com/naturephysics 253
INSIGHT

CONTENTS
C
O
N
T
E
N
T
S
NPG LONDON
The Macmillan Building,
4 Crinan Street, London N1 9XW
T: +44 207 833 4000
F: +44 207 843 4563
naturephysics@nature.com
EDITOR
ALISON WRIGHT
INSIGHT EDITOR
IULIA GEORGESCU
PRODUCTION/ART EDITOR
ALLEN BEATTIE
COPY EDITOR
KEVIN SHERIDAN
SENIOR COPY EDITOR
JANE MORRIS
EDITORIAL ASSISTANT
EMMA BURCH
MARKETING
SARAHJANE BALDOCK
PUBLISHER
RUTH WILSON
EDITORINCHIEF,
NATURE PUBLICATIONS
PHILIP CAMPBELL
INSIGHT

CONTENTS
COVER IMAGE
From science to technology and back again:
stateoftheart tools developed for quantum
technologies, in theory and experiment,
are allowing researchers to revisit the
foundations of quantum theory and to explore
the terra incognita that may lie beyond.
IMAGE: CARTA MARINA, OPUS OLAI MAGNI GOTTI
LINCOPENSIS, EX TYPIS ANTONII LAFRERI SEQUANI,
ROM, 1572 COLOURED ENGRAVING, NATIONAL LIBRARY
OF SWEDEN, MAP COLLECTION, KOB, KARTOR, 1 AB
Foundations of quantum mechanics
T
he felds of quantum information
theory and quantum technology
exploded in the late 1990s
the very decade that marked the rise
of the internet. Labelled the second
quantum revolution, this new wave of
multidisciplinary research was fuelled
by the quest for faster computers and
secure communication. But exploiting
purely quantum mechanical features for
information processing requires a deeper
understanding of their origin and role
in diferent physical systems, as well as
exquisite experimentalcontrol.
More than two decades of research have
resulted in remarkable theoretical progress
and experimental capabilities that now
enable us to revisit the very foundations of
quantum theory. To make a cartographic
analogy, our present understanding of
quantum mechanics is like an island
containing still uncharted regions and with
indistinct coastlines; even less is known of
what may lie beyond the surrounding seas.
Tis Nature Physics Insight covers some of
the exploratory attempts to improve our
map of the quantum world.
Experimental advances in the creation
of macroscopic superposition states are
pushing the limits of quantum theory to
establish whether (or where) the quantum
description eventually breaks down and
the classical one takes over. Such studies
might even betray gravitational corrections
to quantum mechanics and could therefore
be useful in quantum gravity research.
In parallel, photonic experiments are
providing new insight into nonlocality and
complementarity recent work seems to
suggest that these too could be exploited to
test models of quantum gravity, taking that
quest from astrophysical observations to
Earthbased experiments.
On the theoretical side, intriguing
concepts are emerging such as possible
nonlocal correlations that are stronger than
those predicted by quantum mechanics,
or the existence of an indefnite causal
structure. Tese concepts could be
exploited in new quantum information
processing tasks, and they illustrate the
twoway relationship that exists between
quantum information theory and the
foundations of quantum mechanics.
And, as we celebrate ffy years of Bells
theorem this year, it seems timely to
consider entanglement and its previously
unsuspected connections to other areas
of physics, such as thermodynamics and
manybody theory.
It would be impossible to cover all of
the exciting research directions in this
very active feld, hence the aim of this
Insight on the foundations of quantum
mechanics is to provide merely a taste
and to encourage a deeper exploration of
thesubject.
Iulia Georgescu, Associate Editor
COMMENTARY
Gravity in quantum mechanics
Giovanni AmelinoCamelia 254
Quantum entanglement
Vlatko Vedral 256
PROGRESS ARTICLE
Quantum causality
aslav Brukner 259
REVIEW ARTICLES
Nonlocality beyond quantum mechanics
Sandu Popescu 264
Testing the limits of quantum mechanical superpositions
Markus Arndt and Klaus Hornberger 271
Testing foundations of quantum mechanics with photons
Peter Shadbolt, Jonathan C. F. Mathews, Anthony Laing and
Jeremy L. OBrien 278
2014 Macmillan Publishers Limited. All rights reserved
254 NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics
COMMENTARY

INSIGHT
Gravity in quantum mechanics
Giovanni AmelinoCamelia
Gravity and quantum mechanics tend to stay out of each others way, but this might change as we
devise new experiments to test the applicability of quantum theory to macroscopic systems and larger
lengthscales.
T
he remarkable accomplishments of
twentiethcentury physics revolve
around the success of two theoretical
paradigms. On the one hand, we have
phenomena described by quantum
mechanics and involving interactions
that are governed by the standard model
of particle physics. On the other hand,
we have (general) relativity and its
description of gravitational phenomena.
Tese two very diferent theories manage
to share quarters by keeping clear of
each other
1
. Gravity is negligible in
the typical applications of quantum
mechanics, which involve microscopic
particles and relatively short distances.
Analogously, quantum mechanical
efects are usually inconsequential
when we studymacroscopic bodies and
largedistance scales, where gravity is
incharge.
Still, we do not expect gravity to be
truly absent at microscopic scales or that
quantum mechanics should somehow
switch of at macroscopic distances: it is
just that the efects they produce in those
regimes are very small and we have not yet
managed to develop the technologies and
devise the experiments capable of seeing
such small efects. But this frustratingly
leaves us without a clue about how
these very diferent theories manage to
cooperate when they both must be taken
intoaccount.
Te early Universe is the prototypical
example of where we expect both theories
to produce large efects
1,2
. However, with
no direct experimental access to the
conditions in the early Universe we have to
look elsewhere, and our best chances are
in regimes where one of the two theories
dominates the description of the dynamics
and yet the smaller efects of the other
theory might come within the reach of
some highsensitivity experiments. More
simply put, we should then be looking
either for (i) modifcations of gravity by
quantum mechanics or, conversely, for
(ii) modifcations of quantum mechanics
bygravity.
Among the numerous approaches
1
used to defne the interface between
relativity and quantum mechanics, I fnd
it easiest to focus on the one centred
on modifed uncertainty relations and/
or modifed commutator relations. A
wellstudied scenario
3,4
assumes that the
uncertainty relations for measuring the
position coordinates x
j
and momenta p
k
are produced by noncommutativity of the
relevant observables of the form:
[x
j
, p
k
]=i
jk
h(1 +
2
p
2
) (1)
where
jk
is 1 only when j = k (0 otherwise)
and is a lengthscale characteristic
of the modifcation to be determined
experimentally. Te standard Heisenberg
commutator is recovered in the limiting
case where the efects of can be neglected.
In addition, there may also be new
uncertainty principles and nontrivial
commutators involving pairs of position
coordinates of the form
5,6
:
[x
j
, x
k
]=i
jk
+i
m
jk
x
m
(2)
where the matrices
jk
and
m
jk
would
have to be characterized by small length
scales, small enough to explain why
quantum mechanics was so far successful
ignoringthem.
For measurements involving large
distance scales, the term
m
jk
in equation(2)
could be important in two very diferent
ways. Tere will be situations that we
usually describe only in terms of relativity
and gravitational efects and in these cases
the analysis of the spacetime properties
will be afected by the new properties
of spacetime coordinates governed by
m
jk
. And there will be situations that
we usually describe using quantum
theory alone here the analysis of the
quantum uncertainties might receive small
corrections of quantumgravitational origin
governed by
m
jk
.
Te frst of these two possibilities has
already been studied intensely, particularly
over the past decade
2
. Tese eforts focused
on phenomena of a mainly relativistic
and gravitational nature that are studied
with experimental sensitivities for which
one might expect tiny efects originating
from the interface between gravity and
quantum mechanics. Some of the most
interesting opportunities for such tests
concern the description of the propagation
of particles over astrophysical distances.
Relativity makes frm predictions for
these laws of propagation assuming,
however, that spacetime coordinates are
unafected by uncertainty principles.
But new uncertainty principles, such
as the oneencoded through equation
(2), would afect the structure of the
signal in photons and neutrinos seen by
telescopes monitoring distant explosions in
astrophysical bodies.
Such imprints are now being sought
with the Fermi (see Fig. 1), HESS and
MAGIC telescopes for photons, and
with IceCube for neutrinos. Tere is a
determined efort to fnd evidence of
spacetime fuzziness efects. Te data
analysis would be very simple if we could
assume that the astrophysical source
emits a burst of highenergy photons
and neutrinos all in exact simultaneity:
if such a shortduration burst propagates
in a classical spacetime, then all particles
in the burst must reach our telescope
(nearly) simultaneously. One of the
possible implications of the modifed
uncertainty relations is that the signal
would propagate with some fuzziness
and the particles would not reach our
telescope simultaneously the arrival
times might therefore exhibit some sort of
statistical spread.
However, we know that the duration of
particle bursts from astrophysical sources
is not ideally small: in the best cases the
bursts last a few seconds. Tis decreases
the sensitivity of the studies, but we are
learning how to compensate for these
aspects of the emission mechanisms.
Te results so far have been negative,
but the expected pace of improvement
in sensitivities for the next decade or so
2014 Macmillan Publishers Limited. All rights reserved
NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics 255
INSIGHT

COMMENTARY
provides hope that a discovery might be
just around the next corner.
Much less has been done for the
second possibility, the case of phenomena
primarily governed by standard quantum
mechanics, but afected by small
corrections originating from the interface
between gravity and quantum mechanics.
Nevertheless, over the past couple of years
there have been some studies that I believe
might set the stage for quick progress in
this line of research.
In this respect I feel it is very signifcant
that techniques are being developed
for testing some of the most striking
features of quantum mechanics such
as entanglement in experiments
involving the exchange of particles over
truly macroscopic distances, including the
possibility of exchanging particles between
a ground laboratory and a satellite
7
(see also
the Review by Shadboltetal. in this issue
8
).
On the theory side, we will have to catch
up with these experimental opportunities
and we have just recently started to
make progress
9,10
in understanding how
the noncommutativity of coordinates
(equation(2)) could afect entanglement
and other striking aspects of quantum
theory. For instance, we expect that the
presence of further contributions to the
uncertainties that grow with distance
would produce a loss of coherence in the
quantum states that becomes increasingly
signifcant over large distances. Tis
coherence loss would lead to a gradual loss
of entanglement. So with an entangled state
shared between Earth and a distant satellite,
we could fnd an otherwise unexpected
loss of coherence possibly signalling a
quantumgravityefect.
It would be very interesting to look
for such an efect, even though the
theoretical eforts aimed at modelling such
phenomena cannot give us much guidance
yet. We understand qualitatively the
mechanism that should produce the loss
of coherence, but we are still looking for a
satisfactory phenomenological description
for guiding these longdistance quantum
theory tests.
Another promising direction is the
development of experimental techniques
for testing the applicability of quantum
theory to macroscopic systems. In
particular, it is now possible to observe
the quantum behaviour of the observables
of truly macroscopic mechanical
oscillators
11,12
(see also the Review by
Arndt and Hornberger in this issue
13
).
If the gravitational corrections to the
quantum mechanics of macroscopic
bodies are very diferent from those
for microscopic particles, this could be
exploited in the search of manifestations of
the interface between gravity and quantum
mechanics.
It is useful to contemplate an idealized
description
12,14,15
of a macroscopic body
composed of N identical particles in
terms of its centreofmass coordinates
X
j
=
1
N

N
n=1
x
j
n
and P
j
=
1
N

N
n=1
p
j
n
where
x
n
j
and p
n
j
denote the coordinates and
momentum of the nth particle. In the
current formulation of quantum mechanics,
the validity of standard commutators
for the constituent particles [x
m
j
, x
n
k
]=0
and [x
m
j
, p
n
k
]=i
nm
jk
h implies that also
[X
j
,X
k
]=0 and [X
j
, P
k
]=i
jk
h, meaning
that the same commutation relations apply
to the centreofmass degree of freedom
of the macroscopic body. Tis striking
property of standard quantum mechanics
turns out to be fragile, and as soon as any
of the parameters ,
jk
,
m
jk
in equations
(1) and (2) become nonnegligible, the
correspondence between the quantum
properties of the microscopic particles and
those of the macroscopic body is lost
12,14,15
.
Tis should encourage accurate tests
of quantum mechanics with macroscopic
bodies, especially if we can manage to take
diferential measurements that compare
the quantum properties of a macroscopic
body to those of one of its constituents. But
here too theory needs to advance to a level
where it can feed back to experiments: the
sort of descriptions of macroscopic bodies
for which the relevant quantumgravity
scenarios have been so far analysed do
not go much further than the idealized
description of a macroscopic body that
I used here. More realistic theoretical
descriptions of macroscopic bodies could
provide guidance for these experiments.
I, for one, am not at all frustrated by the
fact that theory might have to catch up with
experiments. For a long time a time that
might be eventually viewed as the dark ages
of quantumgravity research it seemed
that the study of the interface between
gravity and quantum mechanics should
be a unique case of puretheory science.
It was not expected that experiments
would ever reach the level of the theory.
But things are now changing, and it would
be extremely exciting if experiments
took the lead in some areas of quantum
gravityresearch.
Giovanni AmelinoCamelia is in the Physics
Department, Sapienza University of Rome,
Rome00185, Italy.
email: Giovanni.AmelinoCamelia@roma1.infn.it
References
1. Carlip, S. Rep. Prog. Phys. 64, 885942 (2001).
2. AmelinoCamelia, G. Living Rev. Rel. 16, 5 (2013).
3. Kempf, A. etal. Phys. Rev. D 52, 11081118 (1995).
4. Ali, A.F. etal. Phys. Lett. B 678, 497499 (2009).
5. Doplicher, S. etal. Phys. Lett. B 331, 3944 (1994).
6. Majid, S. & Ruegg, H. Phys. Lett. B 334, 348354 (1994).
7. Rideout, D. etal. Class. Quant. Grav. 29, 224011 (2012).
8. Shadbolt, P., Mathews, J.C.F., Laing, A. & OBrien, J.L.
Nature Phys. 10, 278286 (2014).
9. Adhikari, S. etal. Phys. Rev. A 79, 042109 (2009).
10. Ghorashi, S.A.A. & Bagheri Harouni, M. Phys. Lett. A
377,952956(2013).
11. Chen, Y.J. Phys. B 46, 104001 (2013).
12. Pikovski, I. etal. Nature Phys. 8, 393397 (2012).
13. Arndt, M. & Hornberger, K. Nature Phys. 10, 271277 (2014).
14. Quesne, C. & Tkachuk, V.M. Phys. Rev. A
81, 012106 (2010).
15. AmelinoCamelia, G. Phys. Rev. Lett. 111, 101301 (2013).
N
A
S
A
/
D
O
E
/
F
E
R
M
I
L
A
T
C
O
L
L
A
B
O
R
A
T
I
O
N
Ursa Major
Leo
GRB 130427A
Figure 1  Searching the skies for the tiny efects that originate from the interface between gravity and
quantum mechanics. Gammaray bursts are shortlived and very bright. The highestenergy light ever
detected from such an event (GRB130427A) was observed in 2013. The image in the left panel taken by
NASAs Fermi Gammaray telescope shows how the northern galactic hemisphere of the gammaray sky
looked just before the GRB130427A burst depicted in the right panel.
2014 Macmillan Publishers Limited. All rights reserved
256 NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics
COMMENTARY

INSIGHT
Quantum entanglement
Vlatko Vedral
Recent advances in quantum information theory reveal the deep connections between entanglement
and thermodynamics, manybody theory, quantum computing and its link to macroscopicity.
Q
uantum physics started with
MaxPlancks act of desperation,
in which he assumed that
energy is quantized in order to explain
the intensity profle of the blackbody
radiation. Some twentyfve years
later, WernerHeisenberg, MaxBorn,
PascualJordan, ErwinSchrdinger and
PaulDirac wrote down the complete laws
of quantum theory. A pertinent question
then immediately came up and was
subsequently hotly debated by the founding
fathers of quantum physics: what features
of quantum theory make it diferent
from classical mechanics? Is it Plancks
quantization, Bohrs complementarity,
Heisenbergs uncertainty principle or the
superposition principle
1
?
Schrdinger felt that the answer was
none of the above. In some sense, each of
these features can also be either present or
mimicked within classical physics: energy
can be coarsegrained classically by
brute force if nothing else; waves can be
superposed; and complementarity and
uncertainty can be found in the tradeof
between the knowledge of the wavelength
and position of the wave. But the one
efect Schrdinger thought had no classical
counterpart whatsoever the characteristic
trait of quantum physics is entanglement
2
.
Te reason entanglement is so
counterintuitive and presents a radical
departure from classical physics can be
nicely explained in terms of modern
quantum information theory mixed with
some of Schrdingers jargon. Te states
of quantum systems are described by
what Schrdinger called catalogues of
information (psiwavefunctions). Tese
catalogues contain the probabilities for all
possible outcomes of the measurements
we can make on the system. Schrdinger
thought it odd that when we have two
entangled physical systems, their joint
catalogue of information can be better
specifed than the catalogue of each
individual system. In other words, the whole
can be less uncertain than either of itsparts!
Tis is impossible, classically speaking.
Imagine that someone asks you to predict
the toss of a single (fair) coin. Most likely
you would not bet too much on it because
the outcome is completely uncertain. But
consider that tossing two coins becomes less
uncertain. Indeed, quantum mechanically,
the state of two coins could be completely
known, whereas the state of each of the coins
is still maximallyuncertain.
In quantum information theory, this
leads to negative conditional entropies.
When it comes to quantum coins, as we
know the outcome, two predictable tosses
have zero entropy. However, if we only
toss one coin, the outcome is completely
uncertain and therefore has one unit of
entropy. If we were to quantify the entropy
of the second toss, given that the frst has
been conducted, we would come up with
one negative bit that is, the entropy
of two tosses minus the entropy of one
toss:01=1bit.
It is precisely because of such
peculiarities that the pioneers of quantum
physics considered entanglement weird and
counterintuitive. However, afer around
twenty years of intense research in this area,
we are now accustomed to entanglement
and, moreover, as we learn more about it
we discover that entanglement emerges in
unexpectedplaces.
Negative entropies have a physical
meaning in thermodynamics. My colleagues
and I have shown
3
that negative entropy
refers to the situation where we can erase
the state of the system, but at the same
time obtain some useful work from it. In
classical physics we need to invest work
in order to erase information a process
known as Landauers erasure
4
, but quantum
mechanically we can have it both ways. Tis
is possible because the system erasing the
information could be entangled with the
system that is having its information erased.
In that case, the total state could have zero
entropy, so it can be reset without doing
work. Moreover, the eraser now also results
in a zeroentropy state and so it can be used
to obtain one unit ofwork.
Furthermore, we realized that
entanglement can exist in manybody
systems (with arbitrarily large numbers of
particles) as well as at fnite temperature
5
.
Entanglement can be witnessed using
macroscopic observables, such as the heat
capacity (see ref. 6 for recent experiments).
In fact, entanglement also serves as an
order parameter characterizing quantum
phase transitions
7
, and there is growing
evidence that quantum topological phase
transitions can only be understood in
terms of entanglement
8
. A quantum phase
transition is a macroscopic change driven
by a variation in the ground state of a many
body system at zero temperature
9
. But,
in contrast to an ordinary phase, no local
order parameter can distinguish between
the ordered and the disordered topological
phases
10
. For instance, because the change
from nonmagnetic to magnetic behaviour
constitutes an ordinary phase transition,
we can check whether an ordinary phase
is magnetic by measuring the state of just
one spin. However, a topological phase
transition cannot be characterized by a local
parameter it requires an understanding of
the global entanglement of the wholestate.
Tis is good news for stable encoding
of quantum information. Te idea is to use
topological phases as quantum memories
11
.
Tis is precisely because topological states
are gapped (that is, the energy gap between
the ground and excited states is fnite) and
no local noise can kick the topological state
out of the protected subspace. Te ground
states are also degenerate, meaning that
there are diferent states with the same
level of robustness that can be used to
encodeinformation.
Quantum information theory has
also expanded our understanding of
entanglement in other areas. Exciting
recent work focuses on ways of quantifying
entanglement. Te most fruitful general idea
is to quantify entanglement by measuring
how diferent quantum states are from
their best possible classical approximations.
But, there are many nonequivalent ways
of capturing this diference, which leads
to a great deal of ongoing research
12
. For
instance, nonlocality
13
, which, strictly
speaking, means that no local realistic model
can be found to explain the outcomes of
measurements performed on entangled
systems, is not the same as inseparability.
Tis is because separable states are still
quantum states, whereas local hidden
variables can be drawn from more general
probabilistic theories. Moreover, quantum
2014 Macmillan Publishers Limited. All rights reserved
NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics 257
INSIGHT

COMMENTARY
nonlocality is just one possible way of
violating Bells inequalities and one can
always imagine more nonlocal theories
(see also the Review by Popescu in this
issue
14
). In addition, there is the notion of
noncontextuality the fact that diferent
quantum measurements do not necessarily
commute
15
and on top of that there
are many diferent types of entanglement
(bipartite, multipartite and global) and they
can all be quantifed in diferentways
16,17
.
Why is the question of quantifying
entanglement important? First, if we can
estimate how close classical states are to
quantum ones, we can tell how easy it would
be to simulate quantum states of manybody
systems. Tis is the logic behind a very
powerful numerical method called matrix
product states that has revolutionized
some aspects of solidstate physics
17
. Te
idea is simple: given as few as twenty half
spins (or qubits), we would need 2
20
bits to
store their quantum state this is already
practically intractable for todays classical
computers. However, if we know that no two
groups of qubits are entangled by more than
one unit of entanglement, the size of the
approximation can be drastically reduced.
Take ten qubits versus the other ten qubits
in principle we need 2
10
states to describe the
entanglement between the two subsystems,
but given that we know that they contain
only one unit of entanglement, only two
states for each subsystem willsufce.
Second, if we think of quantum
entanglement as a resource in quantum
cryptography and protocols such as
quantum teleportation and superdense
coding, then being able to quantify
entanglement is crucial for characterizing
the efciency of such protocols.
Entanglement was initially thought
necessary for facilitating the speedup in
quantum computation. More precisely, if
our quantum computer never contains more
than a certain fnite number of entangled
qubits, then it can never be universal. Tis
is true for computers with registers that are
always pure. It is simple to understand why:
a universal computer should be capable
of preparing any physical state, but if
entanglement must always be bounded, then
those states with more entanglement cannot
be reached
18
. However, when it comes to
Te set of manybody (in this case,
manyqubit) states is broadly divided into
entangled and separable (or disentangled)
states. Among separable states there is a
tiny, nowheredense subset of zero discord,
which is called a classically correlated state.
Classical states can be written as mixtures
of states a
1
a
2
a
N
, b
1
b
2
b
N
, where
{a
1
,b
1
} form an orthogonal basis for
qubit1, {a
2
,b
2
} form an orthogonal basis
for qubit 2 and so on for all qubits up to
N. Te rest of the separable states contain
nonzero discord. Te geometry of this set
is not well understood (other than perhaps
for the special case of two qubits). Te set
of all states is convex and so is the set of
separable states.
Te global entanglement of a given
state can be measured, for instance, by
the relative entropy of entanglement,
which is defned as the relative entropy
S(  )=tr(loglog), between the
state () and the closest separable state
to it, (ref.5). Separable states are those
obtained by mixing product states of the
form
1
N
.
Some important classes of states are
shown in the fgure. Te (manyqubit)
GHZ states, 000+111, are close to
separable states because they always have
one unit of entanglement independent
of the number of qubits. In terms of
global entanglement they are close to, for
instance, the product states 000, which
are disentangled. Another example of a
product state is the state ++++, where
every qubit is in a superposition of the basis
states+=0+1.
Te W state is a symmetric
superposition of states containing a
fxed ratio of zeroes and ones, such as
001+010+100. With respect to global
entanglement, W states are more entangled
than the GHZ states, and entanglement
scales with the logarithm of the number
ofqubits.
Cluster states are even more entangled,
where entanglement scales as N/2, with
N being the number of qubits. Te most
entangled states are typical states, which
are simply random states of N qubits. Teir
entanglement scales as N. Te maximally
mixed state (in which all possible qubit
states are mixed with equal probability) is
completely disentangled and contains no
correlations of any type.
Te dephased GHZ states are mixtures
of the states 000 and 111 and
are classically correlated, but have no
quantum discord (since the states 0 and
1 are orthogonal). Mixtures of the states
0000 and ++++, on the other hand,
are not only classically correlated but
also contain quantum discord, although
they are not entangled. States like this are
thought to still be useful for some quantum
computations, but their exact power is not
fully understood at present.
Te scaling of global entanglement
with the number of qubits does not
necessarily refect other measures
of entanglement, such as quantum
macroscopicity. Tis notion is designed
to capture the state in Schrdinger cat
experiments, namely a superposition
of two, or more, macroscopically
distinct quantum states (as in the
GHZ state, which has a high degree
ofmacroscopicity).
+++... +
2
1
(000... 000... + 111... 111...)
00...0 GHZ W
2
1
(00...0 00...0
+++...+ ++...+)
B
A
C
K
G
R
O
U
N
D
I
M
A
G
E
:
A
L
E
X
E
Y
P
A
V
L
U
T
S
/
A
L
A
M
Y
Box 1  Charting the set of manybody physical states.
2014 Macmillan Publishers Limited. All rights reserved
258 NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics
COMMENTARY

INSIGHT
mixed states (Box1), there are examples
of computations that, despite requiring
a small amount of entanglement (never
more than a single entangled bit), can still
achieve an exponential speedup relative
to their classical counterparts. It has been
suggested that these computers exploit a
more general type of quantum correlation,
known as quantum discord (see Box1 and
ref.12 for a review). Unfortunately, even the
amount of discord is bounded during these
computations, so it is hard to see how it
could make adiference.
Te third point is perhaps the most
intriguing as it touches on the issue of
macroscopicity (see also the Review by
Arndt and Hornberger in this issue
19
).
Namely, is there a limit to how big a system
can be and still exhibit sizable quantum
mechanical features? Here it seems
appropriate to invoke Schrdinger again.
But instead of his deadalive cat thought
experiment, what about superposing 10
18
atoms in two places separated by, say,
a millimetre? I deliberately chose these
numbers since we can just about see
(assuming 20/20 vision) this collection
of atoms (comparable to the size of an
amoeba) and resolve its location with
the naked eye. Now, curiously, this is a
particular quantum mechanical state called
a GreenbergerHorneZeilinger (GHZ)
state, which is written as 000+111,
where the state 0 signifes the atom in
one location and 1 the atom in the other
location and there are 10
18
zeroes and
ones. Based on the proximity of classical
states, it is not very entangled. In fact, it is
just about as close to a classical state as an
entangled state can be (Box1). Te global
entanglement (measured, for instance, by
the relative entropy between this state and
the closest separable state) is always 1 for
GHZ states, no matter how many particles
areinvolved.
GHZ states are examples of states that
are difcult to prepare in practice
20
, but are
very easy to simulate classically. States that
are difcult to simulate are, in general, the
ones where entanglement scales with the
number of particles in the system. Tis is
the case for typical manybody interacting
systems, as well as for cluster states used in
measurementbased quantum computation.
On the other hand, cluster states do not
usually exhibit quantum macroscopicity.
Although it seems to be a problem,
the dichotomy between macroscopicity
and the amount of entanglement could
in fact be fortuitous. It is usually said that
being able to build a largescale universal
quantum computer is tantamount to
testing the macroscopic limits (if any) of
quantum theory. But, it could be that for
whatever unknown reason large GHZ
states cannot be made, yet, at the same time,
quantum computers can be designed that far
outperform the existing classical ones. Tat
would be a curious state of afairsindeed!
Tese research directions have
practical and fundamental implications.
Technologically, it is still not fully
understood how far quantum computers
can be scaled up, nor can the full range of
their applications be easily predicted. On
the fundamental side, the problem is how
to bridge the gap between the micro and
the macro domains. Can thermodynamics
be fully reconciled with quantum
entanglement
21
, and how far into the macro
domain do quantum efects really need to
be taken into account? Tis brings up a
whole new set of exciting questions ranging
from whether living organisms could also
exploit entanglement
22
to whether quantum
efects can ever have an impact in the
gravitationaldomain.
Vlatko Vedral is at the Clarendon Laboratory,
University of Oxford, Oxford OX1 3PU, UK
and the Centre for Quantum Technologies,
National University of Singapore,
Singapore117543,Singapore.
email: vlatko.vedral@gmail.com
References
1. Kumar, M. Quantum: Einstein, Bohr and the Great Debate about
the Nature of Reality (Icon Books, 2008).
2. Schrdinger, E. Proc. Camb. Phil. Soc. 31, 553563 (1935).
3. del Rio, L. etal. Nature 474, 6163 (2011).
4. Maruyama, K., Nori, F. & Vedral, V. Rev. Mod. Phys.
81, 123 (2009).
5. Amico, L. etal. Rev. Mod. Phys. 80, 517576 (2008).
6. Sigh, H. et. al. New J.Phys. 15, 113001 (2013).
7. Osterloh, A. etal. Nature 416, 608610 (2002).
8. Kitaev, A. & Preskill, J. Phys. Rev. Lett. 96, 110404 (2006).
9. Sachdev, S. Quantum Phase Transitions (Cambridge Univ.
Press,2011).
10. Wen, X.G. Phys. Rev. B 65, 165113 (2002).
11. Kitaev, A.Y. Ann. Phys. 303, 230 (2003).
12. Modi, K. etal. Rev. Mod. Phys. 84, 16551707 (2012).
13. Bell, J.S. Speakable and Unspeakable in Quantum Mechanics
(Cambridge Univ. Press, 1987).
14. Popescu, S. Nature Phys. 10, 264270(2014).
15. Grudka, A. et. al. Preprint at http://arxiv.org/
abs/1209.3745(2012).
16. Horodecki, R. etal. Rev. Mod. Phys. 81, 865942 (2009).
17. Eisert, J. etal. Rev. Mod. Phys. 82, 277306 (2010).
18. Jozsa, R. & Linden, N. Proc. R.Soc. A 459, 20112032 (2003).
19. Arndt, M. & Hornberger, K. Nature Phys. 10, 271277 (2014).
20. Korsbakken, J.I., Wilhelm, F.K. & Whaley, K.B. Europhys. Lett.
89, 30003 (2010).
21. Dorner, R. etal. Phys. Rev. Lett. 109, 160601 (2012).
22. Sarovar, M. etal. Nature Phys. 6, 462467 (2010).
2014 Macmillan Publishers Limited. All rights reserved
NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics 259
T
he concept of causality has been the subject of heated debates
in literature about the metaphysics and philosophy of science
for centuries. I have no intention of entering into these discus
sions here, but instead adopt a rudimentary, pragmatic approach.
Causal thinking spontaneously arises in a child at about the time
when she or he realizes that by exerting forces on nearby objects,
the child can make these objects move according to their will.
Causal relations are revealed by observing what would happen in
the world (for instance, with the child's object) if a given parameter
(the child's will) were separated out from the rest of the world and
could be chosen freely.
Te distinction between statistical and causal relations is echoed
in the famous slogan correlation does not imply causation. Whereas
the former are defnable in terms of joint probabilities for observed
variables, the latter require specifcation of conditional probabilities
to provide an analysis of how the probability distribution ought to
change under external interventions
1
. Here, I will say that an A has a
causal infuence on B if conditional probability P(BA) for B observ
ably changes under free variation of A. But how can we be sure that
such a variation is really free? We cannot, but this does not prevent
us from considering a variation to be free whenever we have every
reason to believe that it is. For all practical purposes, it is sufcient
to toss a coin or use a quantum random generator to produce a free
variable. And even if that free variable were produced in a deter
ministic way for example by taking the current temperature in
Celsius, multiplying it by the number of my nextdoor neighbours
children plus one, I would still regard it as being free.
In quantum physics, it is assumed that the background time
or defnite causal structure preexists such that for every pair of
events A and B at distinct spacetime regions one has either A is
in the past of B, or B is in the past of A, or the two are spacelike
separated (see Fig.1a,b). But thanks to the theorems developed by
Kochen and Specker
2
, and by Bell
3
, we know that quantum mechan
ics is incompatible with the view that physical observables possess
preexisting values independent of the measurement context. (Tis
incompatibility still holds if one assigns probabilities to the possible
values of observables independently of the measurement context,
rather than determining which particular result will be obtained in
a single run of the experiment.) Do the theorems extend to causal
structures as well?
If one assumes that quantum mechanical laws can be applied to
causal relations, one might have situations in which the causal order
of events is not always fxed, but is subject to quantum uncertainty,
just like position or momentum. Indefnite causal structures could
correspond to superpositions of situations where, roughly speaking,
A is in the past of B and B is in the past of A jointly. One may spec
ulate that such situations could arise when both general relativity
Quantum causality
aslav Brukner
Traditionally, quantum theory assumes the existence of a xed background causal structure. But if the laws of quantum
mechanics are applied to the causal relations, then one could imagine situations in which the causal order of events is not
always xed, but is subject to quantum uncertainty. Such indenite causal structures could make new quantum information
processing tasks possible and provide methodological tools in quantum theories of gravity. Here, I review recent theoretical
progress in this emerging area.
and quantum physics become relevant. A simple example could
involve a single massive object in a superposition of two or more
distinct spatial positions. Because the object is in a spatial super
position, the gravitational feld it produces will also be in a super
position of states, and so will the spacetime geometry itself. Tis
may lead to situations in which it is not fxed in advance whether a
particular separation between two events is timelike orspacelike.
Te consequences of having indefnite causal order would be
enormous, as this would imply that spacetime and causal order
might not be the truly basic ingredients of nature. It has been repeat
edly pointed out that the notion of time might be at the origin of the
persistent difculties in formulating a quantum theory of gravity
49
.
But how do we formulate quantum theory without the assumption
of an underlying causal structure and background time? What new
phenomenology would be implied by the idea of indefnite causal
structures? If such structures cannot be excluded on a logical basis,
do they exist in nature? And if they do, why have they not yet been
observed in quantum experiments?
In 2005, Lucien Hardy proposed to address these penetrating
questions by developing frameworks in which causal structures may
be considered to be dynamic, as in general relativity, and indefnite,
similar to quantum observables
10,11
. He introduced one such frame
work based on a new mathematical object, the causaloid, which
contains information about the causal relations between diferent
spacetime regions. Since then, researchers, particularly in Pavia
12
,
Vienna
13
and the Perimeter Institute
14
, have applied the powerful
tools and concepts of quantum information to shed new light on
the relation between the nature of time, causality and the formalism
of quantum theory a subject that has been traditionally studied
within the general relativity and quantum gravity communities. In a
similar vein, recent rigorous theorems in quantum information have
been developed, which relate the probabilistic structure of quantum
theory to the threedimensionality of space
15,16
. By making plausi
ble assumptions on how (microscopic) systems are manipulated by
(macroscopic) laboratory devices, it was shown that the structure of
the underlying probabilistic theory cannot be modifed (for exam
ple by replacing quantum theory with a more general probabilistic
theory) without changing the dimensionality of space.
In conventional (causal) formulation of quantum theory, cor
relations between results obtained in causally related and acausally
related experiments are mathematically described in very diferent
ways. For example, correlations between results obtained on a pair
of spacelike separated systems are described by a joint state on the
tensor product of two Hilbert spaces, whereas those obtained from
measuring a single system at diferent times are described by an ini
tial state and a map on a single Hilbert space. (Te causal structure
of quantum theory is unrelated a priori to the causal structure of
Faculty of Physics, University of Vienna, Boltzmanngasse 5, 1090 Vienna, Austria and Institute for Quantum Optics and Quantum Information (IQOQI),
Boltzmanngasse 3, 1090 Vienna, Austria. email: caslav.brukner@univie.ac.at
INSIGHT

PROGRESS ARTICLE
PUBLISHED ONLINE: 1 APRIL 2014DOI: 10.1038/NPHYS2930
2014 Macmillan Publishers Limited. All rights reserved
260 NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics
relativity, as, for example, the measurements on two systems may be
timelike separated in the relativistic sense and still be described
by a tensor product of Hilbert spaces. Here, for simplicity, I will
use causally related and timelike, as well as acausally related
and spacelike, interchangeably.) Very much at the focus of recent
research on causality in quantum theory is the objective of fnding
a unifed way of representing correlations between spacelike and
timelike regions. Once such a representation is found, we may be
able to use it for the description of general quantum correlations,
for which the causal ordering of events and whether they take place
between spacelike or timelike regions is not fxed.
Various results in the past have indicated that a unifed quan
tum description may be possible for experiments involving distinct
systems at one time and those involving a single system at distinct
times. For example, it has been shown that there is an isomor
phism between spatial and temporal quantum correlations
1720
. Tis
has conceptual and practical implications for the correspondence
between quantum bounds on violation of the Bell inequality
21
, and
its temporal analogue, the LeggettGarg inequality
2225
. Te cor
respondence between the communication costs in the classical
simulation of spatial correlations and the memory costs in the simu
lation of temporal correlations is an example of this
17,26
. Eventually
all these developments led to several approaches towards a caus
ally neutral formulation of quantum theory: the causaloid
5
already
mentioned above, and further developments in terms of duoten
sors
27
, the quantum combs
28
, quantum processes
13
and quantum
conditionalstates
14
.
Notwithstanding the diferences among the approaches, they all
make use of the ChoiJamiolkowski (CJ) isomorphism
29,30
to pro
vide a unifed framework for representing a composition of opera
tions as well as tensor products of system states. If an operation is
performed on a quantum state described by a density matrix , M()
describes the updated state afer the operation (up to normaliza
tion), where M is a completely positive (CP), tracenonincreasing
map (because we want our maps to lead to positive probabilities not
larger than one) from the space of matrices over the input Hilbert
space A to the one over the output Hilbert space B, which we
write as M:L(A)L(B) (the two Hilbert spaces can have difer
ent dimensions, as the operation may involve additional quantum
systems). Te CJ isomorphism enables us to represent the opera
tions by operators rather than maps. It associates a bipartite state
M
AB
L (AB) to a CP map, as given by M
AB
=JM(
+
+
),
where indicates tensor product, 
+
=
d
A
j=1
jjAB is a (non
normalized) maximally entangled state, the set of states {j
d
A
j=1
} is an
orthonormal basis of A
with dimension d
A
, and J is the iden titymap.
In the comb
28
and duotensor
27
framework, one associates CJ
operators with arbitrary regions of spacetime in which an observer
might possibly perform a quantum operation. Tese operators can
be combined to obtain the operator for a bigger, composite, region,
using methods that are motivated in part by the graphical repre
sentation of categorical quantum mechanics
31
. In the framework of
quantum conditional states, Matt Leifer and Rob Spekkens
14
have
developed a causally neutral formulation of quantum theory using
a quantum generalization of Bayesian conditioning
32
. Tey intro
duce a conditional state
BA
, playing an analogous role to condi
tional probability P(BA) in classical probability theory. If A and B
are spacelike separated regions, their joint state
AB
is inferred from
the conventional formalism, and the conditional state is derived
from the joint, whereas if they are timelike separated it is the con
ditional state
BA
that is inferred from the conventional formalism
(for example through a map
BA
= M(
A
)), and the joint state is
derived. In either case the relation between the conditional and joint
state is given by
AB
=
BA
*
A
where the
*
product is a particular
product (defned by
BA
*
BA
A
, where I have dropped the
identity operators and tensor products), which is analogous to the
Bayes relation in classical probability theory, P(AB)=P(BA)P(A).
But the approach has limitations, for example in treating multiple
temporal correlations, mostly owing to the fact that the
*
product is
noncommutative and nonassociative. (Te approaches
1214,27,28
dif
fer among themselves in the insertion of partial transposes in the
defnition of CJ operators.)
With the notable exception of ref. 14, which has an epistemo
logical favour, all other approaches are typically formulated opera
tionally; instead of using the notions of traditional physics such as
position, momentum or energy, the focus lies on instrument set
tings and the outcomes of measurements. Te operational idea of
a causal infuence is best illustrated by considering two scientists,
Alice and Bob, who work in two separate laboratories. At every
run of the experiment, each of them receives a physical system and
performs an operation on it, afer which they send their respective
system out of the laboratory. During the operations of each experi
menter, the laboratory is shielded from the rest of the worldit is
only opened for the system to come in and to go out, but except for
these two events, it is kept closed and a signal can neither enter into
nor leak out of the laboratory. Each laboratory features a device with
an input and an output connector. If input a is chosen on Alices side
(or, respectively, b on Bobs side), she will perform an operation on
the system and send it out of the laboratory. Te device will output
measurement result x (respectively y) according to a certain prob
ability distribution p(x,ya,b). Te operations a and b, for example,
could be the fip of a classical bit in the classical world or the unitary
transformation, or in general a CP tracenonincreasing map in the
quantum world.
Te correlations are nonsignalling if no observable change
can take place in Alices laboratory as a consequence of any
thing that may be done in Bobs laboratory and vice versa. More
Future
or
Past
a b c d e
10
BA
AB
AB
BA
1
0
0
0
Figure 1  Diferent causal relations between events in Alice's and
Bobs laboratories. In a denite causal structure, a global background
time determines whether a, Alice is before Bob, b, Bob is before Alice, or
c,the two are causally neutral. Whereas in a and b signalling is always
oneway, from the past to the future, there is no signalling in c because
the two laboratories are spacelike separated. The latter is a typical
situation in tests of Bells inequalities on entangled states. d, In a closed
timelike causal structure, the signalling is twoway, which gives rise to
the grandfather paradox. To illustrate this paradox, consider the following
example. Alice performs an identity operation on her input bit of value 0.
The unchanged bit leaves her laboratory and is sent to Bob, who performs
a bit ip and outputs a bit of value 1. The bit travels backwards in time to
enter Alices laboratory as her input. Hence, the logical contradiction 10
for the value of Alices input arises. This can be seen as an instance of the
grandfather paradox if the bit values 1 and 0 are taken to represent killing
Alices grandfather and not killing Alices grandfather, respectively. e,In
an indenite causal structure, Bob can, by choosing his measurement basis,
end up before or after Alice with a certain probability. The vector on
the circle next to Bobs laboratory represents a resource a process
which gives rise to quantum correlations with indenite causal order. If
he performs a measurement in the red (green) basis, he projects the
process such that his actions occur after (before) Alices operations.
NATURE PHYSICS DOI: 10.1038/NPHYS2930 PROGRESS ARTICLE

INSIGHT
2014 Macmillan Publishers Limited. All rights reserved
NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics 261
specifcally, this implies that Alices marginal distributionwhich
is obtained by summing up the joint probability distribution over
Bobs resultsis independent of Bobs input choice and vice versa:
y
p(x,ya,b)=p(x,a) and
x
p(x,ya,b)=p(y,b) for all a, b, x and y. Te
correlations are oneway signalling if one of the two conditions does
not hold, and twoway signalling if neither condition holds.
It can easily be seen that a fxed causal order will impose restric
tions on the ways in which Alice and Bob can communicate.
Imagine that Alice exists in Bobs past. She can act on her system
and encode her input a into it before sending it to Bob. Tat way, his
device can output y=a and the signalling is perfect. Because each
party receives each system only once, Bob cannot signal to Alice.
Consequently, twoway signalling is impossible. I will denote by
p
A B
(x,ya,b) (by p
B A
(x,ya,b)) the general probability distribution in
which signalling from Alice to Bob (from Bob to Alice) is possible. In
a defnite causal structure, it may still be the case that the causal rela
tion between events is not known with certainty. A situation where
Alice exists before Bob with a probability of 01 and Bob exists
before Alice with a probability of 1 is represented by a probabil
ity of the form of p(x,ya,b)=p
A B
(x,ya,b)+(1)p
B A
(x,ya,b).
Are more general causal structures possible? Can Alice and Bob
have twoway communication even though the exchanged system
enters each of the two laboratories only once? At frst sight this seems
impossible, except in a world with closed causal loops, where a signal
may go back and forth from Alice to Bob. Such closed timelike curves
(CTCs) were frst proposed by Kurt Gdel in 1949. Gdel was an
Austrian logician who discovered, surprisingly, that general relativity
equations allow CTC solutions
33
. But the existence of CTCs seems to
imply logical paradoxes, most notably the grandfather paradox in
which an agent goes back in time to kill his grandfather (see Fig.1d).
Possible solutions have been proposed in which quantum mechanics
and CTCs can coexist and such paradoxes are avoided, but not with
out modifying quantum theory into a nonlinear one
(refs 3437, and
unpublished results by C.H.Bennett and B.Schumacher). Nonlinear
theories themselves are problematic
38
. Te question remains: is it
possible to keep the (linear) framework of quantum theory, have no
paradoxes and still go beyond defnite causal structures?
One such framework was proposed recently by OgnyanOreshkov,
Fabio Costa and I
13
. Tere, we assumed that operations in local
laboratories are described by quantum mechanics (that is, they
are CP maps). Using the CJ isomorphism, the probability for
a pair M
x,a
A
1
A
2
and M
y,b
B
1
B
2
of local CP maps performed by Alice
between the local input A
1
and output A
2
Hilbert spaces, and by
Bob between the Hilbert spaces B
1
and B
2
, are represented as a
bilinear function of the corresponding CJ operators as follows:
p(x,ya,b) = Tr[W
A
1
B
1
A
2
B
2
(M
x,a
A
1
A
2
M
y,b
B
1
B
2
)]. Here W
A
1
B
1
A
2
B
2
belongs
to the space of matrices over the tensor product of the input A
1
,B
1
and the output A
2
,B
2
Hilbert spaces of two parties. It is the cen
tral object of the formalism and represents a new resource called
process a generalization of the notion of state. Te matrix W
is a positive matrix, and it returns unit probability for CP trace
preserving maps. Just like in the aforementioned approaches, it
provides a unifed way to represent correlations in casually related
and acausally related experiments. Although the notion of causal
structure is built within the local laboratories insofar as the output
of an operation is causally infuenced by the input, no reference is
made to any global causal relations between the operations in two
laboratories. Most interestingly, we have found situations where two
operations are neither causally ordered nor in a probabilistic mix
ture of defnite causal orders: that is, one cannot say that one opera
tion is either before or afer the other. In these cases the process
is not a probabilistic mixture of the processes with defnite causal
order: W
AB
W
A B
+(1
+)W
B A
, where 01, and W
A B
is the
process in which Alice can signal to Bob, and W
B A
is that in which
Bob can signal to Alice.
In terms of probability distributions, this can be written as
p(a,bx,y)p
A B
(a,bx,y)+(1)p
B A
(a,bx,y). Because the correla
tions are incompatible with any underlying defnite causal structure,
we call them quantum correlations with indefnite causal order.
Te existence of the new correlations can be demonstrated
in a theoryindependent way on the basis of recorded data in an
experimental test. Tese new correlations violate a causal inequal
ity which is satisfed if events take place in a causal sequence. Tis
stands in direct analogy to the famous violation of Bells inequality
in quantum mechanics, which is satisfed if the measured quan
tities have predefned local values
3
. Te causal inequality is best
explained in terms of a game involving two players, Alice and Bob
again, each of whom receives a random input bit value, 0 or 1.Te
point of the game is that each player tries to guess the input of the
other player. One of the players, say Bob, receives an additional
random bit, which specifes who will need to guess whose bit in
a given run of the game. It can easily be seen that in every causal
scenario the success probability of the game is bounded by 3/4.
Without loss of generality, consider that Alice is in Bobs past, as
illustrated in Fig. 1a. Ten she can always send her input bit to
him and they will accomplish their task perfectly if he is required
to guess her bit, whereas if she is asked to guess his bit, she cannot
do better than giving a random answer. Tis gives an overall suc
cess probability of 3/4. But if Alice and Bob share quantum cor
relations with indefnite causal order, they can achieve a success
probability as high as
13
(2 +

2)/4. Whereas causal correlations
allow signalling in no more than one fxed direction, correlations
with indefnite causal order allow Bob, depending on his choice of
measured observable, to efectively end up before or afer Alice
with a probability of 1/

2 (see Fig.1e). All causal loops and para
doxes are avoided: in every single run, only oneway signalling is
realized, but the signalling direction, from Alice to Bob or from
Bob to Alice, is in Bobs control and may vary from run to run. It
is intriguing that both the classical bound and the quantum vio
lation of the causal inequality match the corresponding numbers
in the ClauserHorneShimonyHolt version
39
of Bells inequality.
Most recently, a process for three parties has been found in which
perfect signalling correlations among three parties are possible,
whereas the same is impossible in any causal scenario
40
(this is
analogous to the all versus nothing type of argument against local
hiddenvariables
41
).
Te possibility of indefnite causal orders has also been discussed
in the context of quantum computation
42
. Te idea of (causal) quan
tum computation, or quantum circuits, may be illustrated as a set
of gates physically connected by wires through which quantum
systems propagate. As the systems pass those gates, they change
their states. Tis is repeated in succession until the computation is
A
0
B
1
Figure 2  Superposition of quantum circuits. The causal succession in
which the physical boxes A
and B

2

1
(0
+
and B
=
+

2

1
(0 + 1), the output of the algorithm is:

2

1
(A
0+B
1)=

2

1
A
(0
+

2

1
(0
+

2)/4, as proved
by Cirelson
29
. Tat this is the case is a simple consequence of the
Hilbertspace structure of quantum mechanics. But the deeper
question is, why?
5
Is there a deep principle of nature that limits the
amount ofnonlocality?
Te frst guess was that stronger nonlocal correlations would
be forbidden by relativistic causality; perhaps the randomness that
provides the umbrella under which nonlocality can coexist with
relativistic causality is not enough to allow for stronger nonlocality.
So the very frst question to ask is: could theoretically nonlocal
b a
y x
Figure 1  The blackbox model of two experiments. Each black box is
a whole laboratory. The inputs, x and y, are instructions indicating the
experiment to be performed in the box and a and b are the outcomes of
theexperiments.
A
N
N
A
I
.
P
O
P
E
S
C
U
INSIGHT

REVIEW ARTICLES NATURE PHYSICS DOI: 10.1038/NPHYS2916
2014 Macmillan Publishers Limited. All rights reserved
266 NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics
correlations stronger than quantum mechanical ones exist, without
violating relativity? When Bell discovered nonlocality, the problem
was not formulated in a modelindependent way but by using the
specifc language of quantum mechanics: entangled quantum states,
Hermitian operators, eigenvalues and so on. From this point of view,
the very question of whether or not nonlocal correlations stronger
than the quantum mechanical ones could exist was very difcult
to even envisage, let alone to answer. In the above box framework,
however, the question and its answer are almost trivial: as long as
locally a and b are 0 or 1 with equal probability, there is nothing that
prevents the game from being won with certainty. Tese particular
correlations are now known as PopescuRohrlich (PR)boxes
5,72
.
Superquantum correlations
Te existence of superquantum nonlocal correlations shows that
quantum mechanics cannot be deduced from the two axioms of
(1) relativistic causality and (2) the existence of nonlocal correla
tions. Something else is needed. But what? What could be a supple
mentary, very natural, axiom that could rule out suchcorrelations?
Te statement that superquantum correlations could in
principle exist is very far from a fully fedged physical theory.
Terefore, it may seem very unlikely that one could make further
progress in answering the above question before such a full theory,
which could explain all the known results hydrogen atoms and
so on but also incorporate superquantum correlations, is for
mulated. Surprisingly enough, it turns out that there is a lot one can
do with even just the above particular example. Help came at frst
from computer science, and now this is one of the hottest areas in
the foundations of physics. Various very interesting situations have
been discussed, including communication complexity
73,74
, nonlocal
computation
75
, information causality
76
, macroscopic locality
77
, local
orthogonality
78
and nonlocality swapping
79
. In this Review, I will
discuss only a fewexamples.
Communication redundancy
Almost all of our communication is redundant, and that is not only
because some of us like to talk too much, but also because it is a law
of nature. Indeed, consider the following problem. Suppose Alice
and Bob would like to meet, but are both very busy. Tey speak
on the telephone and try to fnd a day this year when they could
meet. To make the problem more interesting, suppose that they do
not want to fnd out a precise day, but frst they want to establish
whether the number of days when they could meet is even or odd
(zero counting as even). To make the problem simpler, suppose it is
only Bob that sends information to Alice, and Alice has to decide
the result. Te question is, how much information must Bob send
toAlice?
We have now a problem in which the result is a single bit, a sin
gle yes or no answer: yes = even, no = odd. On the other hand, it
is obvious that Bob needs to inform Alice about the status of each
day of the year in his calendar. Indeed, one of the possible situa
tions is that Alice is free only one single day. To decide whether they
can meet or not, she has to know whether Bob is free that day; as
Bob doesnt know anything about Alices calendar, he has to tell her
about each of his days. He has therefore to send Alice 365 bits of
information, a yes=Im free or no=Im not free for each day of
the year; all this for Alice to fnd out a single bit of information. Very
redundantindeed.
Clearly, in the process Alice learns much more than what she
wanted to know. Indeed, not only will she fnd out if the total num
ber of days when they could meet is even or odd, but also she will
know the precise days they can meet. She didnt want to learn that,
but there is no otherway.
Wim van Dam
73
observed in his PhD thesis, however, that if
Alice and Bob have access to PR boxes, they could reduce the com
munication to a single bit, eliminating therefore the entire redun
dancy. Tey can do this by not attempting to directly communicate
information about their calendars, but using this as input to their
boxes and communicating information about theiroutputs.
In particular, all Alice and Bob have to do is to associate with
each day i a variable x
i
(y
i
) that is equal to 0if the day is busy and to
1 if the day is free and use them as inputs for their PR boxes. Te
sum of their outputs is even (odd) if the number of days when they
can meet is even (odd). For Alice to fnd out whether the sum of
their outputs is even or odd, Bob only needs to inform her whether
the sum of his outputs is even or odd, that is, a single bit of com
munication (Box 1).
Te result is particularly important, as the above calendar
problem is not just some silly communication task; it is in fact
the most difcult communication task possible (technically called
the inner product problem). Indeed, every other communication
problem can be mapped onto this one, so removing the redundancy
from this calendar problem means removing the redundancy from
all communicationproblems.
Crucially, quantum mechanical nonlocal correlations cannot
help with this task
80
(though they can help in easier communica
tion problems
81
), hence, they cannot eliminate all redundancy from
communication. Quantum nonlocal correlations (Box 2) are there
fore dramatically diferent from PRboxes.
Prompted by the above result, Brassard et al.
74
raised a tanta
lizing possibility: maybe not only the perfect PR boxes, which are
the strongest nonlocal correlations possible, but all superquantum
nonlocal correlations could eliminate all redundancy from com
munication. If that were the case, it would single out quantum
mechanics as the maximal nonlocal theory that doesnt make all
communicationefcient.
Brassardetal.
74
took the frst steps towards answering their ques
tion. Recall that quantum mechanical boxes can yield outputs a
and b such that ab=xy with a probability of success of at most
(2+

2)/4 0.85, whereas perfect PRboxes have a probability of suc
cess of 1. Using errorcorrection techniques, they showed that even
imperfect PR boxes can eliminate all communication redundancy,
as long as their probability of success is larger than approximately
0.91. However, there is still a gap, from 0.85 to 0.91, about which we
know nothing. Hence, we dont know yet if the task of eliminating
communication redundancy can single out quantummechanics.
Nonlocal computation
While the status of communication complexity (as the above general
problem is technically known) versus quantum mechanics is yet
unsettled, a diferent task, nonlocal computation
75
, has for the frst
time singled out the quantumsuperquantum transition.
Suppose Alice associates a variable x
i
with each of her days,
i = 1 ... 365 with x
i
=
1 if she is free.
Similarly, Bob defnes y
i
. Now, Alice and Bob could meet on the
ith day if and only if the product x
i
y
i
=
b
i
is even (odd) if the product x
i
y
i
is even (odd). Hence,
the sum of the products,
i
x
i
y
i
, is even (odd) if and only if the
sum of all outputs
i
a
i
+b
i
is even (odd). To fnd this out, all Alice
needs to know from Bob is if the sum of his outputs,
i
b
i
is even
or odd, that is, a single bit of information.
Box 1  Eliminating communication redundancy.
REVIEW ARTICLES

INSIGHT NATURE PHYSICS DOI: 10.1038/NPHYS2916
2014 Macmillan Publishers Limited. All rights reserved
NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics 267
Consider an ordinary computation in which the input consists
of Nbits, z
1
, ..., z
N
and the output is a single bit, c=f(z
1
, ..., z
N
). To
this computation we can associate a nonlocal computation in the
following way. Te computation is carried out by two devices, one
at Alices location and one at Bobs. To each bit z
i
of the original
computation we associate two bits, x
i
given to Alice and y
i
given to
Bob, such that z
i
=x
i
y
i
. For each value of z
i
, there are two possible
combinations of x
i
and y
i
: x
i
=0, y
i
=
0 and x
i
=1, y
i
=
1 for z
i
=
0 and
x
i
=
1, y
i
=
0 and x
i
=
0, y
i
=
1 for z
i
=
z
1
z
2
, the best linear
approximation is f
L
AND
= 0. Indeed, by always yielding 0, f
L
AND
= f
AND
in 3 out of 4 cases, the exception being z
1
=
z
2
=
1.
To better understand nonlocal correlations, a geometric repre
sentation is very useful
28,73
. For any given pair of boxes, the entire
physics is encapsulated in the joint probabilities P(a,bx,y). We
can think of these joint probabilities as coordinates of a point
in an n dimensional space (16 dimensional space in the simple
example considered here, corresponding to all combinations of
a,b,x,y = 0,1). Te set of all possible correlations flls a polytope,
the intersection of the hypercube defned by the linear inequalities
0 P(a,bx,y) 1and the hyperplanes corresponding to the prob
ability normalization constraints:
a,b
P(a,bx,y)=1 (2)
Furthermore, we are only interested in the nonsignalling boxes,
which do not allow Alice to signal instantaneously to Bob or vice
versa, that is, the boxes that do not violate special relativity. For
this to be the case, the probabilities of Alices box outputs must be
independent of Bobs input and vice versa:
b
P(a,bx,y)=
b
P(a,bx,y) (3)
for any y and y,
and:
a
P(a,bx,y)=
a
P(a,bx,y) (4)
for any x and x. Te nonsignalling constraints defne hyper
planes; the intersection of these hyperplanes with the polytope of
all correlations defnes the polytope of nonsignalling correlations
illustratedbelow.
Each point of the fgure represents an entire physical setup.
Te big polytope, including the purple, red and green regions,
constitutes the set of all nonsignalling boxes. Te internal green
polytope represents the set of local correlations; boxes acting
according to classical mechanics can produce all the local corre
lations, and only these correlations. Te vertices of the local poly
tope are deterministic correlations in which Alices box outcome
depends deterministically on her income (such as a = x) and
similar for Bob. (Obviously these deterministic boxes are local
what Alices box does is independent of Bobs box input and vice
versa.) All other points of the classical polytope are obtained as
mixtures of deterministic probabilities; more precisely, one can
prepare the boxes to act, with preprescribed probability, accord
ing to a diferent deterministic strategy. Te faces of the classical
polytope are defned by the Bell inequalities; every correlation
that is outside the local polytope is nonlocal. Te round body
consisting of the red and green parts represents all the quantum
correlations. Tis body is rounded as quantum correlations obey
Schwartz inequalities, due to the vector nature of the Hilbert
space. All points in the red region represent nonlocal boxes, as
they are outside the local polytope. Te boundary of quantum
mechanics is a generalized Cirelson inequality. Incidentally, one
of the great unsolved problems of fundamental quantum mechan
ics is to determine the boundary of quantum correlations
2729,31,32
.
In fact, it is even difcult to determine if a given correlation (that
is, a point in the big polytope) is quantum or not. As the complete
nonsignalling set is a polytope, whereas the quantum one is a
round body, it is clear that points outside quantum mechanics
that are nevertheless nonsignalling exist the purple region.
Tese are the nonsignalling superquantum correlations. Te
vertices of this polytope other than the local deterministic ones
are maximal nonlocal correlations; in the simplest case of boxes
with two inputs and two outputs, these are the perfect PR boxes.
Te challenge is to fnd fundamental properties by which the pur
ple points diferentiate from all others. In the process, we learn
more about what all the others that is, the quantum mechani
cal ones reallyare.
Box 2  The polytope of nonsignalling correlations.
b a
y x
P(a,bx,y)
INSIGHT

REVIEW ARTICLES NATURE PHYSICS DOI: 10.1038/NPHYS2916
2014 Macmillan Publishers Limited. All rights reserved
268 NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics
Nonlinearity is the core of computation, so, in some sense, a lin
ear approximation means no computation at all. Now, it turns out
that if Alice and Bob have at their disposal only devices functioning
according to the laws of classical physics, the best they can do is the
best linear approximation of the desired computation. Even more
surprisingly, although quantum mechanical nonlocal correlations
are, in general, stronger than the classical ones afer all, this is the
whole point of nonlocality these correlations do not help nonlo
cal computation: quantum devices cannot do better than the best
linear approximation either. On the other hand, the very moment
we allow for superquantum correlations, we can do nonlocal com
putation better than the best linear approximation. Hence, as far
as nonlocal computation is concerned, there is a sharp transition
between quantum and superquantumcorrelations.
Information causality
Suppose Alice sends to Bob a message consisting of a single binary
digit (0 or 1). By this procedure, Alice cannot send Bob more than
one bit of classical information, even if they also share some nonlo
cal particles and perform measurements on them according to the
information they wish to transmit or receive. Indeed, if by such a pro
cedure Alice could communicate to Bob more than one bit of infor
mation, they could also communicate superluminally. Tis is easy to
prove Bob wouldnt actually need to wait for Alices message; he
could simply guess it, perform his measurements according to the
guess, simultaneously with those of Alice, and learn, with a success
probability of 1/2, more than one bit of information. Tis can then
easily be converted into learning some information withcertainty.
An interesting possibility emerges, however. Suppose Alice has
two bits that she wants to communicate to Bob. Even though by
sending a onebit message she cannot communicate both bits to Bob,
perhaps Bob could choose which bit to learn, even though he can
make the decision at the last moment, long afer Alice has already
sent her message. Surprisingly, if Alice and Bob share a PR box,
this is possible. Indeed, let x
0
and x
1
be Alices two bits. She inputs
x=x
0
x
1
into her box and sends Bob the message m=x
0
a. If
Bob wants to learn x
0
he inputs in his box y=0, whereas if he wants
to learn x
1
he inputs y = 1. Bob then calculates m b. He obtains
mb=x
0
ab=x
0
xy=x
0
(x
0
x
1
)y. It is easy to see that
if y=0 then mb=x
0
and if y=1 then mb=x
1
.
On the other hand, one may feel uneasy with this result. Indeed,
although Bob cannot fnd both x
0
and x
1
,
one may consider that even
the ability of Bob to choose which bit to learn should be unphysi
cal. Indeed, the message sent by Alice consists of just one binary
digit; how can it allow Bob to retrieve information about two bits,
even if he cannot read both of them? Imposing the restriction that
this is impossible yields a new principle, which was proposed by
Pawlowskietal.
76
and called informationcausality.
As shown above, PR boxes violate information causality.
However, it turns out that both classical physics and quantum
mechanics obey information causality. And here comes the really
exciting thing: for a restricted class of nonlocal correlations (namely
the unbiased ones, where the local probabilities of all outcomes
are equal), information causality breaks exactly at the boundary
between quantum and superquantum nonlocal correlations. Tat
is, suppose we make the PR boxes weaker by adding white noise
until they become only as strong as quantum mechanical corre
lations. Exactly here information causality ceases to be violated.
Information causality is, therefore, yet another example that singles
out part of the quantumsuperquantumboundary.
Quantum mechanics is special (or maybe not)
So what is the status of this research now? In this Review, I have
discussed only a few examples; there is, however, intense, ongoing
efort along similarlines
8296
.
Although it is early days, one can already see that quantum
mechanics is special. Starting from various completely unrelated
tasks that have nothing to do with the dynamics of microscopic par
ticles, but are general purpose questions, such as nonlocal computa
tion, information causality, macroscopic locality, the possibility of
nonlocality swapping and so on, quantum mechanics emerges. It is
precisely at the boundary between quantum mechanical and super
quantum correlations that qualitative changes in the performance
of the above tasks occur. True, these are only glimpses there is
no known task yet that completely diferentiates quantum corre
lations from superquantum ones; only part of the boundary has
emerged so far. Indeed, it is now known that any task that would be
able to completely single out quantum mechanics has to be multi
partite, as opposed to the bipartite tasks discussed here
19
. However,
it is remarkable that parts of the quantum boundary appeared at
all there was no a priori reason whatsoever for this to happen.
Yet, quantum mechanics starts to appear from the fog. Tat quan
tum mechanics has special signifcance in at least some of such
tasks means that quantum mechanics is special, and one should
not expect that the ultimate theory of nature should be some slight
deviation from quantum mechanics there are basic statements
about nature that have to be changed. It also means that quantum
mechanics is probably here to stay at least much longer than one
would haveimagined.
At the same time, one can legitimately question the relevance of
such computer scienceinspired tasks in the grand scheme of things.
Why should we care about such things as communication complex
ity, nonlocal computation or information causality? Why should we
let our quest for a new theory of nature or the justifcation for the
present one be guided by suchideas?
Te very frst indication that this line of thought is good is the
simple fact that it seems to work. Quantum mechanics appears
unexpectedly in various contexts. Te fact that it does so is fascinat
ing, and certainlynontrivial.
Second, whereas the tasks discussed here may appear quite ran
dom and completely insignifcant from the point of view of hard
core physics certainly they tell us nothing about the spectra of
atoms or about phase transitions from the point of view of infor
mation theory they are actually fundamental. (A pair of perfect PR
boxes is a device that transforms the basic nonlinear function, the
product, into a linear one, xy=ab. At the same time, it can be
viewed as the maximal zerocapacity communicationchannel.)
Yet again, it might not be quantum mechanics that we see emerg
ing, but something altogether diferent. A few years ago Navascus
and collaborators
30
discovered a hierarchy of sets of selfconsistent
nonlocal correlations, each set is larger than quantum mechanics,
but their boundaries coincide with quantum mechanics in some
places. Maybe it is one of these sets that we are starting to see. Tese
sets were discovered based on some rather obscure mathematical
considerations, going opposite to the direction of considering natu
ral tasks, which was the whole point of the research discussedabove.
But recently, quantum gravity led to a tantalizing result: moti
vated by considerations of quantum gravity, a class of generalized
theories was proposed by GellMann and Hartle
97,98
, which was
further developed by Sorkin
99
. And in a very recent (yet unpub
lished) paper
100
, it was shown that these theories lead to stronger
thanquantum correlations, namely to the NavascusPironioAcn
set known as Q(1+AB), which is known to coincide with quantum
mechanics in most of the places where the information tasks indi
cated quantum mechanics. Hence, maybe what those tasks indicate
is Q(1+AB), not quantum mechanics. Te jury is stillout.
To conclude, all the above is great fun. Each answer raises new
questions, completely diferent in nature from the ones one started
with; this, more than anything else, indicates that fnally we might
be on the righttrack.
REVIEW ARTICLES

INSIGHT NATURE PHYSICS DOI: 10.1038/NPHYS2916
2014 Macmillan Publishers Limited. All rights reserved
NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics 269
Received 26 November 2013; accepted 12 February 2014;
published online 1 April 2014
References
1. Aharonov, Y. & Bohm, D. Signifcance of electromagnetic potentials in the
quantum theory. Phys. Rev. 115, 485491 (1959).
2. Bell, J.S. On the EinsteinPodolskyRosen paradox. Physics
1, 195200 (1964).
3. Aharonov, Y. & Rohrlich, D. Quantum Paradoxes: Quantum Teory for the
Perplexed (WileyVCH, 2005).
4. Shimony, A. in Proc. Int. Symp. Foundations of Quantum Mechanics (eds
Kamefuchi, S. et al.) 225230 (Physical Society of Japan, 1983).
5. Popescu, S. & Rohrlich, D. Causality and nonlocality as axioms for quantum
mechanics. Found. Phys. 24, 379385 (1994).
6. Horodecki, R., Horodecki, P., Horodecki, M. & Horodecki, K. Quantum
entanglement. Rev. Mod. Phys. 81, 865942 (2009).
Extensive review of the characterization of quantum nonlocality.
7. Gisin, N. Bell inequality holds for all nonproduct states. Phys. Lett. A
154,201202 (1991).
8. Popescu, S. & Rohrlich, D. Generic quantum nonlocality. Phys. Lett. A
166,293297 (1992).
9. Barrett, J., Collins, D., Hardy, L., Kent, A. & Popescu, S. Quantum nonlocality,
Bell inequalities and the memory loophole. Phys. Rev. A 66, 42111 (2002).
10. Brunner, N., Gisin, N. & Scarani, V. Entanglement and nonlocality are
diferent resources. New J.Phys. 7, 88 (2005).
11. Cerf, N.J., Gisin, N., Massar, S. & Popescu, S. Simulating maximal quantum
entanglement without communication. Phys. Rev. Lett. 94, 220403 (2005).
12. Almeida, M., Pironio, S., Barrett, J., Tth, G. & Acn, A. Noise robustness
of the nonlocality of entangled quantum states. Phys. Rev. Lett.
99, 040403 (2007).
13. Brunner, N., Gisin, N., Popescu, S. & Scarani, V. Simulation of partial
entanglement with nosignaling resources. Phys. Rev. A 78, 52111 (2008).
14. Allcock, J., Brunner, N., Pawlowski, M. & Scarani, V. Recovering part of the
boundary between quantum and nonquantum correlations from information
causality. Phys. Rev. A 80, 040103(R) (2009).
15. Acn, A. etal. Unifed framework for correlations in terms of local quantum
observables. Phys. Rev. Lett. 104, 140404 (2010).
16. Barnum, H., Beigi, S., Boixo, S., Elliott, M.B. & Wehner, S. Local quantum
measurement and no signaling imply quantum correlations. Phys. Rev. Lett.
104, 140401 (2010).
17. Oppenheim, J. & Wehner, S. Te uncertainty principle determines the
nonlocality of quantum mechanics. Science 330, 10721074 (2010).
18. Barrett, J. & Gisin, N. How much measurement independence is needed to
demonstrate nonlocality? Phys. Rev. Lett. 106, 100406 (2011).
19. Gallego, R., Wrfinger, L.E., Acn, A. & Navascus, M. Quantum
correlations require multipartite information principles. Phys. Rev. Lett.
107, 210403 (2011).
20. Bancal, JD. etal. Quantum nonlocality based on fnitespeed causal
infuences leads to superluminal signalling. Nature Phys. 8, 867870 (2012).
21. Liang, YC., Masanes, L. & Rosset, D. All entangled states display some
hidden nonlocality. Phys. Rev. A 86, 052115 (2012).
22. Palazuelos, C. Superactivation of quantum nonlocality. Phys. Rev. Lett.
109,190401 (2012).
23. Vrtesi, T. & Brunner, N. Quantum nonlocality does not imply entanglement
distillability. Phys. Rev. Lett. 108, 030403 (2012).
24. Cavalcanti, D., Acn, A., Brunner, N. & Vrtesi, T. All quantum states useful
for teleportation are nonlocal resources. Phys. Rev. A 87, 042104 (2013).
25. Hirsch, F., Quintino, M.T., Bowles, J. & Brunner, N. Genuine hidden
quantum nonlocality. Phys. Rev. Lett. 111, 160402 (2013).
26. Khalfn, L.A. & Tsirelson, B.S. Quantum/classical correspondence in the
light of Bells inequalities. Found. Phys. 22, 879948 (1992).
27. Tsirelson, B.S. Some results and problems on quantum Belltype inequalities.
Hadronic J.Suppl. 8, 329345 (1993).
28. Tsirelson, B.S. Quantum analogues of the Bell inequalities. Te case of two
spatially separated domains. J.Sov. Math. 36, 557570 (1987).
29. Tsirelson, B.S. Quantum generalizations of Bells inequality. Lett. Math. Phys.
4,93100 (1980).
30. Navascus, M., Pironio, S. & Acn, A. Bounding the set of quantum
correlations. Phys. Rev. Lett. 98, 10401 (2007).
31. Navascus, M., Pironio, S. & Acn, A. A convergent hierarchy of semidefnite
programs characterizing the set of quantum correlations. New J.Phys.
10,73013 (2008).
32. Barnum, H., Barrett, J., Leifer, M.S. & Wilce, A. Generalized nobroadcasting
theorem. Phys. Rev. Lett. 99, 240501 (2007).
33. Barrett, J. Information processing in generalized probabilistic theories.
Phys.Rev. A 75, 32304 (2007).
34. Barrett, J. & Short, A.J. Strong nonlocality: a tradeof between states and
measurements. New J.Phys. 12, 33034 (2010).
35. Barnum, H. etal. Entropy and information causality in general probabilistic
theories. New J.Phys. 12, 033024 (2010).
36. Short, A.J. & Wehner, S. Entropy in general physical theories. New J.Phys.
12,33023 (2010).
37. Gross, D., Mueller, M., Colbeck, R. & Dahlsten, O.C.O. All reversible
dynamics in maximally nonlocal theories are trivial. Phys. Rev. Lett.
104,80402(2010).
38. Janotta, P., Gogolin, C., Barrett, J. & Brunner, N. Limits on nonlocal
correlations from the structure of the local state space. New J.Phys.
13,063024(2011).
39. Barrett, J., Hardy, L. & Kent, A. No signaling and quantum key distribution.
Phys. Rev. Lett. 95, 10503 (2005).
40. Acn, A., Gisin, N. & Masanes, L. From Bells theorem to secure quantum key
distribution. Phys. Rev. Lett. 97, 120405 (2006).
41. Scarani, V. etal. Secrecy extraction from nosignaling correlations.
Phys. Rev. A 74, 42339 (2006).
42. Acn, A. etal. Deviceindependent security of quantum cryptography against
collective attacks. Phys. Rev. Lett. 98, 230501 (2007).
43. Hnggi, E. & Renner, R. Deviceindependent quantum key distribution with
commuting measurements. Preprint at http://arxiv.org/abs/1009.1833 (2010).
44. Franz, T., Furrer, F. & Werner, R.F. Extremal quantum correlations and
cryptographic security. Phys. Rev. Lett. 106, 250502 (2011).
45. Masanes, L., Pironio, S. & Acn, A. Secure device independent quantum
key distribution with causally independent measurement devices.
Nature Commun. 2, 238 (2011).
46. Barrett, J., Colbeck, R. & Kent, A. Unconditionally secure device
independent quantum key distribution with only two devices. Phys. Rev. A
86,062326(2012).
47. Vazirani, U. & Vidick, T. Certifable quantum dice. Phil. Trans. R.Soc. A.
370,34323448 (2012).
48. Barrett, J., Colbeck, R. & Kent, A. Memory attacks on deviceindependent
quantum cryptography. Phys. Rev. Lett. 110, 010503 (2013).
49. Huber, M. & Pawlowski, M. Weak randomness in deviceindependent
quantum key distribution and the advantage of using highdimensional
entanglement. Phys. Rev. A 88, 032309 (2013).
50. Colbeck, R. Quantum And Relativistic Protocols For Secure MultiParty
Computation PhD thesis, Univ. Cambridge (2007).
51. Pironio, S. etal. Random numbers certifed by Bells theorem. Nature
464,10211024 (2010).
52. Colbeck, R. & Kent, A. Private randomness expansion with untrusted devices.
J.Phys. A 44, 095305 (2011).
53. Acn, A., Massar, S. & Pironio, S. Randomness versus nonlocality and
entanglement. Phys. Rev. Lett. 108, 100402 (2012).
54. Colbeck, R. & Renner, R. Free randomness can be amplifed. Nature Phys.
8,450454 (2012).
55. Vazirani, U. & Vidick, T. in Proc. 44th Symp. Teory Comput. STOC 2012 (eds
Pitassi, T. et al.) 6176 (ACM Press, 2012).
56. Gallego, R. etal. Full randomness from arbitrarily deterministic events.
NatureCommun. 4, 2654 (2013).
57. Pironio, S. & Massar, S. Security of practical private randomness generation.
Phys. Rev. A 87, 012336 (2013).
58. Rabelo, R., Ho, M., Cavalcanti, D., Brunner, N. & Scarani, V. Device
independent certifcation of entangled measurements. Phys. Rev. Lett.
107,050502 (2011).
59. Bancal, JD., Gisin, N., Liang, YC. & Pironio, S. Deviceindependent witnesses
of genuine multipartite entanglement. Phys. Rev. Lett. 106, 250404 (2011).
60. Moroder, T., Bancal, JD., Liang, YC., Hofmann, M. & Ghne, O.
Deviceindependent entanglement quantifcation and related applications.
Phys.Rev.Lett. 111, 030501 (2013).
61. Brunner, N., Cavalcanti, D., Pironio, S., Scarani, V. & Wehner, S.
Bellnonlocality. Preprint at http://arxiv.org/abs/1303.2849 (2013).
62. Hardy, L. Quantum theory from fve reasonable axioms. Preprint at
http://arxiv.org/abs/quantph/0101012 (2001).
63. Chiribella, G., DAriano, G.M. & Perinotti, P. Probabilistic theories with
purifcation. Phys. Rev. A 81, 062348 (2010).
64. Paterek, T., Daki, B. & Brukner, C. Teories of systems with limited
information content. New J.Phys. 12, 053037 (2010).
65. Chiribella, G., DAriano, G.M. & Perinotti, P. Informational derivation of
quantum theory. Phys. Rev. A 84, 012311 (2011).
66. Masanes, L. & Mller, M.P. A derivation of quantum theory from physical
requirements. New J.Phys. 13, 063001 (2011).
67. de la Torre, G., Masanes, L., Short, A.J. & Mller, M.P. Deriving
quantum theory from its local structure and reversibility. Phys. Rev. Lett.
109,090403(2012).
INSIGHT

REVIEW ARTICLES NATURE PHYSICS DOI: 10.1038/NPHYS2916
2014 Macmillan Publishers Limited. All rights reserved
270 NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics
68. Masanes, L., Mller, M.P., Augusiak, R. & PrezGarca, D. Existence of an
information unit as a postulate of quantum theory. Proc. Natl Acad. Sci. USA.
110, 1637316377 (2013).
69. Mller, M.P. & Masanes, L. Treedimensionality of space and the quantum
bit: an informationtheoretic approach. New J.Phys. 15, 053040 (2013).
70. Popescu, S. Dynamical quantum nonlocality. Nature Phys.
6, 151153 (2010).
71. Clauser, J.F., Horne, M.A., Shimony, A. & Holt, R.A. Proposed experiment to
test local hiddenvariable theories. Phys. Rev. Lett. 23, 880 (1969).
72. Barrett, J., Linden, N., Massar, S., Pironio, S., Popescu, S. & Roberts, D.
Nonlocal correlations as an information theoretic resource. Phys. Rev. A
71,22101(2005).
73. van Dam, W. Implausible consequences of superstrong nonlocality. Preprint at
http://arxiv.org/abs/quantph/0501159 (2005).
74. Brassard, G. etal. Limit on nonlocality in any world in which communication
complexity is not trivial. Phys. Rev. Lett. 96, 250401 (2006).
75. Linden, N., Popescu, S., Short, A.J. & Winter, A. Quantum nonlocality and
beyond: limits from nonlocal computation. Phys. Rev. Lett. 99, 180502 (2007).
76. Pawlowski, M. etal. Information causality as a physical principle. Nature
461,11011104 (2009).
77. Navascus, M. & Wunderlich, H. A glance beyond the quantum model.
Proc.R.Soc. A 466, 881 890 (2010).
78. Fritz, T. etal. Local orthogonality as a multipartite principle for quantum
correlations. Nature Commun. 4, 2263 (2013).
79. Skrzypczyk, P., Brunner, N. & Popescu, S. Emergence of quantum correlations
from nonlocality swapping. Phys. Rev. Lett. 102, 110402 (2009).
80. Cleve, R., van Dam, W., Nielsen, M. & Tapp, A. Quantum Computing
and Quantum Communication, Lecture Notes in Computer Science
Vol.1509(Springer, 1999).
81. Cleve, R. & Buhrman, H. Substituting quantum entanglement for
communication. Phys. Rev. A 56, 12011204 (1997).
82. Barrett, J. & Pironio, S. PopescuRohrlich correlations as a unit of nonlocality.
Phys. Rev. Lett. 95, 140401 (2005).
83. Methot, A.A. & Scarani, V. An anomaly of nonlocality.
QuantumInform.Compu. 7, 157170 (2007).
84. Forster, M., Winkler, S. & Wolf, S. Distilling nonlocality. Phys. Rev. Lett.
102,120401 (2009).
85. Brunner, N. & Skrzypczyk, P. Nonlocality distillation and postquantum
theories with trivial communication complexity. Phys. Rev. Lett.
102, 160403 (2009).
86. Pawlowski, M. & Brukner, C. Monogamy of Bells inequality violations in
nonsignaling theories. Phys. Rev. Lett. 102, 30403 (2009).
87. Cavalcanti, D., Salles, A. & Scarani, V. Macroscopically local correlations can
violate information causality. Nature Commun. 1, 136 (2010).
88. Fitzi, M., Hnggi, E., Scarani, V. & Wolf, S. Te nonlocality of n noisy
PopescuRohrlich boxes. J.Phys. A 43, 465305 (2010).
89. Brunner, N., Cavalcanti, D., Salles, A. & Skrzypczyk, P. Bound nonlocality and
activation. Phys. Rev. Lett. 106, 20402 (2011).
90. Gallego, R., Wrfinger, L.E., Acn, A. & Navascus, M. Operational
framework for nonlocality. Phys. Rev. Lett. 109, 70401 (2012).
91. Borsten, L., Bradler, K. & Duf, M.J. Tsirelsons bound and supersymmetric
entangled states. Preprint at http://arxiv.org/abs/1206.6934 (2012).
92. Short, A.J., Popescu, S. & Gisin, N. Entanglement swapping for generalized
nonlocal correlations. Phys. Rev. A 73, 12101 (2006).
93. Branciard, C., Gisin, N. & Pironio, S. Characterizing the nonlocal correlations
created via entanglement swapping. Phys. Rev. Lett. 104, 170401 (2010).
94. Bancal, JD., Brunner, N., Gisin, N. & Liang, YC. Detecting genuine
multipartite quantum nonlocality: a simple approach and generalization to
arbitrary dimensions. Phys. Rev. Lett. 106, 020405 (2011).
95. Branciard, C., Rosset, D., Gisin, N. & Pironio, S. Bilocal versus nonbilocal
correlations in entanglement swapping experiments. Phys. Rev. A
85,032119(2012).
96. Bancal, JD., Barrett, J., Gisin, N. & Pironio, S. Defnitions of multipartite
nonlocality. Phys. Rev. A 88, 014102 (2013).
97. GellMann, M. & Hartle, G. B. in Complexity, Entropy and the Physics of
Information: SFI Studies in the Sciences of Complexity Vol VIII (ed. Zurek, W.)
150173 (Addison Wesley, 1990).
98. Hartle, J. B. in Proc. 1992 Les Houches Summer School on Gravitation and
Quantization (eds ZinnJustin, J. & Julia, B.) 285480 (NorthHolland, 1995).
99. Sorkin, R.D. Quantum mechanics as quantum measure theory.
Mod.Phys.Lett.A 9, 31193128 (1994).
100. Dowker, F. Henson, J. & Wallden, P. A histories perspective on characterising
quantum nonlocality. Preprint at http://arxiv.org/abs/1311.6287 (2013).
Additional information
Reprints and permissions information is available online at www.nature.com/reprints.
Correspondence and requests for materials should be addressed to S.P.
Competing nancial interests
Te authors declare no competing fnancial interests.
REVIEW ARTICLES

INSIGHT NATURE PHYSICS DOI: 10.1038/NPHYS2916
2014 Macmillan Publishers Limited. All rights reserved
INSIGHT

REVIEW ARTICLES
PUBLISHED ONLINE: 1 APRIL 2014  DOI: 10.1038/NPHYS2863
Testing the limits of quantum mechanical
superpositions
Markus Arndt
1
* and Klaus Hornberger
2
Quantum physics has intrigued scientists and philosophers alike, because it challenges our notions of reality and locality
concepts that we have grown to rely on in our macroscopic world. It is an intriguing open question whether the linearity of
quantum mechanics extends into the macroscopic domain. Scientic progress over the past decades inspires hope that this
debate may be settled by tabletop experiments.
T
he past three decades have witnessed what has been termed
1
the second quantum revolution: a renaissance of research
on the quantum foundations, hand in hand with growing
experimental capabilities
2
, revived the idea of exploiting quantum
superpositions for technological applications, from information
science
35
to precisionmetrology
68
. Quantummechanics has passed
all precision tests with flying colours, but it still seems to be
in conflict with our common sense. As quantum theory knows
no boundaries, everything should fall under the sway of the
superposition principle, including macroscopic objects. This is at
the bottom of Schrdingers thought experiment of transforming
a cat into a state that strikes us as classically impossible. And yet,
Schrdinger kittens of entangled photons
9
and ions
10
have been
realized in the lab.
So why are the objects around us never found in superpositions
of states that would be impossible in a classical description? One
may emphasize the smallness of Plancks constant, or point to
decoherence theory, which describes how a system will eectively
lose its quantum features when coupled to a quantum environment
of sucient size
11,12
. The formalism of decoherence, however, is
based on the framework of unitary quantum mechanics, implying
that some interpretational exercise is required not to become
entangled in a multitude of parallel worlds
13
. More radically, one
may ask whether quantummechanics breaks down beyond a certain
mass or complexity scale. As will be discussed below, such ideas can
be motivated by the apparent incompatibility of quantum theory
and general relativity. It is safe to state, in any case, that quantum
superpositions of truly massive, complex objects are terra incognita.
This makes them an attractive challenge for a growing number of
sophisticated experiments.
We start by reviewing several prototypical tests of the
superposition principle, focusing on the quantum states of motion
exhibited by material objects. Particle position and momentum
variables have a welldefined classical analogue, and they are
therefore particularly suited to probe the macroscopic domain.
We note that aspects of macroscopicity can also be addressed in
experiments with photons
1416
, with the phonons of ion chains
17
,
and by squeezing pseudospins
8,18
.
State of the art
Superconducting quantum interference devices (SQUIDs) have
recently attracted a lot of interest, because they are promising
elements of quantum information processing
19
. A SQUID is
a superconducting loop segmented by Josephson junctions.
Its electronic and transport properties are determined by a
macroscopic wavefunction ordering the Cooper pairs. To exploit
this macroscopicity it is appealing to consider a flux qubit
20
(Fig. 1a):
the singlevaluedness of the wavefunction means that the magnetic
flux encircled by a closedloop supercurrent must be quantized. In
particular, one can define a symmetric and an antisymmetric linear
combination of two supercurrents, which circulate simultaneously
in opposing directions. Billions of electrons may contribute
coherently to the wavefunction over mesoscopic dimensions. The
dierence between the clockwise and anticlockwise currents
21
can reach about 2 A, amounting to a local magnetic moment
of about 10
10
Bohr magnetons. This is an impressive number,
which has led to the suggestion that SQUIDs may exhibit the most
macroscopic quantum superposition to date. However, only a few
thousand of the Cooper pairs carrying the dierent currents are
distinguishable
22
, which points to the need for an objective measure
of macroscopicity (Box 1).
Historically, perfectcrystal neutron quantum optics
23
made
many interference experiments with atoms and photons possible. As
the de Broglie wavelength of thermal neutrons is comparable to the
lattice constant of silicon, quantum diraction o the nuclei may
split the neutron wavefunction at large angles. As of today, neutron
interferometry still realizes the widest delocalization of any massive
object
24
. With an arm separation up to 7 cm, enclosing an area of
80cm
2
, it allows one to stick a hand between the two branches of a
quantum state that describes a single microscopic particle (Fig. 1b).
Even though neutrons are very light neutral particles, they are
prime candidates for emergent tests of postNewtonian gravity at
short distances
25,26
. With an electrical polarizability twenty orders of
magnitude smaller than for atoms, neutrons are much less sensitive
to electrostatic perturbations, such as charges, patch eects or van
der Waals forces.
Much better control and signal to noise can be achieved by using
atoms. Atominterferometry (Fig. 1c) started about 30 years ago
2729
.
The development of Raman
30
beamsplitters then transformed the
tools of basic science into highprecision quantumsensors that split,
invert and recombine the atomic wavefunction in three short laser
pulses (Fig. 1c). In particular, inertial forces such as gravity and
Coriolis forces
31,32
have been measured with stunning precision in
experiments that also promise new tests of general relativity
33
.
1
Faculty of Physics, University of Vienna, QuNaBioS, VCQ, Boltzmanngasse 5, Vienna 1090, Austria,
2
Faculty of Physics, University of DuisburgEssen,
Lotharstrae 1, Duisburg 47048, Germany, *email: markus.arndt@univie.ac.at
email: klaus.hornberger@unidue.de
NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics 271
2014 Macmillan Publishers Limited. All rights reserved.
REVIEW ARTICLES

INSIGHT NATURE PHYSICS DOI: 10.1038/NPHYS2863
1
3
8
11
15
T
O
F
(
m
s
)
t = 0 t = T t = 2T
650 m
b a
c
e
d
0 100 200 300 400 500 600 700 800 900 1,000 1,100
400
800
1,200
1,600
2,000
G
3
grating position (nm)
G
1
G
2
G
3
S
i
g
n
a
l
(
c
o
u
n
t
s
i
n
3
s
)
Figure 1  Superposition experiments. a, A ux qubit realizes a quantum superposition of left and rightcirculating supercurrents
21
with billions of
electrons contributing to the quantum state. b, Neutron interferometry with perfect crystal beamsplitters holds the current record in matterwave
delocalization
24
, separating the quantum wave packet by up to 7 cm. c, Modern atom interferometry achieves coherence times beyond two seconds
with wavepacket separations up to 1.5 cm (refs 3638). d, Interference of two clouds of BoseEinstein condensed diatomic lithium molecules
101
.
e, KapitzaDiracTalbotLau interferometer for macromolecules
44,54,57
. Figures reproduced with permission from: a, ref. 20, 2008 NPG;
b, ref. 24, 2002 Elsevier; d, ref. 101, C. Kohstall and R. Grimm, University of Innsbruck, Austria; e, ref. 57, 2010 RSC.
The mass in these experiments is always limited to that of a single
atom, in practice to the caesium mass of 133 AMU. A degree of
macroscopicity can still be reached in the spatial extension of the
wavefunction and in coherence time. The achievable delocalization
depends on the momentum transfer in the beamsplitting element,
whereas the coherence time is essentially determined by the
duration of free fall in the apparatus. Both impressively wide
angle beam splitters
34,35
and very long coherence times
36
have
been demonstrated separately, and been recently combined in an
experiment with rubidiumatoms, whose wave packets get separated
for 2.3 s with a maximal distance of 1.4 cm (ref. 37). Future
quantumsensors are expected to increase the sensitivity of quantum
metrology by several orders of magnitude. The coherence time
grows only with the square root of the device length, so that it will
be practically limited to several seconds in Earthbound devices,
even in highdrop towers. Progress in matterwave beam splitting
will depend on improved wavefront control of the beam splitting
lasers and other technological breakthroughs. If it were possible to
build interferometers of 100 m length with beamsplitters capable
of transferring a hundred grating momenta
38
, atomic matter would
be delocalized over distances of metres. Even though designed
for testing the eects of general relativity
33,39
, such experiments
would also test the linearity of quantum mechanics
40
as well as the
homogeneity of spacetime
41
.
It is frequently suggested that ultracold atomic ensembles may
serve to test the linearity of quantum physics even better, as
all atoms can be described by a joint manybody wavefunction
once they are cooled below the phase transition to BoseEinstein
condensation (Fig. 1d). Billions of noninteracting atoms may
be united in a quantum degenerate state, which is, however,
a product of singleparticle states (0 + 1)
N
, so that
interference of Bosecondensed atoms depends only on the de
Broglie wavelength of single atoms. A genuinely entangled many
particle state 0
N
+1
N
akin to a Schrdinger cat state
272 NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics
2014 Macmillan Publishers Limited. All rights reserved.
NATURE PHYSICS DOI: 10.1038/NPHYS2863 INSIGHT

REVIEW ARTICLES
Box 1  Measuring macroscopicity.
How can one compare dierent experimental approaches towards
establishing large mechanical superposition states? Various
measures are on oer for attributing a size to a given state
79,95100
.
They presuppose a distinguished partitioning of the manyparticle
Hilbert space into single degrees of freedom, and most of them
rely on distinguished measurement or decoherence bases. Such
approaches work well if the examined systems and states are of
the same kind, but they do not allow us to compare disparate
mechanical superposition states in an unbiased way; for example,
superconducting ring currents with an interfering buckyball.
0 10 20 30
4.8 Neutron interference (1962)
Persistent current superpositions in SQUIDs (2000)
Fareld interference of Na atoms (1988)
Fareld interference of C
60
(1999)
MachZehnder interference of Cs (2009)
TalbotLau interference of PFNS8 (2011)
52
6.8
10.6
10.6
12.1
11.5
14.5
14.5
19.0
20.5
23.3
Membrane phonons
Hypothetical giant SQUID
TalbotLau interf. (10
5
AMU)
Oscillating micromirror (10
15
AMU)
Nanosphere interference (10
7
AMU)
OTIMA nanoparticle interference (10
8
AMU)
Figure B1  Macroscopicities of diferent superposition experiments.
Macroscopicities reached in past experiments (top) and proposed
tests (bottom) of the superposition principle as evaluated in ref. 40.
To circumvent this problem, a recent macroscopicity measure
40
quantifies the empirical relevance of the concrete experiment at
hand, rather than an abstract state in Hilbert space. Ultimately,
any such experiment tests the hypothesis that the superposition
principle is no longer valid at a certain scale. Thus, the more
macroscopic a superposition state is, the better its demonstration
rules out even minimal modifications of quantum mechanics that
lead to classical behaviour on the macroscale.
To turn this into a definite measure one needs to parametrize
the class of minimal classicalizing modifications. This can be
done without looking at specific realizations, such as the continu
ous spontaneous localization model, by focusing on their
observational consequences on the level of the density operator.
Demanding the modification to obey basic symmetry and con
sistency requirements (Galilean and scale invariance, consistent
treatment of identical and of uncorrelated particles), the scope of
falsified theories can be characterized in the end by a single bound,
a coherence time parameter
e
. Given two experiments, the one
implying a larger value of
e
is thus more macroscopic, and one
may define its degree of macroscopicity as =log
10
(
e
/1s). The
electron is taken as reference, such that the experiment confirms
quantummechanics as strongly as anelectronbehaving like a wave
for longer than 10
s (ref. 40).
Figure B1 shows the macroscopicities for a selection of past
and proposed experiments. The superconducting loop currents of
ref. 21 feature as relatively low owing to the small electron mass
and coherence time. It would be much higher in a hypothetical
large SQUIDwith a length of 20 mmand 1 ms coherence time. For
the oscillating micromembrane we assume that the device from
ref. 84 can be kept in a superposition of the zero and onephonon
states for 1,000 oscillation periods.
would be required to reduce the fringe spacing. Such macroscopic
cat states with regard to the particle motion have remained an
open challenge, even though entanglement in other degrees of
freedom has been demonstrated between dozens of atoms
7,8,42
. In
contrast to that, macromolecules and clusters open a new field
involving strongly bound particles with internal temperatures up to
1,000 K. When N atoms are covalently linked into a single molecule
they act as a single object in quantum interference experiments.
The entire Natom system is then delocalized over two or more
interferometer arms.
Macromolecule interferometry started originally with the far
field diraction of fullerenes
43
and works with highmass objects
in currently two dierent settings: the KapitzaDiracTabotLau
interferometer (KDTLI) and an alloptical interferometer in the
time domain with pulsed ionization gratings (OTIMA). Both
concepts were developed and implemented at the University of
Vienna
44,45
and are based on similar ideas. In highmass matter
wave interference we face de Broglie wavelengths between 10 fm
and 10 pm for objects between 10
10
and 10
3
AMU. This is more
than six orders of magnitude smaller than in all experiments with
ultracold atoms. Macromolecules are not susceptible to established
laser cooling techniques, although first steps towards the cavity
cooling of 10
10
AMU objects have been taken
46,47
. The particles
therefore start out in rather mixed states, requiring nearfield
interference schemes
48
.
The KDTLI interferometer is sketched in Fig. 1e. It accepts a
large variety of nanoparticles, because it uses only nonresonant
gratings to split (G
1
), diract (G
2
) and probe (G
3
) matterwaves.
The first grating (G
1
) implements a spatially periodic transmission
function. The size of the slits and the separation between G
1
and
G
2
are chosen such that the positionmomentum uncertainty in
each slit is sucient to expand each particles wavefunction to
cover more than two slits in G
2
downstream. To achieve this,
G
1
must be an absorptive mask, here realized as a silicon nitride
nanostructure. Grating G
2
, a nonresonant standing light wave,
imprints a spatially periodic phase onto the matterwave. A near
field resonance eect rephases the wavefunctions to a molecular
density pattern at the position of G
3
. Although one might capture
the emerging quantum fringe pattern on a substrate for subsequent
highresolution microscopy
49,50
, it is often convenient to scan the
absorptive mask G
3
across the nanopattern: a plot of the number
of transmitted particles as a function of the masks position reveals
the molecular interferogram (Fig. 1e).
In contrast to the KDTLI, an OTIMA interferometer relies on
three pulsed gratings that ionize and thus remove the molecules
at the antinodes of an ultraviolet standingwave laser beam
51
.
Such alloptical gratings can handle of highly polarizable or polar
particles, and their pulsed nature allows us to profit from working
in the time domain. All particles exposed to the spatially extended
nanosecond laser pulses then see the same grating for the same time,
regardless of their velocity. This eliminates numerous dispersive
dephasing phenomena, which is particularly beneficial for quantum
tests at high masses
52,53
. KDTLI and OTIMA are universal in the
sense that they can accept a wide class of dierent objects and both
avoid the detrimental eect of van der Waals forces in G
2
by using
nonresonant optical beamsplitters.
Experiments in the KDTLI currently hold the mass
record in matterwave interference, with a functionalized
NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics 273
2014 Macmillan Publishers Limited. All rights reserved.
REVIEW ARTICLES

INSIGHT NATURE PHYSICS DOI: 10.1038/NPHYS2863
tetraphenylporphyrin molecule that combines 810 atoms into
one particle with a molecular weight exceeding 10,000 AMU
(ref. 54). Even at an internal temperature of 500 K this object
can be delocalized over a hundred times its own diameter and
for more than 1 ms. Very recently, the OTIMA concept has been
demonstrated
45
with clusters of molecules. It will soon be used
to explore quantum coherence at unprecedented masses
52
. Both
interferometers also share a high potential for quantumassisted
metrology targeting internal properties, which reveal themselves
in de Broglie experiments owing to the phase shift induced by
external fields
5557
.
Physics beyond the Schrdinger equation?
The experimental tests discussedso far confirmquantummechanics
impressively, as do highprecision spectroscopic measurements
58,59
and tests of nonlocality
6062
. Many physicists take for granted that
quantumtheory is valid on macroscopic scales, the more so because
environmental decoherence explains why macroscopic objects seem
to assume the classically distinguished states we observe in our
everyday life
11,12
(Fig. 2).
Yet, there are good reasons to take seriously the possibility that
quantum theory may fail beyond some scale. A compelling one is
the diculty of reconciling quantumtheory with the nonlinear laws
of general relativity, which treats spacetime as a dynamical entity.
Most theories of quantum gravity suggest that there is a minimal
observable length scale, often associated with the Planck length.
One way to account for this phenomenologically is to postulate
modifiedcommutator relations for the canonical observables, which
might be testable by monitoring the motion of massive pendulums
at the quantum level
6367
. The granularity of spacetime might
manifest itself also in a fundamentally nonunitary time evolution
of the quantum system, which would be observable as an intrinsic
decoherence process
41,6870
.
The alternative that gravity is not to be quantized, but
fundamentally described by a classical field, suggests one should
extend the Schrdinger equation nonlinearly to account for the
gravitational selfinteraction
71,72
. This idea is formalized in the
SchrdingerNewton equation, which can be obtained as the non
relativistic limit of selfgravitating KleinGordon fields
73
. It has
been hypothesized that this equation defines the timescale and
the basis states of a fundamental collapse mechanism. Indeed,
an additional collapselike stochastic process is required for any
such nonlinear extension of the Schrdinger equation to ensure
that the time evolution maps any initial state linearly to an
ensemble described by a proper density operator. Otherwise an
entangled particle pair would admit superluminal signalling
that is, violate causality because the nonlinearity would imprint
the basis of a distant measurement onto the reduced local state
74
.
A gravitationallyinspired nonlinear modification of quantum
mechanics
75
can be made consistent with causality and observations
at the price of a fictitiously large blurring of the involved
mass density
71
.
The best studied nonlinear modification of quantum mechanics
is the continuous spontaneous localization (CSL) model
76,77
. It
augments the Schrdinger equation for elementary particles with
a Gaussian noise term that gives rise to a continuous stochastic
collapse of wavefunctions delocalized beyond about 100 nm. The
origin of the stochastic process remains unspecified; one may viewit
either as a fundamental trait of nature, or as the repercussion of an
inaccessible underlying dynamics
78
. The CSL eect would be very
weak and practically unobservable on the atomic level, but it would
get strongly amplified for bound atoms forming a solid, such as
the pointer of a measurement device. Any superposition of macro
scopically distinct positions would rapidly collapse, in agreement
with Borns rule, to a classical state characterized by a localized,
objective wavefunction. This way the model serves its purpose of
10
8
AMU
10
7
AMU
10
6
AMU
1 200 400 600 800 1,000
Temperature (K)
P
r
e
s
s
u
r
e
(
m
b
a
r
)
10
14
10
13
10
12
10
11
10
10
10
9
10
8
10
7
Figure 2  Accounting for environmental decoherence. The theory of
decoherence accounts for the impact of a quantum system on practically
unobservable environmental degrees of freedom
11,12
. It can thus explain the
efective superselection of distinguished system states and the emergence
of classical dynamics. From a practical point of view, decoherence theory
tells us how strongly a quantum system must be isolated from its
surroundings to be still expected to show quantum interference. The gure
gives the ambient temperature and pressure requirements for observing
OTIMA interference with gold clusters of 10
6
, 10
7
and 10
8
AMU. Similarly
demanding conditions for shielding environmental decoherence apply to
the other described superposition tests. Figure adapted with permission
from ref. 52, 2011 APS.
restoring objective classical reality on the scale of everyday objects,
allowing one to dispense with the measurement postulate.
It is a contentious issue whether such macrorealism
79
is required
in a plausible description of physical reality. Independent of that,
the CSL model serves as a cautionary tale. It proves that there are
competing descriptions of nature, which predict strongly dierent
eects at macroscopic scales, even though they are compatible
with all experiments and cosmological observations carried out so
far
71,80
. One may invoke metaphysical arguments in favour of one or
another theory, but empirically their status is equal, and only future
experiments will be able to tell them apart.
Venturing towards macroscopic quantum superpositions
Various dierent systems have been suggested for probing the
quantum superposition principle at mesoscopic or even macro
scopic scales. This raises the question how to objectively assess the
degree of macroscopicity reached in dierent experiments
40
(Box 1).
The gravitational collapse hypothesis
81
inspired a proposal to
create a quantum superposition in the centreofmass motion of a
micromirror
82
(Fig. 3a). Alightweight (picogram) mirror suspended
froma cantilever can close a cavity acting as one armof a Michelson
interferometer. A single photon entering the interferometer excites
a superposition of the two cavity modes. The radiation pressure of
the single photon induces a deflective oscillation of the small mirror
by approximately the width of the zeropoint motion. Which
path information is thus left behind once the photon escapes
from the cavities, unless this occurs at a multiple of the cantilever
oscillation period, when the original state of the mirror reappears.
Observing the recurrence of optical interference after one such
oscillation period would therefore prove that the mirror was in a
superposition state
82,83
.
This is a dicult experiment because a relatively massive
oscillator with an eigenfrequency in the low kilohertz regime is
required for probing gravitational collapse. This implies that the
oscillator ground state is reached only at microkelvin temperatures.
Groundstate cooling is easier withlighter andmore rigidmegahertz
or gigahertz oscillators, and by addressing normal modes with
274 NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics
2014 Macmillan Publishers Limited. All rights reserved.
NATURE PHYSICS DOI: 10.1038/NPHYS2863 INSIGHT

REVIEW ARTICLES
a
b
c
t = 0 t = T t = 2T
Figure 3  Interference schemes for large masses. a, The superposition of
a micromechanical oscillator can be triggered by scattering a single photon
in a Michelson interferometer. b, Timedomain matterwave interferometry
of nanoparticles with pulsed laser gratings is expected to be scalable to
high masses. c, Fareld interference of nanospheres at a
measurementinduced double slit may be observed by correlating the
detected positions with a phase measurement.
stronger optomechanical coupling. This feat has been achieved
recently with the flexural mode of a circular aluminium micro
membrane using optical sideband cooling
84,85
. Many groups
worldwide have embarked on studying such nanomechanical
oscillators
86
, which can serve as an interface between quantum
systems. However, it has been dicult to observe genuine quantum
eects in optomechanical systems because they still lack the strong
nonlinear coupling required to generate quantum states of motion
that dier qualitatively from classical ones. As a first step in this
direction a piezoelectric resonator was coupled coherently to a
superconducting loop
87
.
The distinctive feature of micromechanical devices compared
with other quantum systems is their very high mass. However, the
quantum delocalization of the oscillatory ground state, which is a
collective degree of freedom involving all the atoms, will reach at
most about one picometre in conceivable setupsa tiny fraction
of the size of an atom. This indicates why some matterwave
experiments will reach beyond the macroscopicity of a possible
superposition of the micromembrane (Box 1).
As any clamped nanostructure will be prone to damping, recent
proposals
8890
consider levitating dielectric nanoparticles in the
focus of an intense laser beam. Cooling the centreofmass motion
to the ground state should be feasible, owing to their lower mass and
the high trap frequencies. Moreover, the nanosphere position can be
coupled nonlinearly to a resonator light field by placing the optical
trap at the node of a FabryProt cavity. This opens the possibility to
create distinctively nonclassical states, and to probe the wave nature
of the nanospheres, for example, by implementing an eective
doubleslit
91
. In this scheme one would drop the nanosphere once
it has been cooled to the ground state of a dipole trap. After the
wave packet is suciently dispersed, a laser pulse passing through a
FabryProt cavity reveals the square of the position by a homodyne
measurement of the cavity light field. One thus learns the distance
of the sphere from the cavity centre, but not whether it is on
the left or right, thus eectively projecting its wavefunction to a
spatial superposition state. An interference pattern should then be
observable after a further free evolution of the sphere, and after
many repetitions, if one correlates the detected positions with the
results of the homodyne measurements (Fig. 3b). The nanosphere
position would be delocalized by approximately the diameter of the
sphere, which should be suciently large to test the eects of the
CSL collapse model.
A straightforward strategy for probing the wave nature of
nanometresized objects is to push established matterwave
interference schemes to the limits of large masses. The OTIMA
interferometer (Fig. 3c) should allow us to probe the quantum
nature of 10
5
AMU particles if the source ejects them with a velocity
of about 10ms
1
(ref. 53). Objects with a diameter up to 10 nm
would get delocalized over 80 nm. In the future, even nanoparticles
in the mass range of 10
8
AMU might be diracted with an OTIMA
scheme, for example gold clusters with a diameter of 22 nm.
Successful interference at these masses would falsify all current
CSL predictions
52
. However, it would require us to counteract the
gravitational acceleration, by noisefree levitation techniques or by
going to a microgravity environment, to allow the wavefunction
to expand over a coherence time of many seconds. Moreover,
environmental decoherence would need to be suppressed by
setting the ambient pressure to below 10
11
mbar and by cooling
the apparatus to cryogenic temperatures
92
; (Fig. 2). The biggest
challenge, both for OTIMA interferometry and the realization of
a projective double slit, is the preparation of sizeselected neutral
particles in ultrahigh vacuum at low internal and motional
temperatures. Some promising first steps have been achieved
by recent demonstrations of optical feedback cooling
93,94
and
cavity cooling
46,47
.
Perspectives
Will the quantum superposition principle stand the test of time?
We have emphasized that this question is neither crazy nor
heretical. Objective modifications of quantum mechanics can be
set up that agree with all observations and experiments so far,
while describing a tangible breakdown of quantum theory at the
macroscale. Whether quantum mechanics is universally valid is
thus not an issue of conviction or metaphysical reasoning, but an
empirical question, to be answered only by future experiments.
A great variety of quantum systems may be used to demonstrate
mechanical superposition states, whose mass, geometric size and
delocalization scales may vary by orders of magnitude. Any such
quantum test, if carried out successfully, will rule out a generic
class of objective modifications of quantum mechanics. Using
the scope of this falsified class as a yardstick, it is remarkable
that totally dierent experimental approaches lead to comparable
degrees of macroscopicity (Fig. B1). This suggests that there is no
single golden strategy to be pursued, and much will depend on
experimental advances and ideas. It is thus a long and exciting
journey into the realm of large quantum superpositions, and one
worth taking.
Received 8 August 2013; accepted 9 December 2013;
published online 1 April 2014
NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics 275
2014 Macmillan Publishers Limited. All rights reserved.
REVIEW ARTICLES

INSIGHT NATURE PHYSICS DOI: 10.1038/NPHYS2863
References
1. Dowling, J. P. & Milburn, G. J. Quantum technology: The second quantum
revolution. Phil. Trans. A 361, 16551674 (2003).
2. Zeilinger, A. Experiment and the foundations of quantum physics. Rev. Mod.
Phys. 71, S288S297 (1999).
3. Trabesinger, A. Quantum simulation. Nature Phys. 8, 263263 (2012).
4. Bennett, C. H. & DiVincenzo, D. P. Quantum information and computation.
Nature 404, 247255 (2000).
5. Southwell, K. Quantum coherence. Nature 453, 10031003 (2008).
6. Giovannetti, V., Lloyd, S. & Maccone, L. Advances in quantum metrology.
Nature Phys. 5, 222229 (2011).
7. Riedel, M. F. et al. Atomchipbased generation of entanglement for quantum
metrology. Nature 464, 11701173 (2010).
8. Gross, C., Zibold, T., Nicklas, E., Estve, J. & Oberthaler, M. K. Nonlinear
atom interferometer surpasses classical precision limit. Nature 464,
11651169 (2010).
9. Haroche, S. Nobel Lecture: Controlling photons in a box and exploring the
quantum to classical boundary. Rev. Mod. Phys. 85, 10831102 (2013).
10. Wineland, D. J. Nobel Lecture: Superposition, entanglement, and raising
Schrdingers cat. Rev. Mod. Phys. 85, 11031114 (2013).
11. Joos, E. et al. Decoherence and the Appearance of a Classical World in Quantum
Theory 2nd edn (Springer, 2003).
12. Zurek, W. H. Decoherence, einselection, and the quantum origins of the
classical. Rev. Mod. Phys. 75, 715775 (2003).
13. Lalo, F. Do We Really Understand Quantum Mechanics? (Cambridge Univ.
Press, 2012).
14. Fickler, R. et al. Quantum entanglement of high angular momenta. Science
338, 640643 (2012).
15. Ma, X. S. et al. Quantum teleportation over 143 kilometres using active
feedforward. Nature 489, 269273 (2012).
16. Kirchmair, G. et al. Observation of quantum state collapse and revival due to
the singlephoton Kerr eect. Nature 495, 205209 (2013).
17. Monz, T. et al. 14qubit entanglement: Creation. Phys. Rev. Lett. 106,
130506 (2011).
18. Julsgaard, B., Kozhekin, A. & Polzik, E. S. Experimental longlived
entanglement of two macroscopic objects. Nature 413, 400403 (2001).
19. Devoret, M. H. & Schoelkopf, R. J. Superconducting circuits for quantum
information: An outlook. Science 339, 11691174 (2013).
20. Clarke, J. & Wilhelm, F. K. Superconducting quantum bits. Nature 453,
10311042 (2008).
21. Friedman, J., Patel, V., Chen, W., Tolpygo, S. & Lukens, J. Quantum
superposition of distinct macroscopic states. Nature 406, 4346 (2000).
22. Korsbakken, J., Wilhelm, F. & Whaley, K. The size of macroscopic
superposition states in flux qubits. Europhys. Lett. 89, 30003 (2010).
23. Rauch, H., Treimer, W. & Bonse, U. Test of a single crystal neutron
interferometer. Phys. Rev. A 47, 369371 (1974).
24. Zawisky, M., Baron, M., Loidl, R. & Rauch, H. Testing the worlds largest
monolithic perfect crystal neutron interferometer. Nucl. Instrum. Methods
Phys. Res. A 481, 406413 (2002).
25. Nesvizhevsky, V. V. et al. Quantum states of neutrons in the earths
gravitational field. Nature 415, 298300 (2002).
26. Jenke, T., Geltenbort, P., Lemmel, H. & Abele, H. Realization of a
gravityresonancespectroscopy technique. Nature Phys. 7,
468472 (2011).
27. Gould, P. L., Ru, G. A. & Pritchard, D. E. Diraction of atoms by light: The
nearresonant KapitzaDirac eect. Phys. Rev. Lett. 56, 827830 (1986).
28. Keith, D. W., Schattenburg, M. L., Smith, H. I. & Pritchard, D. E. Diraction of
atoms by a transmission grating. Phys. Rev. Lett. 61, 15801583 (1988).
29. Bord, C. Atomic interferometry with internal state labelling. Phys. Lett. A
140, 1012 (1989).
30. Kasevich, M. & Chu, S. Atomic interferometry using stimulated Raman
transitions. Phys. Rev. Lett. 67, 181184 (1991).
31. Peters, A., YeowChung, K. & Chu, S. Measurement of gravitational
acceleration by dropping atoms. Nature 400, 849852 (1999).
32. Stockton, J. K., Takase, K. & Kasevich, M. A. Absolute geodetic rotation
measurement using atom interferometry. Phys. Rev. Lett. 107,
133001 (2011).
33. Hohensee, M., Chu, S., Peters, A. & Mller, H. Equivalence principle and
gravitational redshift. Phys. Rev. Lett. 106, 151102 (2011).
34. Mller, H., Chiow, Sw., Long, Q., Herrmann, S. & Chu, S. Atom
interferometry with up to 24photonmomentumtransfer beam splitters.
Phys. Rev. Lett. 100, 180405 (2008).
35. Chiow, S., Kovachy, T., Chien, H. & Kasevich, M. 102 hk large area atom
interferometers. Phys. Rev. Lett. 107, 130403 (2011).
36. Mntinga, H. et al. Interferometry with BoseEinstein condensates in
microgravity. Phys. Rev. Lett. 110, 093602 (2013).
37. Dickerson, S. M., Hogan, J. M., Sugarbaker, A., Johnson, D. M. S. &
Kasevich, M. A. Multiaxis inertial sensing with longtime point source atom
interferometry. Phys. Rev. Lett. 111, 083001 (2013).
38. Dimopoulos, S., Graham, P., Hogan, J. & Kasevich, M. Testing general
relativity with atom interferometry. Phys. Rev. Lett. 98, 14 (2007).
39. Bouyer, P. & Landragin, A. Interfromtrie atomique et gravitation: du sol
lespace. Journes de laction spcifique GRAM (Gravitation, Rfrences,
Astronomie, Mtrologie) (Nice, France, 2010).
40. Nimmrichter, S. & Hornberger, K. Macroscopicity of mechanical quantum
superposition states. Phys. Rev. Lett. 110, 160403 (2013).
41. Percival, I. C. & Strunz, W. T. Detection of spacetime fluctuation by a model
interferometer. Proc. R. Soc. Lond. A 453, 431446 (1997).
42. Sherson, J. et al. Quantum teleportation between light and matter. Nature 443,
557560 (2006).
43. Arndt, M. et al. Waveparticle duality of C
60
molecules. Nature 401,
680682 (1999).
44. Gerlich, S. et al. A KapitzaDiracTalbotLau interferometer for highly
polarizable molecules. Nature Phys. 3, 711715 (2007).
45. Haslinger, P. et al. A universal matterwave interferometer with optical
ionization gratings in the time domain. Nature Phys. 9, 144148 (2013).
46. Kiesel, N. et al. Cavity cooling of an optically levitated nanoparticle. Proc. Natl
Acad. Sci. USA 110, 1418014185 (2013).
47. Asenbaum, P., Kuhn, S., Nimmrichter, S., Sezer, U. & Arndt, M. Cavity cooling
of free silicon nanoparticles in highvacuum. Nature Commun. 4, 2743 (2013).
48. Clauser, J. in Experimental Metaphysics (eds Cohen, R. S., Horne, M. &
Stachel, J.) 111 (Kluwer Academic, 1997).
49. Jumann, T. et al. Wave and particle in molecular interference lithography.
Phys. Rev. Lett. 103, 263601 (2009).
50. Jumann, T. et al. Realtime singlemolecule imaging of quantum
interference. Nature Nanotech. 7, 297300 (2012).
51. Reiger, E., Hackermller, L., Berninger, M. & Arndt, M. Exploration of gold
nanoparticle beams for matter wave interferometry. Opt. Commun. 264,
326332 (2006).
52. Nimmrichter, S., Hornberger, K., Haslinger, P. & Arndt, M. Testing
spontaneous localization theories with matterwave interferometry. Phys. Rev.
A 83, 043621 (2011).
53. Nimmrichter, S., Haslinger, P., Hornberger, K. & Arndt, M. Concept of an
ionizing timedomain matterwave interferometer. New J. Phys. 13,
075002 (2011).
54. Eibenberger, S., Gerlich, S., Arndt, M., Mayor, M. & Txen, J. Matterwave
interference of particles selected from a molecular library with masses
exceeding 10 000 amu. Phys. Chem. Chem. Phys. 15, 1469614700 (2013).
55. Berninger, M., Stfanov, A., Deachapunya, S. & Arndt, M. Polarizability
measurements in a molecule nearfield interferometer. Phys. Rev. A 76,
013607 (2007).
56. Gerlich, S. et al. Matterwave metrology as a complementary tool for mass
spectrometry. Angew. ChemInt. Ed. 47, 61956198 (2008).
57. Txen, J., Gerlich, S., Eibenberger, S., Arndt, M. & Mayor, M. De Broglie
interference distinguishes between constitutional isomers. Chem. Commun.
46, 41454147 (2010).
58. Niering, M. et al. Measurement of the hydrogen 1S 2S transition frequency by
phase coherent comparison with a microwave cesium fountain clock. Phys.
Rev. Lett. 84, 54965499 (2000).
59. Odom, B., Hanneke, D., DUrso, B. & Gabrielse, G. New measurement of the
electron magnetic moment using a oneelectron quantum cyclotron. Phys.
Rev. Lett. 97, 030801 (2006).
60. Freedman, S. J. & Clauser, J. F. Experimental test of local hiddenvariable
theories. Phys. Rev. Lett. 28, 938941 (1972).
61. Aspect, A., Dalibard, J. & Roger, G. Experimental test of Bells inequalities
using time varying analyzers. Phys. Rev. Lett. 49, 18041807 (1982).
62. Giustina, M. et al. Bell violation with entangled photons, free of the
fairsampling assumption. Nature 497, 227230 (2013).
63. Abbott, B. et al. Observation of a kilogramscale oscillator near its quantum
ground state. New J. Phys. 11, 073032 (2009).
64. Das, S. & Vagenas, E. C. Universality of quantum gravity corrections. Phys.
Rev. Lett. 101, 221301 (2008).
65. Bojowald, M. & Kempf, A. Generalized uncertainty principles and localization
of a particle in discrete space. Phys. Rev. D 86, 085017 (2012).
66. Pikovski, I., Vanner, M. R., Aspelmeyer, M., Kim, M. & Brukner, . Probing
Planckscale physics with quantum optics. Nature Phys. 8, 393397 (2012).
67. Marin, F. et al. Gravitational bar detectors set limits to Planckscale physics on
macroscopic variables. Nature Phys. 9, 7173 (2012).
68. Gambini, R., Porto, R. A. & Pullin, J. Realistic clocks, universal decoherence,
and the black hole information paradox. Phys. Rev. Lett. 93, 240401 (2004).
69. Milburn, G. J. Lorentz invariant intrinsic decoherence. New J. Phys. 8,
96 (2006).
276 NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics
2014 Macmillan Publishers Limited. All rights reserved.
NATURE PHYSICS DOI: 10.1038/NPHYS2863 INSIGHT

REVIEW ARTICLES
70. Wang, C. HT., Bingham, R. & Mendonca, J. T. Quantum gravitational
decoherence of matter waves. Class. Quantum Gravity 23,
L59L65 (2006).
71. Bassi, A., Lochan, K., Satin, S., Singh, T. P. & Ulbricht, H. Models of
wavefunction collapse, underlying theories, and experimental tests. Rev.
Mod. Phys. 85, 471527 (2013).
72. Yang, H., Miao, H., Lee, DS., Helou, B. & Chen, Y. Macroscopic
quantum mechanics in a classical spacetime. Phys. Rev. Lett. 110, 170401
(2013).
73. Giulini, D. & Groardt, A. The SchrdingerNewton equation as a
nonrelativistic limit of selfgravitating KleinGordon and Dirac fields. Class.
Quantum Gravity 29, 215010 (2012).
74. Gisin, N. Stochastic quantum dynamics and relativity. Helv. Phys. Acta 62,
363371 (1989).
75. Disi, L. A universal master equation for the gravitational violation of
quantum mechanics. Phys. Lett. A 120, 377381 (1987).
76. Ghirardi, G. C., Pearle, P. & Rimini, A. Markov processes in Hilbert space and
continuous spontaneous localization of systems of identical particles. Phys.
Rev. A 42, 7889 (1990).
77. Bassi, A. & Ghirardi, G. Dynamical reduction models. Phys. Rep. 379,
257426 (2003).
78. Adler, S. L. Quantum Theory as an Emergent Phenomenon (Cambridge
Univ. Press, 2004).
79. Leggett, A. J. Testing the limits of quantum mechanics: Motivation, state of
play, prospects. J. Phys. Condens. Mater. 14, R415R451 (2002).
80. Feldmann, W. & Tumulka, R. Parameter diagrams of the GRW and CSL
theories of wavefunction collapse. J. Phys. A 45, 065304 (2012).
81. Penrose, R. On gravitys role in quantum state reduction. Gen. Relativ. Gravit.
28, 581600 (1996).
82. Marshall, W., Simon, C., Penrose, R. & Bouwmeester, D. Towards quantum
superpositions of a mirror. Phys. Rev. Lett. 91, 130401 (2003).
83. Bose, S., Jacobs, K. & Knight, P. Scheme to probe the decoherence of a
macroscopic object. Phys. Rev. A 59, 32043210 (1999).
84. Teufel, J. D. et al. Sideband cooling of micromechanical motion to the
quantum ground state. Nature 475, 359363 (2011).
85. Chan, J. et al. Laser cooling of a nanomechanical oscillator into its quantum
ground state. Nature 478, 8992 (2011).
86. Aspelmeyer, M., Kippenberg, T. J. & Marquardt, F. Cavity optomechanics.
Preprint at http://arxiv.org/abs/1303.0733 (2013).
87. OConnell, A. D. et al. Quantum ground state and singlephonon control of a
mechanical resonator. Nature 464, 697703 (2010).
88. Chang, D. E. et al. Cavity optomechanics using an optically levitated
nanosphere. Proc. Natl Acad. Sci. USA 107, 10051010 (2010).
89. RomeroIsart, O., Juan, M. L., Quidant, R. & Cirac, J. I. Toward quantum
superposition of living organisms. New J. Phys. 12, 033015 (2010).
90. Barker, P. F. & Shneider, M. N. Cavity cooling of an optically trapped
nanoparticle. Phys. Rev. A 81, 023826 (2010).
91. RomeroIsart, O. et al. Large quantum superpositions and interference of
massive nanometersized objects. Phys. Rev. Lett. 107, 020405 (2011).
92. Hornberger, K., Gerlich, S., Haslinger, P., Nimmrichter, S. & Arndt, M.
Colloquium: Quantum interference of clusters and molecules. Rev. Mod. Phys.
84, 157173 (2012).
93. Li, T., Kheifets, S. & Raizen, M. G. Millikelvin cooling of an optically trapped
microsphere in vacuum. Nature Phys. 7, 527530 (2011).
94. Gieseler, J., Deutsch, B., Quidant, R. & Novotny, L. Subkelvin parametric
feedback cooling of a lasertrapped nanoparticle. Phys. Rev. Lett. 109,
103603 (2012).
95. Dr, W., Simon, C. & Cirac, J. I. Eective size of certain macroscopic quantum
superpositions. Phys. Rev. Lett. 89, 210402 (2002).
96. Bjrk, G. & Mana, P. A size criterion for macroscopic superposition states.
J. Opt. B 6, 429436 (2004).
97. Korsbakken, J. I., Whaley, K. B., Dubois, J. & Cirac, J. I. Measurementbased
measure of the size of macroscopic quantum superpositions. Phys. Rev. A 75,
042106 (2007).
98. Marquardt, F., Abel, B. & von Delft, J. Measuring the size of a quantum
superposition of manybody states. Phys. Rev. A 78, 012109 (2008).
99. Lee, CW. & Jeong, H. Quantification of macroscopic quantum superpositions
within phase space. Phys. Rev. Lett. 106, 220401 (2011).
100. Frwis, F. & Dr, W. Measures of macroscopicity for quantum spin systems.
New J. Phys. 14, 093039 (2012).
101. Kohstall, C. et al. Observation of interference between two molecular
BoseEinstein condensates. New J. Phys. 13, 065027 (2011).
Acknowledgements
We thank S. Nimmrichter for helpful discussions, and we acknowledge support by the
European Commission within NANOQUESTFIT (No. 304886). M.A. is supported
by the Austrian FWF (Wittgenstein Z149N16) and by the ERC (AdvG 320694
Probiotiqus), K.H. by the DFG (HO 2318/41 and SFB/TR12). We thank the
WE Heraeus Foundation for supporting the physics school Exploring the Limits
of the Quantum Superposition Principle.
Additional information
Supplementary information is available in the online version of the paper. Reprints and
permissions information is available online at www.nature.com/reprints.
Correspondence and requests for materials should be addressed to M.A. or K.H.
Competing nancial interests
The authors declare no competing financial interests.
NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics 277
2014 Macmillan Publishers Limited. All rights reserved.
278 NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics
L
ight has featured in tests of fundamental physics during times that
witnessed key advances in our understanding of nature. Newtons
investigation of the nature of light, using prisms to reveal the vis
ible spectrum, is iconic of the scientifc revolution of the sixteenth
and seventeenth centuries. Early tests of Einsteins general relativity
involved observations of starlight passing close to the Sun during a
solar eclipse. Scientifc advances have led to more convincing (and
striking) observations of relativistic efects, with images from the
Hubble Space Telescope revealing the gravitational lensing of galactic
light. And we now know how the electromagnetic spectrum extends
beyond the visible range and is quantized into single photons.
Because the primary detection apparatus in early experimental
physics consisted of the physicists themselves, light made a natu
ral observable. As our understanding of quantum photonics deep
ened, the utility of photons in tests of foundational concepts in
physics became more evident. Photons are robust to environmental
noise, have low decoherence properties, and are easily manipulated
anddetected.
Te frst half of this Review discusses tests of waveparticle dual
ity with one photon, and the second half looks at experimental tests
of nonlocality with two or more photons.
Waveparticle duality
Te doubleslit experiment has famously been said to contain the
entire mystery of quantum mechanics. It provides a concise demon
stration of the fact that single quanta are neither waves nor particles,
and that in general they are neither in one single place, nor in two
places at once.
Te experiment begins with a source of single quanta. Here we
will consider only photons, but qualitatively identical results have
been observed in a wide variety of quantum systems including elec
trons
13
, atoms
4
, and even large molecules such as C
60
(ref.5). Single
photons are sent towards a mask into which two slits have been
cut. On the far side of the mask, the spatial distribution of single
photon events is measured by a sensitive detector. For each photon,
the detector registers a click at positionx. Simultaneous detection
of two clicks never occurs, and the photon initially seems to travel
and arrive as a discrete particle. But afer many photons have been
detected, the observed probability distribution p(x) can only be
explained by wave interference due to components of the photon
that travel through both slits simultaneously. Confoundingly, when
detectors are placed directly inside the two slits, the photon is only
ever detected at one slit or the other never at both.
Testing foundations of quantum mechanics
with photons
Peter Shadbolt, Jonathan C. F. Mathews*, Anthony Laing and Jeremy L. OBrien
Quantum mechanics continues to predict efects at odds with a classical understanding of nature. Experiments with light at
the singlephoton level have historically been at the forefront of fundamental tests of quantum theory and the current develop
ments in photonic technologies enable the exploration of new directions. Here we review recent photonic experiments to test
two important themes in quantum mechanics: waveparticle duality, which is central to complementarity and delayedchoice
experiments; and Bell nonlocality, where the latest theoretical and technological advances have allowed all controversial loop
holes to be separately addressed in diferent experiments.
Where was the photon when it travelled through the mask? If it
passed through one slit and not the other, wave interference efects
would not be observed. If it passed through both slits at once, it
should be possible to detect it at both simultaneously this never
occurs. If it passed through neither slit, we should not detect it at
all but we do. In this way, the doubleslit experiment reveals the
inadequacy of classical language when describing quantum systems.
In 1909, GeofreyTaylor used a sewing needle to split a beam of
light into two paths, and observed interference fringes in the result
ing pattern of light and shadow
6
. He used an incandescent source
of feeble light with roughly the intensity of a candle held at a dis
tance of one mile. Since then, single photons have played a pivotal
role in tests of waveparticle duality. Tis is largely due to the ease
with which quantum states of light can be generated, manipulated
and measured under ambient laboratory conditions, that is, at room
temperature and pressure. Many of these experiments are based on
a very natural question: what do we know, and how much can we
measure, of the state of the photon as it passes through theslits?
Te light source used by Taylor was thermal it did not gener
ate photons one by one and his experiment consequently admits
a classical model. Te fact that true single photons are not detected
at both slits simultaneously (antibunching) was confrmed experi
mentally by Clauser
7
, who used a more sophisticated light source,
based on atomic cascades in mercury atoms. A similar source was
used by Grangieretal.
8
, who observed both antibunching and wave
interference efects analogous to those of the doubleslitexperiment.
In the quantummechanical description, detection of the pho
ton at one slit collapses the singlephoton wavefunction and pre
cludes detection at the other slit. Collapse is instantaneous, even
when the slits are very far apart, and it was emphasized by Einstein
at the Solvay conference
9
that the efect is thus seemingly nonlocal.
A recent experiment by Guerreiro et al.
10
tested Einsteins thought
experiment for the frst time, using spacelikeseparated (causally
independent)detectors.
Te notion of wavefunction collapse originates from the
Copenhagen interpretation of quantum mechanics, and encom
passes NielsBohrs principle of complementarity. Bohr maintained
that in order to observe complementary properties of a quantum
system, an experimentalist must necessarily use mutually incom
patible arrangements of the measurement apparatus. In the context
of the double slit, this means that any experiment that fully reveals
the wavelike properties of the photon must obscure its particlelike
character, and viceversa.
Centre for Quantum Photonics, H.H.Wills Physics Laboratory and Department of Electrical and Electronic Engineering, University of Bristol, Merchant
Venturers Building, Woodland Road, Bristol, BS8 1UB, UK. *email: jonathan.matthews@bristol.ac.uk
REVIEW ARTICLES

INSIGHT
PUBLISHED ONLINE: 1 APRIL 2014DOI: 10.1038/NPHYS2931
2014 Macmillan Publishers Limited. All rights reserved
NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics 279
Bohrs principle has only very recently been formalized in uni
versal complementarity relations, such as those due to Ozawa and
Hall
1113
. Tese relations formalize the notion that although the inac
curacy in either of two complementary observables can individually
be made arbitrarily small, one cannot measure both to an arbitrary
degree of accuracy using a single confguration of the measuring
device. Tis is distinguished from the Heisenberg uncertainty prin
ciple, which places more general limits on the precision of ensemble
measurements, and does not demand any such limitation on the
measuring device(s). Very recently, Westonetal.
14
used a spontane
ous parametric downconversion source together with a linearopti
cal circuit to test these new relations experimentally. Making use
of entanglement generated by the photon source, the authors were
able to test complementarity under conditions in which previously
discovered, nonuniversal complementarity relations fail.
On frst encountering the doubleslit experiment, it is natural
to wonder about the trajectory of the photon during its path from
source to detector. Complementarity implies that a single experi
mental setup cannot simultaneously obtain precise values of the
position and momentum of the photon. Indeed, a naive experi
ment hoping to track the route of the photon by measurement of
its position will destroy all wavelike efects. However, Kocsisetal.
recently demonstrated
15,16
that quantum weak measurement
17
can be used to approximately reconstruct the average trajectory
of ensembles of photons as they undergo doubleslit interference.
Weak measurement allows approximate information to be obtained
on a particular observable without appreciably disturbing strong
measurement outcomes on a complementary variable. Te authors
sent single photons from a GaAs quantum dot through a double
slit interferometer, in which a piece of birefringent calcite imposes a
weak polarization rotation depending on the angle of incidence
and thus the momentum of the photon. Ten, by simultaneous
detection of the lateral position and polarization using a highres
olution CCD (chargecoupled device) camera, weak measurement
of the photons momentum was accomplished at the same time as
strong measurement of position. Te trajectories measured in this
experiment hold particular signifcance in the de BroglieBohm
interpretation of quantum mechanics, where they are literally inter
preted as the path taken by a single particlelike photon.
John Wheelers famous delayedchoice thought experiment
18,19
also addresses the question of the position or trajectory of the pho
ton in a twopath setup. Considering the doubleslit experiment,
one might attempt to sidestep the uncomfortable implications of
waveparticle duality by means of a pseudoclassical explanation
in which the photon decides in advance to behave as a particle or
wave, depending on the choice of measurement setup. If the pho
ton notices that a particlelike measurement is planned, it dispenses
with all wavelike properties and passes through one slit at random,
and vice versa. Wheeler proposed an elegant test of this comfort
ing (if pathological) model, in which the decision to measure wave
like or particlelike behaviour is delayed until afer the photon has
passed the slits, but before it reaches the measuring apparatus.
Delayedchoice experiments have been performed in a variety of
physical systems
2023
, all of which confrm the quantum predictions
and refute the notion that the photon decides in advance to behave
as a particle or awave.
Of particular signifcance is a recent result
24
of Jacquesetal., in
which relativistic spacelike separation between the random choice
of measurement setting and slits was achieved for the frst time. Tis
ensures that there can be no causal link between the free choice of
measurement setting and the behaviour of the photon at the slits.
Here, a nitrogenvacancy colour centre in diamond was used as the
source of single photons, ensuring extremely close approximation
to the singlephoton Fock state 1. An electrooptic modulator,
driven by a quantum randomnumber generator at 4.2 MHz, was
used to implement the choice of measurement setting. A similar
experimental setup was more recently used by the same group
25
to refute the controversial claims due to Afsharetal.
26,27
that Bohrs
complementarity principle could be violated in a subtle variation on
the doubleslitexperiment.
In delayedchoice experiments, the selection of measurement set
ting is generally implemented using a classical optical switch, driven
by a randomnumber generator, which rapidly inserts or removes
an optical beamsplitter in the path of the photon (Fig. 1a). If the
beamsplitter is present, whichway information is erased and full
contrast wavelike interference is observed. If the beamsplitter is
instead absent, each detection event yields full whichway informa
tion, but no interference is seen. A recent proposal by Ionicioiu and
Terno
28
suggested that the classical random bit might be replaced
by a quantum bit (a qubit), and the classically controlled beamsplit
ter by a quantumcontrolled beamsplitter, or controlledHadamard
(CH) gate (Fig.1b). By preparing the ancilla qubit in the superposi
tion state cos()0+sin()1, the beamsplitter is efectively placed
into a coherent superposition of being present and being absent.
One can then continuously tune between particlelike and wave
like measurement settings, in close analogy with the weak measure
ment technique of ref.16. Tis idea was quickly implemented by a
number of groups
2931
, two of which used photon pairs generated
by spontaneous parametric downconversion (SPDC). Te result of
Peruzzoetal.
30
exploits recent developments in integrated quantum
photonics
32
, with Wheelers interferometer and the CH gate both
implemented onchip
33
. In the latter experiment, entanglement
generated by the CH gate allows for deviceindependent refutation
of hidden variable models in which the photon decides in advance
to behave as a particle or a wave, by violation of the BellCHSH
(ClauserHorneShimonyHolt) inequality
34
.
In the scheme of Ionicioiu and Terno, the whichway information
is carried by the ancillary particle. Tis possibility was previously
emphasized by Scully and Drhl
35
, who pointed out that the choice
of the measurement basis for the entangled ancilla determines the
contrast of wave interference observed, and that this choice can be
made even afer the system photon has been detected. Only a sub
set of allowed measurement settings completely and irrevocably
erase all whichway information, resulting in highcontrast inter
ference fringes. In 2000, Kim et al.
23
implemented this socalled
R
N
G
BS
2
a
BS
1
C
H
BS
1
b
+1
1
c b
Figure 2  A nonlocality experiment and associated loopholes. a, The
detection loophole can be opened by optical loss if there is a sufciently high
proportion of inconclusive outcomes . b, A spacelike separation prohibits
signalling between the various events occurring for each observer and closes
the locality loophole. For example, Alices measurement M
a
and results R
a
are outside the lightcone of inuence from Bobs measurement choice C
b
.
Furthermore, because C
a
and C
b
are causally disconnected from detection
events and the source, Alice and Bob are free to choose their measurement
settings without inuence. c, If the observers are not spacelike separated, it
is possible for signalling to occur between events. In this example, M
a
and M
b
can respectively inuence R
b
and R
a
, and C
b
can inuence both M
a
and R
a
.
REVIEW ARTICLES

INSIGHT NATURE PHYSICS DOI: 10.1038/NPHYS2931
2014 Macmillan Publishers Limited. All rights reserved
NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics 281
(
=(1/

2)(0110) then a choice of measurement settings
z
and
x
for Alice, and
z
+
x
and
z
x
for Bob, leads to a violation
of equation(2), with S=2

2.
Te implications of rigorously violating this inequality have a pro
found efect on our intuition of how the Universe works, for it sug
gests that the two particles are instantaneously communicating with
one another, even though they are far apart. Although the random
ness of outcomes to measurements means that no communication
can occur between Alice and Bob, these nonlocal efects seem to be in
contradiction with the spirit, if not the letter, of special relativity. Tese
farreaching implications have motivated particular scrutiny of the
possible ways in which nature might somehow fake nonlocality, with
focus mainly falling on experimental limitations. An apparent experi
mental violation S>2 could be attributed to assumptions exploited by
LHVs known as loopholes, the more famous of which are the local
ity, detection and freedom of choice loopholes (Fig.2). A completely
unambiguous experimental demonstration of Bell nonlocality requires
the simultaneous obstruction of every possible loophole. Although
this milestone is yet to be reached in experimental physics, photons
have been used to address each of these loopholes individually.
Te detection loophole. Optical tests of nonlocality have sufered
from low detection efciency. With an experimental efciency
of < 100% there exist, in addition to +1 and 1, inconclusive
L P1 P2
P2
HWP1
HWP2
TC NLC
HWP3
HWP3
PBS
PBS
IF
IF
SMF
TDC TES
TDC
SMF
TES
144 km freespace link
La Palma Tenerife
1.2 km RF link
QRNG
A
Bob
0 0 1 1 1 1 00
Transmitter
EOM
Delay
Polarization analyser
HWP
QWP
6 km SMF
Pol. comp.
PBS
D
R
D
T
RF receiver RF transmitter
Quantum random
number generator
(QRNG)
QWP
PBS
D
R
D
T
EOM
OGS telescope
Polarization analyser
LED
BS
PM 0
PM 1
LD
PBS
ppKTP
Entangled photon source
10 m SMF
Alice
Bob
QRNG
A
Logic circuit
GPS TTU Computer
Source
QRNG
B
Sampling
circuit
HWP
1.2 km
144 km quantum link
Source and
Alice
Delay
Logic circuit
GPS TTU Computer
a
b
Figure 3  Experiments for closing the Bell nonlocality loopholes. a, High detection efciency can be achieved with TES to close the detection loophole.
L,laser; PC, Pockels cell; P1 and P2, crossed polarizers; PBS, polarizing beamsplitter; NLC, paired nonlinear BiBO crystals; TC, BBO crystal; HWP, halfwave
plate; IF, interference lters; TDC, timetodigital converter; SMF, singlemode bre. b, Spacelike separation of the quantum randomnumber generators
that choose the random measurement settings and the measurement apparatus. This enables the locality and freedom of choice loopholes in the
experiment to be closed. LD, laser diode; ppKTP, periodically poled potassium titanyl phosphate crystal; QWP, quarterwave plate; EOM, electrooptical
modulator; D
T
, D
R
, photodetectors; QRNG, quantum random number generator; LED, lightemitting diode; PM, photomultiplier; TTU, timetagging unit;
GPS, global positioning system. Figure reproduced with permission from: a, ref.60 2013 APS; b, ref.68 2010 PNAS; Geographic pictures taken from
Google Earth, 2008 Google, Map Data 2008 Tele Atlas.
INSIGHT

REVIEW ARTICLES NATURE PHYSICS DOI: 10.1038/NPHYS2931
2014 Macmillan Publishers Limited. All rights reserved
282 NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics
measurement outcomes that represent the failure to detect an
emitted photon (Fig.2a). Te outcome can be ignored by includ
ing only measurements that register photon detection. But this
relies on the assumption of a fair sampling, because otherwise, local
models may skew the detection statistics of +1 and 1 to falsify
violation of equation (2). Tis has been illustrated experimentally
through the use of side channels to intentionally falsify signatures of
nonlocality in experimental setups that are otherwise considered as
standard Bellinequality experiments
5557
.
When including outcomes, violation of CHSH (equation(2))
only occurs when experimental efciencies are beyond the thresh
old of > 82.8%. Remarkably, Eberhard discovered that lower
ing the amount of entanglement by controlling the r parameter in
(r01 10)/

1

+

r

2
reduces the threshold efciency to > 66.7%
in testing nonlocality
58
. Denoting as n
k,l
(a
i
b
j
) the number of photon
pairs with outcome k {+1,1,} and l {+1,1,} when using
measurement settings i{0,1} on one particle and j{0,1} on the
other, Eberhards inequality (which holds for LHVs) is written as
J = n
+1,+1
(a
1
,b
1
) n
+1,+1
(a
0
,b
0
)+n
+1,1
(a
0
,b
1
) (3)
n
+1,
(a
0
,b
1
)+n
1,+1
(a
1
,b
0
)+n
,+1
(a
1
,b
0
) 0
Notably, each observer needs only one detector, as the decrease in
efciency of detectors responsible for 1 outcomes causes out
comes nominally 1 to be included as , mapping n
+1,1
(a
0
,b
1
) to
n
+1,
(a
0
,b
1
) and n
1,+1
(a
1
,b
0
) to n
,+1
(a
1
,b
0
) (ref.59). Furthermore, test
ing nonlocality with equation (3) is robust to the Poissonian nature
of photoncounting measurements on SPDC sources, and removal
of the vacuum state through postselection is notrequired.
Two recent experiments
59,60
report violation of Eberhards inequal
ity to close the detection loophole. Both experiments use transi
tionedge sensors
61
(TES), which are highefciency singlephoton
detectors, and highcollectionefciency photon sources to surpass
Eberhards efciency threshold, and each obtain >70%: ref.59 uses
a highcollectionefciency photon source based on a Sagnac con
fguration
62,63
; ref.60 uses a noncollinear SPDC photon source con
fguration (Fig.3a). Although both demonstrations are of sufcient
efciency to close the detection loophole, the experiment in ref.59 is
still open to the coincidencetime loophole
64
. Tis experiment relies
(like numerous others) on using timing windows defned by single
photon detection events to perform coincident detection analysis:
when one photon detector registers a singlephoton event, a coinci
dence event is recorded if a second singlephoton event is recorded
within a prescribed window of time. Te coincidencetime loophole
allows the detection time to be shifed by the local measurement set
tings in or out of the coincidence window, so that a completely local
process can match quantum mechanical expectation values. But this
loophole can be avoided by using a coincidence window defned
around a system clock: ref.60 achieves this with a chopped laser pulse
that drives the SPDC to create photon pairs in welldefned events.
Te work in ref.60 also highlights the productionrate loophole
where nonrandom drifing of the pump laser power or detection
efciency can be exploited by local realistic models. Te experi
mental drifs in ref. 59 have, however, been shown
65
not to be suf
fcient for this loophole. Alternatively, a quantum randomnumber
generator can be used to choose measurement settings randomly
in order to close the productionrate loophole
60
. Furthermore, sat
isfying the more stringent requirement of randomly chosen meas
urement settings for every entangled particle pair in order to close
the freedom of choice loophole simultaneously addresses the pro
ductionrate loophole.
Freedom of choice and locality loophole. Two famous experi
ments attempted to close the locality loophole through spacelike
separation by fast measurement settings chosen during the time of
fight of the entangled photons
66,67
. But the settings of ref. 66 were
chosen using periodic sinusoids and were therefore predictable and
susceptible to infuence by hidden variables created at the source,
so failed to close the freedom of choice loophole the possible
infuence of measurement settings either by other measurement
apparatus or by hidden variables created at the source of photons.
Te random settings of ref. 67 were chosen within the forwards
lightcone of the emission point of the entangled photons, so could
also have been infuenced by hidden variables created at the source.
Improving on these experiments, the authors of ref. 68 separated
their randomnumber generators in a spacelike way, to remove
the possibility of transmitting any physical signal between entan
gled particle emission and the random measurement settings. Tis
Bell test was performed between two Canary Islands, La Palma and
Tenerife, separated by 144km, with the quantum randomnumber
generator used to choose measurement bases spacelike separated
from the rest of the experiment (Fig.3b).
EPRsteering. Almost 80 years afer Schrdinger referred to the
efects of entanglement as piloting or steering of one quantum
state by the measurement of another, the concept of EPRsteering
was formalized
69,70
and was swifly followed by an EPRsteering ine
quality based on local models
71
. Steering sits strictly between entan
glement witnesses
72
and Bell nonlocality. Te idea of entanglement
witnesses relies entirely on assumptions that quantum mechanics is
correct, to test for the presence of nonseparability, whereas in Bell
nonlocality ideally no assumptions are made about the experimen
tal setup or the model of physics. Te concept of steering (Fig.4)
assumes that one half of the system, an observer Bob, fully trusts
his measurement apparatus and that any states in his possession
adhere to the laws of quantum mechanics. A second party (Alice)
is tasked with convincing Bob that she can steer a quantum state
that she has already sent to him. Importantly, no assumptions are
made about the physics to which Alice has access, so she is free to
use any means to carry out her task. Assuming local models, this
experiment is constrained by the inequality
71
S
n
k=1
A
k
k
B
C
n
n
(4)
in which Alice and Bob compare n measurement results;
k
B
is the
kth
of nmeasurements performed by Bob in conjunction with Alice
declaring a measurement result A
k
{1,+1}. C
n
is the maximum
value that can be obtained for the quantity S
n
, provided Bob has
preexisting states known to Alice. Tis inequality is violated when
Alice instead shares entanglement with Bob, and, through her own
measurements, afects Bobsresults.
Saunders et al.
73
performed the frst experimental demonstra
tions of violating the steering inequalities with polarizationentan
gled photons, showing that increasing the number of measurements
n (testing up to n=6) increases the robustness of this nonlocality
test to experimental noise. Just like Belllike inequalities, however,
local models can also exploit loopholes to explain steering. Steering
has less stringent requirements than the aforementioned nonlocal
ity tests, owing to the asymmetry of the experiment, and has an
experiment efciency threshold of > 1/3 for closing the detec
tion loophole when using n=3 measurements. Tree experiments
published around the same time collectively address loophole
free steering
7476
. All three experiments use Sagnac entanglement
sources (see Box1) to increase experiment efciency and close the
detection loophole; in addition, Smithetal.
75
use TES singlephoton
detectors. Bennetetal.
74
use up to n=16 measurement settings, and
they show that this allows them to measure violation of equation(4)
without assuming fair sampling, despite high loss (87%) induced by
1 km of coiled optical fbre between the entanglement source and
REVIEW ARTICLES

INSIGHT NATURE PHYSICS DOI: 10.1038/NPHYS2931
2014 Macmillan Publishers Limited. All rights reserved
NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics 283
one of the measurement apparatus; this explores the conditions for
closing the freedom of choice and locality loopholes over a lossy
channel. In addition to closing the detection loophole, Wittmann
76
enforces strict Einstein locality conditions, with spacelike separa
tion over 48m of optical fbre. Tis closes the locality loophole. In
addition, they use a spacelikeseparated quantum randomnumber
generator, closing the freedom of choice loophole that would other
wise allow the photon pair source to infuence the choice of meas
urement setting. Tis is the frst time a nonlocal quantum efect has
been explored while simultaneously closing three important loop
holes. Collectively, these experiments
7476
mark important progress
towards loopholefree, deviceindependent tests of Bell nonlocality.
Referenceframeindependent nonlocality tests. Traditionally,
nonlocality tests take place within a shared reference frame. Tat
is to say, Alice and Bob are able to align their measurement appa
ratus with respect to one another. Tis may be problematic for
experiments using long optical fbre or freespace orbital commu
nications. One solution is to harness decoherencefree subspaces
77
,
and this has been recently implemented using rotationally invariant
entanglement in which photon pairs are entangled in their orbital
angular momentum states and their polarization states, to violate a
Bell inequality in alignmentfree settings
78
. Tis requires an increase
in dimension of the quantum system investigated. Remarkably, it
has been shown that sharing a complete reference frame is not
required for two remote parties to violate a Bell inequality: provided
the parties share one measurement direction perfectly, they have
high probability of violating a Bell inequality perfectly with a maxi
mally entangled state by each choosing maximally complementary
measurements in the plane orthogonal to the shared direction in the
Bloch sphere
79
. By increasing the complexity of the measurements
that each party makes, observers can always violate a Bell inequality
without sharing any information about their reference frames
80,81
.
Tis potentially removes the need for establishing reference frames
for future nonlocality tests, in particular when taking nonlocality
tests into orbit to help address the locality loophole.
Multipartite locality tests. Most nonlocality tests have been
focused towards using bipartite entanglement. Greenberger, Horne
and Zeilinger extended nonlocality tests to that of threeparty
entanglement
82
; this was formulated into an inequality to test for
multipartite nonlocality by Mermin
83
. Te frst threephoton GHZ
entanglement was demonstrated 15years ago using a pulsed SPDC
source
84
and was then subsequently used to violate Mermins ine
quality
85
; fourphoton GHZ states
86,87
have also been used for local
realism tests
88
. Until recently, however, no multiphoton experiment
has succeeded in addressing loopholes that can be exploited by
LHVs. Te main contributing factor is the typically low brightness
of multiphoton entangled sources. Recently, Ervenetal.
89
reported
the generation of heralded threephoton GHZ entanglement at suf
fcient rates (40 Hz) to distribute the three photons using optical
fbre and freespace links to independent measurement stations to
violate Mermins inequality. With sufciently separated measure
ment stations and entanglement source, the authors address the
locality loophole, while the freedom of choice loophole is closed
by spatially separating a randomnumber generator that defnes
the measurement basis settings. But experiment efciencies below
the threshold required to close the loophole of Mermins inequality
mean that the detection loophole is not closed, and fair sampling is
assumed. Tis leaves open the possibility of using highefciency
photon detectors and developing efcient collection in multiphoton
entangled states for loopholefree multipartite nonlocality tests in
thefuture.
Outlook
Photonic experiments over the past four decades have answered
many important debates in the fundamental theory of quantum
mechanics, and new photonic technologies continue to create
opportunities to close loopholes, answer old questions and even
inspire new theoretical research. Experimental confrmation of
the predictions of quantum physics during the previous century
forced a reevaluation of the understanding of the operation of the
Universe as a classical machine, at least at the microscopic scale.
Over the coming decades, as we increase our capabilities to harness
the efects of quantum mechanics to build quantum computers
90
, we
will test the extent to which quantum efects persist at a macroscopic
scale, with further potential consequences for our understanding of
the Universe. Famously, the extended ChurchTuring thesis (ECT)
says that all computational problems that are efciently solvable
with realistic physical systems can be efciently solved with a classi
cal machine a statement clearly in confict with our hopes for the
capabilities of quantum computers
91
. Although we might have to
wait some time for a universal quantum computer to operate at the
scale that challenges the ECT, recent theoretical
92
and technological
Figure 4  EPR steering. Here, one observer (Bob) trusts that his system works according to quantum mechanics (denoted by a clear box), while another
party (Alice) is tasked with supplying Bob with a quantum state and demonstrating that she can afect his measurement results by any means (black box).
Assuming local laws of physics, an inequality for this experiment is derived, which is violated when Alice chooses to share entanglement between herself
and Bob. Figure reproduced with permission from ref.73, 2010 NPG.
Alice Bob
Classical communication Output Entangled pair Key :
1
2
3
4
a
or
b
Measurement Analyser Detector
A
k
k
B
S
n
k=1
A
k
k
B
n
k
Pure state
INSIGHT

REVIEW ARTICLES NATURE PHYSICS DOI: 10.1038/NPHYS2931
2014 Macmillan Publishers Limited. All rights reserved
284 NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics
Sources of entangled photons
For nearly two decades, spontaneous parametric downconver
sion (SPDC) based on nonlinear crystals has been the most widely
applied source of entanglement in quantum optics
98
. A confgura
tion that has recently advanced collection efciency is that of col
linear SPDC, coherently pumped in both directions with a laser
split by a beamsplitter, in a polarization Sagnac interferometer
62,63
(Fig. B1a). Tis allows inherent stability without the need for
active stabilization. By eliminating the transverse walkof efect
by periodically poling in the nonlinear crystal (that is, alternating
the orientations of the birefringent material), high collection ef
ciency into singlemode fbre is obtained. Tis confguration has
been demonstrated to operate with continuouswave or pulsed
regimes
99
with at least 80% coupling efciency and for a number
of nonlinear materials
100
. For future experiments, a more compact
source of entangled photons would probably use an integrated
architecture, where stabilized pathentangled photons
101
and
polarizationentangled photons
102
can be generated (Fig. B1b).
Increasing detector efciency
Singlephoton detectors
103
underpin the measurements made by the
observers in any photonic nonlocality experiment. Transitionedge
sensors (TES) are fabricated using a thin tungsten flm embedded
in an optical stack of materials to enhance the absorption
61
. With
the voltage biased at their superconducting transition, absorbed
photons cause a measurable change in the current fowing through
the tungsten flm that is efciently measured with a superconduct
ing quantum interference device (SQUID) amplifer. TES require
cooling to about 100 mK using adiabatic demagnetization refrig
erators, and detection efciencies of about 95% are now routinely
reported. Nanowire superconducting singlephoton detectors
104
have emerged as a promising alternative for both freespace and
integrated applications: here a single photon absorbed by a super
conductor biased just below its critical current I
c
creates a local
resistive hotspot, generating a voltage pulse. Superconducting
detectors based on NbN nanowires operate at about 4K tempera
tures and are capable of very fast counting rates (up to gigahertz)
and low dark counts (<1 Hz)
104
. Such NbN nanowire detectors
can operate in commercial cryocoolers
103
. Recent NbN nanowire
detectors using a travellingwave design
105
(Fig. B1c) have demon
strated onchip detection efciency above 90%. In addition, recent
realization of NbTiN nanowire singlephoton detectors on SiN
(ref. 106)
extends the operating wavelength from infrared to vis
ible, and reduces the dark count rate to millihertz.
Box 1  Enabling technology for current and future nonlocality tests.
PC
ppKTP
L1
LD
PC
PC
PBS
HWP
HWP
Alice
L2
L2
L2
LP
LP
BP
Bob
TES
TES
SPAD
HWP
QWP
Source
BD
PBS
QWP
Inductor
Capacitor
RF signal
Gnd
I
n
t
e
n
s
i
t
y
Current source
Ampliers
Silicon waveguide
Silicon
Buried oxide
NbN
110nm
3 m
500 m
4nm
D
i
r
e
c
t
i
o
n
o
f
p
r
o
p
a
g
a
t
i
o
n
a b
c
45 polarized
pump pulse
SiO
x
N
y
second
core 840(W) x 840(H) nm
2
Si wire 200(W) x 200(H) nm
2
30 m
Si wire waveguides
400(W) x 200(H) nm
2
1.5 mm long
Silicon
polarization
rotator
( TE,TE
s,i
+e
i
TM,TM
s,i
)
Polarizationentangled photon pairs
2
1
Figure B1  Current and future photonics for nonlocality tests. a, A polarization Sagnac interferometer conguration can ofer >80% collection
efciency of generated photon pairs into optical bre. When used together with highefciency (>95%) TES singlephoton detectors, this can be used
to address the detection loophole in nonlocality tests. The example here depicts Alice using TES to address the detection loophole in EPR steering in
the experiment reported in ref.75. SPAD, singlephoton avalanche diode; BD, polarization beam displacer. Other abbreviations as in Fig.3. b,Waveguide
entangled sources ofer potentially repeatable, high brightness and highefciency sources of entanglement. s, single photon; i, idler photon; TE,
horizontally polarized; TM, vertically polarized. c, Superconducting singlephoton detectors ofer a lowtemperature (4 K) alternative in highefciency
and fast singlephoton detection that can be monolithically integrated into waveguide structures for potentially compact photonics measurement
apparatus. Figures reproduced with permission from: b, ref.102, 2012 NPG; c, ref.105, 2012 NPG.
REVIEW ARTICLES

INSIGHT NATURE PHYSICS DOI: 10.1038/NPHYS2931
2014 Macmillan Publishers Limited. All rights reserved
NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics 285
advances in quantum photonics
93
have developed a path to chal
lenging the ECT on a nearterm timescale, with a nonuniversal
quantum photonic device that performs a task known as boson
sampling
9497
. If experiments confrm the prediction, as we believe
they will, that our Universe cannot be efciently simulated by a
classical machine, then there may be other confounding features of
quantum mechanics currently hidden from us and apparent only
through simulations on a quantum computer. It is therefore pos
sible that, rather than confrming existing theory, future photonic
experiments might be the frst to reveal new and complex quantum
phenomena, requiring innovative theoreticalexplanations.
Received 14 January 2014; accepted 25 February 2014; published
online 1 April 2014.
References
1. Tonomura, A., Endo, J., Matsuda, T., Kawasaki, T. & Ezawa, H. Demonstration
of singleelectron buildup of an interference pattern. Am. J.Phys.
57, 117120 (1989).
2. Bach, R., Pope, D., Liou, S.H. & Batelaan, H. Controlled doubleslit electron
difraction. New J.Phys. 15, 033018 (2013).
3. Jnsson, C. Electron difraction at multiple slits. Am. J.Phys. 42, 411 (1974).
4. Carnal, O. & Mlynek, J. Youngs doubleslit experiment with atoms: A simple
atom interferometer. Phys. Rev. Lett. 66, 26892692 (1991).
5. Arndt, M. et al. Waveparticle duality of C
60
molecules. Nature
401, 680682 (1999).
6. Taylor, G. I. Interference fringes with feeble light. Proc. Camb. Phil. Soc.
15, 114115 (1909).
7. Clauser, J. F. Experimental distinction between the quantum and classical
feldtheoretic predictions for the photoelectric efect. Phys. Rev. D
9, 853860 (1974).
8. Grangier, P., Roger, G. & Aspect, A. Experimental evidence for a photon
anticorrelation efect on a beam splitter: A new light on singlephoton
interferences. Europhys. Lett. 1, 173 (1986).
9. Jammer, M. Te Philosophy of Quantum Mechanics (Wiley, 1974).
10. Guerreiro, T., Sanguinetti, B., Zbinden, H., Gisin, N. &
Suarez, A. Singlephoton spacelike antibunching. Preprint at
http://arxiv.org/quantph/1204.1712 (2012).
11. Hall, M. J. W. Prior information: How to circumvent the standard joint
measurement uncertainty relation. Phys. Rev. A 69, 052113 (2004).
12. Ozawa, M. Universally valid reformulation of the Heisenberg uncertainty
principle on noise and disturbance in measurement. Phys. Rev. A
67, 042105 (2003).
13. Erhart, J. et al. Experimental demonstration of a universally valid error
disturbance uncertainty relation in spin measurements. Nature Phys.
8, 185189 (2012).
14. Weston, M. M., Hall, M. J. W., Palsson, M. S., Wiseman, H. M. & Pryde, G. J.
Experimental test of universal complementarity relations. Phys. Rev. Lett.
110, 220402 (2013).
15. Wiseman, H. M. Grounding Bohmian mechanics in weak values and
Bayesianism. New J.Phys. 9, 165 (2007).
16. Kocsis, S. et al. Observing the average trajectories of single photons in a two
slit interferometer. Science 332, 11701173 (2011).
17. Aharonov, Y., Albert, D. & Vaidman, L. How the result of a measurement
of a component of the spin of a spin1/2 particle can turn out to be 100.
Phys. Rev. Lett. 60, 13511354 (1988).
18. Wheeler, J. A. in Mathematical Foundations of Quantum Teory
(ed.Marlow,A. R.) 948 (Academic Press, 1978).
19. Wheeler, J. A. & Zurek, W. H. Quantum Teory and Measurement
(Princeton Univ. Press, 1984).
20. Alley, C. O., Jacubowicz, O. G. & Wickes, W. C. in Proc. Second Int. Symp.
Foundations of Quantum Mechanics (Narani, H. ed.) 36 (Physics Society of
Japan, 1987).
21. Hellmuth, T., Walther, H., Zajonc, A. & Schleich, W. Delayedchoice
experiments in quantum interference. Phys. Rev. A 35, 25322541 (1987).
22. LawsonDaku, B. J. et al. Delayed choices in atom SternGerlach
interferometry. Phys. Rev. A 54, 50425047 (1996).
23. Kim, YH., Yu, R., Kulik, S. P., Shih, Y. & Scully, M. O. Delayed choice
quantum eraser. Phys. Rev. Lett. 84, 15 (2000).
24. Jacques, V. et al. Experimental realization of Wheelers delayedchoice
gedanken experiment. Science 315, 966968 (2007).
25. Jacques, V. et al. Illustration of quantum complementarity using single photons
interfering on a grating. New J.Phys. 10, 123009 (2008).
26. Afshar, S. S. Violation of the principle of complementarity, and its
implications. Proc. SPIE 5866, 229244 (2005).
27. Afshar, S. S., Flores, E., McDonald, K. F. & Knoesel, E. Paradox in
waveparticle duality. Found. Phys. 37, 295305 (2007).
28. Ionicioiu, R. & Terno, D. R. Proposal for a quantum delayedchoice
experiment. Phys. Rev. Lett. 107, 230406 (2011).
29. Kaiser, F., Coudreau, T., Milman, P., Ostrowsky, D. B. &
Tanzilli, S. Entanglementenabled delayedchoice experiment. Science
338, 637640 (2012).
30. Peruzzo, A., Shadbolt, P., Brunner, N., Popescu, S. & OBrien, J. L. A quantum
delayedchoice experiment. Science 338, 634637 (2012).
31. Roy, S. S., Shukla, A. & Mahesh, T. S. NMR implementation of a quantum
delayedchoice experiment. Phys. Rev. A 85, 022109 (2012).
32. Politi, A., Matthews, J. C. F., Tompson, M. G. & OBrien, J. L.
Integrated Quantum Photonics. IEEE J.Select. Top. Quant, Electron,
15, 16731684 (2009).
33. Shadbolt, P. J. et al. Generating, manipulating and measuring entanglement
and mixture with a reconfgurable photonic circuit. Nature Photon.
6, 4549 (2012).
34. Clauser, J. F., Horne, M. A., Shimony, A. & Holt, R. A. Proposed experiment to
test local hiddenvariable theories. Phys. Rev. Lett. 23, 880884 (1969).
35. Scully, M. O. & Drhl, K. Quantum eraser: A proposed photon correlation
experiment concerning observation and delayed choice in quantum
mechanics. Phys. Rev. A 25, 22082213 (1982).
36. Ma, XS. et al. Quantum erasure with causally disconnected choice.
Proc. Natl Acad. Sci. USA 110, 12211226 (2013).
37. Kochen, S. & Specker, E. P. Te problem of hidden variables in quantum
mechanics. J.Math. Mech. 17, 5987 (1967).
38. Cabello, A. & GarcaAlcaine, G. Proposed experimental tests of the
BellKochenSpecker theorem. Phys. Rev. Lett. 80, 17971799 (1998).
39. Meyer, D. A. Finite precision measurement nullifes the KochenSpecker
theorem. Phys. Rev. Lett. 83, 37513754 (1999).
40. Greenberger, D., Horne, M., Shimony, A. & Zeilinger, A. Bells theorem
without inequalities. Am. J.Phys 58, 11311143 (1990).
41. Simon, C., Zukowski, M., Weinfurter, H. & Zeilinger, A. Feasible
KochenSpecker experiment with single particles. Phys. Rev. Lett.
85, 17831786 (2000).
42. Cabello, A. All versus nothing inseparability for two observers.
Phys. Rev. Lett. 87, 010403 (2001).
43. Amselem, E., Rdmark, M., Bourennane, M. & Cabello, A. State
independent quantum contextuality with single photons. Phys. Rev. Lett.
103, 160405 (2009).
44. Michler, M., Weinfurter, H. & Zukowski, M. Experiments towards
falsifcation of noncontextual hidden variable theories. Phys. Rev. Lett.
84, 54575461 (2000).
45. Huang, YF., Li, CF., Zhang, YS., Pan, JW. & Guo, GC. Experimental
test of the KochenSpecker theorem with single photons. Phys. Rev. Lett.
90, 250401 (2003).
46. Lapkiewicz, R. et al. Experimental nonclassicality of an indivisible quantum
system. Nature 474, 490493 (2011).
47. Klyachko, A. A., Can, M. A., Biniciolu, S. & Shumovsky, A. S. Simple test for
hidden variables in spin1 systems. Phys. Rev. Lett. 101, 020403 (2008).
48. Abrams, D. S. & Lloyd, S. Nonlinear quantum mechanics implies
polynomialtime solution for NPcomplete and #P problems. Phys. Rev. Lett.
81, 39923995 (1998).
49. Sinha, U., Couteau, C., Jennewein, T., Lafamme, R. & Weihs, G. Ruling out
multiorder interference in quantum mechanics. Science 329, 418421 (2010).
50. Bell, J. S. On the Einstein Podolsky Rosen paradox. Physics 1, 195200 (1964).
51. Pan, JW. et al. Multiphoton entanglement and interferometry. Rev. Mod. Phys.
84, 777838 (2012).
52. Brunner, N., Cavalcanti, D., Pironio, S., Scarani, V. &
Wehner, S. Bell nonlocality. Rev. Mod. Phys. (in the press); preprint at
http://arxiv.org/quantph/1303.2849 (2013).
53. Freedman, S. J. & Clauser, J. F. Experimental test of local hiddenvariable
theories. Phys. Rev. Lett. 28, 938 941 (1972).
54. Aspect, A., Grangier, P. & Roger, G. Experimental realization of
EinsteinPodolskyRosenBohm Gedankenexperiment: A new violation of
Bells inequalities. Phys. Rev. Lett. 49, 9194 (1982).
55. Tasca, D. S., Walborn, S. P., Toscano, F. & Souto Ribeiro, P. H. Observation of
tunable PopescuRohrlich correlations through postselection of a Gaussian
state. Phys. Rev. A 80, 030101(R) (2009).
56. Gerhardt, I. et al. Experimentally faking the violation of Bells inequalities.
Phys. Rev. Lett. 107, 170404 (2011).
57. Pomarico, E., Sanguinetti, B., Sekatski, P., Zbinden, H. & Gisin, N.
Experimental amplifcation of an entangled photon: what if the detection
loophole is ignored? New J.Phys. 13, 063031 (2011).
INSIGHT

REVIEW ARTICLES NATURE PHYSICS DOI: 10.1038/NPHYS2931
2014 Macmillan Publishers Limited. All rights reserved
286 NATURE PHYSICS  VOL 10  APRIL 2014  www.nature.com/naturephysics
58. Eberhard, P. H. Background level and counter efciencies required for
a loopholefree EinsteinPodolskyRosen experiment. Phys. Rev. A
47, 747750 (1993).
59. Giustina, M. et al. Bell violation using entangled photons without the fair
sampling assumption. Nature 497, 227 230 (2013).
60. Christensen, B. G. et al. Detectionloopholefree test of quantum nonlocality,
and applications. Phys. Rev. Lett. 111, 130406 (2013).
61. Lita, A. E., Miller, A. & Nam, S. W. Counting nearinfrared singlephotons with
95% efciency. Opt. Express 16, 3032 (2008).
62. Kim, R., Fiorentino, M. & Wong, F. Phasestable source of polarization
entangled photons using a Sagnac interferometer. Phys. Rev. A
73, 12316 (2006).
63. Fedrizzi, A., Herbst, T., Poppe, A., Jennewein, T. & Zeilinger, A. A wavelength
tunable fbrecoupled source of narrowband entangled photons. Opt. Express
15, 1537715386 (2007).
64. Larsson, JA. & Gill, R. D. Bells inequality and the coincidencetime loophole.
Europhys. Lett. 67,707713 (2004).
65. Kofer, J., Ramelow, S., Giustina, M. & Zeilinger, A. On Bell violation
using entangled photons without the fairsampling assumption. Preprint at
http://arxiv.org/abs/1307.6475 (2013).
66. Aspect, A., Dalibard, J. & Roger, G. Experimental test of Bells inequalities
using timevarying analyzers. Phys.Rev. Lett. 49, 18041807 (1982).
67. Weihs, G., Jennewein, T., Simon, C., Weinfurter, H. & Zeilinger, A. Violation
of Bells inequalities under strict Einstein locality conditions. Phys. Rev. Lett.
81, 5039 5034 (1998).
68. Scheidl, T. et al. Violation of local realism with freedom of choice.
Proc. Natl Acad. Sci. USA 107, 1970819713 (2010).
69. Wiseman, H. M., Jones, S. J. & Doherty, A. C. Steering, entanglement,
nonlocality, and the EinsteinPodolskyRosen paradox. Phys. Rev. Lett.
98, 140402 (2007).
70. Jones, S. J., Wiseman, H. M. & Doherty, A. C. Entanglement,
EinsteinPodolskyRosen correlations, Bell nonlocality, and steering.
Phys. Rev. A 76, 052116 (2007).
71. Cavalcanti, E. G., Jones, S. J., Wiseman, H. M. & Reid, M. D. Experimental
criteria for steering and the EinsteinPodolskyRosen paradox. Phys. Rev. A
80, 032112 (2009).
72. Plenio, M. B. & Virmani, S. An introduction to entanglement measures.
Quant. Inf. Comput. 7, 151 (2007).
73. Saunders, D. J., Jones, S. J., Wiseman, H. M. & Pryde, G. J. Experimental
EPRsteering using Belllocal states. Nature Phys. 6, 845849 (2010).
74. Bennet, A. J. et al. Arbitrarily losstolerant EinsteinPodolskyRosen steering
allowing a demonstration over 1km of optical fbre with no detection
loophole. Phys. Rev. X 2, 031003 (2012).
75. Smith, D. H. et al. Conclusive quantum steering with superconducting
transitionedge sensors. Nature Commun. 3, 625 (2012).
76. Wittmann, B. et al. Loopholefree EinsteinPodolskyRosen experiment via
quantum steering. New J.Phys. 14, 053030 (2012).
77. Cabello, A. Bells theorem without inequalities and without alignments.
Phys. Rev. Lett. 91, 230403 (2003).
78. DAmbrosio, V. et al. Complete experimental toolbox for alignmentfree
quantum communication. Nature Commun. 3, 961 (2012).
79. Palsson, M. S., Wallman, J. J., Bennet, A. J. & Pryde, G. J. Experimentally
demonstrating referenceframeindependent violations of Bell inequalities.
Phys. Rev. A 86, 032322 (2012).
80. Shadbolt, P. J. et al. Guaranteed violation of a Bell inequality without aligned
reference frames or calibrated devices. Sci. Rep. 2, 470 (2012).
81. Wallman, J. J. & Bartlett, S. D. Observers can always generate nonlocal
correlations without aligning measurements by covering all their bases.
Phys. Rev. A 85, 024101 (2012).
82. Greenberger, D. M., Horne, M. A. & Zeilinger, A. Bells Teorem, Quantum
Teory, and Conceptions of the Universe (ed. Kafatos, M.) 6972 (Kluwer, 1989)
83. Mermin, N. D. Extreme quantum entanglement in a superposition of
macroscopically distinct states. Phys. Rev. Lett. 65, 18381840 (1990).
84. Bouwmeester, D., Pan, J. W., Daniell, M., Weinfurter, H. & Zeilinger, A.
Observation of threephoton GreenbergerHorneZeilinger entanglement.
Phys. Rev. Lett. 82, 1345 1349 (1999).
85. Pan, JW., Bouwmeester, D., Daniell, M., Weinfurter, H. & Zeilinger, A.
Experimental tests of quantum nonlocality in threephoton Greenberger
HorneZeilinger experiment. Nature 403, 515519 (2000).
86. Pan, JW., Daniell, M., Gasparoni, S., Weihs, G. & Zeilinger, A. Experimental
demonstration of fourphoton entanglement and highfdelity teleportation.
Phys. Rev. Lett. 86, 44354438 (2001).
87. Eibl, M. et al. Experimental observation of fourphoton entanglement from
parametric downconversion. Phys.Rev. Lett. 90, 200403 (2003).
88. Zhao, Z. et al. Experimental violation of local realism by fourphoton
GreenbergerHorneZeilinger entanglement. Phys. Rev. Lett. 912, 180401 (2003).
89. Erven, C. et al. Experimental threeparticle quantum nonlocality under strict
locality conditions. Nature Photon. 8, http://dx.doi.org/nphoton.2014.50 (2014).
90. Nielsen, M. A. & Chuang, I. L. Quantum Computation and Quantum
Information (Cambridge Univ. Press, 2010).
91. Shor, P. W. in Proc. 35th Ann. Symp. Found. Comput. Sci. 124134 (IEEE, 1994).
92. Aaronson, S. & Arkhipov, A. in STOC 11: Proc. 43rd Ann. ACM Symp.Teory
Comput. 333342 (ACM, 2011).
93. Politi, A., Cryan, M. J., Rarity, J. G., Yu, S. & OBrien, J. L. Silicaonsilicon
waveguide quantum circuits. Science 320, 646649 (2008).
94. Broome, M. A. et al. Photonic boson sampling in a tunable circuit. Science
339, 794798 (2013).
95. Spring, J. B. et al. Boson sampling on a photonic chip. Science
339, 798801 (2013).
96. Crespi, A. et al. Experimental boson sampling in arbitrary integrated photonic
circuits. Nature Photon. 7, 545549 (2013).
97. Tillmann, M. et al. Experimental boson sampling. Nature Photon.
7, 540544 (2013).
98. Kwiat, P. G. et al. New highintensity source of polarizationentangled photon
pairs. Phys. Rev. Lett. 75, 43374341 (1995).
99. Predojevic, A., Grabher, S. & Weihs, G. Pulsed Sagnac source of polarisation
entangled photon pairs. Opt. Express 20, 2502225029 (2012).
100. Guerreiro, T. et al. High efciency coupling of photon pairs in practice.
Opt. Express 21, 2764127651 (2013).
101. Silverstone, J. et al. Onchip quantum interference between two silicon
waveguide sources. Nature Photon. 8, 104108 (2013).
102. Matsuda, N. et al. A monolithically integrated polarization entangled photon
pair source on a silicon chip. Sci. Rep. 2, 817 (2012).
103. Hadfeld, R. H. Singlephoton detectors for optical quantum information
applications. Nature Photon. 3, 696705 (2009).
104. Goltsman, G. N. et al. Picosecond superconducting singlephoton optical
detector. Appl. Phys. Lett. 79, 705 (2001).
105. Pernice, W. H. P. et al. Highspeed and highefciency travelling wave single
photon detectors embedded in nanophotonic circuits. Nature Commun.
3, 1325 (2012).
106. Schuck, C., Pernice, W. H. P. & Tang, H. X. Waveguide integrated low noise
NbTiN nanowire singlephoton detectors with milliHz dark count rate.
Sci. Rep. 3, 1893 (2013).
Acknowledgements
We are grateful for fnancial support from EPSRC, ERC, NSQI (S.G.). J.C.F.M. is
supported by a Leverhulme Trust EarlyCareer Fellowship. J.L.O.B. acknowledges a Royal
Society Wolfson Merit Award and a Royal Academy of Engineering Chair in Emerging
Technologies. We thank P. Birchall, N. Brunner and C. Sparrow for helpful comments.
Additional information
Reprints and permissions information is available online at www.nature.com/reprints.
Correspondence and requests for materials should be addressed to J.C.F.M.
Competing nancial interests
Te authors declare no competing fnancial interests.
REVIEW ARTICLES

INSIGHT NATURE PHYSICS DOI: 10.1038/NPHYS2931
2014 Macmillan Publishers Limited. All rights reserved