the NMR
596: spectra are shown in Figures \ref{sample} and \ref{fig4}. The
597: difference of frequency between C1 and C3 is
598: $\Delta\nu_{13}=905.3$Hz. The coupling constants are
599: $J_{13}=103.1$Hz, $J_{12}=200.9$Hz, and $J_{23}=9.16$Hz. Because
600: of the strongly coupled carbons \cite{Miquel} we describe the
601: Hamiltonian of the three-qubit system as
602: \begin{equation}\label{HTCE}
603: H=-\pi\sum_{i=1}^3 \nu_{i}\sigma_{z}^{i}
604: +\frac{\pi}{2} J_{12}\sigma_{z}^{1}\sigma_{z}^{2}
605: +\frac{\pi}{2} J_{23}\sigma_{z}^{2}\sigma_{z}^{3}
606: +\frac{\pi}{2} J_{13}(\sigma_{x}^{1}\sigma_{x}^{3}
607: +\sigma_{y}^{1}\sigma_{y}^{3}
608: +\sigma_{z}^{1}\sigma_{z}^{3}) .
609: \end{equation}
610: Since we use this system to simulate a linear chain with nearest neighbor
611: and three-body interactions, we do not use the coupling between qubits
612: 1 and 3, which represent the end of the chain.
613:
614: Because our quantum register contains only one proton spin, we can
615: implement the rotations $e^{\pm i\frac{\pi}{4}\sigma_{x/y}^{2}}$
616: by hard $\pi/2$ proton pulses, which are selective for qubit H2.
617: We denote rotations along the $\pm x$ or $\pm y$ axis as
618: $[\pm\frac{\pi}{2}]_{x/y}^{2}$. The widths of such pulses are so
619: short that they can be considered as ideal rotations. Figures
620: \ref{seq} show the actual pulse sequences that we used to
621: implement $U_{C}$ and $U_{D}$.
622:
623: Implementing spin-selective operations on the carbon spins turned
624: out to be difficult. We minimized experimental errors by replacing
625: selective pulses with non-selective pulses and free precession
626: periods \cite{Ryan}, using, e.g.,
627: \begin{equation}\label{rx1or3}
628: e^{\pm i\frac{\pi}{4}\sigma_{x}^{m}}=e^{\mp i\frac{\pi}{4}\sigma_{y}^{1,3}}
629: e^{i\frac{\pi}{4}\sigma_{z}^{m}}e^{\pm i\frac{\pi}{4}\sigma_{y}^{1,3}},
630: \end{equation}
631: and
632: \begin{equation}\label{ry1or3}
633: e^{\pm i\frac{\pi}{4}\sigma_{y}^{m}}=e^{\pm i\frac{\pi}{4}\sigma_{x}^{1,3}}
634: e^{i\frac{\pi}{4}\sigma_{z}^{m}}e^{\mp i\frac{\pi}{4}\sigma_{x}^{1,3}},
635: \end{equation}
636: with $m=1$ or $3$.
637: The $\pi/2$ rotations $e^{\pm i\frac{\pi}{4}\sigma_{x/y}^{1,3}}$
638: act on both carbon spins C1 and C3 and were realized by hard $\pi/2$ pulses.
639: The z-rotations $e^{i\frac{\pi}{4}\sigma_{z}^{m}}$ of individual qubits were
640: implemented by the "chemical shift rotation" method of Linden et al. \cite{Linden}.
641:
642: The $\pm$ signs in Eqs. (\ref{UCfTs}) and (\ref{UDfTs}) refer to
643: two formally different expressions that represent the same overall
644: transformation. Implementing both forms and summing over the
645: result turned out to be very useful for suppressing experimental
646: artifacts arising from nonideal gate operations. When the
647: operations $U_C$ and $U_D$ are concatenated, it is possible to
648: combine the last operation of $U_C$ with the first of $U_D$ and
649: realize them as a hard pulse $[-\frac{\pi}{2}]_{x}^{1,3}$.
650:
651: \subsection{Experimental transfer of $(|0\rangle+|1\rangle)/\sqrt{2}$}
652:
653: As discussed in section \ref{2b} and shown in Figure \ref{tQST13},
654: the transfer from qubit 3 to 1 is always speeded up by the
655: three-spin interaction for $\lambda >0$. We therefore start with
656: this transfer, initializing the system to the state
657: $(|000\rangle+|001\rangle)/\sqrt{2}$.
658: To calculate its time evolution, we note that,
659: according to Eq. (\ref{Um}), the state $|000\rangle$
660: is an eigenstate of the Hamiltonian, $U(t)|000\rangle=|000\rangle$.
661: Also from Eq. (\ref{Um}), we find
662: \begin{equation}\label{purex}
663: U(t)|001\rangle=
664: \frac{(2kc)^{2}-(\lambda s)^{2}}{4k^{2}}|001\rangle
665: -i\frac{2kcs-\lambda s^{2}}{k^{2}}|010\rangle
666: -\frac{2s^{2}+k\lambda cs}{k^{2}}
667: |100\rangle.
668: \end{equation}
669: We monitor the progress of the state transfer by the amplitudes of
670: the states $|001\rangle$ and $|100\rangle$: In a superposition
671: with state $|000\rangle$, they correspond to $x$- magnetization of
672: the qubits C$3$ and C$1$, respectively.
673:
674: As discussed in sections \ref{2c} and \ref{2d}, we can observe the
675: transfer from the 4 initial states
676: $|00\rangle(|0\rangle+|1\rangle)/\sqrt{2}$,
677: $|01\rangle(|0\rangle+|1\rangle)/\sqrt{2}$,
678: $|10\rangle(|0\rangle+|1\rangle)/\sqrt{2}$ and
679: $|11\rangle(|0\rangle+|1\rangle)/\sqrt{2}$ to their respective
680: final states in parallel by preparing their sum as a mixed state
681: $I^{1}I^{2}\sigma_{x}^{3}$ using pulse sequence
682: $$[\frac{\pi}{2}]_{y}^{2}-[grad]_{z}-[\frac{\pi}{2}]_{x}^{1}
683: -[grad]_{z}-[\frac{\pi}{2}]_{y}^{3}$$ where $[grad]_{z}$ denotes a
684: gradient pulse along $z$- axis. As usual \cite{deviation}, we
685: describe these mixed states in an operator notation that refers
686: only to the traceless part of the density operator. Since $[D,
687: \sigma_x^3] = [C, D] = 0$, the evolution of this initial condition
688: is determined by $C$ alone,
689: \begin{eqnarray}\label{px}
690: \rho_{1}(t)&=&U(t)\sigma_{x}^{3} U(t)^{\ddag}
691: =U_{C}(t)\sigma_{x}^{3}U_{C}(t)^{\ddag}\nonumber\\
692: &=&\frac{(2kc)^{2}-(\lambda s)^{2}}{4k^{2}}\sigma_{x}^{3}
693: -\frac{2kcs-\lambda s^{2}}{k^{2}}\sigma_{y}^{2}\sigma_{z}^{3}
694: -\frac{2s^{2}+k\lambda cs}{k^{2}}
695: \sigma_{x}^{1}\sigma_{z}^{2}\sigma_{z}^{3} .
696: \end{eqnarray}
697:
698: The first and last term in Eqn.(\ref{px}) correspond to directly
699: observable magnetization. We can therefore monitor the progress of
700: the quantum state transfer by simply recording the free induction
701: decay (FID) signal and calculating its Fourier transform.
702: %To monitor the transfer and
703: Figures \ref{expres} show the corresponding $^{13}$C NMR spectra
704: observed before and after the QST, using TCE. The initial
705: condition shows that the signal is concentrated on qubit C3 shown
706: as Figures \ref{expres} (a-c). After the transfer C3 $\rightarrow$
707: C1, the system is in state
708: $-\sigma_{x}^{1}\sigma_{z}^{2}\sigma_{z}^{3}$. The main signal is
709: on qubit C1, shown as Figures \ref{expres} (d-f) corresponding to
710: $\lambda=0$, $1.5$, and $4$, respectively. The different resonance
711: lines indicate that the magnetization on qubit C1 is aligned along
712: the positive or negative $x$- axis, depending on the state of
713: qubits H2 and C3. This agrees well with the prediction of Eqs.
714: (\ref{px1}-\ref{px4}). After the transfer C3 $\rightarrow$ C1
715: $\rightarrow$ C3, the system is in state $\sigma_{x}^{3}$. The
716: main signal returns to qubit C3, shown as Figures \ref{expres}
717: (g-i) corresponding to $\lambda=0$, $1.5$, and $4$, respectively.
718: The time of QST C3 $\rightarrow$
719: C1 is measured to be $t=1.00$, $t=0.62$, and $t=0.50$, and the
720: time of QST C3 $\rightarrow$ C1 $\rightarrow$ C3 is measured to be
721: $t=2.00$, $t=1.75$, and $t=1.13$, when $\lambda=0$, $1.5$, and
722: $4$, respectively. Here we also use $t_{0}$ as the time unit.
723: Compared with the case of $\lambda=0$, the speed of QST is
724: increased by the three- spin interactions.
725: % The corresponding simulated results without decoherence are shown in Figures
726: %\ref{simres}, where the time of QST is chosen as the theoretical
727: %expectations. By comparing the experimental and simulated results
728: The experimental errors mainly result from the strong coupling
729: between the two carbons and the effects of decoherence. Moreover
730: the imperfection of the pulses, especially the $\pi$ pulses for
731: refocusing is another error source.
732:
733: Similarly the process of transferring
734: $(|0\rangle+i|1\rangle)/\sqrt{2}$ can be observed by choosing the
735: initial state as $I^{1}I^{2}\sigma_{y}^{3}$ where
736: $\sigma_{y}=(|0\rangle+i|1\rangle)(\langle0|-i\langle1|)-I$. Using
737: $[C, \sigma_y^3] = 0$ we then have
738: \begin{eqnarray}\label{py}
739: \rho_{2}(t)&=&U(t)\sigma_{y}^{3}U(t)^{\ddag}
740: = U_{D}(t)\sigma_{y}^{3}U_{D}(t)^{\ddag}\nonumber\\
741: &=&\frac{(2kc)^{2}-(\lambda s)^{2}}{4k^{2}}\sigma_{y}^{3}
742: +\frac{2kcs-\lambda s^{2}}{k^{2}}\sigma_{x}^{2}\sigma_{z}^{3}
743: -\frac{2 s^{2}+k\lambda cs}{k^{2}}
744: \sigma_{y}^{1}\sigma_{z}^{2}\sigma_{z}^{3}.
745: \end{eqnarray}
746:
747: For these initial conditions, it is thus sufficient to consider
748: only part of the evolution operator, generating either $U_C(t)$ or
749: $U_D(t)$.
750:
751: \subsection{General initial conditions }
752:
753: For other initial conditions, the full evolution
754: operator $U(t)$ is required.
755: As an example, we choose
756: $I^{1}I^{2}\sigma_{z}^{3}$ as the initial state, and obtain
757: \begin{eqnarray}\label{ptz}
758: \rho_{3}(t)&=&U(t)\sigma_{z}^{3}U^{\dag}(t) \nonumber\\
759: &=&\frac{(2kc)^{2}-(\lambda s)^{2}}{4k^{2}}[\frac{(2kc)^{2}-(\lambda s)^{2}}{4k^{2}}\sigma_{z}^{3}
760: -\frac{2kcs-\lambda s^{2}}{k^{2}}\sigma_{x}^{2}\sigma_{y}^{3}
761: +\frac{2s^{2}+k\lambda cs}{k^{2}}\sigma_{y}^{1}\sigma_{z}^{2}\sigma_{y}^{3}] \nonumber\\
762: &+&\frac{2kcs-\lambda s^{2}}{k^{2}}[\frac{(2kc)^{2}-(\lambda s)^{2}}{4k^{2}}\sigma_{y}^{2}\sigma_{x}^{3}
763: +\frac{2kcs-\lambda s^{2}}{k^{2}}\sigma_{z}^{2}
764: +\frac{2s^{2}+k\lambda cs}{k^{2}}\sigma_{y}^{1}\sigma_{x}^{2}]\nonumber\\
765: &+&\frac{2s^{2}+k\lambda cs}{k^{2}}[\frac{(2kc)^{2}-(\lambda s)^{2}}{4k^{2}}\sigma_{x}^{1}\sigma_{z}^{2}\sigma_{x}^{3}
766: -\frac{2kcs-\lambda s^{2}}{k^{2}}\sigma_{x}^{1}\sigma_{y}^{2}+\frac{2s^{2}+k\lambda
767: cs}{k^{2}}\sigma_{z}^{1}].
768: \end{eqnarray}
769:
770: Noting that $J_{23}$ is much smaller than $J_{12}$, one finds that
771: $U_{D}$ requires a longer time to complete than $U_{C}$.
772: For
773: example, when $\lambda=1.5$, $U_{C}$ requires about 340 ms for QST
774: from C3 to C1, while $U_{D}$ requires about 420 ms. The effective
775: $T_{2}$ ($T_{2}^{*}$) of the current sample is measured to be
776: $0.35s$, $0.26s$, and $0.23s$ for C1, H2, and C3, respectively.
777: When $U_{D}$ or the full $U$ is applied, decoherence results in a
778: significant degradation of the experimental data.
779: We therefore show here only the results of the simulation.
780: For this purpose, we also neglect the small strong-coupling effects
781: between qubits C1 and C3.
782: The initial states are chosen as $I^{1}I^{2}\sigma_{x}^{3}$,
783: $I^{1}I^{2}\sigma_{y}^{3}$ and $I^{1}I^{2}\sigma_{z}^{3}$ as
784: respectively. Because the relevant terms in Eq. (\ref{ptz}),
785: are not directly observable, we apply readout pulses
786: $[\frac{\pi}{2}]_{y}^{1}$ and $[\frac{\pi}{2}]_{y}^{3}$
787: to $\rho_{3}(t)$ to obtain the observable signals of C1
788: and C3, respectively.
789: %Compared with the results shown in Figures
790: %\ref{simres} we find that the results are be improved greatly.
791:
792: Figures \ref{QSTx}-\ref{QSTz} show the progress of the QST.
793: For each initial state, the results for
794: $\lambda=0$, $1.5$, and $4$ are given. The data points can be
795: well fitted by the corresponding theoretical graphs. Points A, B, and C denote
796: the maxima corresponding to the
797: time of QST C$3\rightarrow$ C$1$; points D, E, and F denote the
798: maxima corresponding to the time of QST C$3\rightarrow$ C$1$
799: $\rightarrow$ C$3$. Obviously the time required for the QST
800: decreases with the increase of $\lambda$.
801: %The simulated data without decoherence for $\lambda=0$, $1.5$,
802: %$2.5$ and $4$ are marked by "*", "+", "$\triangle$", and
803: %"$\times$", respectively. The experimental data are denoted by the
804: %corresponding circled marks. The graphs are the theoretical
805: %results, used to fit the corresponding experimental data. (XX
806: %explain QST ... XX)
807:
808:
809:
810:
811: % We choose
812:
813: %\begin{equation}\label{inixy}
814: % \rho_{ini}=\sigma_{x}^{3}+\sigma_{y}^{3}
815: %\end{equation}
816: % as the initial state to implement $U(T)$. Using Eqs. (\ref{UC})
817: % and (\ref{UD}), we obtain
818:
819:
820: %When $t=t_{QST_{1\rightarrow 3}}$, $\rho(t_{QST_{1\rightarrow
821: %3}})=-(\sigma_{x}^{1}+\sigma_{y}^{1})\sigma_{z}^{2}\sigma_{z}^{3}$,
822: %whicn menas that the state has been transferred from spin 3 to
823: %spin 1.
824:
825:
826:
827: % We implement the QST when $\lambda$ changes from $0$ to $8$. The experimental results
828: %are shown in figure \ref{tQST13}, where $T_{QST}=\sqrt{2}t_{QST}$.
829: %The data for $T_{QST_{3\rightarrow1}}$, and
830: %$T_{QST_{3\rightarrow1\rightarrow3}}$ are marked by "*",
831: % and "+", respectively. We also give the data for $T_{QST_{1\rightarrow3}}$, denoted by "$\times$",
832: % through subtracting $T_{QST_{3\rightarrow1}}$ from
833: % $T_{QST_{3\rightarrow1\rightarrow3}}$.
834: % The data can be fitted by the corresponding
835: %theoretical graphs, which are blue, green, and red for
836: %$T_{QST_{3\rightarrow1}}$, $T_{QST_{1\rightarrow3}}$, and
837: %$T_{QST_{3\rightarrow1\rightarrow3}}$, respectively. The
838: %experimental results agree with the theoretical expectations quite
839: %well. Figures xx....xx show the experimental spectra when the
840: %QST occurs. Each spectrum is obtained through a hard readout pulse
841: %$[\frac{\pi}{2}]_{y}^{1,3}$. (xx .... xx)
842:
843: % The corresponding experimental data measured from Figure \ref{expres}
844: %and simulated data measured on the strong coupling system are also
845: %shown in Figure \ref{QSTx} for contrast.
846:
847: \section{Discussion}
848: The QST can also be implemented by a series of SWAP operations.
849: For the three- spin chain, the state of spin 1 can be transferred
850: to spin 3 through
851: \begin{equation}\label{s13}
852: S_{13}=S_{12}S_{23}S_{12}=
853: \left (\begin{array}{cccccccc}
854: 1 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\
855: 0 & 0 & 0 & 0 & 1 & 0 & 0 & 0 \\
856: 0 & 0 & 1 & 0 & 0 & 0 & 0 & 0 \\
857: 0 & 0 & 0 & 0 &0 & 0& 1 &0 \\
858: 0 & 1 & 0 & 0 & 0 & 0 & 0 & 0 \\
859: 0 & 0 & 0 &0 & 0 & 1 & 0 & 0 \\
860: 0 & 0 & 0 & 1 & 0 & 0 & 0 & 0 \\
861: 0 & 0 & 0 & 0 & 0 & 0 & 0 & 1 \\
862: \end{array}\right ).
863: \end{equation}
864: When $\lambda=0$ Eqs.(\ref{UQST31}) and (\ref{UQST13}) are
865: equivalent to $S_{13}$ (up to some phase factors), just as
866: discussed in Ref. \cite{Zhang05}. When $\lambda\neq0$, however,
867: neither Eq.(\ref{UQST31}) nor (\ref{UQST13}) is equivalent to
868: $S_{13}$. The difference between the stepped-up QST and the SWAP
869: operation comes from the three- spin interaction, which breaks the
870: symmetry for exchanging spins $1$ and $3$. One can prove that when
871: spin $1$ and spin $3$ are exchanged, the three- spin terms in Eq.
872: (\ref{xy}) are changed from
873: $\frac{\lambda}{2}(\sigma_{x}^{1}\sigma_{z}^{2}\sigma_{y}^{3}-
874: \sigma_{y}^{1}\sigma_{z}^{2}\sigma_{x}^{3})$ to
875: $-\frac{\lambda}{2}(\sigma_{x}^{1}\sigma_{z}^{2}\sigma_{y}^{3}-
876: \sigma_{y}^{1}\sigma_{z}^{2}\sigma_{x}^{3})$. Such asymmetry
877: can also explain why $t_{QST_{3\rightarrow 1}}$ differs from
878: $t_{QST_{1\rightarrow 3}}$.
879:
880: While we have considered here only single-qubit states, it is also possible
881: to transfer multi-qubit states through the Heisenberg $XY$ spin
882: chain \cite{Christandl,Subrahmanyam}, even entangled ones.
883: Such transfers can
884: also be speeded up by three-spin interactions.
885: For example,
886: the four Bell-states $(|00\rangle\pm|11\rangle)/\sqrt{2}$,
887: $(|01\rangle\pm|10\rangle)/\sqrt{2}$ can be transferred from spins
888: 2 and 3 to spins 1 and 2 by
889: \begin{equation}\label{Bell12}
890: U(t_{QST_{1\rightarrow 3}})(|0\rangle_{1}|00\rangle\pm|11\rangle)_{23}/\sqrt{2}
891: =(|00\rangle\mp|11\rangle)_{12}|0\rangle_{3}/\sqrt{2}
892: \end{equation}
893: \begin{equation}\label{Bell34}
894: U(t_{QST_{1\rightarrow 3}})(|0\rangle_{1}|01\rangle\pm|10\rangle)_{23}/\sqrt{2}
895: =(|01\rangle\pm|10\rangle)_{12}|0\rangle_{3}
896: (\frac{\lambda^{2}-4}{\lambda^{2}+4}+i\frac{4\lambda}{\lambda^{2}+4})/\sqrt{2}.
897: \end{equation}
898: Using the analysis of section \ref{sect2}, one finds that when
899: $\lambda<0$, the speed of transferring the entangled states is
900: increased by the three-spin interactions.
901:
902: \section{conclusion}
903: We simulated a spin XY chain with three-spin interactions,
904: using a three qubit NMR system.
905: Compared to the case where the system contains only two-spin interactions,
906: the three-spin interaction increases the speed of the operation.
907: Our results [Eqs. (\ref{UQST31}) and (\ref{UQST13})]
908: show that when the three- spin interactions exist, the QST is not
909: equivalent to the SWAP operation any more. Unlike the SWAP
910: operation, not all rows in the unitary evolution to realize the
911: QST have only one nonzero terms.
912: %in
913: %the matrix have 1. The subspace is further reduced. (xx see Ref
914: %\cite{Christandl} xx)
915:
916:
917: % a Simple QST model: $XY$ chain + $\sigma\sigma-\sigma\sigma$, to
918: %explain the effects of the three- interactions, and the network
919: %when N>3.
920:
921: The simulation of the XY chain with three- spin interactions
922: offers a possible laboratory to study the problems related to
923: three- spin interactions. Our techniques can simulate the chain
924: with arbitrary $\lambda$. In fact $\lambda$ represents the ratio
925: of the three-body and two- body coupling constants, because we
926: have set the two- body coupling constants to $1$. In a practical
927: sense $\lambda$ can be enhanced through increasing the three-body
928: couplings or decreasing the two-body couplings to speed up the
929: QST.
930: Although our results are obtained using three spin system, they are helpful
931: for the case of more than three spins.
932:
933: \section{Acknowledgment}
934: We thank Prof. Guilu Long, Prof. Jiangfeng Du and Mr. Bo Chong for helpful discussions. The
935: experiments were performed at the Interdisciplinary Center for
936: Magnetic Resonance. This work is supported by the Alexander von
937: Humboldt Foundation, the National Natural Science Foundation of
938: China under grant No. 10374010, and the DFG.
939:
940: \begin{thebibliography}{}
941:
942:
943: \bibitem{Shor94} P. W. Shor, in Proceedings of the 35th Annual Symposium on the
944: Foundations of Computer Science, Santa Fe, NM, 1994 (IEEE Computer
945: Society Press, New York 1994).
946:
947: \bibitem{Grover97} L. K. Grover, Phys. Rev. Lett, {\bf 79}, 325(1997);
948: L. K. Grover, {\it ibid}. {\bf 80}, 4329(1998);
949: L. K. Grover, {\it ibid}. {\bf 85}, 1334(2000)
950:
951: \bibitem{Feynman82} R. P. Feynman, Int. J. Theor. Phys, {\bf 21}, 467(1982);
952: S. Lloyd, Nature, {\bf 273}, 1073(1996);
953: B. M. Boghosian, and W. Taylor IV, Physica D, {\bf 120}, 30(1998);
954:
955: \bibitem{Somma02} R. Somma, G. Ortiz, J. E. Gubernatis, E. Knill, and R. Laflamme,
956: Phys. Rev. A, {\bf 65}, 042323(2002)
957:
958: \bibitem{Bose03} S. Bose, Phys. Rev. Lett, {\bf 91}, 207901(2003);
959:
960: \bibitem{Christandl} M. Christandl, N. Datta, A. Ekert, and A. J. Landahl, Phys. Rev. Lett, {\bf 92},
961: 187902(2004); M. Christandl, N. Datta, T. C. Dorlas, A. Ekert,
962: A. Kay, and A. J. Landahl, Phys. Rev. A, {\bf 71},
963: 032312(2005)
964:
965: \bibitem{Bennett} C. H. Bennett, G. Brassard, C. Cr$\acute{e}$peau, R.
966: Jozsa,
967: A. Peres, and W. K. Wootters, Phys. Rev. Lett, {\bf 70},
968: 1895(1993)
969:
970: \bibitem{Pan} D. Bouwmeester, J. Pan, K. Mattle, M. Eibl, H. Weinfurter,
971: and A. Zeilinger, Nature (London) {\bf 390}, 575
972: (1997);
973: D. Boschi, S. Branca1, F. De Martini, L. Hardy, and S.
974: Popescu, Phys. Rev. Lett, {\bf 80},
975: 1121(1998)
976:
977: \bibitem{Nielsen98} M. A. Nielsen, E. Knill, and R. Laflamme, Nature (London), {\bf 396}, 52
978: (1998)
979:
980:
981:
982: \bibitem{Karbach} P. Karbach and J. Stolze, Phys. Rev. A, {\bf 72}, 030301(R)
983: (2005);
984: G. D. Chiara, D. Rossini, S. Montangero, and R. Fazio,
985: {\it ibid}. {\bf 72}, 012323(2005)
986:
987: \bibitem{Bose} S. C. Benjamin, and Bose, Phys. Rev. A, {\bf 70}, 032314(2004);
988: M.-H. Yung, S. C. Benjamin, and Bose, e-print quant-ph/0508165
989:
990: \bibitem{DiVincenzo} D. P. DiVincenzo, D. Bacon, J. Kempe, G. Burkard,
991: and K. B. Whaley, Nature, {\bf 408}, 339(2000)
992:
993: \bibitem{Sun} X.-F. Qian, Y. Li, Y. Li, Z. Song, and C. P. Sun,
994: Phys. Rev. A, {\bf 72} 062329(2005)
995:
996: \bibitem{physica} G. Brassarda, S. L. Braunstein, and R. Cleve, Physica
997: D, {\bf 120}, 43(1998)
998:
999: \bibitem{Deutsch85}D. Deutsch, Proc. R. Soc. A, {\bf 400}, 97(1985);
1000: D. Deutsch, {\it ibid}. {\bf 425}, 73(1989);
1001: D. Deutsch, A. Barenco, and A. Ekert, {\it ibid}. {\bf 449},
1002: 669(1995); R. Cleve, A. Ekert, C. Macchiavello and M. Mosca, {\it ibid}.
1003: {\bf 454} 339(1998); M. J. Bremner, C. M. Dawson, J. L.Dodd,
1004: A. Gilchrist, A. W. Harrow, D. Mortimer, M. A. Nielsen and T. J. Osborne
1005: Phys. Rev. Lett, {\bf 89} 247902(2002); A. Barenco , C. H. Bennett, R. Cleve, D. P. DiVincenzo, N. Margolus,
1006: P. Shor, T. Sleator, J. A. Smolin and H. Weinfurter, Phys. Rev. A
1007: {\bf 52}, 3457(1995); M. D. Bowdrey, J. A. Jones, E. Knill,
1008: and R. Laflamme, {\it ibid}. {\bf 72}, 032315(2005)
1009:
1010: \bibitem{Baxter} R. J. Baxter, and F. Y. Wu, Phys. Rev. Lett, {\bf
1011: 31}, 1294(1973)
1012:
1013: \bibitem{PachosPRL} J. K. Pachos, and M. B. Plenio, Phys. Rev. Lett, {\bf
1014: 93}, 056402(2004)
1015:
1016: \bibitem{Tsvelik} A. M. Tsvelik, Phys. Rev. B, {\bf 42}, 779(1990)
1017:
1018: \bibitem{Frahm} H. Frahm, J. Phys. A, {\bf 25}, 1417(1992)
1019:
1020: \bibitem{Cruz} C. D'Cruz, and J. K. pachos, eprint
1021: cond-mat/0506247
1022:
1023: \bibitem{Lou} P. Lou, W.-C. Wu, and M.-C. Chang, Phys. Rev. B, {\bf 70}, 064405(2004)
1024:
1025: \bibitem{Yang} M.-F. Yang, Phys. Rev. A, {\bf 71}, 030302(R)(2005)
1026:
1027: \bibitem{Tseng99}C. H. Tseng, S. Somaroo, Y. Sharf, E. Knill, R. Laflamme, T. F.
1028: Havel, and D. G. Cory, Phys. Rev. A, {\bf61},
1029: 012302(1999); N. Khaneja, S. J. Glaser, and R. Brockett, {\it ibid}.
1030: {\bf 65}, 032301(2002); R. Somma, G. Ortiz, E. Knill, and J. Gubernatis,
1031: e-print quant-ph/0304063
1032:
1033: \bibitem{Datta05} E. Knill and R. Laflamme, Phys. Rev. Lett. {\bf 81}, 5672 (1998);
1034: A. Datta, S. T. Flammia, and C. M. Caves, Phys.
1035: Rev. A, {\bf 72}, 042316(2005);
1036: R. Stadelhofer, D. Suter, and W. Banzhaf,
1037: {\it ibid}. {\bf71}, 032345 (2005).
1038:
1039: \bibitem{Vandersypen} L. M. K. Vandersypen, and I. L. Chuang,
1040: {\bf 76}, Rev. Mod. Phys, {\bf 76}, 1037(2004);
1041: X.-H. Peng, J.-F Du, and D. Suter, Phys. Rev. A, {\bf 71}, 012307
1042: (2005)
1043:
1044:
1045: \bibitem{Zhang05} J.-F. Zhang, G. L. Long, W. Zhang, Z.-W. Deng, W.-Z. Liu, and
1046: Z.-H. Lu, Phys. Rev. A, {\bf 72}, 012331(2005)
1047:
1048:
1049:
1050:
1051:
1052: %\bibitem{Negrevergne05} C. Negrevergne, R. Somma, G. Ortiz, E. Knill, R. Laflamme, Phys. Rev. A,
1053: % {\bf 71}, 032344(2005)
1054:
1055:
1056:
1057:
1058: %\bibitem{Vandersypen04} L. M. K. Vandersypen, and I. L. Chuang,
1059: % {\bf 76}, Rev. Mod. Phys, {\bf 76}, 1037(2004);
1060: % N. Boulant, L. Viola, E. M. Fortunato, D. G. Cory, Phys. Rev. Lett, {\bf 94}, 130501(2005);
1061: % J.- F. Du, T. Durt, P. Zou, H. Li, L. C. Kwek, C. H. Lai, C. H. Oh, and A. Ekert,
1062: % {\it ibid}. {\bf 94}, 040505 (2005);
1063: % D. Wei, J. Luo, X. Sun, X. Zeng, M. Zhan, and M. Liu, {\it ibid}.
1064: % {\bf 95}, 020501(2005);
1065: % J. E. Ollerenshaw, D. A. Lidar, and L. E. Kay, {\it
1066: % ibid}. {\bf 91}, 217904(2003);
1067: % C. A. Ryan, M. Laforest, J. C. Boileau, and R.
1068: % Laflamme, e-print quant-ph/0507267; A. Ghosh and A. Kumar,
1069: % e-print quant-ph/0509139; R. Bhattacharyya, R. Das, K. V. Ramanathan, and A.
1070: % Kumar, Phys. Rev. A, {\bf 71}, 052313(2005); L. Xiao and J. A.
1071: % Jones, {\it ibid}. {\bf 72}, 032326(2005);
1072: % X.-H. Peng, J.-F Du, and D. Suter, Phys. Rev. A, {\bf 71}, 012307(2005)
1073:
1074: \bibitem{DUPRA03} J. -F. Du, H. Li, X. -D. Xu, M. -J. Shi, J. -H. Wu, X. -Y. Zhou, R. -D.
1075: Han, Phys. Rev. A {\bf 67}, 042316(2003)
1076:
1077:
1078: %\bibitem{s10} R. R. Ernst, G. Bodenhausen and A. Wokaum, {\it Principles of nuclear
1079: % magnegtic resonance in one and two dimensions},
1080: % (Oxford University Press, Oxford, 1987).
1081:
1082:
1083: %\bibitem{cory} D. G. Cory, M. D. Price, and T. F. Havel, Physical D, {\bf 120}, 82(1998)
1084:
1085: \bibitem{Miquel} C. Miquel, J. P. Paz, M. Saraceno, E. Knill,
1086: R. Laflamme, and C. Negrevergne, Nature, {\bf
1087: 418}, 59(2002)
1088:
1089: \bibitem{REBURP} H. Geen and R. Freeman, J. Magn. Reson. (1969-1992) {\bf 93}, 93
1090: (1991)
1091:
1092: \bibitem{Ryan} C. A. Ryan, M. Laforest, J. C. Boileau, and R.
1093: Laflamme, Phys. Rev. A, {\bf 72}, 062317(2005);
1094: J.-F. Zhang, G. L. Long, Z.-W. Deng, W.-Z. Liu, and Z.-H.
1095: Lu, {\it ibid}. {\bf 70}, 062322(2004).
1096:
1097: \bibitem{Linden} N. Linden, B. Herv$\grave{e}$, R. J. Carbajo, and R. Freeman, Chem. Phys. Lett,
1098: {\bf 305}, 28(1999)
1099:
1100:
1101: %
1102:
1103: \bibitem{deviation} I. L. Chuang, N. Gershenfeld, M. G. Kubinec, and D. W. Leung,
1104: Proc. R. Soc. London, Ser. A {\bf 454}, 447(1998).
1105:
1106: \bibitem{Subrahmanyam} V. Subrahmanyam and A. Lakshminarayan,
1107: Phys. Lett. A, {\bf 349}, 164(2006)
1108:
1109: \end{thebibliography}
1110:
1111: %----------------------------------------------
1112: \begin{figure}
1113: %\includegraphics[width=5in]{qpt3}% MatLab file qps3.m
1114: \includegraphics[width=4in]{pst_time.eps}% MatLab file qps3.m
1115: \caption{(Color online) The duration of the QST $t_{QST_{1\rightarrow 3}}$
1116: (green), $t_{QST_{3\rightarrow 1}}$ (blue), and
1117: $t_{QST_{3\rightarrow 1 \rightarrow 3}}$ (red) vs $\lambda$.
1118: The unit of the vertical axes is chosen as
1119: $t_{0}=\frac{\pi}{2\sqrt{2}}$, normalized to the duration for
1120: $\lambda=0$.} \label{tQST13}
1121: \end{figure}
1122: %----------------------------------------------
1123: \begin{figure}
1124: \includegraphics[width=2in]{rotation.eps}
1125: \caption{The frame for operations $U_{C}(t)$ and $U_{D}(t)$. The
1126: vectors ${\bf n_{c}}$ and ${\bf n_{d}}$ denote the directions of
1127: the rotation axes for the two operations, respectively.
1128: They
1129: are tilted from the $z$ axis by the angles $\theta_{c}$ and
1130: $\theta_{d}=\pi-\theta_{c}$.
1131: The projections of ${\bf n_{c}}$ and
1132: ${\bf n_{d}}$ into the $xy$- plane are identical and indicated by a black line.
1133: The angles between the projection and the $x$ and $y$ axes are $\pi/4$.}
1134: \label{rotation}
1135: \end{figure}
1136: %----------------------------------------------
1137:
1138: \begin{figure}
1139: \includegraphics[width=2.5in]{TCE_Struct.eps}%Zhang070106 1 &2,
1140: \caption{The parameters of Carbon-13 labeled trichloroethylene
1141: (TCE).
1142: The diagonal terms in the table are the shifts (in Hz) of
1143: the carbons and protons with respect to the reference frequencies
1144: 500.13MHz and 125.76MHz, respectively.
1145: The non-diagonal terms are the coupling constants, also in Hz.}
1146: \label{sample}
1147: \end{figure}
1148: %----------------------------------------------
1149: \begin{figure}
1150: \includegraphics[width=133mm]{fig4.eps}
1151: \caption{The carbon spectrum (a) and proton spectrum (b) obtained
1152: by applying selective readout pulses to the system in its thermal
1153: equilibrium state.
1154: Each qubit gives rise to four resonance lines, which correspond to
1155: specific states of the other qubits.
1156: The highest frequency lines always correspond to the other qubits
1157: being in the $|00\rangle$ state, the lowest frequency lines to the
1158: $|11\rangle$ state.
1159: }\label{fig4}
1160: \end{figure}
1161: %-------------------------------------------
1162: %\begin{figure}
1163: %\includegraphics[width=7.6in]{inistate}
1164: % The original file is inistate.bmp,NMR: QST_x, \lambda=4,d25=4.376m
1165: %\includegraphics[angle=-90,width=15 cm]{fig5}
1166: %\caption{ The carbon spectrum when the three spin system is in the
1167: %initial state $\sigma _{x}^{3}$, shown as the left column, and in
1168: %$-\sigma_{x}^{1}\sigma_{z}^{2}\sigma_{z}^{3}$, shown as the right
1169: %column, after the completion of the QST. Figures (a) and (c) are
1170: %the results by simulation. Figures (b) and (d) are the
1171: %corresponding experimental results. Figure (c) is the reference
1172: %spectrum for the other spectra.} \label{inis}
1173: %\end{figure}
1174: %-------------------------------------------------
1175:
1176:
1177: \begin{figure}
1178: %\includegraphics[width=7.5in]{seq}
1179: % The original results is in seq.bmp and time.m for delay
1180: \includegraphics[width=5in]{fig3.eps}
1181: \caption{Pulse sequences for the implementation of $U_{C}(t)$ (a)
1182: and $U_{D}(t)$ (b). Steps $1$-$10$ correspond to the ten compound
1183: operations separated by '$\times$' in $U_{C}(t)$ and $U_{D}(t)$,
1184: respectively. The unfilled rectangles denote $\pi/2$ pulses, and
1185: the filled rectangles denote $\pi$ pulses. $X$, $\overline{X}$,
1186: $Y$ and $\overline{Y}$ below the pulses denote the $x$, $-x$, $y$,
1187: and $-y$ directions along which the pulses are applied. Those
1188: $\pi$ pulses for which directions are not denoted are refocusing
1189: pulses. They are applied in pairs in which the two pulses take
1190: opposite directions to reduce experimental errors. The durations
1191: of the pulses applied to H2 and the non-selective pulses applied
1192: to C1 and C3 are so short that they can be ignored. The selective
1193: $\pi$ pulses for C1 and C3, denoted by the green or blue
1194: rectangles, are implemented as RE-BURP \cite{REBURP} and Gauss
1195: shaped pulses with $6.2649ms$ and $2.8252ms$ durations,
1196: respectively. The delays are $d_{1}=\frac{9}{8J_{12}}$,
1197: $d_{2}=\frac{1}{16J_{23}}$, $d_{3}=\frac{1}{4\pi
1198: J_{23}}[\frac{\pi}{2}-\arctan(\frac{2\sqrt{2}}{\lambda})]$,
1199: $d_{4}=\frac{1}{2\pi
1200: J_{12}}(t\sqrt{2+\frac{\lambda^{2}}{4}}+2\pi)$,
1201: $d_{5}=\frac{1}{2\pi
1202: J_{12}}\{\pi-\frac{1}{2}[\frac{\pi}{2}-\arctan(\frac{2\sqrt{2}}{\lambda})]\}$,
1203: $d_{6}=\frac{t}{2\pi J_{23}}\sqrt{2+\frac{\lambda^{2}}{4}}$. For
1204: the case of $\lambda=0$, steps 2, 4, 7, and 9 are omitted.}
1205: \label{seq}
1206: \end{figure}
1207: %----------------------------------------------
1208:
1209: \begin{figure}
1210: \includegraphics[width=7.5in]{fig6.eps}
1211: % The original results is in ExpRes_org and ExpRes.bmp
1212: %Win-plot, then is copied to Origin, export page to fig6.eps,
1213: \caption{ The experimental results demonstrating the QST.
1214: The initial state is $\sigma _{x}^{3}$; the corresponding spectrum is
1215: shown in the left hand column.
1216: The results of the QST C3$ \rightarrow$ C1 are shown as Figures (d-f),
1217: and the results of the cyclic transfer C3 $\rightarrow$ C1 $\rightarrow$ C3
1218: are shown as Figures (g-i).
1219: The three rows correspond to increasing three-qubit coupling strength,
1220: $\lambda=0$, $1.5$, and $4$.
1221: The time required for each transfer is shown in the Figures.
1222: At $t=t_{QST_{3\rightarrow 1}}$, the three-spin system is in state
1223: $-\sigma_{x}^{1}\sigma_{z}^{2}\sigma_{z}^{3}$,
1224: and at $t=t_{QST_{3\rightarrow 1 \rightarrow 3}}$, the
1225: system is in state $\sigma_{x}^{3}$.}
1226: \label{expres}
1227: \end{figure}
1228:
1229: %----------------------------------------------
1230: %\begin{figure}
1231: %\includegraphics[width=7.5in]{fig7}
1232: % The original results is in ExpRes_org and SimRes.bmp
1233: %Win-plot, then is copied to Origin, export page to fig7.eps,
1234: %\caption{ The simulated results corresponding to Figures
1235: %\ref{expres} when $t=t_{QST}$ in theory.} \label{simres}
1236: %\end{figure}
1237: %-------------------------------------------------
1238:
1239: \begin{figure}
1240: %\includegraphics[width=5in]{QST3_C_exp}%MatLab QST3_C_exp.m
1241: \includegraphics[width=6in]{pst_x.eps} %MatLab QST3_C_exp.m
1242: \caption{(Color online) Progress of the QST, starting from $\sigma
1243: _{x}^{3}$, for different strengths of the three-body coupling. The
1244: upper part of the figure shows the overlap of the density operator
1245: with the target state $\sigma_x^1\sigma_z^2\sigma_z^3$ as a
1246: function of time. The unit $t_0$ of the time axis corresponds to
1247: the transfer time in the absence of the three-body interaction.
1248: The data for $\lambda=0$, $1.5$, and $4$ are marked by "*", "+",
1249: and "$\times$", respectively. The solid lines represent the
1250: theoretical results, the individual points correspond to the
1251: simulated data by setting TCE as the weak coupling system without
1252: decoherence. Points A, B, and C indicate the maxima corresponding
1253: to the transfer times C3$ \rightarrow$ C1 and the points D, E, and
1254: F to the transfer times C3 $\rightarrow$ C1 $\rightarrow$ C3. This
1255: clearly demonstrates the speedup of the transfer by the three-body
1256: interaction. } \label{QSTx}
1257: \end{figure}
1258:
1259: %----------------------------------------------
1260: \begin{figure}
1261: %\includegraphics[width=5in]{QST3_D_exp}%MatLab QST3_D_exp.m
1262: \includegraphics[width=6in]{pst_y.eps} %MatLab QST3_D_exp.m
1263: \caption{(Color online) Progress of the QST, starting from $\sigma
1264: _{y}^{3}$, for different strengths of the three-body coupling. The
1265: data
1266: %without decoherence
1267: for $\lambda=0$, $1.5$, and $4$ are marked by "*", "+", and
1268: "$\times$", respectively.
1269: %The experimental data are
1270: %denoted by the corresponding circled marks.
1271: The graphs are the theoretical results, used to fit the
1272: corresponding
1273: %experimental
1274: data. Points A, B, and C indicate the maxima corresponding to the
1275: transfer times C$3$ $\rightarrow$ C$1$ and points D, E, and F to
1276: the transfer times C$3$ $\rightarrow$ C$1$
1277: $\rightarrow$ C$3$.}\label{QSTy}
1278: \end{figure}
1279:
1280: %----------------------------------------------
1281: \begin{figure}
1282: %\includegraphics[width=6in]{QST3_z}%MatLab QST3.m
1283: \includegraphics[width=6in]{pst_z.eps}%MatLab QST3.m
1284: \caption{(Color online) Progress of the QST, starting from $\sigma
1285: _{z}^{3}$, for different strengths of the three-body coupling. The
1286: data
1287: %without decoherence
1288: for $\lambda=0$, $1.5$, and $4$ are marked by "*", "+", and
1289: "$\times$", respectively.
1290: %The experimental data are
1291: %denoted by the corresponding circled marks.
1292: The graphs are the theoretical results, used to fit the
1293: corresponding
1294: %experimental
1295: data. Points A, B, and C indicate the maxima corresponding to the
1296: transfer times C$3$ $\rightarrow$ C$1$ and points D, E, and F to
1297: the transfer times C$3$ $\rightarrow$ C$1$ $\rightarrow$
1298: C$3$.}\label{QSTz}
1299: \end{figure}
1300:
1301: %-------------------------------------------------
1302:
1303:
1304:
1305:
1306: \end{document}
1307: