1:
2:
3:
4: %\documentclass[aps,preprint,groupedaddress,showpacs]{revtex4}
5: %\documentclass[aps,preprint,superscriptaddress]{revtex4}
6: \documentclass[aps,twocolumn,showpacs,floatfix,superscriptaddress]{revtex4}
7: \usepackage{graphicx}
8: \usepackage{amsmath}
9: \usepackage{amsfonts}
10: \usepackage{amssymb}
11: \usepackage{epsfig}
12:
13: \begin{document}
14:
15: \newcommand{\la}{\langle}
16: \newcommand{\ra}{\rangle}
17: \def\be{\begin{equation}}
18: \def\ee{\end{equation}}
19: \def\bea{\begin{eqnarray}}
20: \def\eea{\end{eqnarray}}
21: \def\bma{\begin{mathletters}}
22: \def\ema{\end{mathletters}}
23: \newcommand{\one}{\mbox{$1 \hspace{-1.0mm} {\bf l}$}}
24: \newcommand{\eins}{\mbox{$1 \hspace{-1.0mm} {\bf l}$}}
25: \def\C{\hbox{$\mit I$\kern-.7em$\mit C$}}
26: \newcommand{\tr}{{\rm tr}}
27: \newcommand{\half}{\mbox{$\textstyle \frac{1}{2}$}}
28: \newcommand{\shalf}{\mbox{$\textstyle \frac{1}{\sqrt{2}}$}}
29: \newcommand{\ket}[1]{ | \, #1 \rangle}
30: \newcommand{\bra}[1]{ \langle #1 \, |}
31: \newcommand{\proj}[1]{\ket{#1}\bra{#1}}
32: \newcommand{\kb}[2]{\ket{#1}\bra{#2}}
33: \newcommand{\bk}[2]{\langle \, #1 | \, #2 \rangle}
34: \def\II{I(\{p_k\},\{\rho_k\})}
35: \def\ss{{\cal K}}
36: \tolerance = 10000
37:
38:
39:
40: % You should use BibTeX and revtex.bst for references
41:
42: %\bibliographystyle{apsrev}
43:
44:
45:
46: %Title of paper
47:
48: \title{Adiabatic Path to Fractional Quantum Hall States of a Few Bosonic Atoms}
49: \author{M. Popp}
50: \affiliation{Max--Planck Institute for Quantum Optics,
51: Garching, Germany}
52: \author{B. Paredes}
53: \affiliation{Max--Planck Institute for Quantum Optics, Garching, Germany}
54: \author{J. I. Cirac}
55: \affiliation{Max--Planck Institute for Quantum Optics,
56: Garching, Germany}
57:
58:
59:
60: \begin{abstract}
61: We propose a realistic scheme to create motional entangled states of a few
62: bosonic atoms. It can experimentally be realized with a gas of ultra cold
63: bosonic atoms trapped in a deep optical lattice potential. By
64: simultaneously deforming and rotating the trapping potential on each lattice
65: site it is feasible to adiabatically create a variety of entangled states on
66: each lattice well. We fully address the case of $N=2$ and $N=4$ atoms per
67: well and identify a sequence of fractional quantum Hall states: the Pfaffian
68: state, the $1/2$-Laughlin quasiparticle and the $1/2$-Laughlin state. Exact
69: knowledge of the spectrum has allowed us to design adiabatic paths to these
70: states, with all times and parameters well within the reach of current
71: experimental setups.
72: We further discuss the detection of these states by measuring
73: different properties as their density profile, angular momentum
74: or correlation functions.
75:
76:
77: \end{abstract}
78:
79: \date{\today}
80: \pacs{03.75.Ss, 73.43.-f }
81:
82: \maketitle
83:
84: \section{Introduction}
85: The creation of highly entangled multiparticle states is one of
86: the most challenging goals of modern experimental quantum
87: mechanics. In this respect atomic systems offer a very promising
88: arena in which entangled states can be created and manipulated
89: with a high degree of control. The experimental difficulty
90: increases, however, with the number of particles that are to be
91: entangled, since the system becomes then more sensitive to
92: decoherence. Starting with a small number of particles as a first
93: step, important achievements have been already obtained in the
94: creation of atomic entangled states. For example, in recent
95: experiments with trapped ions, entangled states of up to four ions
96: have been demonstrated \cite{Sackett00}. Moreover, in experiments
97: with neutral bosonic atoms in optical lattices Bell-type states
98: have been created by accurately controlling the interactions
99: between neighbouring atoms \cite{Bloch03}. As a typical feature of
100: most of the experimentally realized entangled states, atoms get
101: entangled through their internal degrees of freedom, keeping
102: separable their motional part.
103:
104: In this article we develop a scheme to create {\em motional}
105: entangled states of a small number of atoms in an actual
106: experimental setup with an optical lattice \cite{Bloch02, Ess04, Ph03}. These
107: states are a sequence of fractional quantum Hall (FQH) states,
108: analogous to the ones that appear in the context of fractional
109: quantum Hall effect \cite{QHEbook}. In contrast to typical atomic
110: entangled states, the particles are here entangled in real space,
111: and not in internal space. This peculiarity makes them specially
112: interesting, for it represents a novel nature of atomic
113: entanglement.
114:
115: The possibility of creating FQH atomic states as the Laughlin
116: state by rapidly rotating the trap confining the atoms has been
117: discussed in several theoretical works \cite{WG00, P01}. However,
118: experiments dealing with typically large number of particles have
119: not yet succeeded in reaching these states. Here, we fully address
120: the case of a small number of particles and design a
121: realistic way of entangling them into FQH states. The experimental
122: setup that we have in mind corresponds to a situation in which a
123: Bose-Einstein condensate is loaded in a deep optical lattice. When
124: the lattice depth is very large tunneling between different sites
125: is strongly suppressed and the system can be treated as a lattice
126: of independent wells, each of them with a small number of
127: particles. By independently rotating each of these 3D wells
128: \cite{Bpriv} the lowest Landau level (LLL) regime can be achieved
129: for each copy. We have studied the problem exactly within the LLL
130: for $N=2$ and $N=4$ particles per well. We have identified a
131: sequence of highly entangled stable ground states, which are the
132: Pfaffian state \cite{Pfaffian91}, the $1/2$-Laughlin quasiparticle \cite{L83} and the
133: $1/2$-Laughlin state \cite{L83}.
134: The $1/2$-Laughlin quasiparticle state (which had never been identified before in an atomic system) is particularly interesting. It is the counterpart of the $1/2$-Laughlin quasihole found in \cite{P01} and contains a $1/2$-anyon.
135: Driving the system into these
136: strongly correlated states is, however, not trivial. By simply
137: increasing the frequency of rotation the system will stay in a
138: trivial non-entangled state with angular momentum zero. Exact
139: knowledge of the spectrum of the system has allowed us to design
140: adiabiatic paths to these states by simultaneously rotating and
141: deforming each of the wells. All parameteres and evolution times
142: lie well within the reach of present experimental setups. We
143: further discuss how to detect these entangled states by measuring
144: different properties as their density profile, angular momentum
145: or correlation functions. In particular, we propose a novel
146: technique to measure the density-density correlation function of
147: these strongly correlated states. Even though the number of atoms per well is small, the lattice setup allows to have multiple copies of the system, so that the experimental signal is highly enhanced.
148:
149: We point out that our findings also show that adiabatically
150: achieving FQH states for rapidly rotating traps with a large
151: number of particles turns out to be very challenging, since the relevant experimental parameters scale linearly with the number of particles.
152: Nevertheless, we hope that our results can shed some light on the
153: problems that these current experiments are dealing with, and even
154: may pave the way to new methods of achieving FQH multiparticle
155: entangled states.
156:
157:
158:
159:
160:
161: \section{Identification of entangled states}
162: We consider a system of bosonic atoms loaded in a 3D optical lattice. We assume a commensurate
163: filling of $N$ atoms per lattice site \cite{Com}, and a large value of the lattice depth $V_0/E_R \gg
164: 1$, where $E_R=\hbar^2k^2/2M$ is the recoil energy, $k$ is the wave vector of the laser
165: lattice light, and $M$ the atomic mass. In this limit the lattice can by treated as a system of
166: independent 3D harmonic wells, each of them having $N$ atoms and a trapping
167: frequency $\omega \approx \sqrt{V_0E_R}$.
168:
169: Let us rotate each of these 3D harmonic
170: wells around the direction $x_3$ with frequency $\Omega$. We will
171: identify a sequence of motional entangled ground states of the
172: $N$ atoms that appear as the frequency $\Omega$ is increased. We
173: will assume the limit of rapid rotation \cite{P01}. In this case
174: the motion in the $x_3$ direction is frozen, and the motion in the
175: plane of rotation $x_1,x_2$ is restricted to the LLL. Note that in
176: order to project the system onto the LLL we do not need to start
177: with a 2D configuration (as it is the case in previous proposals
178: \cite{WG00}), since the fast rotation itself restricts the motion
179: in the direction of the rotation to zero point oscillations. The
180: system is then governed by a two dimensional effective
181: Hamiltonian, which written in units of $\hbar \omega$ has the
182: form:
183: \begin{equation}
184: \label{ham}
185: H=\left(1-\Omega/\omega\right)L+ 2 \pi \ \eta V,
186: \end{equation}
187: where $L=\sum_{m=0}m \, a^{\dagger}_{m}a_{m}$ is the angular
188: momentum operator in the $x_3$ direction, and
189: $V=\sum_{m_1,m_2,m_3,m_4}
190: V_{m_1,m_2}^{m_3,m_4}a^{\dagger}_{m_1}a^{\dagger}_{m_2}a_{m_3}a_{m_4}$
191: is the interaction operator. Here the bosonic operator $a^{\dagger}_{m}
192: (a_{m})$ create (anihilate)
193: an atom in the state $\vert m \rangle $
194: of the LLL with well defined $x_3$ component of the angular momentum $m$. The wave functions of the LLL in complex coordinates read
195: \be
196: \varphi_m(z)=\langle z \vert m
197: \rangle = \frac{1}{\sqrt{\pi m!} \ell} \ z^m e^{-\vert z \vert ^2/2} \ ,
198: \ee
199: where
200: $z=(x_1+ix_2)/\ell$, $\ell=\sqrt{\hbar/M\omega}$, and
201: $m=0,1,\ldots \infty$. Assuming contact interactions between the
202: atoms the interaction coefficients are:
203: \be
204: V_{m_1,m_2}^{m_3,m_4}=\frac{(m_1+m_2)!}{2^{m_1+m_2}\sqrt
205: {m_1!m_2!m_3!m_4!}} \ .
206: \ee
207: In Hamiltonian (\ref{ham}) we have introduced the
208: important interaction parameter $\eta=\sqrt{2/ \pi} a_s/
209: \ell$, with $a_s$ the 3D scattering length. Analytical calculations for
210: scattering potentials of finite size $a_0$ have confirmed that the
211: pseudo-potential approximation is also valid for tight traps with $a_s \ll \ell$ as long as $a_0 \ll \ell$ is fulfilled \cite{PPCunp}.
212:
213: \subsection{ N=2}
214: First we consider the case of two particles per lattice well, which can be
215: solved analytically. The Hamiltonian (\ref{ham}) is
216: diagonal in the states $\vert m_r,m_{cm} \rangle$ of well defined
217: relative ($m_r$) and center of mass ($m_{cm}$) angular momentum:
218: \begin{equation} \label{H2}
219: H=\sum_{m_r,m_{cm}}E_{m_r,m_{cm}}\vert m_r,m_{cm}\rangle \langle m_r, m_{cm} \vert,
220: \end{equation}
221: with $E_{m_r,m_{cm}}=\delta_ {m_r,0}
222: \,\eta+(1-\Omega/\omega)(m_r+m_{cm})$. We note that due to the
223: restriction to s-wave scattering, only particles with zero
224: relative angular momentum feel the interaction energy. It follows
225: that for $\Omega/\omega <1-\eta/2$ the ground state of the system
226: is $\vert 0,0\rangle$ (with total angular momentum $L=0$), which
227: is not entangled, whereas for $\Omega/\omega >1-\eta/2$ the state
228: $\vert 2,0\rangle$ (with $L=2$) becomes energetically favourable.
229: This state, $\langle z_1, z_2 \vert 2,0\rangle \propto
230: \left(z_1-z_2\right)^2 e^{-\vert z_1 \vert ^2/2} e^{-\vert z_2
231: \vert ^2/2}$, is clearly entangled since it cannot be written as a
232: product of two single particle wave functions. It is the Laughlin
233: state $|\psi_L\rangle$ for two particles at filling factor
234: $\nu=1/2$ \cite{L83}. In order to quantify the entanglement of
235: this state we write it in the basis of states $\vert m_1 m_2 \rangle$ with well defined single-particle angular momentum. Then the Laughlin state takes
236: the form of a pure two qutrit state: $|\psi_L\rangle= \frac{1}{2}
237: \left( \vert 02 \rangle + \vert 20 \rangle \right)
238: -\frac{1}{\sqrt{2}} |11\rangle$. This is already the Schmidt decomposition of the state, and the entropy of entanglement \cite{Bennett96} can immediately calculated to be $E(|\psi_L \ra)=1.5$. This value is close to $\log_2 3$, corresponding to a maximally entangled pure two qutrit state.
239: \subsection{\em N=3}
240: The case of three particles per lattice well is very similiar to the situation for $N=2$. The $1/2$-Laughlin state ($L=6$) emerges as ground state after an intermediate state with odd angular momentum $L=3$. As we will explain in the next section, ground states with odd angular momentum cannot be reached using our proposal. Hence we now focus on a setup with four particles per lattice well, for which an interesting sequence of prominent FQH states arises.
241: \subsection{\em N=4}
242: In order to obtain the multi-particle energy spectrum, we have exactly diagonalized the Hamiltonian (\ref{ham}) numerically.
243: As the frequency of rotation $\Omega$ increases the ground state of the system passes through a sequence of states
244: with increasing and well defined total angular momentum $L=0, 4, 8, 12$ (see Fig. \ref{Espec4}).
245: %%%%%%%%%%%%%%%%%%%%%%% Egap spec N=4 %%%%%%%%%%%%%%%
246: \begin{figure}[h]
247: \centering
248: \epsfig{file=States.eps,width=\linewidth}
249: % \includegraphics[scale=0.4]{States-b.eps}
250: \caption{Lowest two eigenenergies (in units of $\hbar \omega$) of the Hamiltonian (\ref{ham}) for 4 particles and $\eta=0.1$ as a function of
251: the trap rotation frequency $\Omega / \omega$.
252: The circles mark the
253: level crossings and $L$ denotes the total angular momentum of the
254: ground state. The ground state sequence can be
255: identified as follows (with fidelity given in brackets): L= 0
256: Gaussian ground state (exact) , L=4 Pfaffian state ($0.95$), L=8
257: quasiparticle state ($0.98$), L=12 Laughlin state (exact).The
258: change of angular momentum can readily be obtained from the
259: increasing width of the density distribution depicted below.}
260: \label{Espec4}
261: \end{figure}
262: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
263: These states can be identified as follows:
264: The state with $L=0$ is a trivial non-entangled state in which all the atoms are condensed in the
265: single particle Gaussian state with angular momentum $m=0$.
266: The first nontrivial ground
267: state is the $L=4$ state. This state is not, as one might expect, a
268: single vortex state, in which all the particles would be condensed in the single particle state $m=1$.
269: In contrast, this state is highly entangled and is very close (fidelity $0.95$) to the
270: well-known Pfaffian state:
271: \be
272: \psi_{Pf}({[z]})= \prod_{i<j}^4 (z_i-z_j) \ \textrm{Pf}\left(
273: \frac{1}{z_i-z_j}\right) \ .
274: \ee
275: This state is specially interesting,
276: also in the context of quantum information theory, because its
277: elementary excitations are known to exhibit non-abelian
278: statistics \cite{Kit97}. The next stable state in row ($L=8$) can be very well
279: characterized (fidelity $0.98$) by a Laughlin quasiparticle state:
280: \be
281: \psi_{QP}([z])=\frac{\partial}{\partial z_1}
282: \ldots \frac{\partial}{\partial z_4} \ \psi_L \ .
283: \ee
284: This state is
285: the counterpart of the quasihole excitation, which has previously
286: been studied in the context of $1/2$-anyons in rotating
287: Bose-Einstein condensates \cite{P01}. Finally, the last stable
288: state is identical to the $\frac{1}{2}$-Laughlin state, which we
289: have already encountered in the case of two particles per well:
290: \be \label{psiL}
291: \psi_L([z])= \prod_{i<j}^4 (z_i-z_j)^2 \prod_k^4 e^{|z_k|^2/2} \ .
292: \ee
293: This state is an exact eigenstate of (\ref{ham}) with zero interaction energy. In
294: Fig. \ref{Espec4} we have plotted the density distribution in the $x_1,x_2$ plane of
295: the different stable ground states. As the frequency of rotation $\Omega/\omega$ increases
296: the wave function spreads, and the interaction between the atoms decreases.
297:
298:
299:
300: \section{Adiabatic paths to entangled states}
301: The sequence of
302: entangled states we have described above cannot be obtained by
303: simply adiabatically increasing the frequency of rotation
304: $\Omega$. For the rotational symmetry leads to level crossings
305: between different angular momentum states (Fig. \ref{Espec4}). In
306: order to pass adiabatically from the zero angular momentum ground
307: state to higher angular momentum states the spherical symmetry of
308: the trapping potential has to be broken. For our optical lattice
309: setup this can be achieved for example by deforming the formerly
310: isotropic trapping potential on each well and letting the
311: deformation rotate with frequency $\Omega$ \cite{Bpriv}. In the rotating frame
312: the new trapping potential has the form $V_p \propto \left(\omega
313: +\Delta \omega\right)^2 x_1^2 + \omega^2 x_2^2$, and the new
314: Hamiltonian is $H+H_\epsilon$, with \be \label{Heps}
315: H_{\epsilon}=\frac{\epsilon}{4} \sum_m \beta_m a_{m+2}^\dagger
316: a_{m}^{} + (m+1) a_{m}^\dagger a_{m}^{} + \textrm{h.c.}, \ee
317: where $\beta_m=\sqrt{(m+2)(m+1)}$ and $\epsilon=\Delta \omega/
318: \omega$ is a small parameter. The perturbation (\ref{Heps}) leads
319: to quadrupole excitations, so that states whose total angular
320: momenta differ by two are coupled.
321: %As a consequence, avoided level
322: %crossings emerge (see Figure...) and states with larger angular
323: %momentum can be reached.
324:
325: In order to design appropriate adiabatic paths to the entangled
326: states described above, we have computed numerically the energy
327: gap between the ground and first excited state as a function of
328: the parameters $\Omega/ \omega$ and $\epsilon$ for $N=2$ and $N=4$ (Fig. \ref{paths}).
329:
330: %%%%%%%%%%%%%%%%%%%%%%% Paths %%%%%%%%%%%%%%%
331: \begin{figure}[h]
332:
333: \begin{center}
334: % \epsfig{file=plots/paths.eps,width=\linewidth}
335: \epsfig{file=pathsN2.eps,width=0.85\linewidth}
336: \epsfig{file=pathsN4-1.eps,width=0.85\linewidth}
337:
338: \end{center}
339: \caption{Energy gap in units of $\hbar \omega$ between the ground and first excited state
340: as a function of the rotation frequency $\Omega / \omega$ and the
341: trap deformation $\epsilon$ for an interaction strength
342: $\eta=0.1$. The black lines mark appropriate paths in parameter
343: space for adiabatic ground state evolution starting from the $L=0$ state. The adiabatic evolution times have
344: been calculated for a typical trapping frequency $\omega \simeq
345: (2 \pi) 30$kHz. Top ($N=2$): For a final fidelity
346: $\mathcal{F}=| \la \psi(T) | \psi_L \ra|^2=0.99$ the Laughlin
347: state (L=2) can be reached within T=6.5 ms. Bottom ($N=4$): Adiabatic path, evolution time T and fidelity
348: $\mathcal{F}$ for the following final states (see Fig.
349: \ref{Espec4}): (a) Pfaffian state: T=8 ms, $\mathcal{F}=0.99$; (b)
350: Quasiparticle state: T=12 ms, $\mathcal{F}=0.99$; (c) Laughlin state:
351: T=215 ms, $\mathcal{F}=0.97$.}
352: \label{paths}
353: \end{figure}
354: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
355:
356:
357: We first note that the isolines of constant energy gap show an
358: approximately linear behavior. This feature can be easily
359: understood from a perturbative treatment of the Hamiltonian
360: (\ref{Heps}). To first order, the energy of states with angular
361: momentum $L$ is shifted by an amount $\epsilon L/4 $. Therefore
362: the gap profile for a given $\epsilon$ is very similar to the one
363: for $\epsilon=0$ but shifted an amount $\sim \epsilon$ to larger
364: rotation frequencies. As expected, we find that for $\epsilon \ne
365: 0$ avoided crossings emerge (see Fig. \ref{crossing}). The energy gap of the avoided
366: crossings does, however, not in general increase monotonically
367: with the deformation $\epsilon$. Due to the interplay with other
368: excited states, ``saddlepoints'' appear in the gap profile, which
369: makes the design of appropriate adiabatic paths a nontrivial task.
370: For the stable entangled states of $N=2,4$ identified above these
371: paths are depicted in Fig. \ref{paths}. The actual time needed for
372: the adiabatic path depends on the number of particles as well as
373: on the state we want to achieve. For a typical trapping frequency
374: $\omega\simeq (2 \pi) 30$kHz and an interaction coupling $\eta=0.1$,
375: the evolution times for the $N=2$ Laughlin state as well as for the
376: $L=4$ and $L=8$ states for $N=4$ are of the order of 10 ms. In
377: contrast, the evolution time for the $N=4$ Laughlin state is one
378: order of magnitude larger. We can understand this result in the
379: following way. For the case of $N=2$ direct coupling of the $L=0$
380: state to the $L=2$ Laughlin state is mediated by (\ref{Heps}). For
381: the case of $N=4$ there is no direct coupling between the ground
382: states, since their angular momenta differ by 4. But, as one can
383: see from the spectrum in the vicinity of the crossing to the state
384: $L=4$ (Fig. \ref{crossing}), there is a state with $L=2$ near
385: the crossing that mediates the coupling between the $L=0$ and the
386: $L=4$ state. A similar situation occurs for the crossing to the
387: $L=8$ state. However, there is no such intermediate state in
388: direct proximity of the crossing to the $N=4$ Laughlin state,
389: which leads to a decrease of the energy gap by one order of
390: magnitude.
391:
392: Let us also comment on the situation $N=3$. Here a ground state with odd angular momentum ($L=3$) arises. From the nature of the perturbation (\ref{Heps}) it is clear that ground state evolution is not possible. However, we have shown \cite{PPCunp} that the $1/2$-Laughlin state can be reached by designing appropriate adiabatic paths via excited levels.
393:
394: %%%%%%%%%%%%%%%%%%%%%%% Crossing %%%%%%%%%%%%%%%
395: \begin{figure}[h]
396:
397: \begin{center}
398: % \epsfig{file=plots/paths.eps,width=\linewidth}
399: \epsfig{file=crossing-co.eps,width=0.49\linewidth}
400: \epsfig{file=gap-co.eps,width=0.49\linewidth}
401: \end{center}
402: \caption{Left side: Energy spectrum (in units $\hbar \omega$) for N=4 and $\eta=0.1$ in the
403: vicinity of the first level crossing from the L=0 to the L=4
404: state (see Fig. \ref{Espec4} left circle). Using quadrupole excitations ($|\Delta L|=2$) coupling
405: between these states is provided by the intermediate state L=2.
406: Right side: Emergence of an avoided level crossing for a trap
407: deformation $ \epsilon =0.06$.} \label{crossing}
408: \end{figure}
409: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
410:
411:
412: \section{ Feasibility}
413: Let us now discuss the experimental
414: feasibility of our proposal for a small number of particles $N$. First of all we have assumed that the
415: lattice wells can be treated independently. This requires that
416: the overlap of the Wannier functions on neighbouring sites is
417: small. We can estimate how intense the laser light should be in order to neglect this overlap by requiring
418: the size of the highest occupied angular momentum single particle state ($\approx \sqrt{2 N-1} \ell$) to be
419: much smaller than the separation between lattice sites, $a=\pi/k$. This leads to
420: the condition: $(V_0/E_R)^{1/4} \gg \sqrt{2N-1}/\pi$. Numerical calculations of overlap integrals between adjacent sites have confirmed that for $N=2(4)$
421: lattice depths of $V_0/E_R \approx 30(50)$ are required. Note that these values also guarantee the validity of the harmonic approximation.
422: Moreover we have assumed that the available single particle states
423: on each well lie within the LLL. This implies that the typical
424: energies per particle have to be much smaller than the energy gap
425: to the next Landau level, $\hbar \omega$. For the limiting cases of the $L=0$ state and the Laughlin state,
426: this leads to the
427: conditions $(N-1) \eta/ 2, (N-1) (1-\Omega/ \omega) \ll 1$, which
428: are easily fulfilled for typical interaction strengths
429: ($\eta\sim 0.1$) and small $N$.
430:
431: Finally, in order to adiabatically achieve the entangled states identified above further conditions are required.
432: We analyze the most restrictive case, which corresponds to the
433: Laughlin state. First of
434: all the frequency of rotation has to be very close to the
435: centrifugal limit. Let us find a lower bound to the critical rotation frequency at which the
436: crossing to the Laughlin state appears. This can be done by calculating the rotation frequency at which the
437: Laughlin quasiparticle state, $\psi_{QP}([z])=\frac{\partial}{\partial z_1}
438: \ldots \frac{\partial}{\partial z_N} \ \psi_L $, becomes equal in energy to the Laughlin state. Since the quasiparticle state has
439: $N$ units of angular momentum less than the Laughlin state and an interaction energy $\lesssim \eta$, it follows that
440: $\Omega_c/ \omega \geq 1- \eta/N$. For the cases of $N=2(4)$ this condition is in agreement with the exact
441: values found above.
442: Secondly, the evolution time required for the adiabatic path has to be much smaller than the typical decoherence time.
443: We can estimate this time in the following way. Given the critical frequency above and that the position of the avoided crossing
444: is displaced to larger rotation frequencies an amount proportional to $\epsilon$ it follows that the maximum $\epsilon$ we can have
445: is $\sim \eta/N$, corresponding to a rotation frequency $\Omega/\omega=1$.
446: Assuming an energy gap $\approx \epsilon$ it follows that the typical evolution time
447: scales as $N\eta$. For the case of $N=2(4)$ and typical $\eta$ and $\omega$ these times
448: are of the order of tens of miliseconds
449: as exactly found above, which is much smaller than the typical life time of the lattice states.
450: Finally, a high
451: degree of control of the parameters $\Omega/\omega$ and $\epsilon$
452: is required to perform the appropriate adiabatic paths. The
453: required precision scales again as $\eta/N$, which for the case of $N=4$
454: means a control of the parameter space up to the second digit.
455:
456:
457: From our analysis it follows that the
458: adiabatic creation of the Laughlin state
459: by means of low angular
460: momentum excitations, as quadrupole excitations, becomes very
461: difficult in samples with large number of particles \cite{Boulder, Dalib04}. Even if the
462: centrifugal limit is possible to achieve, as it happens when
463: including an additional $r^4$ trapping potential \cite{Dalib04},
464: the adiabatic creation of the Laughlin state is still very
465: demanding. One reason is that the rotation frequency and the trap
466: deformation have to be controlled within a precision that also
467: scales linearly with $N$. Furthermore we point out that only the
468: exact knowledge of the multi-particle energy spectrum allows to
469: design adiabatic paths that minimize the evolution time.
470:
471: \section{Detection}
472: We discuss now how the entangled states described
473: above can be detected experimentally by measuring different
474: properties of the states. As an important feature of our lattice
475: setup of independent wells, we note that any signal will be highly
476: enhanced by a factor equal to the number of occupied lattice sites ( $\sim 150,000$ \cite{Bloch02}).
477:
478: i) \emph{Density profiles}. A very characteristic feature of the
479: entangled states with large angular momentum that we have
480: described is that they exhibit a much more extended density
481: distribution than the non-entangled $L=0$ state, in which the
482: particles are much more confined in space. For the $1/2$-Laughlin state the typical radius is given by $\bar r\approx \sqrt{2 N-1} \ \ell$. In the case of
483: $N=2(4)$ this results in a radius that is $\sim 2(3)$ times
484: larger than in the case of the condensate. As proposed in
485: \cite{RC03} the density profile of states within the LLL can be
486: measured in a time of flight (TOF) image of the atomic system,
487: since the momentum distribution coincides with the density profile
488: for LLL states. In our case of independent
489: 3D wells, a TOF absorption picture after expansion time $t$ will exhibit a broad central peak of the form:
490: \be \label{rhotf}
491: \rho({\bf r},
492: t)\approx\frac{N_{s}}{(\omega t)^3 } \ |\rho_0( -i z / (
493: \omega t), x_3/ (\omega t)|^2 \ .
494: \ee
495: Here, $\rho_0(z,x_3)$ is the initial density distribution on a single well. In the TOF image it is enhanced by a factor proportional to the number of lattice sites $N_s$ and rescaled by a factor $\omega t \gg 1$. The $\pi/2$ rotation $z\rightarrow -i z$ leaves isotropic states, like the FQH states described above, unaffected.
496: The underlying assumption of free
497: (interactionless) expansion is justified, since the interaction
498: energy is small compared to the kinetic energy (in the stationary
499: frame).
500:
501:
502: ii)\emph{Angular momentum}. For any state within the
503: LLL integration over the density distribution gives $\int d{\bf
504: r}\ r^2 \rho({\bf r})=L+N$. Thus in the limit of weak interaction the total angular momentum can be extracted directly from the TOF picture.
505:
506: iii)\emph{Correlation functions}. Here we propose a novel technique that
507: makes directly use of the rich possibilities offered by the
508: optical lattice setup and which allows to measure both the
509: $g_1=\langle \psi^\dagger ({\bf r}) \psi ({\bf r'})\rangle$ and
510: $g_2=\langle \psi^\dagger ({\bf r}) \psi^\dagger ({\bf r'})
511: \psi({\bf r})\psi ({\bf r'}) \rangle$ correlation functions. The
512: $g_2$ correlation function is for instance very characteristic for
513: a Laughlin state. Since particles can only be at least in relative
514: angular momentum $m_r=2$ it follows that $g_2\propto \lvert
515: r-r'\vert^4$. This behavior reveals the
516: $\frac{1}{2}$-fractional nature of this Laughlin state.
517:
518: We consider two species $a$ and $b$ (hyperfine levels) of bosonic
519: atoms, which can be coupled via Raman transitions. We start with
520: atoms in level $a$ and create the entangled state of interest $|
521: \Psi_{i} \rangle$ with the method described above. Next we apply a
522: $\pi/2$-pulse with the laser and create an equal superposition of
523: $a$ and $b$ states. Finally, we shift the lattice potential trapping atoms
524: of type $b$ (as proposed in \cite{Jaksch00} and realized in
525: \cite{Bloch03}) by a distance ${\bf r_0}$ small compared to the
526: lattice spacing and perform another $\pi/2$-pulse. In the Heisenberg picture this procedure corresponds to the following transformation of the field operator for species $a$:
527: \be
528: \psi_a({\bf r}) \rightarrow \psi_a({\bf
529: r})+\psi_a({\bf r+r_0}) \ .
530: \ee
531: Thus the density distribution of
532: atoms of type $a$ in this new state $| \Psi_{f} \rangle$ contains information
533: about the $g_1$ correlation function of the original state:
534: \bea
535: & &\langle \Psi_f|\psi_a^\dagger({\bf r})\psi_a({\bf
536: r})|\Psi_f\rangle= \\
537: & &\langle \Psi_i|(\psi_a^\dagger({\bf r})
538: +\psi_a^\dagger({\bf r+r_0})) (\psi_a({\bf r})+ \psi_a({\bf
539: r+r_0}) )|\Psi_i\rangle \nonumber \ .
540: \eea
541: Using this procedure we can also measure
542: higher order correlation functions like $g_2$. In this case
543: measuring the interaction energy of the final state will allow us to
544: calculate the $g_2$ of the initial state. For instance, for the
545: Laughlin state we have:
546: \bea
547: & & E_{int}({\bf r_0})= \\
548: & &\frac{\pi \eta}{4} \int d{\bf r} \langle \Psi_i |
549: \psi_a^\dagger({\bf r})\psi_a^\dagger({\bf r+r_0})\psi_a({\bf
550: r})\psi_a({\bf r+r_0}) | \Psi_i \rangle \ . \nonumber
551: \eea
552: The interaction energy is, unfortunately, not directly accessible
553: experimentally. However, the total energy of the final state can
554: be obtained from integrating over the TOF absorption picture,
555: since energy is conserved during the time of flight. For small
556: coupling $\eta$, however, the measurable effect due to
557: interactions will be small compared to the kinetic part of the
558: energy. In addition, the kinetic energy itself shows a significant
559: dependence on the shifting $r_0$, which has to be distinguished
560: from the interaction. Hence, we propose to tune the scattering
561: length $a_s$ (e.g. via a photo association induced Feshbach
562: resonance \cite{Ketterle99}) and to measure the interaction energy
563: both in the weak and strong scattering limit. The difference would
564: then reveal the characteristic behavior of the $g_2$ correlation
565: function.
566:
567: We finally note that, as a further way of detection for the $N=4$
568: Laughlin state, a strong reduction of the three body losses should
569: be observed.
570: \section{conclusion}
571: In conclusion, we have shown how to motionally entangle a small
572: number of particles into a sequence of interesting FQH states. We
573: have fully addressed the adiabatic creation of these states and
574: proposed new techniques for their experimental detection.
575: \section{Acknowledgements}
576: We acknowledge helpful discussions with I. Bloch, J. Garc\'ia-Ripoll and M. Greiner. This work was supported in part by EU IST projects (RESQ and QUPRODIS), the DFG, and the Kompetenznetzwerk ``Quanteninformationsverarbeitung'' der Bayerischen Staatsregierung.
577:
578:
579: \begin{thebibliography}{10}
580: \bibitem{Sackett00}C. A. Sackett, D. Kielpinski, B. E. King, C. Langer, V. Meyer, C. J. Myatt, M. Rowe, Q. A. Turchette,
581: W. M. Itano, D. J. Wienland, and C. Monroe, Nature (London), {\bf
582: 404}, 256 (2000).
583: \bibitem{Bloch03}O. Mandel, M. Greiner, A. Widera, T. Rom, T. W. H{\"a}nsch, and I. Bloch, Nature, {\bf 425}, 937 (2003).
584: \bibitem{Bloch02} M. Greiner, O. Mandel, T. Esslinger, T.W. H{\"a}nsch and I. Bloch, Nature, {\bf 415}, 39 (2002).
585: \bibitem{Ess04} H. Moritz, T. St{\"o}ferle, M. K{\"o}hl, and T. Esslinger,
586: Phys. Rev. Lett. {\bf 91}, 250402 (2003); T. St{\"o}ferle, H. Moritz, C. Schori, M. K{\"o}hl, and T. Esslinger,
587: Phys. Rev. Lett. {\bf 92}, 130403 (2004).
588: \bibitem{Ph03} S. Peil, J. V. Porto, B. L. Tolra, J. M. Obrecht, B. E. King, M. Subbotin, S. L. Rolston, and W. D. Phillips,
589: Phys. Rev. A {\bf 67}, 051603 (2003).
590: \bibitem{QHEbook}T. Chakraborty, P. Pietil\"{a}inen, The Quantum Hall
591: Effects: Fractional and Integral, Springer, Berlin, (1995); A. H.
592: MacDonald, cond-mat/9410047.
593: \bibitem{WG00} N.K. Wilkin and J.M.F. Gunn, Phys. Rev. Lett. {\bf 84}, 6
594: (2000); N. K. Wilkin and J.M.F. Gunn, Phys. Rev. Lett. {\bf 87},
595: 120405 (2001); J. Sinova, C. B. Hanna, and A. H. MacDonald, Phys.
596: Rev. Lett. {\bf 89}, 030403 (2002); T.-L. Ho and E. Mueller, Phys.
597: Rev. Lett. {\bf 89}, 050401 (2002); N. Regnault and Th. Jolicoeur, Phys.
598: Rev. Lett. {\bf 91}, 030402 (2003); T.K. Ghosh and G. Baskaran, Phys.
599: Rev. A {\bf 69}, 023603 (2004);M. A. Baranov, K. Osterloch,
600: M. Lewenstein, cond-mat/0404329.
601: \bibitem{P01} B. Paredes, P. Fedichev, J.I. Cirac, and P. Zoller, Phys. Rev. Lett. {\bf 87}, 010402 (2001),
602: B. Paredes, P. Zoller, J. I. Cirac, Phys. Rev. A 66, 033609
603: (2002).
604: \bibitem{Bpriv} M. Greiner and I. Bloch, private communication.
605: \bibitem{Pfaffian91} G. Moore and N. Read, Nucl. Phys. B {\bf 360}, 362 (1991).
606: \bibitem{L83} R. B. Laughlin, Phys. Rev. Lett. {\bf 50}, 1395 (1983).
607: \bibitem{Bloch04}A. Widera, O. Mandel, M. Greiner, S. Kreim, T. W. H{\"a}nsch, and I. Bloch,
608: Phys. Rev. Lett. {\bf 92}, 160406 (2004).
609: \bibitem{Com} Typically the filling factor $N$ is not homogeneous,
610: because the Gaussian intensity profile of the laser beams results
611: in an additional weak harmonic confinement of the atoms. However, it has
612: been demonstrated that commensurate filling can be achieved using an entanglement interferometer \cite{Bloch04}.
613:
614: %\bibitem{Blochtbp} I. Bloch et al, to be published
615:
616:
617: \bibitem{PPCunp} M. Popp, B. Paredes, and I. Cirac, unpublished.
618: \bibitem{Kit97} A. Yu. Kitaev, quant-ph/9707021
619: \bibitem{Bennett96} C.H. Bennett, H.J. Bernstein, S. Popescu, and B. Schumacher, Phys. Rev. A. {\bf 53}, 2046 (1996).
620: \bibitem{Dalib04} V. Bretin, S. Stock, Y. Seurin, and J. Dalibard,
621: Phys. Rev. Lett. {\bf 92}, 050403 (2004).
622: \bibitem{Boulder} V. Schweikhard, I. Coddington, P. Engels, V. P. Mogendorff, and E. A. Cornell,
623: Phys. Rev. Lett. {\bf 92}, 040404 (2004).
624: \bibitem{RC03} N. Read and N.R. Cooper, preprint cond-mat/0306378.
625: \bibitem{Jaksch00}D. Jaksch, J. I. Cirac, P. Zoller, S. L. Rolston, R. C\^{0}t\'{e}, and M. D. Lukin
626: Phys. Rev. Lett. 85, 2208-2211 (2000).
627: \bibitem{Ketterle99}J. Stenger, S. Inouye, M. R. Andrews, H.-J. Miesner, D. M. Stamper-Kurn, and W.
628: Ketterle, Phys. Rev. Lett. 82, 2422 (1999).
629:
630:
631:
632:
633:
634:
635:
636:
637:
638: \end{thebibliography}
639:
640:
641:
642: \end{document}
643: