1: \documentstyle[psfig,aps,pra,twocolumn]{revtex}
2:
3: \begin{document}
4: \draft
5:
6: \title{Bell-inequality violation with ``thermal'' radiation\thanks{This
7: paper is dedicated to Professor Jan Pe\v{r}ina in the occasion of
8: his $65^{\rm th}$ birthday.}}
9:
10:
11: \author{Radim Filip,$^1$ Miloslav Du\v{s}ek,$^{2,1}$
12: Jarom\'\i r Fiur\' a\v sek,$^1$ and Ladislav Mi\v sta$^{1}$}
13:
14: \address{$^1$Department of Optics, Palack\'y University, 17.~listopadu
15: 50, 772\,00 Olomouc, Czech~Republic}
16:
17: \address{$^2$Institute of Physics, Slovak Academy of Sciences,
18: D\'{u}bravsk\'{a} cesta 9, 842\,28 Bratislava, Slovakia}
19:
20: \date{\today}
21:
22:
23: \maketitle
24:
25:
26: \begin{abstract}
27: The model of a quantum-optical device for a conditional preparation
28: of entangled states from input mixed states is presented. It is
29: demonstrated that even thermal or pseudo-thermal radiation can be
30: entangled in such a way, that Bell-inequalities are violated.
31: \end{abstract}
32:
33: \pacs{PACS number(s):~03.65.Bz, 42.50.Dv}
34:
35: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
36:
37: \section{Introduction}
38:
39: In last decades, the phenomenon of entanglement between two
40: spatially separated photons was investigated both experimentally
41: and theoretically mainly in order to
42: show that quantum mechanics is not a local realistic theory \cite{micro}.
43: As a counterpart to the particle-like behavior of photons, the
44: entanglement of coherent states, that can be considered as
45: the quantum analogue of deterministic light waves, was examined
46: \cite{CohSt,Gerry99,nas}. In both the cases, the entangled states, that were
47: used to test Bell inequalities, were usually considered to be pure
48: states. Recently, the entanglement of mixed states has been analysed to
49: understand, how the disorder influences on the amount of entanglement
50: \cite{mixx}.
51:
52: In this report, we examine a new situation, when an
53: entangling device prepares the entangled states of radiation from
54: mixed states (thermal or pseudo-thermal light) at the input. Similarly
55: to the idea presented in Ref.~\cite{nas}, the entangling device can
56: produce a four-mode entangled state with two mixed states and two
57: vacuum states.
58: It is shown, that even for very disordered states Bell
59: inequalities can strongly be violated. If there is a narrow
60: frequency portion of thermal radiation in the input of the
61: entangling device then Bell inequalities are violated when the
62: frequency of radiation is ``low'' and the temperature of thermal
63: source is ``high''. For a pseudo-thermal radiation the violation of
64: Bell inequalities is even more significant. In addition,
65: the violation can be enhanced for both the cases of radiations,
66: if a lot of different modes are entangled with vacuum state.
67: Thus almost the maximal Bell inequality violation can be achieved with
68: such thermal states exhibiting a large entropy.
69:
70: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
71:
72: \section{Preparation of mixed entangled state}
73: \label{device}
74:
75: We consider two separate systems $A$ and $B$ which consist locally
76: of two modes $A1$, $A2$ and $B1$ and $B2$. All modes are initially
77: unentangled. We further assume that the density matrices
78: of these four modes are diagonal in orthonormal Fock
79: (number-state) bases $\{ |n\rangle \}$ and that the modes $A2$ and $B2$
80: are in vacuum states,
81: \begin{eqnarray}
82: \hat{\rho}_{A}&=&\sum_{n}p_{n}|n\rangle_{A1}\langle n|\otimes
83: |0\rangle_{A2}\langle 0|,\nonumber\\
84: \hat{\rho}_{B}&=&\sum_{m}r_{m}|m\rangle_{B1}\langle m|\otimes
85: |0\rangle_{B2}\langle 0|.
86: \end{eqnarray}
87: The density matrix of the total system has
88: a factorized form $\hat{\rho}_{\rm in}=\hat{\rho}_{A}
89: \otimes \hat{\rho}_{B}$. Now, one can consider a conditional
90: operation which enables to prepare the following
91: entangled states (for $n\ne 0$ or $m\ne 0$)
92: \begin{eqnarray}
93: \label{oper}
94: |\psi_{nm}\rangle=\frac{1}{\sqrt{2}}\left(
95: |n\rangle_{A1}|0\rangle_{A2}
96: |0\rangle_{B1}|m\rangle_{B2}\right.\nonumber\\
97: \left.-|0\rangle_{A1}|n\rangle_{A2}
98: |m\rangle_{B1}|0\rangle_{B2}\right).
99: \end{eqnarray}
100: The entangling device prepares, for each $m,n$, the analogue of a
101: singlet state, that was often employed to test Bell-type
102: inequalities.
103: Thus the initial density matrix $\hat{\rho}_{\rm in}$ is
104: transformed into the form
105: \begin{equation}
106: \label{final}
107: \hat{\rho}_{\rm out}= N\sum_{mn}p_{n}r_{m}
108: (1-\delta_{n0}\delta_{m0})
109: |\psi_{nm}\rangle\langle \psi_{nm}|,
110: \end{equation}
111: where $N=\left[ \sum_{nm} p_n r_m (1-\delta_{n0}\delta_{m0})
112: \right]^{-1} = (1- p_0 r_0)^{-1}$.
113:
114: If there is at least one $n>0$ and one $m>0$ such that $p_n\ne 0$ and
115: $r_m\ne 0$ then state (\ref{final}) is entangled. It can be proved in a
116: very straightforward way using the so called transposition criterion
117: \cite{Transp}. This criterion says that if operator
118: ${\hat{\rho}}^{T_B}$, obtained from $\hat{\rho}$ by partial
119: transposition in subsystem $B$, is not positive the state $\hat{\rho}$
120: is entangled.
121: Partial transposition of
122: $$
123: \hat{\rho}_{\rm out} = \sum_{ijklmnst} \!\! \rho_{ijklmnst}\,
124: |i_{A1} j_{A2} k_{B1} l_{B2} \rangle \langle
125: m_{A1} n_{A2} s_{B1} t_{B2}|
126: $$
127: in basis $|i_{A1} j_{A2} k_{B1} l_{B2} \rangle \equiv
128: |i\rangle_{A1}|j\rangle_{A2} |k\rangle_{B1}|l\rangle_{B2}$ gives
129: $$
130: \hat{\rho}_{\rm out}^{T_B} = \sum_{ijklmnst} \!\! \rho_{ijklmnst}\,
131: |i_{A1} j_{A2} s_{B1} t_{B2} \rangle \langle
132: m_{A1} n_{A2} k_{B1} l_{B2}|.
133: $$
134: Now, let us suppose vector
135: \begin{eqnarray*}
136: |\phi_{mn}\rangle=\frac{1}{\sqrt{2}}\left(
137: |0\rangle_{A1}|m\rangle_{A2}
138: |0\rangle_{B1}|n\rangle_{B2}\right. \\
139: \left.+|m\rangle_{A1}|0\rangle_{A2}
140: |n\rangle_{B1}|0\rangle_{B2}\right)
141: \end{eqnarray*}
142: where $m,n>0$ and calculate the following mean value
143: \begin{equation}
144: \langle \phi_{mn} | \,
145: {\hat{\rho}}_{\rm out}^{T_B}
146: \, | \phi_{mn}\rangle
147: = - N {p_m r_n \over 2}.
148: \label{RedCr}
149: \end{equation}
150: If $p_m \ne 0$ and $r_n \ne 0$,
151: this quantity is {\it negative}.
152: On the other hand, the entanglement of discussed states can often be
153: ``masked'' by the noise of original mixed states. E.g., conditional
154: von Neumann entropy, $S({\hat{\rho}}'_A)-S(\hat{\rho})$, is positive
155: for many particular cases here. Nevertheless, we will show that the
156: entanglement is ``strong'' enough to violate CHSH-Bell inequality.
157:
158: The proposed conditional operation can be, in principle, realized
159: in the following way (see Fig.~\ref{fig1}) \cite{nas}: Let us assume a
160: Mach-Zehnder (M-Z) interferometer with equal-length arms and with
161: one photon in its input.
162: Into both the arms of the interferometer we insert a nonlinear Kerr
163: medium effectively described by the following interaction Hamiltonian
164: \begin{equation}
165: \hat{H}_{I,i}=\hbar\kappa\hat{a}^{\dag}\hat{a}\hat{a}_{i1}^{\dag}\hat{a}_{i1},
166: \end{equation}
167: where $\hat{a}^{\dag}$ and $\hat{a}$ are the creation and annihilation
168: operators of the mode corresponding to the left (or right) arm in the
169: M-Z interferometer, $\hat{a}_{i1}^{\dag}$ and
170: $\hat{a}_{i1}$, with $i=A,B$, are the creation
171: and annihilation operators of modes
172: $A1$ (or $B1$), and $\kappa$ is a real interaction constant.
173:
174:
175: \begin{figure}
176: \medskip
177: \centerline{\psfig{width=0.9\hsize,file=fig1.eps,clip=}}
178: \medskip
179: \caption{Preparation device; $SA$ and $SB$ denote the sources of thermal
180: (pseudothermal) radiation, $SPS$ is a single photon source, $QND$ is
181: quantum nondemolition measurement performed by the Kerr interaction
182: and $D_{+}$ and $D_{-}$ are detectors.}
183: \label{fig1}
184: \end{figure}
185:
186:
187: If there is a photon in the left arm of the central M-Z interferometer and
188: the product $\kappa\tau_{\rm int}$, where $\tau_{\rm
189: int}$ is an effective interaction time,\footnote{Quantity $\tau_{\rm
190: int}$ has the meaning of the parameter of the device. It is not a
191: usual time variable. It represents the effective expression
192: of the fact that the nonlinear medium has finite dimensions.}
193: is set to be equal exactly to $\pi$ then the
194: described device realizes the phase shift $\pi$ in the left M-Z
195: interferometer, $A$, and effectively flips the modes $A1$ and $A2$ on the
196: output. On the other hand, if there is no photon in the left arm then
197: the states of modes $A1$ and $A2$ stay unchanged,
198: \begin{eqnarray}
199: \hat{U}_{A}|n\rangle_{A1}|0\rangle_{A2}|1\rangle&=&|0\rangle_{A1}|n\rangle_{A2}|1\rangle,\nonumber\\
200: \hat{U}_{A}|n\rangle_{A1}|0\rangle_{A2}|0\rangle&=&|n\rangle_{A1}|0\rangle_{A2}|0\rangle.
201: \end{eqnarray}
202: The same is true about the right arm of the central M-Z interferometer and
203: modes $B1$ and $B2$.
204: These unitary transformations $\hat{U}_{i}$, $i=A,B$, can be expressed as
205: \begin{equation}
206: \hat{U}_{i}=\hat{U}_{BS,i}^\dagger\hat{U}_{I,i}\hat{U}_{BS,i},
207: \end{equation}
208: where $\hat{U}_{BS,i}$ is the $50:50$ beam splitter transformation
209: and $\hat{U}_{I,i}$ accounts for the nonlinear interaction in Kerr medium,
210: \begin{eqnarray}
211: \hat{U}_{BS,i}&=&\exp\left[\frac{\pi}{4}(\hat{a}_{i2}^{\dag}\hat{a}_{i1}
212: -\hat{a}_{i1}^{\dag}\hat{a}_{i2})\right], \nonumber\\
213: \hat{U}_{I,i}&=&\exp(i\pi\hat{a}^{\dag}\hat{a}\hat{a}_{i1}^{\dag}\hat{a}_{i1}).
214: \end{eqnarray}
215: So, if the photon goes through
216: the left arm the modes $A1$ and $A2$ are flipped while
217: the state of system $B$ is unchanged. Completely symmetrical
218: situation occurs, if the photon goes through the right arm.
219:
220: Due to the path uncertainty of the photon in the interferometer
221: the state of the whole system after the interaction is given by the
222: formula
223: \begin{eqnarray}
224: |\Psi\rangle=\frac{1}{\sqrt{2}}(|0\rangle|1\rangle\,
225: |n\rangle_{A1}|0\rangle_{A2}
226: |0\rangle_{B1}|m\rangle_{B2}+\nonumber\\
227: +i|1\rangle|0\rangle\,
228: |0\rangle_{A1}|n\rangle_{A2}
229: |m\rangle_{B1}|0\rangle_{B2}),
230: \end{eqnarray}
231: where the kets without any subscript denote possible states of the
232: photon inside the M-Z interferometer situated in the center.
233: Which-way information is finally erased \cite{eraser} by a beam splitter
234: with amplitude reflectivity $i/\sqrt{2}$ (the last one in the M-Z
235: interferometer) followed by two photodetectors $D_{+}$ and $D_{-}$
236: (see Fig.~\ref{fig1}).
237: Depending on which one of these two detectors fires
238: we obtain one of two possible output states of modes $A1$, $A2$,
239: $B1$, and $B2$. Detector $D_{+}$ fires with probability
240: $w_{+}=(1+\delta_{n0}\delta_{m0})/2$ and if it clicks the following
241: state is obtained
242: \begin{eqnarray}
243: |\Psi_{+}\rangle&=&\frac{1}{\sqrt{2}}(
244: |n\rangle_{A1}|0\rangle_{A2} |0\rangle_{B1}|m\rangle_{B2}
245: \nonumber\\
246: &&+ |0\rangle_{A1}|n\rangle_{A2}
247: |m\rangle_{B1}|0\rangle_{B2}.
248: \end{eqnarray}
249: Similarly, detector $D_{-}$ clicks with probability
250: $w_{-}=(1-\delta_{n0}\delta_{m0})/2$ and when it fires one
251: obtains the state
252: \begin{eqnarray}
253: |\Psi_{-}\rangle&=&\frac{1}{\sqrt{2}}(
254: |0\rangle_{A1}|n\rangle_{A2}
255: |m\rangle_{B1}|0\rangle_{B2},
256: \nonumber\\
257: &&-|n\rangle_{A1}|0\rangle_{A2}
258: |0\rangle_{B1}|m\rangle_{B2}
259: \end{eqnarray}
260: which is exactly the considered state (\ref{oper}).
261:
262: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
263:
264:
265: \section{Bell-inequality violation}
266:
267: In order to demonstrate the violation of Bell inequalities one needs
268: local operations analogous to spin rotations. In our particular case
269: the following operations do the job
270: \begin{eqnarray}
271: |n\rangle_{1}|0\rangle_{2} &\rightarrow&
272: \cos\theta\,
273: |n\rangle_{1}|0\rangle_{2} + \sin\theta\,
274: |0\rangle_{1}|n\rangle_{2}
275: \mbox{~for~} n \ne 0,\nonumber\\
276: |0\rangle_{1}|n\rangle_{2}&\rightarrow&
277: -\sin\theta\,
278: |n\rangle_{1}|0\rangle_{2} + \cos\theta\,
279: |0\rangle_{1}|n\rangle_{2}
280: \mbox{~for~} n \ne 0, \nonumber\\
281: |0\rangle_{1}|0\rangle_{2}&\rightarrow& |0\rangle_{1}|0\rangle_{2},
282: \label{operace}
283: \end{eqnarray}
284: where $\theta$ is the parameter of transformation, it does not
285: depend on $n$.
286:
287: Bell-type experiment consists of two ``rotations'' according to
288: recipe (\ref{operace}), performed by two possibly space-like
289: separated observers, followed by realistic yes--no detection on each
290: mode. Each such detection has only two possible outcomes (detector
291: either fires or it does not), that can be described by projectors
292: $|0\rangle\langle 0|$ (for ``no'') and
293: $\hat{1}-|0\rangle\langle 0|=\sum_{n=1}^{\infty}|n\rangle\langle n|$
294: (for ``yes'').
295: Let us assign the following values to these outcomes: $z_i$=0
296: if the detector (in mode $i$) is quiet and $z_i$=1 if it clicks.
297: Then the results $X$ and $Y$ of local two-mode measurements
298: (including ``rotations'') performed by the first and the second
299: observer, respectively, can be expressed as
300: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
301: \begin{eqnarray}
302: \label{observation}
303: X(\theta) &=&
304: z_{A1}(\theta)-z_{A2}(\theta), \nonumber \\
305: Y(\theta) &=&
306: z_{B1}(\theta)-z_{B2}(\theta).
307: \end{eqnarray}
308: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
309: After the experiment is repeated many times and our two observers
310: compare their results, the mean value of Bell operator (for CHSH
311: inequalities) can be estimated,
312: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
313: \begin{equation}
314: \label{def-B}
315: {\cal B}
316: =|C(\theta_A,\theta_B)+C(\theta_A,\theta'_B)+C(\theta'_A,\theta _ B)
317: -C(\theta'_A,\theta'_B)|,
318: \end{equation}
319: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
320: where correlation function
321: \begin{equation}
322: C(\theta_1,\theta_2) \equiv
323: \sum_{j,k} X_j Y_k\, p(X_j,Y_k|\theta_A,\theta_B)
324: \label{corr}
325: \end{equation}
326: (summations go over all possible results).
327: Every local-realistic theory \cite{EPR} must fulfill the following
328: inequality ${\cal B} \le 2$ \cite{Bell}. However, it follows from
329: straightforward quantum-mechanical calculations that
330: for state (\ref{final}) the correlation function (\ref{corr}) reads
331: \begin{equation}
332: \label{corr2}
333: C(\theta_{A},\theta_{B})=-\cos\left[2(\theta_{A} - \theta_{B})\right]
334: \frac{(1-p_0)(1-r_0)}{1-p_0 r_0}.
335: \end{equation}
336: Therefore the results of the above mentioned local measurements
337: performed on state (\ref{final}) {\it violate\/} inequality ${\cal B}
338: \le 2$ in principial. Maximal violation,
339: \begin{equation}\label{max}
340: {\cal B}_{\rm max}=2\sqrt{2} \, \frac{(1-p_0)(1-r_0)}{1-p_0 r_0},
341: \end{equation}
342: occurs for the angles
343: \begin{equation}
344: \label{uhly}
345: \theta_{A}=0,~~\theta'_{A}=\frac{\pi}{4}~~
346: \theta_{B}=\frac{\pi}{8},~~\theta'_{B}=-\frac{\pi}{8}.
347: \end{equation}
348: If both the mixed states have the same overlap with vacuum state
349: $p_{0}=r_{0}$, the condition for the violation of
350: Bell inequality for the considered angles is given in a simple form
351: \begin{equation}
352: p_{0}<\frac{\sqrt{2}-1}{\sqrt{2}+1}\approx 0.1716.
353: \end{equation}
354: As one can see the maximum value of ${\cal B}$ depends on the
355: probability of the presence of the vacuum state in the input density
356: matrices. Thus, if the input density matrices of systems $A1$ and
357: $B1$ do not contain the vacuum state the maximal violation of
358: CHSH-Bell inequality is the same as for the pure EPR maximally
359: entangled state of two spin-half particles. In the opposite case, the
360: mean value of Bell operator decreases as the contribution of the
361: vacuum state increases in the mixtures. It should be noticed
362: that for properly chosen local measurements the violation of
363: CHSH-Bell inequality does not depend on the randomness contained in
364: the mixture but only on the overlaps of the vacuum state and the
365: input density matrices.
366:
367: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
368:
369: \section{Thermal and pseudo-thermal radiation}
370:
371: There are two mixed states of special interest, namely
372: thermal radiation, exhibiting Bose-Einstein statistics, and
373: pseudo-thermal radiation, exhibiting Poissonian statistics.
374: Let us study now the entangled states prepared by the device
375: proposed in Sec.~\ref{device} when thermal and pseudo-thermal states
376: are at the input.
377:
378: A single mode of thermal radiation has the density matrix
379: \begin{equation}
380: \hat{\rho}=\sum_{n}\frac{\langle
381: n\rangle^{n}}{(1+\langle n\rangle)^{1+n}} \,|n\rangle\langle n|,
382: \end{equation}
383: where
384: \begin{equation}
385: \langle n\rangle=
386: \frac{1}{\exp(\frac{\hbar\omega}{k_{B}T})-1}.
387: \end{equation}
388: For example, if
389: the temperature of a radiation source (e.g., incandescent
390: lamp) $T\approx 3000\,$K and the optical frequency
391: $\omega\approx 2.5\times 10^{15}\,$Hz, the mean value of photon
392: number is $\langle n\rangle\approx 1.77\times 10^{-3}$.
393: The probability of the vacuum state in the mixture is
394: \begin{equation}
395: p_{0}=1-\exp\left(-\frac{\hbar\omega}{k_{B}T}\right)=\frac{1}{1+\langle
396: n\rangle},
397: \end{equation}
398: what leads to the value
399: $p_0\approx 0.9982$ for the above given data.
400: Thus in the optical region, the
401: overlap of vacuum and thermal light is too large and the
402: Bell-inequality violation does not occur.
403:
404: The dependence of the
405: maximal Bell-inequality violation on the parameter
406: $\beta_{i}=\hbar\omega_{i}/k_{B}T_{i}$, $i=A,B$
407: of particular modes $A1$, $B1$
408: can be simply evaluated:
409: \begin{eqnarray}
410: {\cal B}_{\rm
411: max}&=&2\sqrt{2}\frac{1}{\exp(\beta_{A})+\exp(\beta_{B})-1}=\nonumber\\
412: & &=2\sqrt{2}\frac{1}{1+\langle n\rangle_{A}^{-1}+\langle n\rangle_{B}^{-1}}
413: \end{eqnarray}
414: and
415: it is displayed in Fig.~\ref{fig2}. Only for very small $\beta_{A}$
416: and $\beta_{B}$, i.e., for high temperatures and small frequencies,
417: CHSH-Bell inequality is violated. Thus for the given temperature $T$
418: of both the thermal sources the infrared component of radiation
419: gives better results than the ultra-violate one. On the other hand,
420: for the fixed frequency $\omega$ of both the sources the higher
421: temperature leads to the stronger violation of Bell inequality. If
422: both the sources are identical the Bell-inequality violation occurs
423: only if the dimensionless parameter $\beta$ satisfies relation
424: $\beta<\ln\frac{\sqrt{2}+1}{2}\approx 0.1882$
425: or the mean number $\langle n\rangle$
426: is sufficiently large $\langle n\rangle> 2(\sqrt{2}+1)\approx 4.828$.
427: Consequently, for the visible component of
428: radiation the thermal sources must have an ``astronomical'' temperature
429: $T>101000\mbox{K}$, whereas for the infrared component with
430: $\omega\approx 5\times 10^{13}\mbox{Hz}$, temperature
431: $T>2021\mbox{K}$ is sufficient to obtain Bell-inequality violation.
432:
433:
434: \begin{figure}
435: \medskip
436: \centerline{\psfig{width=0.9\hsize,file=fig2.eps,clip=}}
437: \medskip
438: \caption{The maximal violation of the CHSH-Bell inequality for
439: thermal light as the function of parameters
440: $\beta_{A}=\hbar\omega_{A}/k_{B}T_{A}$ and
441: $\beta_{B}=\hbar\omega_{B}/k_{B}T_{B}$}
442: \label{fig2}
443: \end{figure}
444:
445:
446: Another interesting kind of mixed state is that corresponding to
447: pseudo-thermal light \cite{pseudoth}. Its density matrix can be written as
448: \begin{equation}\label{pseudo}
449: \hat{\rho}=\sum_{n}\frac{\langle n\rangle^{n}}{n!}e^{-\langle
450: n\rangle}|n\rangle\langle n|.
451: \end{equation}
452: Pseudo-thermal radiation can be obtained from an intensity-stabilized
453: single mode laser with the phase uniformly distributed in the
454: interval $\langle 0,2\pi)$.
455: In contrast to thermal radiation, the
456: maximally probable state in the mixture (\ref{pseudo}) is not vacuum
457: state but it is state $|n\rangle$, where $n$
458: corresponds approximately to the mean number of photons $\langle n
459: \rangle$. Thus the overlap of pseudo-thermal light with the vacuum
460: state is much less than for thermal light. The probability of the
461: vacuum state in the density matrix (\ref{pseudo}) is $p_{0}=\exp(-\langle
462: n\rangle)$. This leads to maximal Bell-inequality violation
463: \begin{equation}
464: {\cal B}_{\rm max}=2\sqrt{2}\frac{[1-\exp(-\langle
465: n\rangle_{A})][1-\exp(-\langle n\rangle_{B})]}{1-\exp[-(\langle
466: n\rangle_{A}+\langle n\rangle_{B})]}.
467: \end{equation}
468: From Fig.~\ref{fig3} one can see that in the case of
469: pseudo-thermal light the Bell-inequality violation is achieved for
470: less $\langle n\rangle_{A}$ and $\langle n\rangle_{B}$ than in
471: the case of thermal light. If one considers two identical pseudothermal
472: sources, then the Bell inequality is violated if $
473: \langle n\rangle>\ln\frac{\sqrt{2}+1}{\sqrt{2}-1}$.
474: As the pseudo-thermal light with such a mean photon number can
475: be experimentally achieved from the laser light in optical frequencies,
476: the violation can be obtained more simply than for the thermal light.
477:
478:
479: \begin{figure}
480: \medskip
481: \centerline{\psfig{width=0.9\hsize,file=fig3.eps,clip=}}
482: \medskip
483: \caption{The border of violation of CHSH-Bell inequality for thermal
484: and pseudo-thermal light
485: in dependence on mean photon numbers $\langle n\rangle_{A}$
486: and $\langle n\rangle_{B}$.}
487: \label{fig3}
488: \end{figure}
489:
490:
491: Real light sources emit to a large amount of different independent modes.
492: The density matrix of this multi-mode state is given
493: in the following form
494: \begin{equation}\label{multi}
495: \hat{\rho}=\prod_{\mu}\,\,\sum_{n_{\mu}=0}^{\infty}\frac{\langle
496: n_{\mu}\rangle^{n_{\mu}}}{(1+\langle n_{\mu}\rangle)^{1+n_{\mu}}}
497: \,|n_{\mu}\rangle\langle n_{\mu}|,
498: \end{equation}
499: where $n_{\mu}$ is photon number for particular mode $\mu$ and
500: $|n_{\mu}\rangle$ is the Fock state of the corresponding mode.
501: Let us suppose that this multi-mode
502: thermal state is feeded to the inputs $A1$ and $B1$ and the multi-mode
503: vacuum states are present in the inputs $A2$ and
504: $B2$.\footnote{In reality there could be a problem to set the proper
505: parameters of Kerr interaction for all the frequancy components
506: together.}
507: The analysis presented in Sec. III may be
508: generalized to multi-mode light in a straightforward way. We define the
509: ``rotations'' of the multi-mode vacuum $|\{0\}\rangle$ and any excited
510: multi-mode state $|\{n\}\rangle$ as follows,
511: % where the multi-mode index $\{n\}$ differs from the zero for every mode,
512: \begin{eqnarray}
513: |\{n\}\rangle_{1}|\{0\}\rangle_{2} &\rightarrow&
514: \cos\theta\,
515: |\{n\}\rangle_{1}|\{0\}\rangle_{2} + \sin\theta\,
516: |\{0\}\rangle_{1}|\{n\}\rangle_{2},\nonumber\\
517: |\{0\}\rangle_{1}|\{n\}\rangle_{2}&\rightarrow&
518: -\sin\theta\,
519: |\{n\}\rangle_{1}|\{0\}\rangle_{2} + \cos\theta\,
520: |\{0\}\rangle_{1}|\{n\}\rangle_{2},\nonumber\\
521: \end{eqnarray}
522: for $\{n\}\not=0$, and for multi-mode vacuum in both the modes:
523: $|\{0\}\rangle_{1}|\{0\}\rangle_{2}\rightarrow
524: |\{0\}\rangle_{1}|\{0\}\rangle_{2}$.
525: Detection that discriminates between the field
526: vacuum and other states has two possible outcomes described by
527: projectors $|\{0\}\rangle\langle\{0\}|$ and
528: $\hat{1}-|\{0\}\rangle\langle\{0\}|$.
529: It can be shown that the maximal violation of Bell inequality exhibits
530: the same form (\ref{max}) as in the case of single-mode radiation,
531: but with the following notation
532: \begin{equation}
533: p_{0}=\prod_{\mu}p_{0,\mu},\hspace{0.2cm}r_{0}=\prod_{\mu}r_{0,\mu}.
534: \end{equation}
535: With increasing number of the modes of thermal radiation
536: the effective overlap of vacuum state and such a multi-mode field
537: decreases and, consequently, the maximal violation of Bell
538: inequality is enhanced. In this way, the Bell inequality violation can
539: be achieved for every thermal radiation, if the sufficient number of
540: modes is taken into account.
541:
542:
543: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
544: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
545:
546: \section{Conclusion}
547:
548: An entangling device employing non-linear dynamics and postselection
549: has been proposed and it has been shown that two mixed states can be
550: entangled in such a way that the entanglement of the resulting state
551: is strong enough to violate Bell inequalities (when proper local
552: measurements are chosen). The disorder due to the statistical nature
553: of the density matrices of input states is irrelevant -- it does not
554: influence the violation of Bell inequality. The only parameters
555: affecting the maximum of the mean value of Bell operator are overlaps
556: $p_{0}=\langle 0|\rho_{A1}|0\rangle$ and $r_{0}=\langle 0|\rho_{B1}|0\rangle$.
557: %
558: %
559: This is also the reason of a contra-intuitive behavior when the
560: entanglement increases as the input thermal state becomes more
561: `classical' ($\beta\rightarrow 0$), whereas in the `quantum' limit
562: ($\beta\rightarrow\infty$) the entanglement vanishes.
563: Another contra-intuitive aspect of this phenomena appears if the
564: multi-mode thermal radiation is considered. Since the overlap with
565: multimode vacuum becomes smaller as the number of modes increases,
566: the multi-mode thermal radiation can violate
567: Bell inequality more notably,
568: irrespective of its larger entropy. Thus this ``classical-like''
569: radiation can be strongly entangled in
570: the ideal case and even exhibit the pronounced quantum nonlocality.
571: Unfortunately, like the other kinds of mesoscopic states, the described
572: quantum superpositions are very sensitive to the descructive influence of
573: decoherence and losses.
574:
575:
576:
577:
578: \section*{Acknowledgments}
579:
580: This research was supported under the project LN00A015 of the
581: Ministry of Education of the Czech Republic and the European Union project
582: EQUIP (contract IST-1999-11053). R.F. thanks to
583: J. \v Reh\' a\v cek for stimulating discussions.
584:
585:
586: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
587:
588: \begin{thebibliography}{99}
589:
590: \bibitem{micro}
591: A.~Aspect, P.~Grangier, and G.~Roger, Phys.\ Rev.\ Lett.\ {\bf 48},
592: 91 (1982); A.~Aspect, J.~Dalibard, and G.~Roger, Phys.\ Rev.\ Lett.\ {\bf 49}
593: 1804 (1982);
594: C.O.~Alley, and Y.H.~Shih, Phys.\ Rev.\ Lett.\ {\bf 61}, 2921 (1988);
595: Z.Y.~Ou, and L.~Mandel, Phys.\ Rev.\ Lett.\ {\bf 61} 50 (1988);
596: P.~Kwiat, K.~Mattle, H.~Weinfurter, A.~Zeilinger, A.V.~Sergienko, and
597: Y.~Shih, Phys.\ Rev.\ Lett.\ {\bf 75} 4337 (1995).
598: S.~Popescu, Phys.\ Rev.\ Lett.\ {\bf 74}, 2619 (1995);
599: N.~Gisin, Phys.\ Lett.\ A {\bf 210}, 151 (1996);
600: S.~Popescu, L.~Hardy, and M.~\.{Z}ukowski, Phys.\ Rev.\ A {\bf 56}, R4353
601: (1997).
602:
603: \bibitem{CohSt}
604: B.\,C.~Sanders, Phys.\ Rev.\ A~{\bf 45}, 6811 (1992); {\bf 46}, 2966
605: (1992);
606: A.~Mann, B.\,C.~Sanders, and W.\,J.~Munro, Phys.\ Rev.\ A~{\bf 51}, 989
607: (1995);
608: D.\,A.~Rice, G.~Jaeger, and B.\,C.~Sanders, Phys.\ Rev.\ A~{\bf 62}, 012101
609: (2000).
610:
611: \bibitem{Gerry99}
612: C.C. Gerry, Phys. Rev. A {\bf 59}, 4095 (1999).
613:
614: \bibitem{nas}
615: R.~Filip, J.~\v{R}eh\'{a}\v{c}ek, and M.~Du\v{s}ek, arXiv:
616: quant-ph/0011006 (2000).
617:
618: \bibitem{mixx}
619: V.~Vedral, M.B.~Plenio, M.A.~Rippin, and P.L.~Knight, Phys.\ Rev.\
620: Lett.\
621: {\bf 78}, 2275 (1997); V.~Vedral and M.B.~Plenio, Phys.\ Rev.\ A {\bf
622: 57}, 1619 (1998);
623: C.H.~Bennett, D.P.~DiVincenzo, J.A.~Smolin, and W.K.~Wooters, Phys.\
624: Rev.\ A {\bf 54}, 3824 (1996);
625: W.K.~Wooters, Phys.\ Rev.\ Lett.\ {\bf 80}, 2245 (1998);
626: S.~Bose, and V.~Vedral, Phys.\ Rev.\ A {\bf 61}, 040101R (2000);
627: S.~Bose, K.~Jacobs, and P.L.~Knight, Phys.\ Rev.\ A {\bf 59}, 3204
628: (1999);
629: S.~Bose, I.~Fuentes-Guridi, P.L.~Knight, and V.~Vedral, arXiv:
630: quant-ph/0103063 (2001).
631:
632: \bibitem{eraser}
633: M.~Hillery and M.\,O.~Scully, in {\em Quantum Optics, Experimental
634: Gravitation, and Measurement Theory}, eds. P.~Meystre and
635: M.\,O.~Scully (Plenum, New York, 1983), p.~65;
636: M.\,O.~Scully, B.\,G.~Englert, and H.~Walther, Nature {\bf 351}, 111
637: (1991);
638: P.\,G.~Kwiat, A.\,M.~Steinberg, and R.\,Y.~Chiao, Phys.\ Rev.\
639: A~{\bf 45}, 7729 (1992);
640: P.\,G.~Kwiat, A.\,M.~Steinberg, and R.\,Y.~Chiao, Phys.\ Rev.\
641: A~{\bf 49}, 61 (1994).
642:
643: \bibitem{Transp}
644: A.~Peres, Phys.\ Rev.\ Lett.\ {\bf 77}, 1413 (1996).
645:
646:
647: \bibitem{EPR}
648: A.~Einstein, B.~Podolsky, N.~Rosen, Phys.\ Rev.\ {\bf 47}, 777
649: (1935); D.~Bohm, Phys.\ Rev.\ {\bf 85}, 166 (1952);
650: N.~Bohr, Nature {\bf 136}, 65 (1935).
651:
652: \bibitem{Bell}
653: J.\,S.~Bell, Physics {\bf 1}, 195 (1964);
654: J.\,F.~Clauser, M.\,A.~Horne, A.~Shimony, R.\,A.~Holt, Phys.\
655: Rev.\ Lett.\ {\bf 23}, 880 (1969).
656:
657: %%%%%%%%%
658: %
659: % \bibitem{Peres96}
660: % A.~Peres, Phys.\ Rev.\ A {\bf 54} 2685 (1996).
661: %
662: % \bibitem{error-free} I.D.~Ivanovic, Phys.\ Lett.\ A {\bf 123},
663: % 257 (1987); D.~Dieks, Phys.\ Lett.\ A {126}, 303 (1988);
664: % A.~Peres, Phys.\ Lett.\ A {\bf 128}, 19 (1988);
665: % B.~Huttner, N.~Imoto, N.~Gisin, and
666: % T.~Mor, Phys.\ Rev.\ A {\bf 51}, 1863 (1995); L.S.~Phillips, S.M.~Barnett,
667: % and D.T.~Pegg, Phys.\ Rev.\ A {\bf 58}, 3259 (1998).
668: %
669: %%%%%%%%%
670:
671: \bibitem{pseudoth}
672: F.T.~Arecchi, Phys.\ Rev.\ Lett.\ {\bf 15}, 912 (1965);
673: W.~Martienssen, and E.~Spiller, Phys.\ Rev.\ Lett.\ {\bf 16}, 531 (1966).
674:
675: \end{thebibliography}
676:
677:
678: \end{document}
679:
680:
681:
682:
683: