quant-ph0108134/epr.tex
1: 
2: \documentstyle[twocolumn,aps]{revtex}
3: %\documentstyle[multicol,aps]{revtex}
4: %\documentstyle[preprint,aps]{revtex}
5: \begin{document}
6: \draft
7: \title{ A Local Realistic Model for
8:  Two-Particle Einstein-Podolsky-Rosen Pairs }
9: \author{ Zhi-Yuan Li } 
10: \address{Ames Laboratory, US Department of Energy and
11: Department of Physics and Astronomy, Iowa State University, 
12: Ames, Iowa 50011, and
13: Institute of Physics, Chinese Academy of Sciences, P. O. Box 603,
14: Beijing 100080, China}
15: \date{Received~~~~~~~~~June  2001}
16: \maketitle
17: 
18: \begin{abstract}
19:  A local realistic model for quantum mechanics of two-particle
20: Einstein-Podolsky-Rosen pairs is proposed. In this model, it is
21: the strict obedience of  conservation laws  in each  
22: event at the quantum level that uphold the perfect correlation
23: of two spatially-separated particles, instead of nonlocality in
24: the orthodox formulation of quantum mechanics. Therefore, 
25: one can conclude that all components of the spin of two particles, and 
26: the position and momentum of a particle can 
27: be measured simultaneously. The proposed model yields
28: the same  statistical prediction on an
29: ensemble of  individual particles as the orthodox formulation 
30: does. This suggests that 
31: the wave  function is not a complete description of individual particle
32: as assumed in the orthodox  formulation, 
33: but only a statistical description of an ensemble of  particles.
34: 
35:  
36: \end{abstract}
37: \pacs{ PACS numbers:  03.65.-w, 03.65.Ud, 03.65.Ta }
38: \narrowtext
39: \par
40: 
41: The 20th century saw the birth, growth, and prosperity of quantum
42: mechanics, and its influence in every aspect of science and 
43: technology.      The prediction of
44:  quantum theory has been found in excellent agreement with experimental
45: results in an extremely wide range.
46:  However, the interpretation of quantum mechanics has
47: been widely disputed over a long time [1-16], notably coming from 
48: Einstein in one side and Bohr  in the other.
49: The central point of the controversy is whether or not
50: the  wave function is a complete description of an individual system
51: at the quantum level. In a historic paper [1], Einstein, Podolsky, and 
52: Rosen (EPR) analyzed   a system consisting of two spatially
53: separated but quantum-mechanically correlated particles, the so-called
54: EPR pairs or entangled pairs [3], and 
55:  argued that both particles could simultaneously have 
56: predetermined values of non-commuting operators, such as position and 
57: momentum. Thus,  they concluded that quantum mechanics is
58: not complete.  In regard to this  kind of entangled system, 
59: Bell proved an inequality [11] based on which 
60: it is possible to test  quantum mechanics and the
61: opponent deterministic hidden-variable 
62: theory [9-14] in a quantitative manner. 
63: Most experiments to date have favored quantum mechanics [12,17-20]. 
64: Today, it has been well-established  in the orthodox 
65: formulation of quantum mechanics that  spooky 
66: nonlocal characteristics are indeed present in these EPR pairs.
67: This, together with the abandonment of realism (enforced in   
68:  the wave-packet reduction hypothesis in  
69: old quantum measurement problem) represents the central  
70: viewpoint of orthodox quantum mechanics on  physics in
71: microscopic world. Both viewpoints will drastically 
72: change our concepts and philosophy of nature.
73:   
74: 
75: \par
76: Is it possible to avoid such a radical 
77: revolution of  concepts about nature by using some
78:  local realistic viewpoints to understand microscopic world 
79: while pertaining agreement 
80: with experimental observations? In this paper
81:  we will give a positive answer to this question. We will propose
82: a local realistic model which  
83: directly comes from  standard  quantum mechanics for many-particle 
84: systems. Yet, the proposed model starts from a different viewpoint 
85: and finally leads to  different results from the orthodox formulation. 
86: 
87: \par
88: To this end, we first consider in general 
89: the motion of  two  independent  particles under the law of 
90: quantum mechanics. One typical case is that these particles 
91: are  separated in space, and noninteracting  with each other.
92:   The Schr\"odinger equation for
93: this two-particle system is  written as
94: $$ i\hbar \frac{\partial}{\partial t}
95:  \Psi = H\Psi= (H_1+H_2) \Psi. \eqno(1) $$
96: $H_1({\bf r}_1)$ and $H_2({\bf r}_2)$ are  the Hamiltonian 
97: for particles 1 and 2, respectively. 
98: From physical  intuition and simple argument one can find that
99:  the system wave function $\Psi$ is just
100:  the  direct  product of the wave function  
101: of each particle, namely
102: $$\Psi({\bf r}_1,{\bf r}_2)=\psi_1({\bf r}_1) \psi_2({\bf r}_2). \eqno(2)$$
103: $\psi_1$ and $\psi_2$ satisfy 
104: $$i\hbar \frac{\partial}{\partial t}
105:  \psi_1 =  H_1 \psi_1;~~~~~~~i\hbar \frac{\partial}{\partial t}
106:  \psi_2 =  H_2 \psi_2.  \eqno(3) $$
107: The general solution of Eq. (3) is
108: $$  \psi_1=\sum_{\lambda_1} c_{1,\lambda_1}
109:  \psi_1({\bf r}_1,\lambda_1),  ~~~~ 
110: \psi_2=\sum_{\lambda_2} c_{2,\lambda_2}
111:  \psi_2({\bf r}_2,\lambda_2),   \eqno(4)$$
112: where $\psi_1({\bf r}_1,\lambda_1)$ and $\psi_2({\bf r}_2,\lambda_2)$ are
113:  eigenstates of particles 1 and 2 under Hamiltonian $H_1$ and $H_2$,
114:  respectively,  with $\lambda_1$
115: ($\lambda_2$) being their eigenvalues, and
116:  $c_1$ and $c_2$ are the superposition coefficients. 
117: Eqs. (2)-(4) verify that there is no mechanical correlation
118: between particles 1 and 2, as is exactly what we have supposed that
119: these two particles are independent.
120: Any external influence on one particle will not affect the wave
121: function of the other particle.
122: 
123: \par
124: The above analysis is also applicable to a special two-particle system,
125: the free EPR pair, since what we only require in Eqs. (1)-(4)
126: is that two particles are far-away separated and
127: non-interacting, no matter these two particles are 
128:  either  uncorrelated or correlated. 
129:  To be more specific,
130: we consider Bohm's scheme with  a pair of spatially separated spin-1/2 
131: atoms,   which is produced from the dissociation of
132: a molecule in the spin singlet (total spin 0) 
133: state (e.g., $\rm Hg_2$ molecule)[5,6]. 
134: Up to some time $t=0$, the wave function
135: of this combined system is 
136: $$\Psi=\frac{1}{\sqrt{2}} [\psi_{+}(1) \psi_{-}(2)-\psi_{-}(1) \psi_{+}(2)],
137: \eqno(5)$$
138:  where $\psi_{+}(1)$ means that atom 1 has spin $+\hbar /2$, and so on. 
139: As $\Psi$ is rotationally invariant, the spin axis ``+" can be selected
140:  along any  direction  in space. 
141: This antisymmetric spin wave function is very common in
142: molecular bound  states. 
143: A good example is the spin singlet state of a 
144: hydrogenic molecule, which is formed from
145:  the group motion of two electrons under nuclear attraction.
146: It should be  emphasized that two free hydrogenic atoms can 
147: not form a spin singlet state. 
148: 
149: 
150: \par
151: What happens if the two atoms are separated
152: by a method that does not influence the total spin?  
153: In the orthodox formulation, the wave function in Eq. (5) 
154: keeps unchanged even when the two atoms have separated so far away that 
155: they cease to interact, namely, atoms in 
156: both the bound state and free state have the same spin-singlet state.
157: And it is the nonlocality (something of action-at-a-distance) that  
158: uphold the perfect correlation of the spins of the two atoms. Once
159: one measures one spin and finds it is along $\bf \hat n$ direction, 
160: the other spin is immediately known
161: to  be along  $-\bf \hat n$ even if no measurement is
162:  performed. Although spooky, this assumption has found   
163: strong experimental supports for a long time.  
164: These experiments are closely related to 
165: the test of Bell's inequality, and  found to favor
166:  quantum mechanics [10,15-18].
167: The joint  outcome of  the spin of particle 1   
168: along the direction $ {\bf n}_1$ and that of  particle 2  
169: along direction $ {\bf n}_2$ in orthodox quantum mechanics is [12] 
170: $$E^{\Psi}( {\bf n}_1, {\bf n}_2)=<\Psi|({\bf \sigma}_1 \cdot
171:  {\bf n}_1) ~~({\bf \sigma}_2 \cdot  {\bf n}_2)|\Psi>
172: =- {\bf n}_1
173: \cdot  {\bf n}_2, \eqno(6)$$
174: where $\sigma$ is the vector of  Pauli matrices. 
175: The result in Eq. (6)  clearly violates Bell's inequality [11-14].
176: It is such kinds of  violation that  validate orthodox quantum mechanics 
177: and  refutes any hidden-variable theory.
178:  
179: 
180: \par
181: In our model, the system wave function 
182: $\Psi$ has the form  of Eqs. (2) and (4), each atom having its
183: own wave function independent of the other. Then,  because
184: the total spin is conserved in every process of molecular dissociation,
185: the two particles must have a specific spin vector antiparallel to
186: each other, and thus  both are in a pure spin state each time. 
187: This  yields to 
188: $$\Psi=\psi_1 \psi_2, \eqno(7a) $$
189: $$\psi_1= \psi_{+}(1, {\bf n}),~~~~~~ 
190: \psi_2= \psi_{-}(2, {\bf n}),  \eqno(7b)$$
191: where $ {\bf n}$ is a random unit vector of direction  
192: uniformly distributed in space, in accordance with the rotational invariance
193: of original spin-singlet state in the molecule.  
194: The  wave function Eq. (7) changes its form  under axis 
195: rotation if it refers to  a single spin pair. However, 
196: since here  it actually represents the whole ensemble 
197: of spin pairs,  it is rotationally invariant, 
198: as is clear from the fact that $\bf n$
199: is uniformly distributed in space. This ensemble is composed of many
200: spin pairs each of which is composed of two spins
201: with definite but antiparallel directions.
202: 
203: 
204: \par
205: The physical picture involved in 
206: Eq. (7) is quite simple. If we assume the spin-singlet molecule as
207: an atom source, then this source emits  at each time  a pair of
208: atoms who have definite  spin vector antiparallel to each other, but what
209: direction the spins are along is unknown.  This is similar
210: to spontaneous emission of photon from excited atoms. Each time the emitted
211: photon has a definite polarization, but the direction is unspecified, 
212: all  are possible. One can  utilize  this similarity to calculate
213:  the quantum-mechanical expectation value (statistical average) of 
214: all components of each spin, $\bar S_x$, $\bar S_y$, and so on, 
215: which are of course all zero, the same as predicted
216: by the orthodox  formulation. 
217: The joint outcome  can be found according to Fig. 1, which is  
218: $$E( {\bf n}_1, {\bf n}_2)= <\psi_1( {\bf n}_1)|
219: {\bf \sigma}_1 \cdot  {\bf n}_1  |\psi_1( {\bf n}_1)> $$ $$
220: \times  ~~<\psi_2(- {\bf n}_1) |{\bf \sigma}_2 \cdot 
221:  {\bf n}_2|\psi_2(- {\bf n}_1) >=- {\bf n}_1
222: \cdot  {\bf n}_2, \eqno(8)$$
223: also exactly the same as that predicted by the orthodox formulation of 
224: quantum mechanics [shown in  Eq. (6)]. Therefore our proposed model
225:  also passes the test of Bell's inequalities. 
226: Derivation of Eq. (8) implies
227: the application  of Malus's law in classical optics. 
228: It is well-known that when natural light passes through two
229: polarizers with an inclination of $\theta$, we observe 
230:  a  field  amplitude of $E_0 \cos \theta $, 
231: and a light  intensity of $\frac{1}{2}I_0 \cos^2 \theta $,
232: where  $E_0$ and $I_0$ are the amplitude and intensity of incident light,
233: respectively. Natural light is a completely unpolarized photon beam, 
234: which can be
235: decomposed into two independent orthogonal components of field. 
236: Similarly, the spin pair as a  whole is also completely unpolarized.
237: A photon beam passing through two polarizers is equivalent in effect 
238: to  a  spin-pair  beam  of which each spin passing through one 
239: Stern-Gerlach magnet,  therefore, they yield to the same cosine law of 
240: statistical joint outcome.
241: 
242: \par
243: Note that the wave function Eq. (7) is not new. It has been suggested 
244: soon after the EPR historical paper [3,4,12], but  finally it was 
245: refuted and abandoned due to two 
246: difficulties assumed. The first difficulty is that Eq. (7) do not 
247: conserve its form under  rotation [8], the second  difficulty
248: is that the  statistical joint outcome
249: is different from the standard one  shown in Eq. (8) (which is
250: the experimentally correct one) [13].
251: We now see that these two objections do  not stand in our model.  
252: The wave function Eq. (7) is rotationally invariant and yields the same 
253:  joint outcome [Eq. (8)] as the orthodox formulation, when the statistical
254: feature of the whole ensemble of spin pairs is concerned.  
255: Both models predict  definitely that 
256: two spins in each pair is always antiparallel to each other. 
257: If one is interested in the motion of single spin pair  
258: as to what direction the spin is along, then
259: neither our model nor the orthodox formulation can give  definite 
260: results, only statistical results can be predicted definitely.
261: 
262: 
263: \par
264: Several important differences can be  found between our model and
265: the orthodox formulation. 
266:  One  significant  consequence of Eq. (7) is that all
267: components of each spin can be measured simultaneously, since
268: now the two spins are noninteracting while in perfect correlation and thus
269: allow measuring one particle without affecting the other. 
270: This can be  accomplished by means of two spin-measuring 
271:  apparatuses (Stern-Gerlach magnets) oriented perpendicular
272:  to each other, say,
273: along the x-axis and y-axis, respectively.  Particles 1 and 2  passing
274:  through  apparatuses 1 and 2 will have their spin  components 
275: measured to be $S_{1x}$ and $S_{2y}$. Because of perfect antiparallel
276: correlation of the two spins, we at once know $S_{1y}=-S_{2y}$,
277: and  $S_{2x}=-S_{1x}$.  This is in contradiction with Heisenberg's
278:  uncertainty principle in the orthodox formulation, 
279: which asserts that  $S_x$ and $S_y$ of a particle
280:  can not be measured  simultaneously because they are non-commuting 
281: operators. 
282: 
283: \par
284: Another significant point of Eq. (7) is that
285:  the conventional concept of
286: wave-packet  reduction [5,6] is not present in our model,
287: because now each atom  is in a pure state with a definite spin vector. 
288: The measurement does not  determine  
289: what spin state the atom is in, but just tells 
290: us what it is in. From this viewpoint, the measurement is not an
291:  inseparable part  combined with the quantum system considered, 
292: a manner the orthodox  formulation strongly stresses. If one
293: wishes to uphold the concept of wave-packet reduction,
294: then this process must have happened during the process of
295: molecular dissociation. This also has  nothing to do with external
296: measurements by the observer. 
297: 
298: \par
299: In our model the perfect antiparallel correlation of the two spins is
300: maintained by the conservation law,
301:  while in the orthodox formulation
302: it is preserved by the nonlocal entanglement of spin wave function
303:  and realized in measurement by the mechanism of wave-packet reduction. 
304: In experiments  to test Bell's inequalities [17-20], two loopholes
305:  are generally supposed  to leave  the conclusion uncertain.
306: One is that only a small subset of all pairs are detected
307: due to  low detection/collection efficiency of the apparatus, the other
308: is that  the Einstein locality condition might not be maintained 
309: strictly.  These two loopholes are, however, not present in our model. 
310: The measurement process involved is always local,
311: and a small subset always represents fairly the whole ensemble
312: in a statistical manner.
313: It is not the communication between the two particles, but the
314: strict maintenance of conservation law in nature that upholds
315: the perfect correlation. From this point of view, our model
316:  agrees  with experimental observations in a more natural manner
317: than  the orthodox formulation does.
318: 
319: \par
320: Now we can see that our proposed model is realistic and local, 
321: and at the same time it is in as good  
322: agreement with experimental observations as the orthodox formulation,
323: when two-particle EPR pairs are concerned. 
324: From  the viewpoint of the orthodox formulation,  our model  
325: essentially lies inside the framework of hidden-variable theory. Then, since  
326: the hidden-variable theory constructed here  is local, realistic,  and 
327: makes same predictions as  orthodox quantum mechanics does, one
328: is led to conclude that Bell's theorem is not always true in regard to
329: two-particle EPR pairs.
330: 
331: 
332: \par 
333: Following the concept implied in above arguments, we turn to analyze the
334: scheme of entangled pair originally  proposed by EPR
335: [1]. The system they studied consists of two particles, and 
336: lies  in the state
337: $\Psi(x_1,x_2)=\delta(x_1-x_2-a)$, which is the eigenfunction of
338: the operator $x_1-x_2$ with eigenvalue $a$, and of the operator $p_1+p_2$
339: with eigenvalue 0. We emphasize 
340:  that this $\delta$-type wave function implies very strong
341: interaction between the two particles. This pair as a whole can not
342: be subject to a simultaneous measurement of the position and momentum 
343: without affecting each other. But the situation
344: changes completely   when the pair is dissociated  
345: at time $t_0$ and becomes  noninteracting by
346: a process that does not influence the total momentum.
347: Each particle is now  in a free state, with the momentum ($p_1$ and
348: $p_2$) specifying its motion feature (a pure state).
349: The conservation of total 
350: momentum satisfied at each event leads to a perfect correlation 
351: of the momentum  $p_1+p_2=0$,  from which 
352: we derive 
353: $$ x_1(t)- x_2(t)=a +p_1 (\frac{1}{m_1}+\frac{1}{m_2}) (t-t_0) \eqno(9) $$ 
354: under assumption of  free propagation of particles.
355: $m_1$ ($m_2$) is the mass of particle 1 (2), $t_0$ is the time when 
356: the particle dissociation occurs, and $t$ is the time at measurement. 
357: 
358: \par
359: Whether or not the position of the two particles persists a perfect correlation
360:  depends on the accuracy of determining at what time the dissociation of the
361:  molecule takes place.   If this can be accomplished, 
362: we immediately obtain a perfect correlation in both 
363: momentum and position, and  conclude that
364: both  the position and momentum of each  particle  can be measured 
365: simultaneously, and that  Heisenberg's uncertainty principle  
366: is violated.   Even if
367: we can not acquire a perfect correlation in  position, 
368:  we can still simultaneously 
369: measure the position and momentum of one of the two  particles. 
370: For example, we now decide to see  particle 1. We first measure 
371: its position and get a value
372: of $x_1$, at the  same time we measure the momentum of 
373: particle 2 and get a value
374: of $p_2$. Due to the perfect correlation of the  momentum, 
375: we at once know $p_1=-p_2$, and arrive at a position
376: that  the uncertainty principle is violated. 
377: Note that our argument is different from
378: original EPR's, where  the two particles are supposed to be always
379:  in a perfect entangled state (for both position and momentum)
380: when they are subject to measurements. In this entangled state, 
381: the momentum and position of a particle  can not be measured
382: simultaneously, because of the strong entanglement
383:  between the two composites.  We can further argue that 
384: in  Bohm's scheme  the violation of
385: uncertainty principle remains even if we only have a system with 
386: a perfect correlation in one of the spin components 
387: instead of all components. 
388: 
389: 
390: \par
391: In the orthodox  quantum theory, it is assumed that
392: the wave function completely
393: specifies  the physical state of an individual system.
394: According to our local realistic model, it is possible to obtain 
395: through conservation laws in nature 
396: (which hold true in all mechanical theories) 
397: to obtain simultaneous knowledge
398: of non-commuting variables such as different components of a spin, and
399: the position and momentum of a particle. This  violates  
400: Heisenberg's uncertainty principle, one of the
401: basic foundations of orthodox quantum mechanics.  
402: Therefore, it is possible
403: to obtain more  complete information about individual system  
404:  beyond that allowed by  the  wave function  in the orthodox formulation, 
405: while at the same time arrive at the same statistical information
406:  on an ensemble of individual systems as that predicted  
407: by the orthodox formulation. 
408: This suggests that the wave function in orthodox quantum theory is
409: only a complete description of an ensemble of individual particles,
410: but not a complete description of individual particles.  
411: Let us look at the wave functions Eq. (5) and Eq. (7) once again.
412: They are equivalent to each other  when they are referred to an 
413: ensemble of EPR pairs, 
414: while  totally different when referred to an individual pair.
415: In the orthodox formulation, the wave function is always assumed to describe
416: an individual system. Now one finds  it is more likely that
417: the wave function [like Eq. (5)] only describes an ensemble of individual
418: systems.  It is  this difference in the wave function for
419: an individual system  that cause significant results of the proposed model
420: beyond the orthodox formulation.
421: 
422: 
423: \par
424: In summary,  we have proposed a local realistic model for quantum
425: mechanics of two-particle EPR pairs in the framework of standard quantum
426: mechanics for many-body systems. 
427:  In this model, the spooky nonlocal
428: characteristics present in the space-separated free EPR pairs 
429: as assumed in the orthodox formulation is abandoned. Also abandoned 
430: is the concept of wave-packet reduction  in the orthodox 
431: formulation  which is assumed to happen 
432: when external measurement occurs.   It is
433: the strict obedience of  conservation laws  in each  
434: event at the quantum level that maintain the perfect correlation
435: of two spatially-separated particles.
436: With a different starting viewpoint from orthodox quantum mechanics,  
437: the proposed model allows  more information
438: on individual particles beyond  the wave function in 
439: the orthodox formulation to be obtained.  
440:  At the same time it leads to the same statistical information on
441: an ensemble of individual particles as predicted by the orthodox 
442: formulation.
443:  It may be expected that the proposed local realistic model
444:  will help people  to  understand quantum mechanics in a conceptually 
445:  easy  way, as has been the case for classical physics. 
446: 
447: 
448: 
449: \begin{references}
450: 
451: \bibitem{[1]} A. Einstein, B. Podolsky, and N. Rosen, Phys. Rev. {\bf 47}, 
452: 777 (1935). 
453: 
454: \bibitem{[2]} N. Bohr, Phys. Rev. {\bf 48}, 696 (1935). 
455: 
456: \bibitem{[3]} E. Schr\"odinger, Naturwissenschaften {\bf 23}, 807; 823;
457: 844 (1935). 
458: 
459: \bibitem{[4]}  W. F. Furry, Phys. Rev. {\bf 49}, 393, 476 (1936). 
460: 
461: \bibitem{[5]} P. A. M. Dirac, {\it Quantum Mechanics} (Clarendon Press,
462: Oxford, 1958).
463: 
464: \bibitem{[6]} J. von Neumann, {\it Mathematical Foundations of Quantum
465: Mechanics }(Princeton University Press, Princeton, 1955).
466: 
467: \bibitem{[7]} D. Bohm, {\it Quantum Theory} (Englewood Cliffs,  Princeton
468: Hall, NJ, 1951). 
469: 
470: \bibitem{[8]} D. Bohm and Y. Aharonov, Phys. Rev. {\bf 108}, 1070 (1957).
471: 
472: \bibitem{[9]} L. de Broglie,  C. R. Acad. Sci. (Paris) {\bf 184}, 273 (1927);
473: J. Phys. Radium {\bf 8}, 22 (1928).
474: 
475: \bibitem{[10]} D. Bohm, Phys. Rev. {\bf 85}, 166; 180 (1952). 
476:   
477: \bibitem{[11]} J. S. Bell, Physics (Long Island City, NY) {\bf 1}, 195 (1965).
478: 
479: \bibitem{[12]} J. F. Clauser and A. Simony, Rep. Prog. Phys. {\bf 41}, 1881
480:  (1978). 
481: 
482: \bibitem{[13]} D. M. Greenberger, M. A. Horne, A. Shimony, and A. Zeilinger,
483: Am. J. Phys. {\bf 58}, 1131 (1990).
484: 
485: \bibitem{[14]} M. O. Scully and M. S. Zubairy, {\it Quantum Optics}
486:  (Cambridge University Press, Cambridge, 1997).
487: 
488: \bibitem{[15]} J. S. Bell, Phys. World, August 33 (1990).
489: 
490: \bibitem{[16]} K. Gotterfried, Nature (London) {\bf 405}, 533 (2000). 
491: 
492: \bibitem{[17]} S. J. Freeman and J. F. Clauser, Phys. Rev. Lett.
493: {\bf 28}, 938 (1972).
494: 
495: \bibitem{[18]} A. Aspect, J. Dalibard, and G. Roger, Phys. Rev. Lett. 
496: {\bf 49}, 1904 (1982). 
497: 
498: \bibitem{[19]} G. Weihs, T. Jennewein, C. Simon, H. Weinfurter, 
499: and A. Zeilinger, Phys. Rev. Lett. {\bf 81}, 5039 (1998).
500: 
501: \bibitem{[20]} P. G.  Kwiat, K. Mattle, H. Weinfurther, A. Zeilinger,
502: A. V. Sergienko, and Y. Shih,  Phys. Rev. Lett. {\bf 75},
503: 4337 (1995). 
504: 
505: \end{references}
506: 
507: \begin{figure}
508: \caption{ Schematic configuration of an entangled spin-1/2 pair
509: with  antiparallel spins ${\bf s}_1$ and ${\bf s}_2$ 
510:  passed through two Stern-Gerlach
511: apparatuses along the ${\bf n}_1$ and ${\bf n}_2$ directions with an 
512: inclination of $\theta$. Due to rotational invariance and
513: completely unpolarized nature of the spin-pair beam, 
514: ${\bf s}_1$ can be set to be parallel to ${\bf n}_1$.
515: }
516: \end{figure}
517: 
518: 
519: \end{document}
520: