1: \documentclass[showpacs,twocolumn,preprintnumbers,amsmath,amssymb,pra]{revtex4}

2: 

3: \usepackage{graphicx}% Include figure files

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

5: \usepackage{bm}% bold math

6: \usepackage{psfrag}

7: \usepackage{epic}

8: \usepackage{eepic}

9: \usepackage{bm}

10: \usepackage{mathrsfs}

11: 

12: \newcommand{\trace}[1]{\mathrm{Tr}\, #1}

13: \newcommand{\vkt}[1]{{\boldsymbol{#1}}}

14: \newcommand{\matr}[1]{#1}

15: \newcommand{\matrg}[1]{#1}

16: \newcommand{\transp}{\mathrm{T}}

17: \newcommand{\forv}[1]{\langle #1\rangle}

18: \newcommand{\bra}[1]{{\langle #1|}}

19: \newcommand{\ket}[1]{{|#1 \rangle}}

20: \newcommand{\braket}[2]{{\langle #1|#2 \rangle}}

21: \newcommand{\expt}[1]{\left\langle #1\right\rangle}

22: \newcommand{\dd}{\ensuremath{\mathrm{d}}}

23: \newcommand{\imi}{\ensuremath{\mathrm{i}}}

24: \newcommand{\expe}{\ensuremath{\mathrm{e}}}

25: \newcommand{\Tr}{\mathrm{Tr}}

26: 

27: \begin{document}

28: 

29: \newenvironment{smatrix}{\begin{bmatrix}}{\end{bmatrix}}

30: \setlength{\tabcolsep}{4ex}

31: \setlength{\unitlength}{1mm}

32: 

33: \title{Entanglement properties of quantum spin chains}

34: 

35: \author{Stein Olav Skr{\o}vseth}

36: \email{stein.skrovseth@phys.ntnu.no}

37: \affiliation{%

38: Department of Physics,

39: Norwegian University of Science and Technology,

40: N-7491 Trondheim, Norway

41: }%

42: 

43: \date{1 March 2006}

44: 

45: \begin{abstract}

46:   We investigate the entanglement properties of a finite size 1+1

47:   dimensional Ising spin

48:   chain, and show how these properties scale and can be utilized to

49:   reconstruct the ground state wave function. Even at the critical

50:   point, few terms in a Schmidt decomposition contribute to the exact

51:   ground state, and to physical properties such as the

52:   entropy. Nevertheless the entanglement here is prominent due to the

53:   lower-lying states in the Schmidt decomposition.

54: \end{abstract}

55: 

56: \pacs{03.67.Mn, 03.65.Ud, 75.10.Pq, 05.70.Jk}

57: %% Entanglement and quantum nonlocality

58: %% Spin chain models

59: %% Critical point phenomena

60: \keywords{}

61: 

62: \maketitle

63: 

64: \section{Introduction}

65: The EPR argument \cite{EPR35} and the Bell inequalities

66: \cite{Bell64} were, albeit with almost 30  

67: years in between, acknowledgments

68: that quantum theory exhibits the strange correlations known as

69: entanglement. However, only in the latest few years has it been

70: known that entanglement is a resource that can be utilized in quantum

71: computing,\cite{Nielsen&Chuang} and is thus a central subject in the

72: continually expanding field of 

73: quantum information theory. Furthermore, entanglement has been shown

74: to be a fundamental feature in quantum phase transitions, something

75: that has spawned a whole new field of research \cite{Sachdev,

76:   Osterloh:2002, Vidal:2002rm, SOS1, Osborne:2002, Korepin2004, SOSbose,

77:   Calabrese04, calabrese-2005-0504, latorre05}. This article will 

78: focus on entanglement in both quantum critical and non critical

79: systems, the structure of entanglement in condensed matter systems

80: being at its most complex in critical systems. 

81: 

82: The determination of the ground state in a quantum system such as 1D

83: spin chains is highly complex, but the last years have seen the

84: development of techniques such as DMRG \cite{White1992,Schollwoeck2005}

85: and the recent entanglement renormalization \cite{Vidal2005} which

86: solve the problem efficiently. In both cases the entanglement in the

87: problem seems to be the key to the methods' success over traditional

88: renormalization schemes. Moreover, for a pure state, any long range

89: correlations require the existence of entanglement in the state.

90: 

91: A quantum system with wave function $\ket\Psi$ in a Hilbert space

92: $\mathcal H$, which is partitioned

93: into two subspaces $\mathcal H_{\mathscr A}\otimes\mathcal H_{\mathscr

94: B}$, can be written as a Schmidt decomposition 

95: \begin{equation}

96:   \ket\Psi=\sum_{n=1}^\chi\sqrt{\lambda_n}\,

97:   \ket{\psi_n^{(\mathscr A)}}\otimes\ket{\psi_n^{(\mathscr B)}},

98:   \label{Schmidt}

99: \end{equation}

100: where $\ket{\psi_n^{(\mathscr A)}}\in\mathcal H_{\mathscr A}$ and vice

101: versa. The coefficients $\lambda_n$ are real, positive c-numbers and

102: the states are mutually orthogonal, $\braket{\psi_n^{\mathscr

103:     A}}{\psi_m^{\mathscr A}}=\delta_{nm}$. The upper

104: limit $\chi$, the Schmidt number, is a brute measure of the

105: entanglement between the two subsystems, running from one (no

106: entanglement), maximally reaching the dimensionality of the smallest

107: of the two Hilbert spaces. The reduced density matrix of (say) system

108: $\mathscr A$ can then be written

109: \[\rho^{(\mathscr A)}=\trace_{\mathscr

110:   B}\ket\Psi\bra\Psi=\sum_{n=0}^\chi\lambda_n\ket{\psi_n^{(\mathscr

111:     A)}}\bra{\psi_n^{(\mathscr A)}}\]

112: and equivalently for $\mathscr B$. Hence, knowing the Schmidt

113: decomposition of the wave function is equivalent to knowing the basis

114: in which the reduced density matrix is diagonal, modulo phases. We

115: will in this 

116: article show that the effective Schmidt number $\chi_{\mathrm{eff}}$

117: is much smaller than the possible maximum, and thus the number of

118: terms contributing significantly to the wave function is

119: surprisingly small. Remarkably, this also is true at highly entangled

120: points in the phase space, such as at criticality. 

121: 

122: The entropy of the wave function $\ket\Psi$ can be measured as the entropy

123: of the subsystem,

124: \begin{equation}

125:   S_{\mathscr A}=-\trace \rho^{(\mathscr A)}\log_2\rho^{(\mathscr A)}=-\sum_n\lambda_n\log_2\lambda_n,

126:   \label{Sdef}

127: \end{equation}

128: increasing from zero if the two subspaces are entangled. Hence,

129: computing the

130: eigenvalues $\lambda_n$ of the reduced density matrix is vital to the

131: understanding of entangled states.

132: 

133: \section{Ising model}

134: The quantum Ising chain in external field $h$ in 1+1 dimension is a

135: good benchmark tool for 

136: the analysis of entanglement since its properties are extremely

137: well known. The model is defined by the Hamiltonian 

138: \begin{equation}

139:   H_{\mathrm{Ising}}=-\sum_{n=1}^N\left(\sigma_n^x\sigma_{n+1}^x+h\sigma_n^z\right),

140: \end{equation}

141: on $N$ lattice sites.

142: We operate with open boundary conditions (OBC), since this will

143: ease the formalities when partitioning the system. This will weaken

144: the phase transition in the sense that the conformal symmetry is

145: broken compared to periodic boundary conditions (PBC)

146: \cite{Francesco97}, but the phase 

147: transition in the thermodynamic limit will prevail. Also, Calabrese

148: and Cardy have found that one can compute conformal signatures in the

149: OBC case provided the boundary conditions are conformal \cite{Calabrese04}.

150: The model has a phase transition at $h=1$ between the product state

151: $\ket{\Psi_\infty}=\ket{\uparrow\uparrow\cdots\uparrow\,}$ for $h>1$ and

152: the Schr\"{o}dinger cat state

153: $\ket{\Psi_0}=\frac1{\sqrt2}\left(\ket{\rightarrow\rightarrow\cdots\rightarrow}+\ket{\leftarrow\leftarrow\cdots\leftarrow}\right)$ 

154: when $h<1$. Here $\ket{\!\!\uparrow}$ is the eigenstate of $\sigma^z$

155: with eigenvalue 1, and $\ket{\!\!\leftrightarrows}$ are

156: eigenstates of $\sigma^x$ with eigenvalues $\pm1$. Note that the

157: transition is between unit entropy in the low field limit and zero in

158: the high field limit. 

159: 

160: Doing a Jordan-Wigner transform as sketched in Refs. \cite{Latorre:2003kg,

161:   Osborne:2002}, we can map the model onto a string of non interacting

162: fermions, and thus compute the eigenvalues for the Ising model (and

163: a larger class of models) in what resembles the thermodynamic

164: limit, typically a few hundred particles. In essence, we define the

165: $N$ fermionic operators 

166: \begin{equation}

167:   \hat

168:   a_n=\frac12\left(\bigotimes_{k=1}^{n-1}\sigma_k^z\right)\otimes\left(\sigma_n^x+\imi\sigma_n^y\right),

169:   \label{def_fermions}

170: \end{equation}

171: and the $2N$ Majorana fermions 

172: \[\check\gamma_{2n}=\frac1{\imi\sqrt2}\left(\hat a_n-\hat

173:   a_n^\dag\right),\quad

174: \check\gamma_{2n-1}=\frac1{\sqrt2}\left(\hat a_n+\hat a_n^\dag\right).\]

175: These fulfill $\{\check\gamma_m,\check\gamma_n\}=\delta_{mn}$, and

176: are delocalized in terms of the original lattice of fermions. The

177: Majorana fermions

178: diagonalize the Hamiltonian in the sense that

179: \[H=\sum_{mn}C_{mn}\check\gamma_m\check\gamma_n,\]

180: with an Hermitian $2N\times2N$ matrix $C$. Next, define the imaginary

181: and anti symmetric correlation matrix 

182: $\Gamma_{ij}=\expt{[\check\gamma_i,\check\gamma_j]}$. We will consider

183: only the ground state, and thus the expectation

184: values are to be taken in the ground state. In this sense, our model is

185: defined by the matrix $C$ while the state is defined in the matrix

186: $\Gamma$. The Majorana fermions have a two-to-one correspondence to

187: the fermions (\ref{def_fermions}), and hence tracing out a particle

188: from the system amounts to removing the two adjacent rows and columns

189: in $\Gamma$ corresponding to the particle. Tracing out $N-N'$

190: particles this way, we end up with the $2N'\times2N'$ matrix

191: $\bar\Gamma_{ij}$. This 

192: latter matrix can be made block diagonal through an orthogonal

193: transformation $O$, such that we can define

194: new Majorana fermions $\bar\gamma_n$ that fulfill

195: \begin{equation}

196:   O^{\mathrm T}\bar\Gamma O=\left\{\expt{[\bar\gamma_i,\bar\gamma_j]}\right\}=\bigoplus_{k=1}^{N'}\begin{pmatrix}0&\imi\xi_k\\-\imi\xi_k&0\end{pmatrix}

197:   \label{bargammadef}

198: \end{equation}

199: with $0\leq \xi_k\leq1$. The transformation $O$ is the same

200: transformation that block diagonalizes the Hamiltonian matrix $C$.

201: The matrix $\bar\Gamma$ corresponds to the

202: state described by the reduced density matrix $\rho'$, with the $N-N'$

203: particles traced out. The fermion 

204: operators $\bar a_n$ corresponding to $\bar\gamma_n$ will

205: diagonalize the reduced density matrix such that the eigenvalues

206: thereof can be computed. The eigenvalues are

207: determined by the set of 

208: binary occupation numbers $\eta=\{n_k\}$, $k=1,2,\ldots,N'$, and

209: $n_k=\{0,1\}$, $\ket\eta$ being an eigenstate of $\bar a_k\bar

210: a_k^\dag$ with 

211: eigenvalue $n_k$. Thus $\ket\eta$ is also an eigenstate of $\rho'$ and

212: the reduced density matrix is

213: $\rho'=\sum_\eta\lambda_\eta\ket\eta\bra\eta$ with eigenvalues 

214: \begin{equation}

215:   \lambda_\eta=\prod_{k=1}^{N'}\left[\frac12+(-1)^{n_k}\xi_k\right].

216:   \label{lambdafromxi}

217: \end{equation}

218: Finally, the entropy becomes 

219: \begin{equation}

220:   S_{\mathscr A}=\sum_{k=1}^N H\left(\frac12\left(1+\xi_k\right)\right),

221: \end{equation}

222: where $H(x)=-x\log_2x-(1-x)\log_2(1-x)$ is the binary entropy function. 

223: 

224: The entropy of the ground state as measured by

225: Eq. (\ref{Sdef}) when the system is partitioned into two equal parts,

226: is shown in Fig. \ref{fig:SIsing}.

227: \begin{figure}[htbp]

228:   \includegraphics[width=.9\columnwidth]{S_Ising.eps}

229:   \caption{The entropy in the quantum Ising model as function of the

230:     external magnetic field $h$ for $N=10$ (lower line) and $N=100$

231:     particles. In both cases half system is traced out, from the

232:     edge. The letters refer to the places of investigation in 

233:     Fig. \ref{fig:eigvalues_Ising}. The critical point (in the

234:     thermodynamic limit) is shown as a vertical line.} 

235:   \label{fig:SIsing}

236: \end{figure}

237: The increased entropy around the critical point is a hallmark of the

238: quantum phase transition, though this does not show in the small system.

239: 

240: \begin{figure}[htbp]

241:   \includegraphics[width=.9\columnwidth]{eigvals.eps}

242:   \caption{The decreasingly ordered eigenvalues $\lambda_n^\downarrow$

243:     of the reduced density 

244:     matrix of the Ising model's ground state when tracing out half

245:     size. Data for $N=50$. The upper line ($\bullet$) denotes the critical values

246:     when $h=h_c=1$, at (B) in figure \ref{fig:SIsing}; ($\triangle$)

247:     denotes the Scr\"{o}dinger cat state with $h=0.5$, at (A); and

248:     ($\blacktriangle$) denotes the approximate product state with $h=2$, at

249:     (C). The fitting line to the critical values

250:     $\lambda_n^\downarrow\sim e^{-2n}$ is shown. 

251:     Only the 10 largest eigenvalues are shown since numerical

252:     errors become prominent after this.}

253:   \label{fig:eigvalues_Ising}

254: \end{figure}

255: Figure \ref{fig:eigvalues_Ising} shows the magnitude of the

256: decreasingly ordered 

257: eigenvalues $\lambda_n^\downarrow$. Note that there are $2^{N/2}=128$ possible

258: eigenvalue contributions, while this shows that only a very few of

259: these contribute to the wave function, and thus to the entanglement as

260: measured by the entropy $S$.

261: 

262: The main eigenvalues at critical point decrease roughly exponentially,

263: as $\lambda_n^\downarrow\sim e^{-2n}$, while off criticality only the

264: very first contribute significantly.

265: In the low field limit, the 

266: highest eigenvalues are paired due to the Schr\"{o}dinger cat nature

267: of the state, while in the high field limit there is one single main

268: contribution, which is the product state $\ket{\Psi_\infty}$. With

269: PBC, the eigenvalues would be paired also for $h>1$, except for the

270: single main eigenvalue due to the translational symmetry. This pairing

271: also occurs in a conformal bosonic chain \cite{SOSbose}, but is not a

272: general property of a non critical system.

273: 

274: Hence we can approximate the wave function by restricting the sum

275: (\ref{Schmidt}) to some 

276: upper limit $\chi'<\chi$ to make the new wave function 

277: \[\ket{\Psi'}=\sum_{n=1}^{\chi'}\sqrt{\lambda_n'}\ket{\psi_n^{(\mathscr

278:     A)}}\otimes\ket{\psi_n^{(\mathscr B)}},\]

279: where the new coefficients $\lambda_n'$ are determined by

280: normalization, $\lambda_n'=\lambda_n/(1-\varepsilon)$. 

281: We define an error as

282: \[\varepsilon=\sum_{n=\chi'+1}^\chi\lambda_n\]

283: which measures the difference in the eigenvalue sum of the two

284: wave functions. The overlap becomes

285: \[\braket{\Psi'}\Psi=\sqrt{1-\varepsilon}.\]

286: 

287: \section{Scaling of eigenvalues}

288: We investigate how the eigenvalues scale with increasing system

289: size. Off the critical point there

290: is {\it no scaling} with system size of the eigenvalues since the

291: entropy saturates at some value. However, on, or near the critical

292: point the entanglement entropy 

293: diverges as predicted by conformal field theory. Given that we trace

294: out a constant fraction of the 

295: entire system, the entanglement entropy diverges up to an additive

296: constant as \cite{Vidal:2002rm, Calabrese04}

297: \[S_{\mathscr A}\sim\frac c6\log_2(N),\]

298: where $c$ is the central charge of the conformal field theory

299: corresponding to the phase transition. In the Ising case $c=1/2$.

300: 

301: \begin{figure}[htbp]

302:   \includegraphics[width=.9\columnwidth]{eigvals_N.eps}

303:   \caption{From the top, the figures show the four smallest

304:     eigenvalues of the correlation matrix in Eq. (\ref{bargammadef}),

305:     the 

306:     four largest eigenvalues of the density matrix as computed from

307:     Eq. (\ref{lambdafromxi}), and the corresponding entropy

308:     contributions  

309:     $s_n=-\lambda_n\log\lambda_n$. For the two lower graphs, equal

310:     point types refer to the same eigenvalues. System is at the

311:     critical Ising point and is traced at half size.}

312:   \label{fig:scaling_Ising}

313: \end{figure}

314: Figure

315: \ref{fig:scaling_Ising} shows how the largest eigenvalues scale at the

316: critical Ising point.

317: The largest eigenvalue $\lambda^\downarrow_1(N)$ {\it decreases} with system size

318: at criticality, while the other increases. The largest decrease

319: roughly as $\lambda^\downarrow_1(N)\sim-0.027\log N$, while the second increases

320: roughly as $\lambda^\downarrow_2(N)\sim0.021\log N$, both within the range shown

321: in Fig. \ref{fig:scaling_Ising}.

322: However, all entropy contributions increase with $N$, thus

323: contributing to the overall divergence of the entanglement entropy.

324: 

325: \section{Approximating the ground state}

326: There exists successful techniques to approximate the ground state of a

327: quantum system, the most prominent of which is the DMRG

328: scheme \cite{White1992}. Rather than taking only ground

329: states of a 

330: sub lattice  when renormalizing the system, this method takes into

331: account possible entangled states as well, producing unprecedented

332: accuracy. The role of entanglement in DMRG is still under

333: investigation, but it seems clear that it is vital to the success of

334: the scheme \cite{Schollwoeck2005}.

335: The results of our work here are indicative that the main

336: contributions to the ground state are indeed entangled at the critical

337: point, along with few of these terms contributing to the actual

338: wave function. These results are not directly applicable to an

339: improvement of the DMRG algorithm, but rather indications as to

340: the success of DMRG. Furthermore, any prospective technique to find

341: the ground state of a quantum system, even in those areas where known

342: techniques fail, needs to understand the nature of the wave function in

343: the system, and we believe that the entanglement properties would play

344: an important part in such a method. In particular, the entanglement

345: properties is precisely what distinguishes a quantum many-body system

346: from the classical counterpart, and entanglement

347: must therefore be an essential part of any such method.

348: 

349: Having found that only few terms contribute to the entanglement in the

350: ground state, we query how well we could possibly approximate the

351: ground state of the full system by the first few terms in the Schmidt

352: decomposition. To this end, we focus on the few-particle case, where

353: the exact wave functions can be computed explicitly. We split the open

354: spin chain in two partitions, and compute the overlap between the

355: actual wave function and the Schmidt expansion of the two subsystems,

356: ordered decreasingly on those terms with the largest overlaps.

357: \begin{align*}

358:   \mathcal O_n&=\braket{\Psi}{\psi_n^{(\mathscr

359:       A)}}\otimes\ket{\psi_n^{(\mathscr B)}}.

360: \end{align*}

361: {\it A priori} it is clear that in the zero field limit $\mathcal

362: O_1=\mathcal O_2=1/\sqrt2$, while any higher terms vanish. In the

363: high-field limit, $\mathcal O_1=1$ as the only non vanishing

364: overlap. We assume that the wave functions $\ket{\psi_n^{(\cdot)}}$

365: are ordered with decreasing Schmidt number, and the results are shown

366: in Fig. \ref{fig:overlaps}.

367: \begin{figure}[htp]

368:   \includegraphics[width=.9\columnwidth]{ooverlap.eps}

369:   \caption{The four largest overlaps $\mathcal O_n$, $n=1,2,3,4$ (from

370:     top to bottom) with

371:     magnetic field $h$ and system sizes $N=10$ (full lines) and

372:     $N=100$ (dashed lines). Again, the critical point is indicated

373:     with the vertial line.}

374:   \label{fig:overlaps}

375: \end{figure}

376: 

377: Since the fermions $\bar a_k$ corresponding to the Majorana fermions

378: defined in Eq. (\ref{bargammadef}) 

379: diagonalize the reduced density matrix, the overlap follows

380: straightforwardly. The probability of the state $\ket\eta$ given the

381: density matrix $\rho'=\lambda_\eta\ket\eta\bra\eta$ is $\lambda_\eta$,

382: and the overlap between this and the ground state will therefore be

383: $\sqrt{\lambda_\eta}$, or

384: \[\mathcal O_\eta=\prod_k\sqrt{\frac12+(-1)^{n_k}\xi_k}.\] 

385: 

386: Most

387: prominently, the $\mathcal O_1$ overlap increases monotonically as $h$

388: increases. That is, even at criticality this first term consisting of

389: the ground states at each side 

390: approximates the actual wave function even better that it does for the

391: cat-state where the overlap naturally is $1/\sqrt2$. The increased

392: entropy at critical points is mainly due to 

393: other terms rising around criticality, terms that are not part of the

394: entire wave function at non critical points in the phase

395: space.

396: 

397: Investigating the Schmidt coefficients' entropy

398: contributions, the results are shown in Fig. \ref{fig:coeffs}. Hence

399: \begin{figure}[t]

400:   \includegraphics[width=.9\columnwidth]{S_h.eps}

401:   \caption{The three main entropy contributions $s_n$, $n=1,2,3$ (from

402:     top to bottom) in

403:     the Ising model 

404:     with magnetic field $h$. Full lines are for $N=10$, while dashed

405:     lines are for $N=100$.}

406:   \label{fig:coeffs}

407: \end{figure}

408: we see that apart from the two major contributions that are

409: equally prominent at $h\to0$, there is one more contributing at

410: criticality. The effect is mainly pronounced with big systems, but a

411: trace is also seen at $N=10$. The increased entropy at the critical

412: point is contributed to by the largest term, but also by the

413: appearance of terms that are zero at all noncritical points.

414: 

415: The error in the overlap and in the entropy that arises from only

416: choosing $\chi'<\chi$ terms in the Schmidt decomposition is zero (or

417: very close) off criticality, since the Schmidt rank is very

418: small. However, on criticality $\chi$ in principle extends to the

419: dimension of the Hilbert space. However, the errors in the overlap

420: $\delta\mathcal O=1-\sum_{n=1}^{\chi'}\mathcal O_n$ and entropy $\delta

421: S=S-\sum_{n=1}^{\chi'}s_n$ are plotted in Fig \ref{fig:errors}. We see

422: that the errors grow linearly  with $N$ when $N>100$. 

423: \begin{figure}

424:   \includegraphics[width=\columnwidth]{errors.eps}

425:   \caption{The errors $\delta\mathcal O$ ($\circ$) and $\delta S$

426:     ($\bullet$) as defined in 

427:     the text with different system sizes at the critical Ising

428:     point. For $\delta\mathcal O$ we have chosen $\chi'=4$ and for

429:     $\delta S$ $\chi'=3$. In all

430:     cases the system is traced at half-size. The lines 

431:     for $N>100$ are linears

432:     $\delta\mathcal O\simeq5\times10^{-6}N-3.8\times10^{-4}$ and $\delta

433:     S\simeq1.75\times10^{-6}N-8.1\times10^{-5}$. For small systems the linearity

434:     does not hold.}

435:   \label{fig:errors}

436: \end{figure}

437: The number of Schmidt terms needed to compute the entropy up to an

438: error of $10^{-4}$, or the effective Schmidt number is only 4 up to a

439: system of $N=400$ sites.

440: 

441: \section*{Conclusions}

442: We have seen that the complex structure of a wave function at critical

443: point in the Ising model comes from the emergence of several terms in

444: the Schmidt decomposition of the wave function. Nevertheless,

445: remarkably few terms actually contribute to the expansion, and

446: physical properties such as the entropy can be extracted using very

447: few terms, even at criticality. Mainly, the rapid decrease of the

448: Schmidt coefficients on or off criticality identifies this effect. We

449: have detailed the effects with both large and small systems, and the

450: primary conclusions hold in both cases.

451: 

452: The matter that these few eigenstates contribute to physical

453: properties such as the entropy and also the overlap with the true

454: ground state signifies the fact that finding the true ground state, or

455: an approximation to it, is not a complicated task {\it per se}, we

456: known that only a few of the eigenstates of the Hamiltonian would

457: suffice for a sufficiently good description of the state. The results in 

458: this paper also indicate that these states are indeed entangled, and

459: thus any algorithm that does not include entanglement is bound to

460: fail. 

461: 

462: The effective Schmidt number that is needed to find a state within

463: some limit is small, on the order of 10 even up to several hundred

464: particles. Thus the effective Hilbert space is dramatically

465: reduced. However, we have not addressed the problem of determining the

466: unknown ground state of a system, only pointed out that the problem is

467: not consistently hard given the correct (so far unknown) approach.

468: 

469: \section*{Acknowledgments}

470: The author is greatly indebted to K{\aa}re Olaussen for valuable

471: contributions to this work. Susanne Viefers and the NordForsk network

472: on low-dimensional physics is thanked for hosting valuable meetings.

473: The University of Troms{\o} (UiT{\o}) is thanked for providing

474: office space.

475: 

476: \bibliography{art}

477: 

478: \end{document}

479: