Está en la página 1de 6

PR Boxes and Quantum Non-Locality

Sundara Rajan, Suhas Kowshik


Indian Institute of Technology, Madras
ee11b130,ee11b132 @ee.iitm.ac.in
Abstract
Non-locality is one the most intriguing and useful aspects of Quantum Mechanics. Non-locality in nature is
seemingly implausible and there exist some phenomenon which can be explained by the non-locality associated
with quantum mechanics. Entanglement is one such phenomenon which is a part of quantum mechanics. But a
deeper question to ponder is if nature is implicitly non-local, if so is it possible to have non-local correlations
stronger than those predicted by quantum mechanics? Here we try to answer some of these questions, and in the
process unearth a whole bunch of more fundamental questions.

I.

Introduction

As we all know quantum mechanics, as yet, is the most complete and extensively studied theory to
explain nature. It explains almost every known fact. Yet, there are dozens of paradoxes and puzzles
which are yet to be explained. It is safe to assume that we still lack the intuition needed to gain a deep
understanding of quantum mechanics, or that we need a much stronger theory to understand nature.
One reason for this may be that the axioms of quantum mechanics are not natural or physical like those of
say, classical mechanics. The fundamental indeterminism that exists in quantum mechanics makes the
subject one of the most difficult to study and comprehend. We all know the famous Einsteins remark
"God doesnt play dice".
Speaking of indeterminism, it is easy to see that it is certainly possible to have a theory which
is non-deterministic without being non-local. Random processes which exist in nature all fall under
this category, which are very well explained. On the other hand, it is not possible for a theory to be
non-local, but deterministic. If that is the case, a direct consequence of such a theory is superluminous
travel. Putting it more informally, if an event happening here leads to a wiggle elsewhere, relativity
demands that the wiggle be uncertain. This means that non-locality and not non-determinism should be
the foundation over which theories must be built.
Popescu and Rohrlich tried to address this concern, and they stumbled upon something remarkable.
Their research concluded that nature can be more non-local than what quantum mechanics predicts, or
in other words, it is, in principle, possible to have a theory which allows more non-locality than quantum
mechanic, yet be perfectly consistent with relativity.

II.

Preliminaries

We go through some basic definitions and theorems that will be used throughout this paper.
Definition II.1. A box is a device with 2 ends, one of which is held by Alice and the other is held by Bob, who is
far away from Alice. Each end has an input and output behaviour. Given inputs x from Alice and y from Bob,
Alices end will output a and Bobs end b, with a probability P(a,b|x,y). Where x, y, a, b {0,1} [9]
We also assume instantaneity of processing, that is, the outputs appear as soon as the inputs are given.
Definition II.2. A box is non-signalling if the marginals of one persons output doesnt depend on the other
persons input. Mathematically speaking:

P(a, b|x, y) = P(a, b|x, y0 ) = P A (a|x)


b

a, x, y, y0

and

P(a, b|x, y) = P(a, b|x0 , y) = PB (b|y)


a

b, x, y, x 0

(1)

A box is signalling if it is not non-signalling.


We now come to Non-Locality which is defined in [1], as follows
Definition II.3. There exist in nature channels connecting two (or more) distant partners, that can distribute
correlations which can neither be caused by the exchange of a signal (the channel does not allow signalling, and
moreover, a hypothetical signal should travel faster than light), nor be due to pre-determined agreement (because
the correlations violate the so-called "Bells inequalities").
In simple terms a box is local if the output on one side depends only on the input on the same side.
A box is non-local if it is not local. It is useful to quantify non-locality in terms of a Bell-Type inequality
derived by Clauser, Horne, Shimony and Holt(CHSH), which reads as follows:
CHSH = (0, 0) + (0, 1) + (1, 0) (1, 1)

(2)

where (0, 0) is the correlation between the outputs conditioned on the inputs 0 and 0. A system, box in
our case, is local iff its CHSH value is less than or equal to 2. Else it is non-local.
It is also proved that if

a box can be implemented by quantum mechanics, its CHSH value is 2 2.

III.

The PR-Box

We now come to a particular box which seems to have a great significance, the PR-Box.

Alices
end

Bobs
End

Figure 1: The PR-Box

Definition III.1. A PR-Box is a box which satisfies the following conditions:


1. It is Non-Signalling, as defined in Definition 2.2.
2. For the box P( a| x ) = P(b|y) = 21 .
3. The input-output relationship follows a + b = x.y, where the addition is done mod 2.
The third condition states that if at least one of the inputs x or y is 0, then a = b, else a 6= b.
Theorem III.1. The PR-Box violates the CHSH inequality. In fact, the PR-Box reaches upto CHSR
= 4, which is
the maximal possible violation, compared to quantum physics, which can attain a CHSH upto 2 2.
Proof.
CHSHPR = (0, 0) + (0, 1) + (1, 0) (1, 1).
(0, 0) =

E[ ab|00] E[ a|00] E[b|00]


a|00 b|00
2

(3)

Now,
E[ ab|00] = P(00|00) 0 + P(11|00) 1

= P(11|00) = P( a = 1|b = 1, 00) P(b = 1|00) = 1 1/2 = 1/2


as, if (x,y) = (0,0), it is necessary that a = b, and P(b| = 1|00) = P(b = 1|y = 0) = 1/2.
E[ a|00] = E[ a|0] = 1/2 = E[b|00]
a2|00 = E[ a2 |00] E[ a|00]2 = 1/2 1/4 = 1/4
Similarly b2|00 = 1/4.
1/4
Therefore (0, 0) = 1/21/4
= 1.
Similarly it can be shown that (0, 1) = (1, 0) = 1 and (1, 1) = 1. Therefore

CHSHPR = 1 + 1 + 1 (1) = 4

(4)

It can also be seen that this is the algebraic limit of CHSH. This proves that PR-Box attains the maximal
possible violation.

IV.
I.

Applications of the PR-Box

Simulation of the singlet using the PR-Box

in the 1960s, Bell proved that quantum correlations cannot be reproduced by any local hidden variable
theory, the results of which are now widely known as the famous "Bells Inequalities". Hence, there is a
necessity for a non-local resource in order to simulate quantum correlations - especially entanglement.
There is an interesting result in the field of communications, involving the simulation of Bell correlations
- The singlet state can be simulated by supplementing local hidden variables with a single bit of communication [2].
But interestingly enough, the simulation of a singlet state can be achieved with a single use of the
PR-Box. The protocol for the same, proceeds as follows:
Consider that Alice and Bob share a PR-Box and share a pair of normalized random and independent
~1 and ~2 distributed on a sphere. Let v~A and v~B determine the measurement direction of Alice
vectors
and Bob respectively.
Alice inputs

~1 ) + sgn(v~A .~2 )
x = sgn(v~A .

(5)

into the PR-Box, where



sgn( x ) =

1
0

x0
x<0

(6)

~1 ). This is the equivalent of


Alice gets a from the PR-Box and then outputs A = a + sgn(v~A .
measurement outcomes of a singlet state along v~A .
Similarly Bob inputs
y = sgn(v~B .~+ ) + sgn(v~B .~ )
(7)
~1 ~2 . The PR-Box outputs B, and he outputs B = b + sgn(v~B .~+ ) + 1. It can be shown,
where ~ =
as in [3], that this model reduces to that described by [2]. Hence the correlations are exactly same as
that of the singlet state.
3

II.

Simulation of 3-party GHZ

The correlations are defined by


a0 + b0 + c1 = 0
a0 + b1 + c0 = 0
a1 + b0 + c0 = 0
a1 + b1 + c1 = 1

(8)

where a x is the outcome of Alice when her input is x, and similarly for Bob and Charlie. It can be easily
seen that no shared randomness can fulfil all the conditions simultaneously, as adding the first three
equations gives a1 + b1 + c1 = 0.
These correlations can be simulated using a PR-Box between A and B. Suppose Alice, Bob and
Charlie share random variables { x , y , z }. Let these satisfy the first three relations. Now Alice
and Bob input x and y in the PR-Box and get output a and b. Alice outputs a x = a + x , Bob outputs
by = y + b, and Charlie outputs cz = z . So it is observed that
a x + by + cz = x + y + z + xy

(9)

since a + b = xy. This means that unless x = y = 1, the first three relations are satisfied, and when
x = y = 1, the fourth relation is also satisfied.

III.

No Cloning Theorem

The theory of PR-Boxes contains a no-cloning theorem. This can be seen as follows:
Let Alice and Bob share a PR-Box with a + b = xy, as defined earlier, and suppose that Bob can clone
his part of the Box to get another channel between Alice and himself, according to a + b0 = xy0 , where
b0 and y0 represent the cloned part. If this is the case, adding the two equations yields b + b0 = x (y + y0 ),
which implies Bob can compute b + b0 and he knows y and y0 , thus obtaining the value of x. This would
imply signalling, which contradicts the no-signalling property of PR-Boxes. Hence perfect cloning is not
possible.

V.
I.

Features of the PR-Box

Information Theoretical Power

Information theory is one of the most extensively studied aspects of communication. Claude E. Shannon
was the first one to come up with a mathematical theory that places theoretical bounds on the speed
and efficiency of information transfer. But the advent of Quantum Information theory further relaxed
the theoretical bounds, and subsequently the problem of information transfer regained prominence.
Te PR-Box is one such theory, which allows the communication complexity to reduce beyond imaginations, as seen in [4]. This can be demonstrated as follows:
Suppose Alice and Bob each receive a string of n bits, say, ~x and ~y. Bob is required to output a bit
f (~x, ~y). This requires some communication between the two. Let f (~x, ~y) = nk=1 xk yk . Now ideally, one
might think that Alice has to send Bob all the n bits of ~x in order for Bob to compute the given function.
But this task can be simplified with a PR-Box shared between Alice and Bob. The protocol goes as
follows:
f (~x, ~y) = A + B =

a k + bk
k

(10)

since xk yk = ak + bk Now Alice just sends one bit A = k ak to Bob. So Bob can compute the required
function, by receiving just one bit from Alice.

The above problem is not just a random and silly communication task, but is in fact, the most difficult
and practically relevant communication task - known as the inner-product problem. This is equivalent to
finding the inner-product of two vectors. Quantum mechanical correlations cannot help us with this
task, hence they cannot eliminate all the redundancy from communication.

II.

Non Local Computation

In the problem of non-local computation, Alice and Bob must compute a function without knowing
anything about the inputs apriori. Entanglement provides no benefit over local classical strategies for
such tasks, but super-quantum correlations such as that of the PR-Box allow for perfect success [7].
Say we have a function f (z1 , z2 , ..., zn ). The computation is carried out in two places: in Alices and
Bobs. For each zi , we randomly assign xi to Alice and yi to Bob such that xi + yi = zi . Thus for each
value of zi , ( xi , yi ) can take two sets of values with equal probability. The output is written as
a + b = f ( x + y) = f ( x1 + y1 , x2 + y2 , ..., xn + yn )

(11)

We take an extended PR-Box([7]) for which the input-output relationship is given by



P( a, b| x, y) =

1
2

if

a + b = f ( x + y)
else

(12)

This distribution gives rise to both sets of outputs , which means Alice and Bob can compute f
simultaneously without learning about the others input.

VI.

Failures of the PR-Box

The PR-Box theory cannot simulate every non-local multi-partite correlations, even if arbitrarily large
amount of them are used. For example, it has been proved in [1] that the following three-partite
correlations for the bits cannot be simulated:
a0 + b1 = 0
c1 + b0 = 0
c0 + a1 = 0
a0 + b0 + c0 = 0

a1 + b1 + c1 = 1

VII.

(13)

Entanglement vs Boxes

The non-local boxes are built to describe the measurement process, i.e, there is an input and an output.
But a quantum state can be defined without assuming that it is going to be measured. PR-Boxes are
classical channels, that is, there is no definition of super position principle or coherent measurements.
Also entanglement cannot increase if local operations are done, and communicated classically. So,
classical information can be used alongside entanglement. But for PR-Boxes classical information is
forbidden by definition due to the assumption of no-signalling; classical information increases nonlocality. Thus, it can be seen that there are quite a few fundamental differences between entanglement
and PR-Boxes.
5

VIII.

Information Causality and the PR-Box

Suppose Alice has two bits that she wants to communicate to Bob. It can be shown that even though by
sending a one-bit message, Alice cannot communicate both bits to Bob, it is possible for Bob to choose
which bit he wants to learn, when Alice and Bob share a PR-Box. The protocol is as follows:
Let x0 and x1 be Alices two bits. She inputs x = x0 + x1 into the PR-Box. She sends m = x0 + a to Bob.
Now, if Bob wants to learn x0 , he inputs y = 0, whereas if he wants to learn x1 , he inputs y = 1.
Bob then calculates
m + b = x0 + a + b = x0 + xy = x0 + ( x0 + x1 )y
(14)
since a + b = xy. It can be seen that if y = 0 then m + b = x0 + 0 = x0 , and if y = 1, m + b =
x0 + x0 + x1 = 0 + x1 = x1 .
The ability of Bob in this particular scenario to choose which bit he wants to learn violates a principle
known as Information Causality: It is impossible to obtain information about two bits by the transmission
of a single bit. Thus PR-Boxes violate Information Causality. On the other hand, it turns out that
the point at which Information Causality is violated is exactly at the boundary of quantum and
super-quantum Correlations. Thus both classical and quantum mechanics do not violate Information
Causality.

IX.

Conclusion

The usual approach to Quantum Mechanics is one in which we include indeterminism as an axiom,and
non-locality arises as a result of the axioms. The question to ask would be whether non-locality can be
taken as an axiom along with relativistic causality so that we can arrive at quantum mechanics as a result.
But it has been found that by taking relativistic causality and non-locality as axioms we arrive at a set of
theories to which quantum mechanics also belongs [8]. The above discussion shows the existence of an
other theory which satisfies non-locality and relativistic causality yet differs from quantum mechanics,
as can be seen from the violation of CHSH. These violations of CHSH above quantum boundaries are
called super-quantum correlations. If, in future, such super-quantum correlations/non-local boxes are
found in nature then they can have profound impact on how we understand nature, and also would
have a lot of applications and change the faces of computational sciences and communications.

References
[1] Valerio Scarani, Feats, Features and Failures of the PR-box.
[2] B. F. Toner, D. Bacon, Communication Cost of Simulating Bell Correlations.
[3] N. J. Cerf, N. Gisin, S. Massar, S. Popescu, Simulating Maximal Quantum Entanglement without Communication.
[4] W. van Dam Implausible Consequences of Superstrong Nonlocality.
[5] Sandu Popescu, Nonlocality beyond quantum mechanics.
[6] Cleve R, van Dam W, Nielsen M, and Tapp A, Quantum Computing and Quantum Communication, Lecture Notes in Computer
Science
[7] Noah Linden, Sandu Popescu, Anthony J. Short, Andreas Winter, Quantum Nonlocality and Beyond: Limits from Nonlocal
Computation.
[8] Sandu Popescu and Daniel Rohrlich, Quantum Nonlocality as an axiom.
[9] Philippe Lamontagne, Non-Local Boxes.

También podría gustarte