PUBLISHED BY

World's largest Science,

Technology & Medicine
Open Access book publisher

107,000+
3,300+ INTERNATIONAL 113+ MILLION
OPEN ACCESS BOOKS AUTHORS AND EDITORS DOWNLOADS

AUTHORS AMONG
BOOKS TOP 1% 12.2%
DELIVERED TO AUTHORS AND EDITORS
MOST CITED SCIENTIST FROM TOP 500 UNIVERSITIES
151 COUNTRIES

Selection of our books indexed in the

Book Citation Index in Web of Science™
Core Collection (BKCI)

Chapter from the book Advances in Quantum Mechanics

Downloaded from: http://www.intechopen.com/books/advances-in-quantum-
mechanics

Interested in publishing with InTechOpen?

Contact us at book.department@intechopen.com
 Provisional chapter
Chapter 4

The Improvement of the Heisenberg Uncertainty

The Improvement of the Heisenberg Uncertainty
Principle
Principle

L. M. Arévalo Aguilar,
L. M.
C. Arévalo
P. García Aguilar,
Quijas C. P.Robledo-Sanchez
and Carlos García Quijas and
Carlos Robledo-Sanchez
Additional information is available at the end of the chapter
Additional information is available at the end of the chapter
10.5772/54530
http://dx.doi.org/10.5772/54530

1. Introduction
One of the fundamental cornerstone of quantum mechanics is the Heisenberg uncertainty
principle. This principle is so fundamental to quantum theory that it is believed that if
a single phenomenon that could violate it is found then the whole building of quantum
mechanics will fall apart. However, since the formulation of the uncertainty principle until
today there is not clear and universal agreement in its formulation or interpretation. Even
Heisenberg was not clear about the exact meaning of p1 and x1 in their first formulation of
the uncertainty relations [1]:

p1 q1 ∼ h, (1)
nor in the interpretation of the uncertainty principle. According to Heisenberg, in Eq. (1) q1
represents "the precision with which the value of q is know (q1 is, say, the mean error of q), therefore
here the wavelength of light. Let p1 be the precision with which the value of p is determinable;
that is, here, the discontinuous change of p in the Compton effect [1]". He also thought the
uncertainty principle in terms of disturbance produced on an observable when it is measured
its canonical counterpart.
The relevance of the uncertainty principle to Physics is that it introduced for the first time
the indeterminacy in a physical theory, which mean the end of the era of certainty in Physics.
That is to say, what uncertainty principle made evident was the peculiar characteristic of
quantum theory of not being able to predict with certainty a property of a physical system;
in words of Heisenberg: ". . . canonically conjugate quantities can be determined simultaneously
only with a characteristic indeterminacy. This indeterminacy is the real basis for the occurrence of
statistical relations in quantum mechanics [1]".

©2012 Aguilar et al., licensee InTech. This is an open access chapter distributed under the terms of the Creative
Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use,
© 2013 Aguilar
distribution, et al.; licenseeinInTech.
and reproduction This is provided
any medium, an open access article
the original distributed
work under
is properly the terms of the Creative
cited.
Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use,
distribution, and reproduction in any medium, provided the original work is properly cited.
68 Advances in Quantum Mechanics
2 Quantum Mechanics

Since now, you can perceive two different meanings of the Uncertainty Principle in the two
quoted paragraphs above. In the first one, the uncertainty comes from a statistical property
(according with Heisenberg, the mean error) of quantum theory; in the second meaning the
uncertainty is a restriction to simultaneously measure two physical properties.
On the other hand, to elucidate the meaning of the time-energy uncertainty relation [1]
E1 t1 ∼ h is quite difficult, for, contrary to the uncertainty relation given in Eq. (1), it is not
possible to deduce it from the postulates of quantum mechanics, i. e. there is not an operator
for time. In Heisenberg’s paper the meaning of t1 is the "time during which the atoms are under
the influence of the deflecting field" and E1 refers to the accuracy in the energy measurement.
Heisenberg concludes that "a precise determination of energy can only be obtained at the cost of a
corresponding uncertainty in the time [1]".
In this Chapter of the book, we will review the evolution of the Uncertainty Principle
since its inception by Heisenberg until their application to measure entanglement. We
will review some problems (usually untouched by quantum mechanic’s textbooks) that the
usual interpretation of the Uncertainty Principle have in terms of standard deviations and
its dependence of the wave function. Also, we will review the efforts made to clarify the
meaning of the Uncertainty Principle using uncertainty relations.

2. The relation between the Heisenberg Uncertainty Principle and the

Uncertainty Relations
The uncertainty principle is one of the fundamental issues in which quantum theory differs
from the classical theories, then since its formulation has attracted considerable attention,
even from areas normally outside the scientific development. This has lead to create
misunderstandings about the content of the principle. Thus, it is important to mention
that when we say that there is a lower limit on irreducible uncertainty in the result of a
measurement, what we mean is that the uncertainty is not due to experimental errors or to
inaccuracies in the laboratory. Instead, the restriction attributed to the uncertainty principle
is fundamental and inherent to the theory and is based on theoretical considerations in which
it is assumed that all observations are ideal and perfectly accurate.
A reading of the original Heisenber’s paper shows that he writes (i. e. believes) in some
pharagraps that the indeterminacies comes from the observational procedures. For, in
his original paper, Heisenberg stated [1] that the concepts of classical mechanics could be
used analogously in quantum mechanics to describe a mechanical system, however, the use
of such concepts are affected by an indeterminacy originated purely by the observational
procedures used to determine simultaneously two canonically conjugate variables. This could
be contrasted with the called Statistical Interpretation where it is tough that the wave function
represents and ensemble of identical prepared system and, therefore, the indeterminacy
comes form an intrinsic indeterminacy of the physical properties.
Usually, the uncertainty principle is stated in terms of uncertainty relations. One of the first
way to obtain this indeterminacy relation is due to Robertson [2]. Here, instead, we use the
textbooks approach to deduce the uncertainty relations from the quantum postulates [3, 4].
This approach uses both the Schwarz inequality

hφ|φi h ϕ| ϕi ≥ |hφ| ϕi|2 , (2)

 The Improvement of the Heisenberg Uncertainty Principle 69
The Improvement of the Heisenberg Uncertainty Principle 3
http://dx.doi.org/10.5772/54530
10.5772/54530

and the following quantum postulates:

• The state of a quantum system is represented by a wave function Ψ( x, t) ( |Ψi, in Dirac

notation).
• For

every
R observable A there
is a self-adjoint operator Â, its expectation value is given by
 = Ψ∗ ( x ) ÂΨ( x )dx = Ψ| Â|Ψ .


Now, consider the following operators defined as 1 :


∆ Â = Â − Â


∆ B̂ = B̂ − B̂ . (3)

Let them operate on an state |Ψi, given:

∆ Â |Ψi = |ψa i
∆ B̂ |Ψi = |ψb i . (4)

Therefore, using the Schwarz inequality given in the Eq. (2),

hψa |ψa i hψb |ψb i ≥ |hψa |ψb i|2 (5)

we arrive to:
D ED E 
2
∆ Â2 ∆ B̂2 ≥  ∆ Â∆ B̂  , (6)


2
where ∆ Â2 = Â2 − Â = δA2 is the variance, the same for the operator B̂. From the



Eq. (6), it is not difficult to show that:

q
 Â, B̂ 2 +  Â, B̂ 2 ,

  
 
δAδB ≥ (7)


where Â, B̂ = Â B̂ + B̂ Â, and δA and δB are the standard deviation. It is worth to notice
that the association of the standard deviation whit the uncertainty relations was not proposed
by Heisenberg, it was Kennard and Robertson [2] who made this association. Although
Heisenberg endorse it later. As it was stated above, Heisenberg associates p1 and q1 with the
mean error, also in the same paper he associates these quantities with the widths of Gaussian
functions representing the quantum states of the system.
Some problems arises with the textbooks uncertainty relations: i) They are given in terms of
the standard deviation, ii) They depend on the state of the system. Additionally, iii) They
1
There are others forms

to obtain the uncertainty relations, this begin by defining an operator as D̂ = ∆ Â + λ∆ B̂ and,
then, requiring that D̂ † D̂ ≥ 0.

70 Advances in Quantum Mechanics
4 Quantum Mechanics

does not represent the meaning of the impossibilities of simultaneous measurement of two
observables, iv) They does not quantify the role of the disturbance in the state after the
measurement process. Finally, v) They does not address the concept of complementarity.
There have been proposed some criteria to solve this problems, we are going to review this
proposals in the next sections.

3. Reformulations to the uncertainty principle

In this section we will review some proposed solutions to the problems stated in the last
paragraph of the previous section.

3.1. The dependence on the standard deviation

The principal criticism to the dependence of the uncertainty relation on the standard
deviation comes from J. Hilgevoord and J. M. B. Uffink [5, 6]. Their argument is based
on two reason, first, they argue that the standard deviation is an appropriate measure of
the error of a measurement because errors usually follow a Gaussian distribution, and the
standard deviation is an appropriate measurement of the spread of a Gaussian; however, this
is not true for a general distribution. Secondly, they gave as a principal counter argument the
fact that even for simple phenomenon as the single slit the standard deviation of momentum
diverges. Their approach is inside the thinking that the uncertainty relations are the measure
of the spread of the probability distribution, i. e. it is believed that ∆x and ∆p represents
the probability distribution of the possibles properties of the system. In short, it represents
the spread of values (of x̂ or p̂) that are intrinsic in the physical system that are available to
appear after a measurement.
The principal counter argument with regard to the standard deviations comes from the
single-slit experiment. In this case, it is supposed that the state of an income beam of particles
is represented by plane waves. This plane wave represents a particle of precise momentum
p0 . Then, the plane wave arrives at the single-slit and is diffracted by it. Therefore, the wave
function at the screen, according to Hilgevoord and Uffink, is:

(2a)−1/2 , if | x | ≤ a;

ψ( x ) =
0, if | x | < a.
and
sin ap
φ( p) = ( a/π )1/2 . (8)
ap

Now, the problem with the standard deviation, as defined in quantum mechanics, in this
case is that it diverges:
D E
∆p = p̂2 − h p̂i2 → ∞. (9)

Therefore, these authors defined, instead of the standard deviation, the overall width
(Wψ ) and the mean peak width of ψ as the smaller W and w that satisfies the following
 The Improvement of the Heisenberg Uncertainty Principle 71
The Improvement of the Heisenberg Uncertainty Principle 5
http://dx.doi.org/10.5772/54530
10.5772/54530

equations [7]:

Z x0 +W/2
|ψ( x )|2 dx = N
x0 −W/2
Z 2
 ψ∗ ( x ′ )ψ( x ′ − w)dx ′  = M2
 
  (10)

These quantities, i. e. W and w, provides a better characterization of the spread of the

possible values of x̂ and p̂, in particular there is not any divergence in these numbers. Based
in these definitions Hilgevoord and Uffink give the following uncertainty relations, that they
propose as a substitute to the uncertainty relation given by Kennard (∆x∆p ≥ 1/2), [7]:

M+1−N
 
wφ Wψ ≥ arccos
N
M+1−N
 
wψ Wφ ≥ arccos (11)
N

these uncertainty relations works well for the single-slit and double-slit experiments.

3.2. Entropic Uncertainty Relations

In the quantum literature, there are many defined Entropic Uncertainty Relations. Mostly,
they are based in terms of Shannon entropy [8, 9], although in last ten years there has been
extension to other forms of entropy, like Renyi entropy [10]. In reference [11] there is a recent
review of this research area.
One of the important result in this area was the one found by Deutsch [8]. What Deutsch
pursuit was a quantitative expression of the Heisenberg uncertainty principle, he notice that
the customary generalization has the drawback that the lower limit depends on the quantum
state, that is:
1 
 2
∆A∆B ≥ Â, B̂  . (12)
4
Deutsch stress that the right hand side of the Equ. (12) does not has a lower bound but is a
function of the state |ψi, even it vanishes for some choices of |ψi. So, in search of a quantity
that could represent the uncertainty principle Deutsch propose some elementary properties,
like for example that the lower limit must vanishes if the observables have an eigenstate
in common. Based in this considerations he proposed the following entropic uncertainty
relation:
 
2
S Â + SB̂ ≥ 2Ln , (13)
1 + sup{| h a|bi |}

where S Â = − ∑ a |h a|ψi|2 Ln |h a|ψi|2 and SB̂ = − ∑b |hb|ψi|2 Ln |hb|ψi|2 are the Shanon
entropy, and | ai and |bi are, respectively, the eigenstates of  and B̂.
72 Advances in Quantum Mechanics
6 Quantum Mechanics

The next step in this line of research, was quite soon given by Hossein Partovi [12], he
points out that the above uncertainty relation does not take into account the measurement
process. Then, considering that the measuring device realizes a partitioning of the spectrum
of the observable and the assignation of their corresponding probabilities, he proposes the
following definition of entropy [12]:

S A = − ∑ pi ln { pi } . (14)
i

where pi = ψ π̂iA  ψ / hψ|ψi and π̂iA is the projection onto the subspaces spanned by the

  

states corresponding to the partition induced by the measuring apparatus [12]. In this case, pi
gives the probability of obtaining the outcome of a measurement in a subset of the partition
realized by the measuring apparatus. In this approach, the whole spectrum correspond
to the observable  but its partitioning correspond to the measuring device. Using these
considerations Hossein Patrovi proses the following lower bound for the uncertainty relation:

!
2
S Â + SB̂ ≥ 2Ln . (15)
1 + supij {||π̂iA + π̂ jB ||}

In the special case where the partition realized by the measuring device  E D includes
 only one
A B
point of the spectrum of Â, i. e. π̂i = | ai i h ai | and B̂, i. e. π̂ j = b j

b j , then Equ. (15)


reduces to Equ. (13). Finally, it is worth to mention that the Patrovi’s formulation requires
a formulation of the details of the measuring devices, specifically, the kind of partition that
induces (or could be used) in the spectrum of the observable.
There were two additional improvement on the lower bound of the entropic uncertainty
relations defined above. The first one was due to Bialynicki-Birula who presented, based in
his earlier wok [9], a lower bound for the angle-angular momentum pair [13] Sφ + S Lz ≥
−ln(∆φ/2π ) and an improved lower bound for the position-momentum pair S x + S p ≥
1 − ln(2) − ln(γ), where γ = ∆x∆p/h. The second one was proposed by Maasen and Uffink
[14] who demonstrated, based on a previous work of Kraus [15], that

S A + S B ≥ −2ln(c), (16)

D E
where c = max jk  a j |bk .
 

3.3. Simultaneous measurement

Whereas in the previous two subsection we treated the face of the Uncertainty Principle
that is related with the probability distribution of observables of a given wave function, in
this subsection we talk a bout a second version of The Uncertainty Principle. This version
is related with the fact that it is not possible to determine simultaneously, with precision,
two canonically conjugate observable and usually called joint measurement. This is stated,
generally, as: "It is impossible to measure simultaneously two observables like, for example, position
 The Improvement of the Heisenberg Uncertainty Principle 73
The Improvement of the Heisenberg Uncertainty Principle 7
http://dx.doi.org/10.5772/54530
10.5772/54530

and momentum." So, this sub-research area is concerned with the simultaneous measurement
of two observables.
One of the first work in this approach was that of Arthurs and Kelly [16], they analyze
this problems as follows: First, they realize that as the problem is the measurement of
two observables, then it is required two devices to perform the measurement. That is,
the system is coupled to two devises. Then, they consider that as the two meter position
commutes then it is possible to perform two simultaneous measurements of them. Therefore,
the simultaneous measurement of the two meters constitutes a simultaneous measurement
of two non-commuting observables of the system. As the two meters interacts with the
quantum system, they consider the following Hamiltonian:

Ĥint = K q̂ P̂x + p̂ P̂y (17)

where q̂ and p̂ correspond to the position and momentum of the quantum system,
respectively, and P̂x and P̂y are the momentum of the two independent meters. Using two
Gaussian function as the initial wave function of the meters they arrive at the following
uncertainty relation for the simultaneous measurement of two observables:

σx σp ≥ 1. (18)

Therefore, the uncertainty relation of the simultaneous measurement of q̂ and p̂ is greater

(by a factor of two) than the uncertainty relations based on the probability distribution of the
two observables, the topic of the previous two sub-sections.
The next step in this approach was given by Arthurs and Goodman [17]. In this case, the
approach is as follow: To perform a measurement, the system observables, Ĉ = Ĉ1 ⊗ Î2
and D̂ = D̂1 ⊗ Î2 , must be coupled to a measuring apparatus which is represented by the
operators R̂ = Î1 ⊗ R̂2 and Ĉ = Ĉ1 ⊗ Î2 . Then, if we consider that there is access only to the
meter operators then there must exist an uncertainty relations for these operators that puts
a limit to the available information. Based in this consideration, they prove what they call a
generalized uncertainty relation. To prove it they defined a a noise operator by

N̂R = R̂ − GR Ĉ (0),
N̂S = Ŝ − GS D̂ (0) (19)

where Ĉ (0) and D̂ (0) are the system observables and R̂ and Ŝ are the tracking apparatus
observables, the latter obey the commutation rule [ R̂, Ŝ] = 0. Also, it is required that the
correlation between the system observables and the meter has, on average, a perfect match,
that is:




Tr ρ̂ N̂R,S = R̂ − GR Ĉ (0) = 0. (20)

Using the previous condition, i. e. Equ (20), it is possible to show that the noise operator
is uncorrelated with all system operators like Ĉ and D̂. Using all the previous properties of
74 Advances in Quantum Mechanics
8 Quantum Mechanics

the system, meter and noise operators they arrive to the following generalized Heisenberg
uncertainty relation:
  

σξ ση ≥  Tr ρ̂ Ĉ, D̂  , (21)
where ρ̂ is the state of the system, σξ and ση are, respectively the standard deviation of the
normalized operators ξ = R̂/GR and η = R̂/GR . This uncertainty relation is four times
the corresponding uncertainty relation for Ĉ and D̂. Notice that in the left hand side of the
Eq. (21) there is information of the meter operator whereas in the right hand side there is
information of the system operators and that we have access only to the meter system. In
reference [18] there was published an experimental verification of this uncertainty relation.

3.4. Disturbance due to measurement

The disturbance produced on an observable due to the measurement of another observable
is, perhaps, the face of the uncertainty principled most talked about but the least studied.
This comes from the fact that in quantum mechanics any measurement introduces an
unforeseeable disturbance in the measured quantum system. It was only recently that there
have been some research and understanding of this effect.
Originally, the idea that the measuring process disturb observables comes from Heisenberg’s
analysis of the observation of an electron by means of a microscope. This kind of uncertainty
principle is written down, to use recent terminology, as [19]:

1
ǫ( x )η ( p) ≥ |hψ |[ x̂, p̂]| ψi| , (22)
2

where ǫ( x ) is the noise in the measurement in position and η ( p) is the disturbance caused by
the apparatus [19]. Using a general description of measurement Ozawa demonstrated that
the uncertainty relation for disturbance and noise given by the Eq. (22) does not accurately
represent the disturbance process. He has show that this kind of uncertainty relation includes
additional terms not present in Eq. (22). In the measurement process, the quantum system
interacts with a measuring device. He considers that this devices measures observable A
precisely if its experimental probabilty distribution coincides with the theoretical probability
distribution of the observable. In the measurement process, when the interaction have been
turned off, the device is subject to a measurement of an observable M. Then, Âin = Â ⊗ Î is
the input observable, Âout = Û † Â ⊗ Î Û is the observable after the mesaurement, M̂in =

Î ⊗ M̂ is the device observable when the interaction begin, M̂out = Û † ( Î ⊗ M̂ )Û and Û is the
unitary time evolution operator
To show that the original uncertainty relation need additional terms, he introduces the
following noise N ( Â) and disturbance D ( B̂) operators:

N ( Â) = M̂out − Âin ,

D ( B̂) = B̂out − B̂in . (23)
 The Improvement of the Heisenberg Uncertainty Principle 75
The Improvement of the Heisenberg Uncertainty Principle 9
http://dx.doi.org/10.5772/54530
10.5772/54530

Using this operators, and considering that M̂out , B̂out = 0, Ozawa was able to show the
 

following uncertainty relation [19]:

1 Dh iE Dh iE 1 

 N ( Â), B̂in + Âin , D ( B̂)  ≥  ψ  Â, B̂  ψ  .
 
ǫ( A)η ( B) + (24)

2 2

where the noise ǫ( A) was defined by Ozawa as the root-mean-square deviation of the
experimental variable M̂out from the theoretical variable Âin :

 1/2 
ǫ( A) = M̂out − Âin (25)

and the disturbance η ( B) on observable B̂ is the change in the observable caused by the
measurement process:

 1/2 
η ( B) = B̂out − B̂in . (26)

This uncertainty relation has been recently experimentally tested, see reference [20]

4. Entanglement determination using entropic uncertainty relations

Nowadays entanglement is considered as an useful resource to make non-clasical task. As
a resource it is convenient to have adequate measures to quantify how much entanglement
are in a given entangled state. However, until recently the most known proposed measures
have the unwanted fact of being difficult to apply in experimental settings. Therefore, it was
necessary to find out new ways of entanglement determination that enable that the amount
of entanglement in a quantum state could be experimentally tested.
Recently there has been much research to proposed new entanglement determination based,
mostly, in uncertainty relations. In this case, the entropic uncertainty relations helps to realize
this task. Recently, Berta et. al. [21] have proposed a new uncertainty relation (based on that
proposed in references [8, 14]) to test the entanglement:

S( R| B) + S(S| B)log2c + S( A| B) (27)

to propose this equation Berta et. al. consider that the system, with observables S and R,
is entangled with a memori, with observable B, so in equation S( R, B) is the von Neumann
entropy and gives the uncertainty about the measurement of R given information stored in a
quantum memory, B. The term S(A|B) quantifies the amount of entanglement between the
particle and the memory. This relation was experimentally tested in reference [22].
76 Advances in Quantum Mechanics
10 Quantum Mechanics

5. Conclusions
In this chapter we review some of the most important improvements of the Heisenberg
uncertainty relation. Although there are advances in their understanding and formulation,
it remains yet as an open research area, specially in the quantification of entanglement.

Acknowledgements
We thanks Consejo Nacional de Ciencia y Tecnologia (CONACYT). L. M. Arévalo Aguilar
acknowledge the support from Vicerrectoria de Investigación y Posgrado VIEP-BUAP
under grand ARAL-2012-I. P. C. Garcia Quijas acknowledges CONACYT for a posdoctoral
scholarship at the Universidad Autonoma de Guadalajara.

Author details
L. M. Arévalo Aguilar1,⋆ ,
C. P. García Quijas2 and Carlos Robledo-Sanchez3
⋆ Address all correspondence to: olareva@yahoo.com.mx
1 Facultad de Ciencias Físico Matemáticas, Benemérita Universidad Autónoma de Puebla,
Puebla, México
2 Departamento de Fśica, Universidad de Guadalajara, Guadalajara, Jalisco, México
3 Facultad de Ciencias Fisico Matemáticas, Benemérita Universidad Autónoma de Puebla,
Puebla, México

References
[1] W. Heisenberg, Zeitschrift fur Physik 43, 172 (1927). Translated in the book, Quantum
Measurement, Weeler and Zurek editors.

[2] H. P. Robertson, Phys. Rev. 46, 794 (1934).

[3] D. Griffiths, Introduction to Quantum Mechanics, Addison-Wesley; 2nd edition (April 10,
2004).

[4] N. Zettili, Quantum Mechanics: Concepts and Applications, Wiley; 2 edition (March 4,
2009).

[5] J. Hilgevoord and J. M. M. Uffink, Eur. J. Phys. 6, 165 (1985).

[6] J. Hilgevoord and J. M. M. Uffink, The mathematical expression of the uncertainty

principle, in the book Microphysical Reality and Quantum Formalism, 91-114, A. van der
Merve et. al. Eds.

[7] J. M. M. Uffink and J. Hilgevoord, Found. Phys. 15, 925 (1985).

[8] D. Deutsch, Phys. Rev. Lett. 50, 631 (1983).

[9] I. Bialynicki-Birula and J. Mycielski, Commun. math. Phys. 44, 129 (1975).
 The Improvement of the Heisenberg Uncertainty Principle 77
The Improvement of the Heisenberg Uncertainty Principle 11
http://dx.doi.org/10.5772/54530
10.5772/54530

[10] I. Bialynicki-Birula, Phys. Rev. A 74, 052102 (2006).

[11] S. Wehner, New Jour. of Phys. 12, 025009 (2010).

[12] M. Hossein Partovi, Phys. Rev. Lett. 50, 1883 (1983).

[13] I. Bialynicki-Birula, Phys. Lett. A 103, 253 (1984).

[14] H. Maasen and J. B. Uffink, Phys. Rev. Lett. 60, 1103 (1988).

[15] Krauss, Phys. Rev. D 35, 3070 (1987).

[16] E. Arthurs and J. L. Kelly JR., Bell. Syst. Tech. J. 44, 725 (1965).

[17] E. Arthurs, M. S. Goodman, Phys. Rev. Lett. 60, 2447 (1988).

[18] A. Trifonov, G. Bjȯrk and J. Sȯderholm, Phys. Rev. Lett. 86, 4423 (2001).

[19] M. Ozawa, Phys. Rev. A67, 042105 (2003).

[20] J. Erhart, S. Sponar, G. Sulyok, G. Badurek, M. Ozawa and Y. Hasegawa, Nature Physics
8, 185Ð189 (2012).

[21] M. Berta, M. Christandl, R. Colbeck, J. M. Renes and R. Renner, Phys. Nature Physics 6,
659 (2010).

[22] Chuan-Feng Li, Jin-Shi Xu, Xiao-Ye Xu, Ke Li & Guang-Can Guo, Nature Physics 7, 752
(2011).