1: \documentclass[twocolumn,showpacs,floatfix,aps]{revtex4}

2: \usepackage{graphicx}% Include figure files

3: \usepackage{dcolumn}% Align table columns on decimal point

4: \usepackage{bm}% bold math

5: \newcommand{\n}{\mbox{\boldmath $\nabla$}}

6: \renewcommand\baselinestretch{1.0}

7: \begin{document}

8: 

9: %\draft

10: 

11: \title{Two-particle localization and antiresonance \\ in disordered spin

12:  and qubit chains}

13: 

14: \author{L. F. Santos}

15: \email{santos@pa.msu.edu}

16: \author{M. I. Dykman}

17: \email{dykman@pa.msu.edu}

18: \affiliation{Department of Physics and Astronomy

19: and the Institute for Quantum Sciences,

20: Michigan State University, East Lansing, Michigan 48824}

21: 

22: \date{\today}

23: 

24: \begin{abstract}

25: We show that, in a system with defects, two-particle states may

26: experience destructive quantum interference, or antiresonance. It

27: prevents an excitation localized on a defect from decaying even

28: where the decay is allowed by energy conservation. The system studied

29: is a qubit chain or an equivalent spin chain with an anisotropic ($XXZ$)

30: exchange coupling in a magnetic field. The chain has a defect with an

31: excess on-site  energy. It corresponds to a qubit with the

32: level spacing different from other qubits.  We show that, because of

33: the interaction between excitations, a single defect may lead to

34: multiple localized states. The energy spectra and localization lengths

35: are found for two-excitation states. The localization of

36: excitations facilitates the operation of a quantum

37: computer. Analytical results for strongly anisotropic coupling are

38: confirmed by numerical studies.

39: \end{abstract}

40: 

41: \pacs{03.67.Lx,75.10.Pq,75.10.Jm,73.21.-b}

42: 

43: \maketitle

44: %\newpage

45: 

46: \section{Introduction}

47: 

48: One of the most important potential applications of quantum computers

49: (QC's) is studies of quantum many-body effects. It is particularly

50: interesting to find new many-body effects in condensed-matter systems

51: that could be easily simulated on a QC. In the present paper we

52: discuss one such effect: antiresonance, or destructive quantum

53: interference between two-particle excitations in a system with

54: defects. We also study interaction-induced two-particle localization on

55: a defect and discuss implications of the results for quantum

56: computing.

57: 

58: The basic elements of a QC, qubits, are two-state systems. They are

59: naturally modeled by spin-1/2 particles. In many suggested

60: realizations of QC's, the qubit-qubit interaction is ``on'' all the time

61: \cite{recent_reviews}. In terms of spins, it corresponds to exchange

62: interaction. The dynamics of such QC's and spin systems in solids have

63: many important similar aspects that can be studied together.

64: 

65: In most proposed QC's the energy difference between the qubit states is

66: large compared to the qubit-qubit interaction. This corresponds to a

67: system of spins in a strong external magnetic field. However, in

68: contrast to ideal spin systems, level spacings of different qubits

69: can be different. A major advantageous feature of QC's is that the qubit

70: energies can be often individually controlled

71: \cite{Makhlin99,JJ-all,mark}. This corresponds to controllable

72: disorder of a spin system, and it allows one to use QC's for studying a

73: fundamentally important problem of how the spin-spin interaction

74: affects spin dynamics in the presence of disorder.

75: 

76: %The analogy with coupled-spin systems is particularly relevant for the

77: %QC's where the interqubit interaction is permanently on rather than

78: %being turned on and off.

79: 

80: Several models of QC's where the interqubit interaction is permanently

81: ``on'' are currently studied. In these models the effective spin-spin

82: interaction is usually strongly anisotropic. It varies from the

83: essentially Ising coupling $\sigma_n^z\sigma_m^z$ in nuclear

84: magnetic resonance and some

85: other systems

86: \cite{Chuang9802,Yamamoto02,Cory00,Piermarocchi02,Demille02} ($n,m$

87: enumerate qubits, and $z$ is the direction of the magnetic field) to

88: the $XY$-type (i.e., $\sigma_n^x\sigma_m^x + \sigma_n^y\sigma_m^y$) or

89: the $XXZ$-type (i.e., $\Delta_{nm}\sigma_n^z\sigma_m^z+

90: \sigma_n^x\sigma_m^x + \sigma_n^y\sigma_m^y$) coupling in some

91: Josephson-junction based systems \cite{Makhlin99,JJ-all}.

92: 

93: The Ising coupling describes the system in the case where the

94: transition frequencies of different qubits are strongly

95: different. Then $\sigma_n^z\sigma_m^z$ is the only part of the

96: interaction that slowly oscillates in time, in the Heisenberg

97: representation, and therefore is not averaged out. If the qubit

98: frequencies are close to each other, the terms $\sigma_n^x\sigma_m^x +

99: \sigma_n^y\sigma_m^y$ become smooth functions of time as well. They

100: lead to resonant excitation hopping between qubits.  In a multiqubit

101: system with close frequencies, both Ising and $XY$ interactions are

102: present in the general case, but their strengths may be different

103: \cite{Silvestrov01,Kaminsky-Lloyd02}. In this sense the $XXZ$

104: coupling is most general, at least for qubits with high transition

105: frequencies.

106: 

107: The interqubit interaction often rapidly falls off with the distance

108: and can be approximated by nearest neighbor coupling. Many important

109: results on anisotropic spin systems with such coupling have been

110: obtained using the Bethe ansatz. Initially the emphasis was placed on

111: systems without defects \cite{Bethe} or with defects on the edge of a

112: spin chain \cite{Baxter}. More recently these studies have been

113: extended to systems with defects that are described by integrable

114: Hamiltonians \cite{integrable}. However, the problem of a spin chain

115: with several coupled excitations and with defects of a general type

116: has not been solved.

117: 

118: In this paper we investigate interacting 

119: excitations in an anisotropic spin system

120: with defects. We show that the excitation localized on a defect does

121: not decay even where the decay is allowed by energy conservation.  We

122: also find that, in addition to a single-particle excitation, a defect

123: leads to the onset of two types of localized two-particle excitations.

124: 

125: The analysis is done for a system with the $XXZ$ coupling. The coupling

126: anisotropy is assumed to be strong, as in the case of a QC based on

127: electrons on helium, for example \cite{mark}. The ground state of the

128: system corresponds to all spins pointing in the same direction

129: (downwards, for concreteness).  A single-particle excitation

130: corresponds to one qubit being excited, or one spin being flipped. If

131: the qubit energies are tuned in resonance with each other, a QC

132: behaves as an ideal spin system with no disorder. A single-particle

133: excitation is then magnon-type, it freely propagates through the

134: system.

135: 

136: In the opposite case where the qubit energies are tuned far away from

137: each other (as for diagonal disorder in tight-binding models), all

138: single-particle excitations are localized. If the excitation density

139: is high, the interaction between them may affect their localization,

140: leading to quantum chaos, cf. Refs.

141: \onlinecite{Silvestrov01,Shepelyansky00,Izrailev01,Lloyd02}.

142: Understanding the interplay between interaction and disorder is a

143: prerequisite for building a QC. We will consider the case where the

144: excitation density is low, yet the interaction is important. In

145: particular, excitations may form bound pairs (but the pair density is

146: small).

147: 

148: One of the important questions is whether the interaction leads to

149: delocalization of excitations. More specifically, consider an

150: excitation, which is localized on a defect in the absence of other

151: excitations. We now create an extended magnon-type excitation (a

152: propagating wave), that can be scattered off the localized one.

153: % \footnote{In the case of a

154: %finite-length spin chain, we use the term ``extended'' for the states

155: %with a sinusoidal wave function.}.

156: The problem is whether this will

157: cause the excitation to move away from the defect. We show below that,

158: due to unexpected destructive quantum interference, the scattering

159: does not lead to delocalization.

160: 

161: 

162: \subsection{Model and preview}

163: We consider a one-dimensional array of qubits which models a spin-1/2

164: chain. For nearest-neighbor coupling, the Hamiltonian is

165: %

166: \begin{eqnarray}

167: \label{hamiltonian}

168: &&H =  \frac{1}{2}\sum_n \varepsilon^{(n)} \sigma_{n}^{z} +

169: \frac{1}{4}\sum_n \sum_{i=x,y,z}J_{ii}\sigma_n^i\sigma_{n+1}^i\\

170: &&J_{xx}=J_{yy}=J,\;J_{zz}=J\Delta\nonumber .

171: \end{eqnarray}

172: %

173: Here, $\sigma_n^i$ are the Pauli matrices and $\hbar =1$. The

174: parameter $J$ characterizes the strength of the exchange coupling, and

175: $\Delta$ determines the coupling anisotropy. We assume that

176: $|\Delta|\gg 1$; for a QC based on electrons on helium, $|\Delta|$

177: lies between $20$ and $8$, for typical parameter values

178: \cite{mark}.

179: 

180: We will consider effects due to a single defect. Respectively, all

181: on-site spin-flip energies $\varepsilon^{(n)}$ are assumed to be the

182: same except for the site $n=n_0$ where the defect is located, that is,

183: %

184: \begin{equation}

185: \label{chain_energies}

186: \varepsilon^{(n)}= \varepsilon + g\delta_{n,n_0} .

187: \end{equation}

188: 

189: In order to formulate the problem of interaction-induced decay of

190: localized excitations, we preview in Fig.~\ref{fig:energies} a part of

191: the results on the energy spectrum of the system.  In the absence of

192: the defect, the energies of single-spin excitations (magnons) lie

193: within the band ${\varepsilon}_1 \pm J$, where ${\varepsilon}_1

194: =\varepsilon-J\Delta $ (the energy is counted off from the 

195: ground-state energy).  The defect has a spin-flip energy that differs by $g$

196: (a qubit with a transition frequency different from that of other

197: qubits).  It leads to a localized single-spin excitation with no

198: threshold in $g$, for an infinite chain. The energy of the localized

199: state is shown by a dashed line on the left panel of

200: Fig.~\ref{fig:energies}.

201: 

202: \begin{figure}[ht]

203: \includegraphics[width=3.2in]{fig1_XXZ.eps}

204: \caption{Left panel: the one-excitation energy spectrum in an infinite

205: spin chain with a defect.  The energies of extended states (magnons)

206: form a band of width $J$ centered at ${\varepsilon}_1$. The dashed

207: line indicates the energy of the excitation localized on the

208: defect. Right panel: the two-excitation energy spectrum. The band

209: $2({\varepsilon}_1\pm J)$ is formed by uncoupled magnons. The band

210: centered at $2{\varepsilon}_1+g$ is formed by localized-delocalized

211: pairs (LDP's) in which one excitation is localized on the defect and

212: the other is in an extended state. The narrow band centered at

213: $2{\varepsilon}_1 + J\Delta$ is formed by propagating bound pairs (BP's)

214: of excitations.  The dashed lines show the energies of the

215: states where both excitations are bound to the defect.}

216: \label{fig:energies}

217: \end{figure}

218: 

219: We now discuss excitations that correspond to two flipped spins. A

220: defect-free $XXZ$ system has a two-magnon band of independently

221: propagating noninteracting magnons. However, the anisotropy of the

222: exchange coupling leads also to the onset of bound pairs (BP's) of

223: excitations. The BP band is much narrower than the two-magnon band and

224: is separated from it by a comparatively large energy difference

225: $J\Delta$, see the right panel of Fig.~\ref{fig:energies}. In the

226: presence of a defect, there are two-excitation states with one

227: excitation localized on the defect and the other being in an extended

228: state. We call them localized-delocalized pairs (LDP's).  An interplay

229: between disorder and interaction may lead to new types of states where

230: both excitations are localized near the defect. Their energies are

231: shown in the right panel of Fig.~\ref{fig:energies} by dashed lines.

232: 

233: A localized one-spin excitation cannot decay by emitting a magnon, by

234: energy conservation. But  it might experience an

235: induced decay when a magnon is inelastically scattered off the excited

236: defect into an extended many-spin state. Magnon-induced decay is

237: allowed by energy conservation when the total energy of the localized

238: one-spin excitation and the magnon coincides with the energy of

239: another two-particle state.  In the $XXZ$ model the total number of

240: excitations (flipped spins) is conserved, and therefore decay is only

241: possible into extended states of two bound magnons. In other words, it

242: may only happen when the LDP band overlaps with the BP band in

243: Fig.~\ref{fig:energies}.

244: 

245: Decay into BP states may occur directly or via the two-excitation

246: state located next to the defect. The amplitudes of the corresponding

247: transitions turn out to be nearly equal and opposite in sign. As a

248: result of this quantum interference, even though the band of bound

249: magnons is narrow and has high density of states, the LDP to BP

250: scattering does not happen, i.e., the excitation on the defect is not

251: delocalized. The BP to LDP scattering does not happen either, i.e., a

252: localized excitation is not created as a result of BP decay.

253: 

254: In Sec.~II below and in the Appendix we briefly analyze localization of

255: one excitation in a finite chain with a defect, for different boundary

256: conditions. In Sec.~III we discuss the two-excitation states localized

257: near a defect. In Sec.~IV we consider the resonant situation where the

258: energy band of extended bound pair states is within the band of

259: energies of the flipped defect spin plus a magnon, i.e., where the BP

260: band overlaps with the LDP band in Fig.~\ref{fig:energies}.  We find

261: that the localized excitation remains on the defect site in this

262: case. Analytical results for a chain with strong anisotropy $|\Delta|$

263: are compared with numerical calculations. Section V contains concluding

264: remarks.

265: 

266: 

267: 

268: \section{One excitation: localized and extended states}

269: 

270: In order to set the scene for the analysis of the two-excitation case,

271: in this section and in the Appendix we briefly discuss the well-known case

272: of one excitation (flipped spin) in an $XXZ$ spin chain with a defect

273: \cite{Economou} and the role of boundary conditions.  The Hamiltonian

274: of the chain with the defect on site $n_0$ has the form

275: (\ref{hamiltonian}), (\ref{chain_energies}). We assume that the

276: excitation energy $\varepsilon$ largely exceeds both the coupling

277: constant $|J|$ and the energy excess on the defect site $|g|$.

278: In this case the ground-state of the system corresponds to all spins

279: being parallel, with $\langle\sigma_n^z\rangle=-1$

280: irrespective of the signs of $J,g$, and $\Delta$.

281: 

282: 

283: Without a defect, one-spin excitations are magnons. They freely

284: propagate throughout the chain. The term in the Hamiltonian

285: (\ref{hamiltonian}) responsible for one-excitation hopping is

286: $H^{(t)}=\sum\nolimits_n H^{(t)}_n$, with

287: %

288: \begin{eqnarray}

289: \label{transfer_one}

290: &H^{(t)}_n = {1\over 4}J \sum_{i=x,y} \sigma_n^i\sigma_{n+1}^i \equiv {1\over

291: 8}J (t_n^{(l)} + t_n^{(r)}),\\

292: %

293: &t_n^{(r)}= [t_{n}^{(l)}]^{\dagger}=

294: \sigma_{n+1}^{+} \sigma_{n}^{-}.\nonumber

295: \end{eqnarray}

296: %

297: The operators $t_{n}^{(r)}$ and $t_{n}^{(l)}$ cause excitation

298: shifts $n\to n+1$ and $n+1\to n$, respectively.

299: 

300: A defect leads to magnon scattering and to the onset of localized

301: states.  Both propagation of excitations and their localization are

302: interesting for quantum computing. Coherent excitation transitions allow

303: one to have a QC geometry where remote ``working'' qubits are

304: connected by chains of ``auxiliary'' qubits, which form ``transmission

305: lines'' \cite{mark}. Localization, on the other hand, allows one to

306: perform single-qubit operations on targeted qubit.

307: 

308: A QC makes it possible to model spin chains with different boundary

309: conditions. The simplest models are an open spin chain with free

310: boundaries, which is mimicked by a finite-length array of qubits (for

311: electrons on helium, it can be implemented using an array of equally

312: spaced electrodes, cf. Ref.~\onlinecite{Goodkind01}) or a periodic

313: chain, which can be mimicked by a ring of qubits.

314: 

315: An open $N$-spin chain is described by the Hamiltonian

316: (\ref{hamiltonian}), where the first sum runs over $n=1,\ldots,N$ and

317: the second sum runs over $n=1,\ldots,N-1$ (the edge spins have

318: neighbors only inside the chain). In what follows, we count energy off

319: the ground-state energy ${E}_{0} = -(N\varepsilon+g)/2 + (N-1) J

320: \Delta /4 $, i.e., we replace in Eq.~(\ref{hamiltonian}) $H\to H-E_0$.

321: 

322: The eigenfunctions of $H$ in  the case of one

323: excitation can be written as \cite{Economou}

324: 

325: \begin{equation}

326: \psi_1 = \sum_{n=1}^{N} a(n) \phi (n)

327: \label{one_exc} ,

328: \end{equation}

329: %

330: where $\phi (n) $ corresponds to the spin on site $n$ being up

331: and all other spins being down.  The Schr\"odinger equation for

332: $a(n) $ has the form

333: 

334: \begin{eqnarray}

335: && \left( {\varepsilon}_{1} + g \delta_{n,n_0}

336: +\frac{J\Delta }{2} \delta_{n,1}

337: +\frac{J\Delta }{2} \delta_{n,N} \right) a(n)

338: \nonumber \\

339:         &&+

340: \frac{J }{2} [a(n-1) + a(n+1)]=E_1a(n),

341: \label{eq_a}

342: \end{eqnarray}

343: %

344: where ${\varepsilon}_{1} = \varepsilon -J\Delta$ is the energy of a

345: flipped spin in an ideal infinite chain in the absence of excitation

346: hopping and $E_1$ is the one-excitation energy eigenvalue. For an

347: open $N$-spin chain we set $a(0)= a(N+1)=0$ in

348: Eq.~(\ref{eq_a}).

349: 

350: The Hamiltonian of a closed $N$-spin chain has the form

351: (\ref{hamiltonian}), where both sums over $n$ go from 1 to $N$ and the

352: site $N+1$ coincides with the site 1. Here, the defect

353: location $n_0$ can be chosen arbitrarily.

354: The wave function can be sought in the

355: form (\ref{one_exc}). The Schr\"odinger equation then has the form

356: (\ref{eq_a}), except that there are no terms proportional to $\delta_{n,1},\,

357: \delta_{n,N}$ from the end points of the chain. It has to be solved

358: with the boundary condition $a(n+N)=a(n)$.

359: 

360: For an open chain, the solution of the Schr\"odinger equation

361: (\ref{eq_a})  can be sought in the form of plane

362: waves propagating between the chain boundaries and the defect,

363: %

364: \begin{equation}

365: \label{left_right}

366: a (n) =

367:   C_{l,r}e^{i\theta n} + C'_{l,r} e^{- i\theta n}, \quad |n-n_0|\geq 1 .

368: \end{equation}

369: %

370: The subscripts $l$ and $r$ refer to the coefficients for the waves to

371: the left ($n<n_0$) and to the right ($n>n_0$) from the defect. The

372: interrelations between these coefficients and the coefficient $a(n_0)$

373: follow from the boundary conditions and from matching the solutions at

374: $n_0$. They are given by Eqs.~(\ref{bound_cond}) and

375: (\ref{boundary_defect}).

376: 

377: For a closed chain, on the other hand, the solution can be sought

378: in the form

379: %

380: \begin{eqnarray}

381: \label{right}

382: a(n) &=& Ce^{i\theta n} + C'e^{-i\theta n}, \quad n_0<n\leq

383: N, \\

384: %

385: a(n) &=& Ce^{i\theta (n+N)} + C'e^{-i\theta (n+N)}, \quad 1\leq n

386: < n_0.\nonumber

387: \end{eqnarray}

388: 

389: 

390: The energy $E_1$ as a function of $\theta$ can be obtained by

391: substituting Eqs.~(\ref{left_right}) and (\ref{right}) into

392: Eq.~(\ref{eq_a}). Both for the open and closed chains it has the form

393: %

394: \begin{equation}

395: E_1 = {\varepsilon}_{1} +

396: J \cos \theta .

397: \label{energy_one}

398: \end{equation}

399: 

400: The eigenfunctions (\ref{left_right}), (\ref{right}) with real

401: $\theta$ correspond to sinusoidal waves (extended states). From

402: (\ref{energy_one}), their energies lie within the band

403: ${\varepsilon}_{1} \pm J$. In contrast, localized states have complex

404: $\theta$, and their energies lie outside this band. The corresponding

405: solutions for both types of chains are discussed in the Appendix. In a

406: sufficiently long chain, there is always one localized one-excitation

407: state on a defect. Its energy is given by Eq.~(\ref{E_d}). In the case

408: of an open chain with the coupling anisotropy parameter $|\Delta|

409: > 1$, there are also localized states on the chain boundaries. Their

410: energy is given by Eq.~(\ref{E_s}).

411: 

412: 

413: 

414: \section{Two excitations: unbound, bound, and localized states}

415: 

416: A spin chain with a defect displays rich behavior in the presence

417: of two excitations. It is determined by the interplay between

418: disorder and inter-excitation coupling.  Solutions of the

419: two-excitation problem have been obtained in the case of a

420: disorder potential of several special types, where the system is

421: integrable \cite{integrable}. Here we will study the presumably

422: nonintegrable but physically interesting problem where the

423: on-site energy of the defect differs from that of the host sites.

424: 

425: The system is described by Hamiltonian (\ref{hamiltonian}). 

426: In order to concentrate on the effects of

427: disorder rather than boundaries, we will consider a closed chain of

428: length $N$. We will also assume that the anisotropy is strong,

429: $|\Delta| \gg 1$.

430: 

431: The wave function of a chain with two excitations is given by a linear

432: superposition

433: %

434: \begin{equation}

435: \label{two_exc}

436: \psi_2  = \sum_{n<m} a(n,m) \phi (n,m),

437: \end{equation}

438: %

439: where $\phi(n,m)$ is the state where spins on the sites $n$ and $m$

440: are pointing upward, whereas all other spins are pointing downward. In

441: a periodic chain, the sites with numbers that differ by $N$ are

442: identical, therefore we have $a(n,m) = a(m, n+N)$.

443: 

444: From Eq.~(\ref{hamiltonian}), the Schr\"odinger equation for the

445: coefficients $a(n,m)$  is

446: %

447: \begin{widetext}

448: \begin{eqnarray}

449: \label{a(n,m)}

450: &&\left( 2{\varepsilon}_{1} + g \delta_{n,n_0}+ g \delta_{m,n_0}

451: +J\Delta  \delta_{m,n+1}\right)a(n,m)

452: \nonumber \\

453: %

454:        && +\frac{1}{2}J \bigl[a(n-1,m) + [a(n+1,m) +

455: a(n,m-1)](1-\delta_{m,n+1}) + a(n,m+1)\bigr] =E_2a(n,m).

456: \end{eqnarray}

457: \end{widetext}

458: %

459: Here, $E_2$ is the energy of a two-excitation state [${\varepsilon}_1 =

460: \varepsilon - J\Delta$ is the on-site one-excitation energy,

461: cf. Eq.~(\ref{eq_a})]. As before, we assume that the defect is located

462: on site $n_0$.

463: 

464: \subsection{An ideal chain}

465: 

466: In the absence of a defect the system is integrable. The solution of

467: Eq.~(\ref{a(n,m)}) can be found using the Bethe ansatz \cite{Bethe},

468: %

469: \begin{equation}

470: \label{two_bethe_ideal}

471: a(n,m) = C e^{i (\theta_1 n+ \theta_2 m)} +

472: C' e^{i (\theta_2 n +  \theta_1 m)}.

473: \end{equation}

474: %

475: The energy of the state with given $\theta_1,\theta_2$ is obtained by

476: substituting Eq.~(\ref{two_bethe_ideal}) into Eq.~(\ref{a(n,m)})

477: written for $m>n+1$. This gives

478: %

479: \begin{equation}

480: \label{E2}

481: E_2 = 2{\varepsilon}_{1} + J (\cos \theta_1 + \cos \theta_2).

482: \end{equation}

483: 

484: By requiring that the ansatz (\ref{two_bethe_ideal}) apply also for

485: $m=n+1$, we obtain an interrelation between the coefficients $C$ and $C'$,

486: %

487: \begin{equation}

488: \label{CC'}

489: \frac{C}{C'} =

490: -\frac{1 - 2 \Delta  e^{i \theta_1 } + e^{i (\theta_1 + \theta_2)}}

491: {1 - 2 \Delta  e^{i \theta_2 } + e^{i (\theta_1 + \theta_2)}}.

492: \end{equation}

493: %

494: With account taken of normalization, Eqs.~(\ref{two_bethe_ideal}) and

495: (\ref{CC'}) fully determine the wave function.

496: The states with real $\theta_{1,2}$ form a two-magnon band with width

497: $4|J|$, as seen from the dispersion relation (\ref{E2}). The magnons

498: are not bound to each other and propagate independently.

499: 

500: For $|\Delta| > 1$, Eq.~(\ref{CC'}) also has a solution $C=0$, which

501: gives a complex phase $\theta_{2}= \theta_1^* = {1\over 2}\theta

502: -i\kappa$ with $\kappa > 0$. This solution corresponds to the wave

503: function $a(n,m)\propto \exp[i\theta(n+m)/2 - \kappa (m-n)]$;

504: we have $m>n$. From

505: Eqs.~(\ref{two_bethe_ideal})--(\ref{CC'}) we obtain

506: %

507: \begin{eqnarray}

508: &&e^{-\kappa}=\Delta^{-1}\cos(\theta/2),\quad E_{BP} = E_{BP}^{(0)}+

509: {J\over 2\Delta}\cos \theta,\nonumber\\

510: %

511: &&

512: E_{BP}^{(0)}= 2{\varepsilon}_1 + J\Delta + {J\over 2\Delta}.

513: \label{bound}

514: \end{eqnarray}

515: 

516: Eq.~(\ref{bound}) describes a bound pair of excitations. Such a

517: pair can freely propagate along the chain. The wave function is

518: maximal when the excitations are on neighboring sites.  The size of

519: the BP, i.e. the typical distance between the excitations, is

520: determined by the reciprocal decrement $\kappa^{-1}$, and ultimately

521: by the anisotropy parameter $\Delta$. For large $|\Delta|$, the

522: excitations in a BP are nearly completely bound to nearest sites. Then

523: the coefficients $a(n,m)\propto \delta_{n+1,m}$, to the lowest order

524: in $|\Delta|^{-1}$.

525: 

526: The distance between the centers of the BP band and the two-magnon

527: band $E_{BP}^{(0)}-2\varepsilon_1$ is given approximately by the BP

528: binding energy $J\Delta$. The width of the BP band $|J/\Delta|$ is

529: parametrically smaller than the width of the two-magnon band $4|J|$,

530: see Fig.~\ref{fig:energies}.

531: 

532: For nearest-neighbor coupling and for large $|\Delta|$, transport of

533: bound pairs can be visualized as occurring via an intermediate

534: step. First, one of the excitations in the pair makes a virtual

535: transition to the neighboring empty site, and as a result the parallel

536: spins in the pair are separated by one site. The corresponding state

537: differs in energy by $J\Delta$ from the bound-pair state. At the next

538: step the second spin can move next to the first, and then the whole

539: pair moves by one site. From perturbation theory, the bandwidth should

540: be $J^2/J\Delta\equiv J/\Delta$, which agrees with Eq.~(\ref{bound}).

541: 

542: The above arguments can be made quantitative by introducing an

543: effective Hamiltonian $\tilde H^{(t)} = \sum_{n=1}^{N} \tilde

544: H_{n}^{(t)} $ of BP's. It is obtained from the Hamiltonian

545: $H^{(t)}$ (\ref{transfer_one}) in the second order of perturbation

546: theory in the one-excitation hopping constant $J$,

547: %

548: \begin{eqnarray}

549: \label{transfer_bound}

550: \tilde H_n^{(t)}&=&\frac{J }{64 \Delta  }

551: [t^{(r)}_{n-1} t^{(l)}_{n-1} + t^{(l)}_{n+1} t^{(r)}_{n+1}]

552: \\

553: &+& \frac{J }{64 \Delta  }

554: [t^{(l)}_{n} t^{(l)}_{n-1} + t^{(r)}_{n} t^{(r)}_{n+1}].

555: \nonumber

556: \end{eqnarray}

557: %

558: The operators $t^{(r,l)}_n$ of excitation hopping to the right or

559: left are given by Eq.~(\ref{transfer_one}). The first pair of terms in

560: $\tilde H_n^{(t)}$ [Eq.~(\ref{transfer_bound})] describes virtual

561: transitions in which a BP dissociates and then recombines on the same

562: site. This leads to a shift of the on-site energy level of the BP by

563: $J/2\Delta$. The second pair of terms describes the motion of a BP as

564: a whole to the left or to the right.

565: 

566: The action of the Hamiltonian $\tilde H^{(t)}$ on the wave function

567: $a(n,n+1)$ is given by

568: %

569: \begin{eqnarray}

570: \label{transferBP}

571: &&\tilde H^{(t)}a(n,n+1)=\\

572: &&{J\over 2\Delta}a(n,n+1) + {J\over

573: 4\Delta}[a(n+1,n+2) + a(n-1,n)]. \nonumber

574: \end{eqnarray}

575: %

576: The Schr\"odinger equation for a BP is given by the sum of

577: the diagonal part (the first term) of Eq.~(\ref{a(n,m)}) with $m=n+1$

578: and the right-hand side of Eq.~(\ref{transferBP}).  In this

579: approximation a BP eigenfunction is $a(n,m)=\delta_{n+1,m}\exp(i\theta

580: n)$, and the dispersion law is of the form (\ref{bound}).

581: 

582: \subsection{Localized states in a chain with a defect}

583: 

584: We now consider excitations in the presence of a defect. In this section we

585: assume that the defect excess energy $g$ is such that

586: %

587: \begin{equation}

588: \label{nonresonant_cond}

589: |g|\gg |J|, \quad |J\Delta-g| \gg |J| .

590: \end{equation}

591: The first inequality guarantees that the localization length of an

592: excitation on the defect is small [its inverse Im~$\theta_d\approx

593: \ln 2|g/J| \gg 1$ cf. Eq.~(\ref{one_localized})].

594: 

595: The second condition in Eq.~(\ref{nonresonant_cond}) can be understood by

596: noticing that,

597: %, besides the bands of two-excitation states with unbound

598: %and bound excitations of an ideal chain,

599: in a chain with a defect, there is a two-excitation state where one

600: excitation is localized on the defect whereas the other is in an

601: extended magnon-type state. These excitations are not bound

602: together. The energy of such an unbound localized-delocalized pair

603: (LDP) should differ from the energy of unbound pair of magnons by

604: $\approx g$, and from the energy of a bound pair of magnons

605: (\ref{bound}) by $\approx J\Delta - g$. In this section we consider

606: the case where both these energy differences largely exceed the magnon

607: bandwidth $J$ (the case where $|J\Delta - g|\alt |J|$ will be

608: discussed in the following section).

609: 

610: \subsubsection{Unbound localized-delocalized pairs (LDP's)}

611: 

612: In the neglect of excitation hopping, the energy of an excitation

613: pair where one excitation is far from the defect ($|n-n_0|\gg 1$) and

614: the other is localized on the defect is $2{\varepsilon}_1 + g$. [This

615: can be seen from Eq.~(\ref{a(n,m)}) for the coefficients $a(n,m)$ in

616: which off-diagonal terms are disregarded.] At the same time, if one

617: excitation is on the defect and the other is on the neighboring site

618: $n_0\pm 1$, this energy becomes $2{\varepsilon}_1 + g + J\Delta$. The

619: energy difference $J\Delta$ largely exceeds the characteristic

620: bandwidth $J$. Therefore, if the excitation was initially far from the

621: defect, it will be reflected before it reaches the site $n_0\pm 1$.

622: 

623: The above arguments suggest to seek the solution for the wave function

624: of an LDP in a periodic chain in the form

625: %

626: \begin{equation}

627: \label{LDP_wave}

628: a(n_0,m)=Ce^{i\theta m} + C'e^{-i\theta m} ,

629: \end{equation}

630: %

631: with the boundary condition $a(n_0, n_0 +1)= a(n_0,n_0+N-1) = 0$. This

632: boundary condition and the form of the solution are similar to what

633: was used in the problem of one excitation in an open chain.

634: 

635: From Eq.~(\ref{a(n,m)}), the energy of an LDP is

636: %

637: \begin{equation}

638: \label{energy_LDP}

639: E_{LDP}\approx 2{\varepsilon}_1 + g + J\cos\theta .

640: \end{equation}

641: The wave number $\theta$ takes on $N-3$ values $\pi k/(N-2)$, with

642: $k=1,\ldots, N-3$.

643: As expected, the bandwidth of the LDP is $2|J|$, as in the case of

644: one-excitation band in an ideal chain.

645: 

646: We have compared Eq.~(\ref{energy_LDP}) with numerical results

647: obtained by direct solution of the eigenvalue problem

648: (\ref{a(n,m)}). For $N=100, \Delta = 20, g/J=10$ we obtained excellent

649: agreement once we took into account that the energy levels

650: (\ref{energy_LDP}) are additionally shifted by $J^2/2g$. This shift

651: can be readily obtained from Eq.~(\ref{a(n,m)}) as the second-order

652: correction (in $J$) to the energy $E_d$ of the excitation localized on

653: the defect. It follows also from Eq.~(\ref{E_d}) for $|g|\gg |J|$.

654: 

655: We note that the result is trivially generalized to the case of a

656: finite but small density of magnons. The wave function $a(n_0,

657: m_1,m_2,\ldots,m_M)$ of $M$ uncoupled magnons and an excitation on the

658: site $n_0$ is given by a sum of the appropriately weighted

659: permutations of $\exp(\pm i\theta_1m_1 \pm i\theta_2m_2 \ldots \pm

660: i\theta_Mm_M)$ over the site numbers $m_i$ (these numbers can be

661: arranged so that $m_1<\ldots <m_k<n_0<m_{k+1}<\ldots < m_M$), with

662: real $\theta_i$. The weighting factors for large $|\Delta|$ are found

663: from the boundary conditions and the condition that $a(n_0,

664: m_1,m_2,\ldots)=0$ whenever any two numbers $m_i, m_{i+1}$ differ by

665: 1. When the ratio of the number of excitations $M$ to the chain length

666: $N$ is small, the energy is just a sum of the energies of uncoupled

667: magnons and the localized excitation, i.e., it is

668: $(M+1){\varepsilon}_1 + g + J\sum\nolimits_{i=1}^M\cos\theta_i$. For

669: small $M/N$, scattering of a magnon by the excitation on a defect

670: occurs as if there were no other magnons, i.e., the probability of a

671: three-particle collision is negligibly small.

672: 

673: 

674: \subsubsection{Bound pairs localized on the defect: the doublet}

675: 

676: For large $|\Delta|,|g/J|$, a bound pair of neighboring excitations

677: should be strongly localized when one of the excitations is on the

678: defect site. Indeed, if we disregard intersite excitation hopping,

679: the energy of a BP sitting on the defect is

680: $E_D^{(0)}=2{\varepsilon}_1+ g+ J\Delta $. It differs significantly

681: from the energy of freely propagating BP's (\ref{bound}), causing

682: localization.

683: 

684: The major effect of the excitation hopping is that the pair can make

685: resonant transitions between the sites $(n_0, n_0+1)$ and $(n_0-1,

686: n_0)$. Such transitions lead to splitting of the energy level of the

687: pair into a doublet. To second order in $J$ the energies of the

688: resulting symmetric and antisymmetric states are

689: %

690: \begin{equation}

691: \label{doublet}

692: E_{D}^{(\pm)} = E_D^{(0)} + \frac{J(2J\Delta +g)}{4\Delta (J\Delta +g)}

693:  \pm \frac{J^2}{4 (J\Delta + g)} .

694: \end{equation}

695: %

696: The energy splitting between the states is small if $J\Delta$ and $g$

697: have same sign. If on the other hand, $|J\Delta +g| \alt |J|$, the

698: theory has to be modified. Here, the bound pairs with one excitation

699: localized on the defect are resonantly mixed with extended unbound

700: two-magnon states. We do not consider this case in the present paper.

701: 

702: We note that, in terms of quantum computing, the onset of a doublet

703: suggests a simple way of creating entangled states. Indeed, by

704: applying a $\pi$ pulse at frequency $\varepsilon_1+g$ one can

705: selectively excite the qubit $n_0$. If then a $\pi$ pulse is applied

706: at the frequency $E_D^{+}-(\varepsilon_1+g)$, it will selectively

707: excite a Bell state $\left[|01\rangle + |10\rangle\right]/\sqrt{2}$,

708: where $|ij\rangle$ describes the state where the qubits on sites

709: $n_0-1$ and $n_0+1$ are in the states $|i\rangle$ and $|j\rangle$,

710: respectively. The excitations on sites $n_0\pm 1$ can then be

711: separated without breaking their entanglement using two-qubit

712: gate operations.

713: 

714: 

715: \subsubsection{Localized states split off the bound-pair band}

716: 

717: The energy difference $E_D^{(\pm)}-E_{BP}\approx g$ between BP states

718: on the defect site (\ref{doublet}) and extended BP states largely

719: exceeds the bandwidth $|J/\Delta|$ of the extended states. Therefore

720: it is a good approximation to assume that the wave functions of

721: extended BP states are equal to zero on the defect. In other words,

722: such BP's are reflected {\it before} they reach the defect. In this

723: sense, the defect acts as a boundary for them. One may expect that

724: there is a surface-type state associated with this boundary.

725: 

726: The emergence of the surface-type state is facilitated by the

727: defect-induced change of the on-site energy of a BP located next to

728: the defect on the sites $(n_0+1, n_0+2)$ [or $(n_0-2,n_0-1)$]. This

729: change arises because virtual dissociation of a BP with one excitation

730: hopping onto a defect site gives a different energy denominator

731: compared to the case where a virtual transition is made onto a regular

732: site. It is described by an extra term $\delta E_{BP}$ in the

733: expression (\ref{transfer_bound}) for the diagonal part of the

734: Hamiltonian $\tilde H_n^{(t)}$ with $n=n_0+1$ [or $n=n_0-2$],

735: %

736: \begin{equation}

737: \label{delta_BP}

738: \delta E_{BP}= \frac{Jg}{4\Delta(J\Delta - g)}.

739: \end{equation}

740: 

741: Using the transformed Hamiltonian (\ref{transferBP}), one can analyze

742: BP states in a way similar to the analysis of one-excitation states in

743: an open chain, see the Appendix. The Schr\"odinger equation for BP states

744: away from the defect is given by the first term in Eq.~(\ref{a(n,m)})

745: and by Eqs.~(\ref{transferBP}) and (\ref{delta_BP}). The BP wave functions

746: can be sought in the form

747: %

748: \begin{equation}

749: \label{BPfull}

750: a(n,n+1) = C e^{i\theta n} + C' e^{-i \theta n}.

751: \end{equation}

752: %

753: Then the BP energy as a function of $\theta$ is given by $E_{BP}=

754: E_{BP}^{(0)} + (J/2\Delta)\cos\theta$, cf. Eq.~(\ref{bound}).

755: 

756: The values of $\theta$ can be found from the boundary condition

757: that the BP wave function is equal to zero on the defect, i.e.,

758: $a(n_0,n_0+1) = a(n_0-1,n_0) \equiv a(n_0+N-1,n_0+N)= 0$. From the

759: Schr\"odinger equations for $a(n,n+1)$ with $n=n_0+1$ and $n=N+n_0-2$,

760: we obtain an equation for $\theta$ of the form

761: %

762: \begin{eqnarray}

763: \label{theta_BP}

764: f(\theta ) = f(-\theta ),\; f(\theta ) = \left[\delta E_{BP} -

765: {J\over 4 \Delta} e^{i \theta }\right]^2e^{i\theta (N-3)}.

766: \end{eqnarray}

767: 

768: Equation (\ref{theta_BP}) has $2(N-1)$ solutions for $\exp(i\theta)$. The

769: solutions $\exp(i\theta)=\pm 1$ are spurious, in the general case. The

770: roots $\theta$ and $-\theta$ describe one and the same wave

771: function. Therefore there are $N-2$ physically distinct roots

772: $\theta$, as expected for an $N$-spin chain with two excluded BP

773: states located at $(n_0, n_0\pm 1)$.

774: 

775: Depending on the ratio $q=J/(4\Delta\,\delta E_{BP})\equiv (J\Delta -

776: g)/g$, the roots $\theta$ are either all real or there is one or two

777: pairs of complex roots with opposite signs. Real roots correspond to

778: extended states, whereas complex roots correspond to the states that

779: decay away from the defect. The onset of complex roots can be analyzed

780: in the same way as described in the Appendix for one excitation. There is

781: much similarity, formally and physically, between the onset of

782: localized BP states next to the defect and the onset of surface states

783: at the edge of an open chain.

784: 

785: We

786: rewrite Eq.~(\ref{theta_BP}) in the form

787: %

788: \begin{equation}

789: \label{tangent_BP}

790: \tan[\theta(N-1)]={2\sin\theta \,(\cos \theta - q)\over (\cos\theta -

791: q)^2 - \sin^2\theta}.

792: \end{equation}

793: %

794: For $|q| > 1$ all roots of Eq.~(\ref{tangent_BP}) are real. At $|q|=1$

795: there occurs a bifurcation where two real roots with opposite signs

796: merge (at $\theta = 0$, for $q=1$, or at $\theta = \pi$, for

797: $q=-1$). They become complex for $|q| < 1$. For $|q| = 1-2(N-1)^{-1}$

798: two other real roots coalesce at $\theta = 0$ or $\pi$ and another

799: pair of complex roots emerges.

800: 

801: As the length of the chain increases, the difference between the pairs

802: of complex roots decreases. In the limit $N\to \infty$ the roots merge

803: pairwise. The imaginary part of one of the roots is%

804: \begin{equation}

805: \label{BP_s}

806: {\rm Im}~\theta_{BP}^{(s)}= \ln |\delta E_{BP}/(J/4\Delta )|.

807: \end{equation}

808: %

809: The second root has opposite sign.

810: 

811: 

812: One can show from the Schr\"odinger equation for the BP's that the

813: solution (\ref{BP_s}) describes a ``surface-type'' BP state localized

814: next to the defect on sites $(n_0+1, n_0+2)$. The solution with

815: $-\theta_{BP}^{(s)}$ describes the surface state on

816: $(n_0-2,n_0-1)$. The amplitudes of these states exponentially decay

817: away from the defect. For $J\Delta >g> J\Delta/2 > 0$ or $J\Delta <g<

818: J\Delta/2 < 0$ we have Re~$\theta_{BP}^{(s)}=0$.  For $g> J\Delta > 0$

819: or $g< J\Delta < 0$, we have Re~$\theta_{BP}^{(s)} =\pi$, and decay of

820: the localized state is accompanied by oscillations.  The complex roots

821: in a finite chain can be pictured as describing those same states on

822: the opposite sites of the defect. But now the states are

823: ``tunnel'' split because of the overlap of their tails inside the

824: chain, which leads to the onset of two slightly different localization

825: lengths.

826: 

827: The energy of the localized state in a long chain is

828: %

829: \begin{equation}

830: \label{BP_s_energy}

831: E_{BP}^{(s)}= E_{BP}^{(0)}+ \delta E_{BP} +

832: \frac{(J/4\Delta )^2}{\delta E_{BP}}.

833: \end{equation}

834: %

835: It lies outside the band $ E_{BP}^{(0)}\pm (J/2\Delta)$ (\ref{bound})

836: of the extended BP states. The distance to the band edge strongly

837: depends on the interrelation between the defect excess energy $g$ and

838: the BP binding energy $J\Delta$. It is of the order of the BP

839: bandwidth $|J/\Delta|$, except for the range where the difference

840: between $J\Delta$ and $g$ becomes small. In this range the energy of

841: the surface-type state sharply increases in the absolute value. This

842: is illustrated in Fig.~\ref{fig:BP_s}. The localization length $|{\rm

843: Im}~\theta_{BP}^{(s)}|^{-1}$ is large when the state energy is close

844: to the band edge and shrinks down with decreasing $|J\Delta -g|$,

845: i.e., with increasing $|\delta E_{BP}|$.

846: 

847: 

848: \begin{figure}[ht]

849: \includegraphics[width=2.8in]{fig2_XXZ.eps}

850: \caption{The distance between the energy level of the BP localized

851: next to the defect and the center of the band of extended BP states,

852: $\varepsilon_{BP} = (E_{BP}^{(s)} - E_{BP}^{(0)})/(J/4\Delta) $, vs

853: the scaled defect excess energy $g/J\Delta$. The localized state

854: exists in an infinite chain for $g/J\Delta> 1/2$. The vertical and

855: horizontal dot-dashed lines show the asymptotes for $g\to J\Delta$ and

856: $g/J\Delta\to\infty$, respectively. The results refer to the range

857: where the BP and LDP bands are far from each other, compared to the

858: LDP bandwidth $|J|$. These bands are sketched in the inset (the BP band

859: is above the LDP band for $J\Delta/g > 1$). The dashed line in the

860: inset shows where the energy level of the localized BP state is

861: located with respect to the bands.}

862: \label{fig:BP_s}

863: \end{figure}

864: 

865: 

866: \section{Antiresonant decoupling of two-excitation states}

867: 

868: The analysis of the preceding section does not apply if the pair

869: binding energy $J\Delta$ is close to the defect excess energy $g$.

870: When $|g-J\Delta|$ is of the order of the LDP bandwidth $|J|$, the BP

871: and LDP states are in resonance, their bands overlap or nearly overlap

872: with each other. One might expect that there would occur mixing of

873: states of these two bands. In other words, a delocalized magnon in the

874: LDP band might be scattered off the excitation on the defect, and as a

875: result they both would move away as a bound pair. However, as we show,

876: such mixing does not happen.  In order to simplify notations we will

877: assume in what follows that $g,J,\Delta >0$.

878: 

879: 

880: \subsection{A bound pair localized next to the defect}

881: 

882: In the resonant region we should reconsider the analysis of the

883: next-to-the-defect bound pair localized on sites $(n_0+1, n_0+2)$ or

884: $(n_0-2,n_0-1)$. As $|g-J\Delta|$ decreases, the energy of this pair

885: (\ref{BP_s_energy}) moves away from the BP band, see

886: Fig.~\ref{fig:BP_s}. At the same time, the distance between the pair

887: energy and the LDP band $E_{BP}^{(s)}- 2\varepsilon_1-g$ becomes {\it

888: smaller} with decreasing $|g-J\Delta|$ as long as $|J\Delta - g|>

889: |J|/2$. For $|J\Delta - g|\sim J$, the next-to-the-defect pairs are

890: hybridized with LDP's. The hybridization occurs in first order in the

891: nearest-neighbor coupling constant $J$.

892: 

893: To describe the hybridization, we will seek

894: the solution of the Schr\"odinger equation (\ref{a(n,m)}) in the form

895: of a linear superposition of an LDP state (\ref{LDP_wave}) and a

896: pair on the next to the defect sites,

897: %

898: \begin{eqnarray}

899: \label{hybridization}

900: a(n,m)&=& (Ce^{i\theta m} +C'e^{-i\theta m}) \delta_{n,n_0} \nonumber\\ &+&

901: a(n_0+1,n_0+2)  \delta_{n,n_0+1}\delta_{m,n_0+2} \nonumber\\

902: &+&

903: a(n_0-2,n_0-1) \delta_{n,n_0-2}\delta_{m,n_0-1}

904: \end{eqnarray}

905: (we remind that $m > n$). The energy of a state with given $\theta$

906: can be found from (\ref{a(n,m)}) as before by considering $m$ far away

907: from $n_0$. It is given by Eq.~(\ref{energy_LDP}).

908: 

909: The interrelation between the coefficients $C, C', a(n_0+ 1, n_0+2)$,

910: and $a(n_0-2, n_0-1) $, as well as the values of $\theta$ should be

911: obtained from the boundary conditions. These conditions follow from

912: the fact that the energy of a pair on sites $(n_0, n_0\pm 1)$ and the

913: energies of the pairs described by Eq.(\ref{hybridization}) differ by

914: $\sim g\approx J\Delta$, cf. (\ref{doublet}). Therefore the pairs on

915: sites $(n_0, n_0\pm 1)$ are decoupled from the states

916: (\ref{hybridization}), and in the analysis of the LDP's we can set

917: $a(n_0, n_0\pm 1)=0$. Decoupled also are unbound two-excitation states

918: with no excitation on the defect, i.e., two-magnon states. Therefore

919: $a(n,m)=0$ if simultaneously $m-n>1$ and $(n-n_0)(m-n_0)\neq 0$.

920: 

921: From Eq.~(\ref{a(n,m)}) written for $n=n_0,m=n_0+ 2$ and

922: $n=n_0-2,m=n_0$ [with account taken of the relation $a(n,m)=a(m,n+N)$]

923: we obtain

924: %

925: \begin{eqnarray}

926: \label{hybr_interrelation}

927: a(n_0+1, n_0+2)&=& Ce^{i\theta (n_0+1)} +C'e^{-i\theta (n_0+1)},\\

928: a(n_0-2, n_0-1)&=& Ce^{i\theta (n_0+N-1)} +C'e^{-i\theta (n_0+N-1)}\nonumber .

929: \end{eqnarray}

930: 

931: The interrelation between $C$ and $C'$ and the equation for $\theta$

932: follow from Eqs.~(\ref{hybridization}), (\ref{hybr_interrelation}) 

933: and (\ref{a(n,m)}) written

934: for $n=n_0+1,m=n_0+ 2$ and $n=n_0-2,m=n_0-1$. They have the form

935: %

936: \begin{equation}

937: \label{hybr_cc'}

938: C' = -\frac{2(J \Delta -g) e^{i\theta } -J}

939: {2(J \Delta-g) e^{-i\theta } -J} Ce^{2i\theta n_0},

940: \end{equation}

941: %

942: and

943: %

944: \begin{eqnarray}

945: \label{hybr_theta}

946: &\tilde f(\theta)=\tilde f(-\theta),\\

947: &\tilde f(\theta)=e^{i \theta N} \left[2(J \Delta-g)e^{-i\theta} -

948: J \right]^2.\nonumber

949: \end{eqnarray}

950: 

951: Equation (\ref{hybr_theta}) is an $2N$th order equation for

952: $\exp(i\theta)$. Its analysis is completely analogous to that of

953: Eq.~(\ref{theta_BP}). The roots $\theta=0,\pi$ are spurious, and the

954: roots $\theta$ of opposite signs describe one and the same wave

955: function. Therefore Eq.~(\ref{hybr_theta}) has $N-1$ physically

956: distinct roots $\theta$. Real $\theta$'s correspond to extended

957: states. Complex roots appear for $2|J \Delta-g|

958: >J$. These roots, $\theta_{LDP}^{(s)}$, describe localized states. In

959: the limit of a long chain, $N\to \infty$,

960: the imaginary part of one of them is

961: %

962: \begin{equation}

963: \label{hybr_decrem}

964: {\rm Im}\, \theta_{LDP}^{(s)} =  \ln |2(J\Delta - g)/J|,

965: \end{equation}

966: whereas the other root has just opposite sign.

967: 

968: The wave function of the localized state is maximal either on sites

969: $(n_0+1, n_0+2)$ or $(n_0-2,n_0-1)$ and exponentially decays into the

970: chain. For $J\Delta - g < 0$ this decay is accompanied by

971: oscillations, Re~$\theta_{LDP}^{(s)}=\pi$.  The energy of the

972: localized state is

973: %

974: \begin{equation}

975: \label{hybr_s_energy}

976: E_{LDP}^{(s)} =  2{\varepsilon}_1 + J \Delta  + \frac{(J/2)^2}{J\Delta - g}.

977: \end{equation}

978: %

979: 

980: The localized state (\ref{hybridization}), (\ref{hybr_decrem}) is a

981: ``surface-type'' state induced by the defect. It is the resonant-region

982: analog of the localized next-to-the-defect state discussed in

983: Section~III.B3. The wave function of the latter state (\ref{BPfull}),

984: (\ref{BP_s}) was a linear combination of the wave functions of bound

985: pairs. In contrast, the state given by Eqs.~(\ref{hybridization}),

986: (\ref{hybr_decrem}) is a combination of the wave functions of the bound

987: pair located next to the defect and a localized-delocalized pair.

988: 

989: The evolution of the surface-type state is controlled by the

990: difference between the excess energies of binding two excitations in a

991: pair or localizing one of them on the defect $|J\Delta - g|$. As

992: $|J\Delta - g|$ varies, the state changes in the following way. It

993: first splits off the BP band when $|J\Delta - g|$ becomes less than

994: $g$, see Fig.~\ref{fig:BP_s}. Its energy moves away from the band of

995: extended BP states with decreasing $|J\Delta - g|$ and the

996: localization length decreases [cf. Eq.~(\ref{BP_s})]. Well before

997: $|J\Delta - g|$ becomes of order $J$, the state becomes strongly

998: localized on the sites $(n_0+1, n_0+2)$ or $(n_0-2, n_0-1)$.

999: 

1000: In the region $|J\Delta - g|\sim J$ the localized state becomes

1001: stronger hybridized with LDP states than with extended BP states. This

1002: hybridization occurs in first order in $J$, via a transition

1003: $(n_0+1,n_0+2) \to (n_0, n_0+2)$ [or $(n_0-2,n_0-1) \to

1004: (n_0-2,n_0)$]. In this region the localization length increases with

1005: decreasing $|J\Delta - g|$, cf. Eq.~(\ref{hybr_decrem}). Ultimately,

1006: for $|J\Delta - g|=J/2$ the localized surface-type state disappears,

1007: as seen in Fig.~\ref{fig:LDP}. The evolution of the energy of the

1008: localized state with $J\Delta - g$ is shown in Fig.~\ref{fig:LDP}.

1009: 

1010: \begin{figure}[ht]

1011: \includegraphics[width=2.8in]{fig3_XXZ.eps}

1012: \caption{The energy difference between the localized BP state and the

1013: LDP band, $\varepsilon_{LDP} = (E_{LDP}^{(s)} - 2{\varepsilon}_1 -

1014: g)/(J/2)$, vs the energy mismatch $2(J\Delta -g)/J$.

1015: In the region between the dot-dashed lines the BP state is

1016: delocalized. The results refer to the case where the BP and LDP bands

1017: are close to each other or overlap, as sketched in the inset (the BP band

1018: is above the LDP band for $J\Delta > g$).  The dashed line in the

1019: inset shows where the energy level of the localized BP

1020: state is located with respect to the bands.}

1021: \label{fig:LDP}

1022: \end{figure}

1023: 

1024: 

1025: 

1026: 

1027: The crossover from hybridization of the localized surface-type BP

1028: state with extended BP states to that with LDP states occurs in the

1029: region $|g|\gg |J\Delta - g| \gg |J|$. It is described using a

1030: different approach in Ref.~\onlinecite{us_DS}. We note that the

1031: expressions for the energy of the localized states (\ref{BP_s_energy})

1032: and (\ref{hybr_s_energy}) go over into each other for $|J\Delta -

1033: g|\ll |g|$.

1034: 

1035: \subsection{Decoupling of  bound pairs and LDP's

1036: in the resonant region}

1037: 

1038: We are now in a position to consider resonant coupling between

1039: extended states of bound pairs and localized-delocalized pairs. To

1040: lowest order in $\Delta^{-1}$, a BP on sites $(n, n+1)$ far away from

1041: the defect can resonantly hop only to the nearest pair of sites, see

1042: Eqs.~(\ref{transfer_bound}), (\ref{transferBP}). As noted before, the

1043: hopping requires an intermediate virtual transition of the BP into a

1044: dissociated state, which differs in energy by $ J\Delta$.

1045: 

1046: A different situation occurs for the BP on the sites $(n_0+2,n_0+3)$

1047: [or $(n_0-3,n_0-2)$]. Such BP can hop onto the sites $(n_0+1,n_0+2)$

1048: and $(n_0+3,n_0+4)$, as described by (\ref{transferBP}). But in

1049: addition, for $|J\Delta - g|\alt J$ it can make a

1050: transition into the LDP state on sites $(n_0,n_0+3)$

1051: [or $(n_0-3,n_0)$] . Indeed, such

1052: state has the energy $\approx 2{ \varepsilon}_1+g$, which is

1053: close to the BP energy $\approx 2{ \varepsilon}_1+J\Delta$.

1054: 

1055: The transition $(n_0+2,n_0+3) \to (n_0,n_0+3)$ goes through the

1056: intermediate dissociated state $(n_0+1,n_0+3)$, which differs in

1057: energy by $\approx J\Delta$.  It can be taken into account by adding

1058: the term $\delta\tilde H^{(t)}$ to the BP hopping Hamiltonian

1059: (\ref{transferBP}) for the sites $(n_0+2,n_0+3)$,

1060: %

1061: \begin{equation}

1062: \label{extra_H_t}

1063: \delta \tilde H^{(t)}a(n_0+2,n_0+3) = (J/4\Delta)a(n_0,n_0+3).

1064: \end{equation}

1065: 

1066: Extended BP states are connected to LDP states only through a BP on

1067: the sites $(n_0+2,n_0+3)$. We are now in a position to analyze this

1068: connection.  From Eqs.~(\ref{transferBP}) and (\ref{extra_H_t}), we see

1069: that

1070: %

1071: \begin{eqnarray}

1072: \label{coupling_resonant}

1073: &&[\tilde H^{(t)} + \delta \tilde H^{(t)}]a(n_0+2,n_0+3) =

1074: (J/4\Delta)\nonumber\\

1075: &&\times[a(n_0,n_0+3) + a(n_0+1,n_0+2)] + A,

1076: \end{eqnarray}

1077: %

1078: where $A$ is a linear combination of the amplitudes $a(n_0+2,n_0+3)$

1079: and $a(n_0+3,n_0+4)$.

1080: 

1081: The sum $a(n_0,n_0+3)+ a(n_0+1,n_0+2)$ in

1082: Eq.~(\ref{coupling_resonant}) can be expressed in terms of the LDP

1083: wave functions (\ref{hybridization}). With account taken of the

1084: interrelation (\ref{hybr_cc'}) between the coefficients $C,C'$ in

1085: Eq.~(\ref{hybridization}), we have

1086: %

1087: \begin{eqnarray}

1088: \label{amplitude_resonant}

1089: &&a(n_0+1,n_0+2)+a(n_0,n_0+3)\nonumber\\

1090: &&\propto C \sin \theta \cos \theta [J\Delta -g -

1091: J \cos \theta ].

1092: \end{eqnarray}

1093: %

1094: Equations (\ref{coupling_resonant}), (\ref{amplitude_resonant}) describe

1095: the coupling between the BP on the sites $(n_0+2,n_0+3)$ and the LDP

1096: eigenstates with given $\theta$.

1097: 

1098: An important conclusion can now be drawn regarding the behavior of BP

1099: and LDP states in the resonant region. The center of the BP band lies

1100: at $2{\varepsilon}_1 + J\Delta +(J/2\Delta)$ (\ref{bound}), and the BP

1101: band is parametrically narrower than the LDP band $2{\varepsilon}_1+g

1102: +J\cos\theta$ (\ref{energy_LDP}).  When the BP band is inside the LDP

1103: band, this means that, for an appropriate wave number $\theta$ of the

1104: LDP magnon, $J\Delta = g+J\cos\theta$, to zeroth order in

1105: $\Delta^{-1}$ (this is the approximation used to obtain the LDP

1106: dispersion law). It follows from Eqs.~(\ref{coupling_resonant}) and

1107: (\ref{amplitude_resonant}) that, for such $J\Delta -g$ and $\theta$

1108: there is no coupling between the LDP and extended BP states.

1109: 

1110: The above result means that LDP and extended BP states do not experience

1111: resonant scattering into each other, even though it is allowed by the

1112: energy-conservation law. Such scattering would correspond to the

1113: scattering of a magnon off the excitation localized on the

1114: defect, with both of them becoming a bound pair that moves away from

1115: the defect, or an inverse process.

1116: 

1117: Physically, the antiresonant decoupling of BP and LDP states is a

1118: result of strong mixing of a BP on the sites $(n_0+1,n_0+2)$ and

1119: LDP's. Because of the mixing, the amplitudes of transitions of extended

1120: BP states to the sites $(n_0+1,n_0+2)$ and $(n_0,n_0+3)$ compensate

1121: each other, to lowest order in $\Delta^{-1}$.

1122: 

1123: To illustrate the antiresonant decoupling we show in

1124: Fig.~\ref{fig:temporal_nonresonant} two types of time evolution of

1125: excitation pairs. In both cases the initial state of the system was

1126: chosen as a pair on the sites $(n_0+1,n_0+2)$, i.e., $a(n_0+1,n_0+2)=1$

1127: for $t=0$. The time-dependent Schr\"odinger equation was then solved

1128: with the boundary condition that corresponds to a closed chain.

1129: 

1130: The solid lines refer to the case of nonoverlapping BP and LDP bands,

1131: $|J|\ll |g-J\Delta|$ and $|g|<|J\Delta|/2$.  In this case there does

1132: not emerge a localized surface-type BP state next to the

1133: defect. Therefore an excitation pair placed initially on the sites

1134: $(n_0+1,n_0+2)$ resonantly transforms into extended BP states and

1135: propagates through the chain. This propagation is seen from the figure

1136: as oscillations of the return probability and the probability to find

1137: the BP on another arbitrarily chosen pair of neighboring sites

1138: $(n_0+2,n_0+3)$.  The oscillation rate should be small, of the order

1139: of the bandwidth $J/\Delta$. This estimate agrees with the numerical

1140: data. It is seen that the BP state is not transformed into LDP

1141: states. The amplitude of LDP states on sites $(n_0,m\geq n_0+2)$

1142: remains extremely small, as illustrated for $m=n_0+2$.

1143: 

1144: The dotted lines in Fig.~\ref{fig:temporal_nonresonant} show a

1145: completely different picture which arises when the BP band is inside

1146: the LDP band. In this case an excitation pair placed initially on

1147: $(n_0+1,n_0+2)$ hybridizes with LDP rather than BP states. A

1148: transformation of the pair $(n_0+1,n_0+2)$ into LDP's with increasing

1149: time is clearly seen. The period of oscillations is of the order of

1150: the reciprocal bandwidth of the LDP's $J^{-1}$, it is much shorter than in

1151: the previous case.  Remarkably, as a consequence of the antiresonance,

1152: extended states of bound pairs are not excited to any appreciable

1153: extent, as seen from the amplitude of the pair on the sites

1154: $(n_0+2,n_0+3)$.

1155: 

1156: \begin{figure}[ht]

1157: \includegraphics[width=3.2in]{fig4_XXZ.eps}

1158: \caption{Time evolution of a two-excitation wave packet in an $XXZ$

1159: chain with a defect; $|a(n,m)|^2$ is the occupation of sites

1160: $(n,m)$. Initially an excitation pair is placed next to the defect on

1161: sites $(n_0+1,n_0+2)$. The solid and dashed lines refer to the cases

1162: where the BP and LDP bands are, respectively, far away from each other

1163: ($g=J\Delta/4$) and overlapping ($g=J\Delta$). In the first case a

1164: bound pair slowly oscillates between neighboring sites $(n,n+1)$ and

1165: does not dissociate [$a(n_0,n_0+2)$ remains very small]. In the case

1166: of overlapping bands, the pair at $(n_0+1,n_0+2)$ is hybridized with

1167: LDP's, but practically does not mix with bound pairs on other sites

1168: [$a(n_0+2,n_0+3)$ remains very small]. The results refer to a ten-site

1169: closed chain with $\Delta =10$.}

1170: \label{fig:temporal_nonresonant}

1171: \end{figure}

1172: 

1173: 

1174: \section{Conclusions}

1175: 

1176: We have analyzed the dynamics of a disordered spin chain with a

1177: strongly anisotropic coupling in a magnetic field. A defect in such a

1178: chain can lead to several localized states, depending on the number of

1179: excitations. This is a consequence of the interaction between

1180: excitations and its interplay with the disorder. We have studied

1181: chains with one and two excitations.

1182: 

1183: The major results refer to the case of two excitations. Here, the

1184: physics is determined by the interrelation between the excess on-site

1185: energy of the defect $g$ and the anisotropic part of the exchange

1186: coupling $J\Delta$.  Strong anisotropy leads to binding of excitations

1187: into nearest-neighbor pairs that freely propagate in an ideal

1188: chain. Because of the defect, BP's can localize. A simple

1189: type of a localized BP is a pair with one of the excitations located

1190: on a defect.  A less obvious localized state corresponds to a pair

1191: localized next to a defect. It reminds of a surface state split off

1192: from the band of extended BP states, with the surface being the defect

1193: site.  We specified the conditions where the localization occurs and

1194: found the characteristics of the localized states.

1195: 

1196: Our most unexpected observation is the antiresonant decoupling of 

1197: extended BP states from localized-delocalized pairs. The LDP's are

1198: formed by one excitation on the defect site and another in an extended

1199: state. The antiresonance occurs for $g\approx J\Delta$, when the BP and

1200: LDP bands overlap. It results from destructive quantum interference of

1201: the amplitudes of transitions of BP's into two types of resonant

1202: two-excitation states: one is an LDP, and the other is an excitation

1203: pair on the sites next to the defect. As a result of the

1204: antiresonant decoupling, extended BP's and LDP's do not \

1205: scatter into each other, even though the

1206: scattering is allowed by energy conservation. This means that an

1207: excitation localized on the defect does not delocalize as a result of

1208: coupling to other excitations.

1209: 

1210: The occurrence of multiple localized states in the presence of other

1211: excitations is important for quantum computing.  It shows that, even

1212: where the interaction between the qubits is ``on'' all the time, we

1213: may still have well-defined states of individual qubits that can be

1214: addressed and controlled. One can prepare entangled localized pairs of

1215: excitations, as we discussed in Sec.~III~B~2, or more complicated

1216: entangled excitation complexes.  The results of the paper also provide

1217: an example of new many-body effects that can be studied using quantum

1218: computers with individually controlled qubit transition energies.

1219: 

1220: \begin{acknowledgments}

1221: This research was supported in part by the NSF through Grant

1222: No. ITR-0085922 and by the Institute for Quantum Sciences at Michigan

1223: State University.

1224: \end{acknowledgments}

1225: 

1226: 

1227: \appendix

1228: 

1229: \section{One excitation}

1230: 

1231: In an infinite spin chain in a magnetic field, the anisotropy of the

1232: spin-spin interaction does not affect the spectrum and wave functions

1233: of one excitation.  The matrix element of the term

1234: $\sum_n\sigma_n^z\sigma_{n+1}^z$ in the Hamiltonian

1235: (\ref{hamiltonian}) is just a constant. However, the situation becomes

1236: different for a chain of finite length, because the coupling

1237: anisotropy can lead to surface states. In the case of two excitations,

1238: analogs of surface-type states emerge near defects 

1239: in an infinite chain, as discussed in Sec.~III. 

1240: Here, for completeness and keeping

1241: in mind a reader with the background in quantum computing, we briefly

1242: outline the results of the standard analysis of a finite-length spin

1243: chain with one excitation.

1244: 

1245: \subsection{Localized surface and defect-induced states in an open chain}

1246: 

1247: The Schr\"odinger equation for an excitation in an open chain has the

1248: form (\ref{eq_a}), and its solution $a(n)$ to the left and to the

1249: right from the defect can be written in the form of a superposition of

1250: counterpropagating plane waves, Eq.~(\ref{left_right}). The relation

1251: between the amplitudes of these waves $C'_{l,r}$ and $C_{l,r}$ follows

1252: from the boundary conditions $a(0)=a(N+1)=0$. By substituting

1253: Eq.~(\ref{left_right}) into Eq.~(\ref{eq_a}) with $n=1$ and $n=N$, we

1254: obtain

1255: %

1256: \begin{eqnarray}

1257: \label{bound_cond}

1258: &&C'_l = - D C_l, \quad C'_r = -D^{-1} e^{2i\theta (N+1)} C_r,\\

1259: &&D = [1-\Delta \exp (i\theta )]/[1-\Delta \exp (-i \theta )].\nonumber

1260: \end{eqnarray}

1261: %

1262: 

1263: The relations between $C'_{l,r}$, $C_{l,r}$, and the amplitude of the

1264: wave function on the defect site $a(n_0)$ follow from

1265: Eqs.~(\ref{eq_a}) and (\ref{left_right}) for $n=n_0\pm 1$,

1266: %

1267: \begin{eqnarray}

1268: \label{boundary_defect}

1269: a(n_0)&&= C_{l} e^{i\theta n_0}\left[1 - D e^{-2i\theta n_0}\right]\nonumber\\

1270: &&= C_{r} e^{i\theta n_0}\left[1 - D^{-1}

1271: e^{2i\theta (N+1-n_0)}\right] .

1272: \end{eqnarray}

1273: %

1274: With (\ref{bound_cond}) and (\ref{boundary_defect}), all coefficients in

1275: the wave function (\ref{left_right}) are expressed in terms of one

1276: number, $a(n_0)$. It can be obtained from normalization.

1277: 

1278: In a finite chain, the values of $\theta$ are quantized. They can be

1279: found from Eq.~(\ref{eq_a}) with $n=n_0$. With account taken of

1280: (\ref{boundary_defect}), this equation can be written as

1281: %

1282: \begin{eqnarray}

1283: \label{theta}

1284: f_N (\theta ) - D^2 f_N (-\theta ) = -(ig/J\sin \theta )

1285: \qquad\qquad\qquad\nonumber\\

1286:  \times [

1287: f_N (\theta ) + D^2 f_N (-\theta ) - 2D

1288: \cos \theta (N-2 n_0 +1) ],

1289: \end{eqnarray}

1290: %

1291: where

1292: %

1293: \begin{equation}

1294: f_N (\theta ) = \exp[i\theta (N+1)] .

1295: \end{equation}

1296: 

1297: %

1298: The analysis of the roots of Eq.~(\ref{theta}) is standard. This is a

1299: $2(N+1)$-order equation for $\exp(i\theta)$, but its solutions for

1300: $\theta$ come in pairs $\theta$ and $-\theta$. Each pair gives one

1301: wave function, as seen from Eq.~(\ref{left_right}). In addition,

1302: Eq.~(\ref{theta}) has roots $\theta = 0,\pi$; they are spurious

1303: (unless $|\Delta| = 1$ and $g= 0$) and appear as a result of algebraic

1304: transformations. Therefore Eq.~(\ref{theta}) has $N$ physically

1305: distinct roots, as expected for a chain of $N$ spins.  We note that,

1306: for $g=0$ the position of the impurity $n_0$ drops out from

1307: Eq.~(\ref{theta}), and then the equation goes over into the result for

1308: an ideal chain.

1309: 

1310: Solutions of Eq.~(\ref{theta}) with real $\theta$ correspond, in the

1311: case of a long chain, to delocalized magnon-type excitations

1312: propagating in the chain. Their bandwidth is $2|J|$.

1313: 

1314: Along with delocalized states, Eq.~(\ref{theta}) describes also

1315: localized states with complex $\theta$. Complex roots of

1316: Eq.~(\ref{theta}) can be found for a long chain, where $|{\rm

1317: Im}\,\theta|(N-n_0),\, |{\rm Im}\,\theta|n_0 \gg 1$. They describe

1318: surface states, which are localized on the chain boundaries, and a

1319: state localized on the defect. The localization length of the states

1320: is given by $|1/{\rm Im}\,\theta|$.

1321: 

1322: The surface states arise only for the

1323: anisotropy parameter $|\Delta|>1$. The corresponding values of

1324: $\theta$ are

1325: %

1326: \begin{equation}

1327: \label{bound_local}

1328: \theta_s = \pm i\ln|\Delta| + \pi\Theta(-\Delta).

1329: \end{equation}

1330: %

1331: Here, the signs $+$ and $-$ refer to the states localized on the

1332: left and right boundaries, respectively, and $\Theta(x)$ is the step

1333: function.

1334: 

1335: From Eq.~(\ref{energy_one}), the energy of the surface state is

1336: %

1337: \begin{equation}

1338: \label{E_s}

1339: E_s={\varepsilon}_1 + J(\Delta^2 +1)/2\Delta

1340: \end{equation}

1341: %

1342: It lies outside the energy band of delocalized excitations. We note

1343: that, for $\Delta > 1$ the surface states decay monotonically with the

1344: distance from the boundary (Re~$\theta=0$). For sufficiently large

1345: negative $\Delta$, on the other hand, the decay of the wave function

1346: is accompanied by oscillations, and $a(n)$ changes sign from site to site.

1347: 

1348: A defect in a long chain gives rise to a localized one-spin excitation

1349: for an arbitrary excess energy $g$ \cite{Economou}. The amplitude

1350: $a(n)$ decays away from the defect as

1351: %

1352: \begin{eqnarray}

1353: \label{one_localized}

1354: &a(n)=a(n_0)\exp(i\theta_d|n-n_0|),\\

1355: &\theta_d=i\sinh^{-1}(|g/J|)+

1356: \pi\Theta(-g/J).\nonumber

1357: \end{eqnarray}

1358: %

1359: The energy of the localized state is

1360: %

1361: \begin{equation}

1362: \label{E_d}

1363: E_d=\varepsilon_1+

1364: (g^2+J^2)^{1/2}{\rm sgn}\,g.

1365: \end{equation}

1366: 

1367: For small $|g/J|$, we have Im~$\theta_d \approx |g/J|$, i.e., the

1368: reciprocal localization length is simply proportional to the defect

1369: excess energy $|g|$. In the opposite case of large $|g/J|$ we have

1370: Im~$\theta_d = \ln |2g/J|$. In this case the amplitude of the

1371: localized state rapidly falls off with the distance from the defect,

1372: $a(n)\propto (2|g/J|)^{-|n-n_0|}$.

1373: 

1374: If the localization length is comparable to the chain length, the

1375: notion of localization is not well defined. However,

1376: when discussing numerical results, one can formally call

1377: a state localized if its wave function exponentially decays away from

1378: the defect and is described by a solution of Eq.~(\ref{theta}) with

1379: complex $\theta$. This is equivalent to the statement that the state

1380: energy $E_d$ lies outside the band of magnons in the infinite

1381: chain. In an open finite chain such localized state may emerge

1382: provided the localization length is smaller than the distance from the

1383: defect to the boundaries. This means that the defect excess energy

1384: $|g|$ should exceed a minimal value that depends on the size of the

1385: chain. The comparison of Eq.~(\ref{one_localized}) with the numerical

1386: solutions of the full equation (\ref{theta}) for a finite chain is shown in

1387: Fig.~\ref{fig:loc_length}.

1388: 

1389: \begin{figure}[ht]

1390: \includegraphics[width=3.2in]{fig5_XXZ.eps}

1391: \caption{ The reciprocal localization length Im~$\theta_d$ for an

1392: infinite chain [Eq.~(\protect\ref{one_localized})] as a function of

1393: the defect excess energy $g$ for $J=1$ (solid line). Also shown are

1394: the results for an open chain with

1395: the number of sites $N=6$ and 12 and $n_0=N/2$ (stars and crosses,

1396: respectively) and a closed chain with 6 and 12 sites (circles and

1397: squares, respectively). They are obtained from

1398: Eq.~(\protect\ref{theta}) with $\Delta = 10$ and from

1399: Eq.~(\protect\ref{dispersion}). In an open chain, solutions with

1400: nonzero Im~$\theta_d$ emerge starting with a certain

1401: $|g/J|>|g/J|_{\min}$. In a closed chain with even $N$ there is no

1402: threshold in $|g/J|$ for the onset of states with Im~$\theta_d\neq

1403: 0$.}

1404: \label{fig:loc_length}

1405: \end{figure}

1406: 

1407: \subsection{One-excitation states in a closed chain}

1408: 

1409: As pointed out in Sec.~II, for a closed chain the solution of the

1410: Schr\"odinger equation can be also sought in the form of

1411: counterpropagating waves with different amplitudes,

1412: Eq.~(\ref{right}). Clearly, the phases $\theta$ and $-\theta$ describe

1413: one and the same wave function.  The one-excitation energy $E_1$ is

1414: given by Eq.~(\ref{energy_one}).

1415: 

1416: The interrelation between the amplitudes of the waves $C, C'$ in

1417: Eq.~(\ref{right}) and the amplitude of the wave function on the defect

1418: site $a(n_0)$ can be obtained from Eq.~(\ref{eq_a}) with $n=n_0\pm

1419: 1$. This equation has two solutions,

1420: %

1421: \begin{equation}

1422: \label{untouched}

1423: e^{i\theta N}=1,\quad a(n_0)=Ce^{i\theta n_0}+C'e^{-i\theta n_0}

1424: \end{equation}

1425: %

1426: and

1427: %

1428: \begin{eqnarray}

1429: \label{touched}

1430: a(n_0)&=&Ce^{i\theta n_0}\left(1+e^{i\theta N}\right)\\

1431: &=&C'e^{-i\theta n_0}\left(1+e^{-i\theta N}\right)\quad [\exp(i\theta

1432: N)\neq 1].\nonumber

1433: \end{eqnarray}

1434: %

1435: In order to fully determine the wave function $a(n)$ (\ref{right}),

1436: Eqs.~(\ref{untouched}), (\ref{touched}) should be substituted into the

1437: Schr\"odinger equation (\ref{eq_a}) for $n=n_0$.

1438: 

1439: Equations (\ref{eq_a}) and (\ref{untouched}) can be satisfied provided that

1440: either $g=0$, which means that there is no defect, or $a(n_0)=0$. The

1441: first condition describes excitations in an ideal closed chain and is

1442: not interesting for the present paper. The condition $a(n_0)=0$

1443: corresponds to the wave function $a(n)\propto \sin \theta (n-n_0)$,

1444: which has a simple physical meaning. It is a standing wave in an ideal

1445: chain with a node at the location of a defect. Because of the node,

1446: the corresponding state ``does not know'' about the defect, and

1447: therefore it is exactly the same as in an ideal chain.

1448: 

1449: In the presence of a defect, the solutions of Eq.~(\ref{untouched}) in

1450: the range of interest $0<\theta < \pi$ are $\theta = 2\pi k/N$ with

1451: $k=1,2,\ldots,(N-1)/2$ for odd $N$, or $k=1,2,\ldots,N/2-1$ for even

1452: $N$.

1453: 

1454: The equation for $\theta$ that follows from Eqs.~(\ref{eq_a}) and

1455: (\ref{touched}) has the form

1456: %

1457: \begin{equation}

1458: \label{dispersion}

1459: \exp (i\theta N) - 1 = -\frac{ig}{J \sin \theta} [\exp (i\theta N)+1] .

1460: \end{equation}

1461: %

1462: For $g\neq 0$ this equation has either $(N+1)/2$ (for odd $N$) or

1463: $N/2+1$ (for even $N$) solutions for $\pm \theta$. Therefore the total

1464: number of solutions for $\theta$ that follow from

1465: Eqs.~(\ref{untouched}) and (\ref{dispersion}) is $N$, as expected.

1466: 

1467: By rewriting Eq.~(\ref{dispersion}) as $\tan(\theta N/2) =

1468: -(g/J\sin\theta)$ and plotting the left- and right-hand sides as

1469: functions of $\theta$ (cf. Ref.~\onlinecite{Economou}), one can see

1470: that all physically distinct roots of this equation but one are real

1471: and lie in the interval $0< \theta < \pi$ [except for one case, see

1472: below]. Such solutions describe delocalized states with

1473: sinusoidal wave functions.

1474: 

1475: The complex root of Eq.~(\ref{dispersion}), $\theta=\theta_d$,

1476: describes a state localized on the defect. For a long chain,

1477: Im~$\theta_d N\gg 1$, the solution has the form (\ref{one_localized}),

1478: as expected. An interesting situation occurs for a shorter chain. If

1479: $N$ is even or if $g/J>0$, a localized solution with complex

1480: $\theta_d$ emerges for any defect excess energy $g$.  Thresholdless

1481: localization does not happen in an open chain. In a closed chain, it

1482: arises because there is no reflection from boundaries. For small

1483: positive $g/J$ one obtains the complex solution of

1484: Eq.~(\ref{dispersion}) in the form $\theta_d\approx i(2g/NJ)^{1/2}$.

1485: The square-root dependence of Im~$\theta_d$ on $g$ is seen from

1486: Fig.~\ref{fig:loc_length}.

1487: 

1488: Equation (\ref{dispersion}) has a complex

1489: solution also for even $N$ and small negative $g/J$.  In this case

1490: $\theta_d\approx i(-2g/NJ)^{1/2} + \pi$, i.e., the decay of the wave

1491: function $a(n)$ is accompanied by sign flips, $a(n+1)/a(n) <

1492: 0$. Such oscillations cannot be reconciled with the

1493: periodicity condition for odd $N$. Therefore, for odd $N$ and negative

1494: $g/J$ a decaying solution arises only when $-g/J$ exceeds a threshold

1495: value. One can show from Eq.~(\ref{dispersion}) that this value is

1496: $|g/J|_{\min}=2/N$, which has also been confirmed numerically.

1497: 

1498: 

1499: 

1500: 

1501: 

1502: 

1503: 

1504: 

1505: 

1506: \begin{thebibliography}{99}

1507: 

1508: 

1509: \bibitem{recent_reviews} M.A. Nielsen and I.L. Chuang,

1510: {\it Quantum Computation and Quantum Information} (Cambridge

1511: University Press, Cambridge, 2000).

1512: 

1513: %\bibitem{Kane98} B. Kane, Nature {\bf 393}, 133 (1998).

1514: 

1515: \bibitem{Makhlin99} Y. Makhlin, G. Sch\"on, and A. Shnirman,

1516: Rev. Mod. Phys. {\bf 73}, 357 (2001).

1517: 

1518: \bibitem{JJ-all}

1519: %V.D. Averin, Solid State Commun. {\bf 105}, 659 (1998);

1520: %Y. Nakamura, Y.A. Pashkin, and H.S. Tsai, Nature {\bf 398}, 786 (1999);

1521: D. Vion, A. Aassime, A. Cottet, P. Joyez, H. Pothier,

1522: C. Urbina, D. Esteve, and M.H. Devoret, Science {\bf 296}, 886 (2002);

1523: Y. Yu, S.Y. Han, X. Chu, S.I. Chu, and Z. Wang, {\it ibid} {\bf 296}, 889

1524: (2002); 

1525: Y.A.~Pashkin, T.~Yamamoto, O.~Astafiev, Y.~Nakamura, D.V.~Averin, and

1526: J.S.~Tsai, Nature (London) {\bf 421}, 823 (2002);

1527: A.J. Berkley, H. Xu, R.C. Ramos, M.A. Gubrud, F.W.  Trauch,

1528: P.R. Johnson, J.R. Anderson, A.J. Dragt, C.J. Lobb, and

1529: F.C. Wellstood, Science {\bf 300}, 1548 (2003).

1530: %D.A. Lidar, L.A. Wu, and A. Blais,

1531: %Quant. Info. Proc. {\bf 1}, 155 (2002).

1532: 

1533: 

1534: \bibitem{mark} P.M. Platzman and  M.I. Dykman, Science {\bf 284},

1535: 1967 (1999);  M.I. Dykman and P.M. Platzman, Fortschr. Phys.

1536: {\bf 48}, 9 (2000);

1537: M.I. Dykman, P.M. Platzman, and P. Seddighrad, Phys. Rev. B {\bf 67},

1538: 155402 (2003).

1539: 

1540: \bibitem{Chuang9802} I.L. Chuang, L.M.K. Vandersypen,

1541: X. Zhou, D.B. Leung, and S. Lloyd,

1542: Nature (London) {\bf 393}, 143 (1998);

1543: L.M.K. Vandersypen, M. Steffen, G. Breyta, C.S. Yannoni,

1544: M.H. Sherwood, and I.L. Chuang, {\it ibid} {\bf 414}, 883 (2001).

1545: 

1546: \bibitem{Yamamoto02} T.D. Ladd, J.R. Goldman, F. Yamaguchi,

1547: Y. Yamamoto, E. Abe, and K.M. Itoh,

1548: Phys. Rev. Lett. {\bf 89}, 017901 (2002).

1549: 

1550: \bibitem{Cory00} D.G. Cory, R. Laflamme, E. Knill, L. Viola, T.F. Havel,

1551: N. Bouland, G. Boutis, E. Fortunato, S. Lloyd, R. Martinez,

1552: C. Negrevergne, M. Pravia, Y. Sharf, G. Teklemariam,

1553: Y.S. Weinstein, and W.H. Zurek,

1554: Fortschr. Phys. {\bf 48}, 875 (2000).

1555: 

1556: \bibitem{Piermarocchi02} P.~Chen, C.~Piermarocchi, and L.J.~Sham,

1557: Phys. Rev. Lett.  {\bf 87}, 067401 (2001); X.~Li, Y.~Wu, D.G.~Steel,

1558: D.~Gammon, T.H.~Stievater, D.S.~Katzer, D.~Park, L.J.~Sham, and

1559: C.~Piermarocchi, Science {\bf 301}, 809 (2003).

1560: 

1561: \bibitem{Demille02} D. DeMille, Phys. Rev. Lett. {\bf 88}, 067901 (2002).

1562: 

1563: \bibitem{Silvestrov01} P.G. Silvestrov, H. Schomerus, and

1564: C.W.J. Beenakker, Phys. Rev. Lett.  {\bf 86}, 5192 (2001);

1565: 

1566: \bibitem{Kaminsky-Lloyd02} W.M. Kaminsky and S. Lloyd quant-ph/0211152.

1567: 

1568: \bibitem{Bethe} H.A. Bethe, Z. Phys. {\bf 71}, 205 (1931);

1569: C.N. Yang and C.P. Yang, Phys. Rev. {\bf 150}, 321, 327 (1966).

1570: 

1571: \bibitem{Baxter} F.C. Alcaraz, M.N. Barber, M.T. Batchelor, R.J.

1572: Baxter and G.R.W. Quispel, J. Phys. A. {\bf 20}, 6397 (1987).

1573: 

1574: \bibitem{integrable} P. Schmitteckert, P. Schwab, and U. Eckern,

1575: Europhys. Lett. {\bf 30}, 543 (1995);

1576: H.-P. Eckle, A. Punnoose, and R.A. R\"omer,

1577: Europhys. Lett. {\bf 39}, 293 (1997);

1578: A. Zvyagin, J. Phys. A {\bf 34}, R21 (2001).

1579: 

1580: %\bibitem{Golding_unpublished} B. Golding, unpublished.

1581: 

1582: %\bibitem{Romer99}

1583: 

1584: 

1585: %\bibitem{Wellstood} P.R. Johnson, F.W. Strauch, A.J. Dragt,

1586: %R.C. Ramos, C. J. Lobb, J. R. Anderson, and F. C. Wellstood,

1587: %Phys. Rev. B {\bf 67}, 020509 (2003); A.J. Berkley, H. Xu, R.C. Ramos,

1588: %M.A. Gubrud, F.W.  Trauch, P.R. Johnson, J.R. Anderson, A.J. Dragt,

1589: %C.J. Lobb, and F.C. Wellstood, Science {\bf 300}, 1548 (2003).

1590: 

1591: 

1592: \bibitem{Shepelyansky00} B. Georgeot and D.L. Shepelyansky,

1593: Phys. Rev. E {\bf 62}, 3504, 6366 (2000).

1594: 

1595: \bibitem{Izrailev01} G.P. Berman, F. Borgonovi, F.M. Izrailev,

1596: and V.I. Tsifrinovich, Phys. Rev. E {\bf 64}, 056226 (2001);

1597: {\it ibid} {\bf 65}, 015204 (2002).

1598: 

1599: \bibitem{Lloyd02} V.V. Flambaum, Aust. J. Phys. {\bf 53}, 489 (2000);

1600: %P.G. Silvestrov, H. Schomerus, and C.W.J. Beenakker, Phys. Rev. Lett.

1601: %{\bf 86}, 5192 (2001);

1602: J. Emerson, Y.S. Weinstein, S. Lloyd, and D.G. Cory,

1603: Phys. Rev. Lett. {\bf 89}, 284102 (2002).

1604: 

1605: \bibitem{Economou} G.F. Koster and J.C. Slater,

1606: Phys. Rev. {\bf 95}, 1167 (1954); E.N. Economou,

1607: {\it Green's function in quantum physics}, (Springer-Verlag,

1608: Berlin, New York, 1979).

1609: 

1610: \bibitem{Goodkind01} J.M. Goodkind and S. Pilla,  Quantum

1611: Information and Computation {\bf 1}, 108 (2001).

1612: 

1613: \bibitem{us_DS} M.I. Dykman and L.F. Santos, J. Phys. A 

1614: {\bf 36}, L561 (2003).

1615: 

1616: 

1617: \end{thebibliography}

1618: 

1619: \end{document}

1620: