See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/358284821

Three-Qubit-Embedded Split Cayley Hexagon is Contextuality Sensitive

Preprint · January 2022

CITATIONS READS
0 180

3 authors:

Frederic Holweck Henri de Boutray

Université de Technologie de Belfort-Montbéliard Institut FEMTO-ST
74 PUBLICATIONS 524 CITATIONS 17 PUBLICATIONS 41 CITATIONS

SEE PROFILE SEE PROFILE

Metod Saniga
Astronomical Institute of the Slovak Academy of Sciences
213 PUBLICATIONS 2,056 CITATIONS

SEE PROFILE

All content following this page was uploaded by Metod Saniga on 03 February 2022.

The user has requested enhancement of the downloaded file.

 Three-Qubit-Embedded Split Cayley Hexagon is
Contextuality Sensitive

Frédéric Holweck1,2 , Henri de Boutray3 and Metod Saniga4

1
Laboratoire Interdisciplinaire Carnot de Bourgogne, ICB/UTBM, UMR 6303 CNRS,
Université de Bourgogne Franche-Comté, 90010 Belfort Cedex, France
2
Department of Mathematics and Statistics, Auburn University, Auburn, AL, USA
3
ColibriTD, La Défense, Paris, France
4
Astronomical Institute, Slovak Academy of Sciences, SK-05960 Tatranská Lomnica,
Slovak Republic
arXiv:2202.00726v1 [quant-ph] 1 Feb 2022

(29 January 2022)

Abstract: It is known that there are two non-equivalent embeddings of the split Cayley
hexagon of order two into W(5, 2), the binary symplectic polar space of rank three, called
classical and skew. Labelling the 63 points of W(5, 2) by the 63 canonical observables of
the three-qubit Pauli group subject to the symplectic polarity induced by the (commutation
relations between the elements of the) group, the two types of embedding are found to be
quantum contextuality sensitive. In particular, we show that the complement of a classically-
embedded hexagon is not contextual, whereas that of a skewly-embedded one is.

Keywords: finite geometry, quantum contextuality, Pauli groups, generalized polygons,

Kochen-Specker Theorem, split Cayley hexagon of order two

1 Introduction
Quantum contextuality is one of the most counter-intuitive notions in quantum physics
that is assumed to be of main importance in quantum information [1, 2, 3, 4]. Roughly
speaking, quantum contextuality rules out any Hidden Variables Theory unless one assumes
that the pre-existing values of a measurement depend on the context, i.e. on the set of
mutually commuting measurements in which a given experience takes place. First proved
mathematically by Kochen and Specker [5], several alternative proofs [6, 7] and experimental
validations [8, 9] have been provided since then. For a recent comprehensive survey about
contextuality, see [10].
Among various formulations of quantum contextuality we will deal in this paper with
observable-based proofs of quantum contextuality as proposed by Mermin [11] and Peres
[12]. In this formulation, see Section 2, a proof of the Kochen-Specker (KS) Theorem is
provided by a configuration of sets of mutually commuting observables, called contexts, such
that all the constrains imposed by the configuration on the eigenvalues of those observables
cannot be satisfied by a classical function unless that function is context-dependent. We
will call such a configuration contextual when it furnishes a proof of the KS Theorem. These
proofs of contextuality can also be tested experimentally [13, 14, 15]. Interestingly, this type
of contextual configurations, which involves N -qubit Pauli observables, can be embedded
as subgeometries of the geometric realization of the N -qubit Pauli group known as the
symplectic polar space of rank N and order two, usually denoted as W(2N − 1, 2), see
Section 2. This geometric perspective on N -qubit observables was introduced some 15 years
ago [16, 17, 18] and led to several surprising connections between physics and geometry
[19] and also between different branches of modern physics, like, for instance, the so-called
black-hole/qubit correspondence [20, 21].
In W(5, 2) one can find a number of copies of a very special configuration made of 63
points and 63 lines, called the split Cayley hexagon of order two. This configuration is a
generalized hexagon is the sense that it is m-gon free for m ≤ 5. The fact that it can be
embedded into the three-qubit polar space, W(5, 2), was first employed in [22] to establish

1
connections between three-qubit observables and black-hole entropy formulas. Later it was
also proved that this configuration has, in fact, two distinguished embeddings in W(5, 2), one
called classical, the other skew [23]. In this article one shows that these two distinguished
embeddings behave differently in terms of contextuality. More precisely, if one denotes the
two embeddings by HC and HS , respectively, then one demonstrates that the configuration
HC = W(5, 2) \ HC , i.e. the configuration made of all the line-contexts that are in W(5, 2)
but not in HC , is not contextual, while HS = W(5, 2) \ HS is contextual.
The paper is organized as follows. In Section 2 we recall the basic notions on observable-
based proofs of contextuality as well as the notion of symplectic polar spaces where corre-
sponding configurations live. We also explain how we prove that a given configuration is
contextual by solving a particular linear system. In Section 3 one details the differences
between the two distinguished symplectic embeddings of the split Cayley hexagon of order
two and recall some known results about these specific configurations. Finally in Section 4
we collect our results and state precisely our hexagon-related contextuality findings. Section
5 concludes this paper with general remarks and perspectives.

2 Observable-based proofs of contextuality and the sym-

plextic polar spaces over the two-element field
Let us recall the principle of the Mermin and Peres [11, 12] operator-based proofs of quantum
contextuality. Consider the configuration of two-qubit observables depicted in Figure 1 and
known as a Peres-Mermin magic square.

XI XZ IZ +

IX ZX ZI +

XX YY ZZ −

+ + +

Figure 1: The Peres-Mermin magic square: All two-qubit observables on the same
row/column mutually commute. The product of observables in each row/column is equal
to ±I4 as indicated by the signs. The measurement of each two-qubit observable yields
±1. One observes that there is no deterministic classical function that would predict the
outcomes for each node and satisfy all the signs constraints.

Each node of the square is labelled by a two-qubit observable with a shorthanded notation
AB ≡ A ⊗ B. Here A, B ∈ {I, X, Y, Z} with X, Y, Z being the usual Pauli matrices and I
the unit matrix. In each row/column of the grid, two-qubit observables mutually commute
and form what we call a context, i. e. a set of compatible measurements whose product is, up
to a sign, the identity operator. In the Peres-Mermin square the product of the observables
on each context is ±I4 as indicated by the signs in Figure 1. In each context, the product of
the measurements, which are eigenvalues of the observables, should be an eigenvalue of the
product of the observables. The eigenvalues of each node are ±1 and the constraint on each
context is ±1 according to the signs prescription on rows/columns. Because there is an odd
number of negative contexts in the square, it is clear that there exists no classical function
that can assign predefinite values to each node and satisfy all the constrains imposed by the
signs of the contexts (rows/columns). If such a classical function existed, it would necessarily
be context-dependent, i. e. the predefined value assigned to a given node would depend on

2
the considered context. This elementary configuration of observables furnishes a proof of the
KS Theorem by showing that there is no Non-Contextual Hidden Variable (NCHV) theory
that can reproduce the outcomes of quantum physics. Another elementary proof based on
three-qubit observables was proposed by Mermin in [11] and is reproduced in Figure 2.

Y II

XXX YYX Y XY XY Y

IIX IIY

XII

IY I IXI

Figure 2: The Mermin pentagram [11]. Each node is labelled by a three-qubit Pauli ob-
servable. The lines of the pentagram are sets of mutually commuting observables whose
product is +I8 (each thin line) or −I8 (the double line). The fact that there is only one
(odd number) negative line allows us to use the same argument as for the Peres-Mermin
magic square. This configuration therefore also furnishes a proof of the KS Theorem.

In [24] it was proved that Peres-Mermin magic squares and Mermin pentagrams are the
smallest possible proofs of the KS Theorem that can be achieved in terms of the number of
observables and contexts: 9 observables and 6 contexts for the Peres-Mermin magic square
and 10 observables and 5 contexts for the pentagram.
If one considers two-qubit Pauli observables up to a global phase (±1, ±i), there are 10
copies of the magic square. One explains now how one can naturally embed these 10 grids
into the symplectic polar space of rank N = 2 over the two-elements field F2 = {0, 1}. In
full generality let us consider the group of N -qubit Pauli observables, i.e.

PN = {sA1 A2 . . . AN , s ∈ {±1, ±i}, Ai ∈ {I, X, Y, Z}}, (1)

employing again the shorthand notation A1 A2 . . . AN ≡ A1 ⊗ A2 ⊗ · · · ⊗ AN . The center of

PN is CN = {±I2N , ±iI2N }. The abelian group PN /CN is isomorphic to the 2N -dimensional
vector space VN = F2N
2 as we now explain.
Up to a phase, each observable Ai can be represented by a doublet (µi , νi ) ∈ F22 as follows

Ai = Z µi · X νi where · denotes the ordinary product of matrices; (2)

more precisely, we have the following relations:

I ↔ (0, 0), X ↔ (0, 1), Z ↔ (1, 0), iY ↔ (1, 1). (3)

Hence, to a given class O = {A1 A2 . . . AN , −A1 A2 . . . AN , iA1 A2 . . . AN , −iA1 A2 . . . AN }

in PN /CN one can associate a unique 2N -plet (µ1 , ν1 , . . . , µN , νN ) ∈ F2N
2 . Moreover, the
multiplicative structure of PN /CN is mapped to the additive structure of VN . Considering
now the projective space associated with VN , P G(2N − 1, 2), one obtains a map between
non-trivial N -qubit observables, up to a global phase, and points of P G(2N − 1, 2)

(PN /CN ) \ I2N → P G(2N − 1, 2),
π: (4)
O 7→ [µ1 : ν1 : µ2 : µ2 : · · · : µN : νN ].

This mapping represents (classes of) observables as points in P G(2N − 1, 2), but tells us
nothing about the commutation relations in PN . In order to recover this information, one

3
introduces the following symplectic form on P G(2N − 1, 2):
N
X
hp, qi = pi qN +i + pN +i qi , (5)
i=1

where p = [p1 : · · · : p2N ] and q = [q1 : · · · : q2N ] ∈ P G(2N −1, 2). Then, if one considers two
non-trivial classes Op and Oq of N -qubit Pauli observables in PN /CN such that π(Op ) = p
and π(Oq ) = q, then a straightforward calculation shows that

Op and Oq commute ⇔ hp, qi = 0. (6)

Employing the above-defined symplectic form leads to the definition of W(2N − 1, 2), the
symplectic polar space of rank N and order 2.
Definition 1. The space of totally isotropic subspaces1 of P G(2N − 1, 2) endowed with a
nondegenerate symplectic form h, i is called the symplectic polar space of rank N and order
2, W(2N − 1, 2).
Note that because of the symplectic form, the group Sp(2N, 2) of symplectic matrices
acts transitively on W(2N − 1, 2). This group is spanned by the transvections Tp for all
p ∈ W(2N − 1, 2):


W(2N − 1, 2) → W(2N − 1, 2),
Tp : (7)
q 7 → q + hp, qip.

If we consider the action of transvections in terms of the labelling of W(2N −1, 2) by N -qubit
Pauli observables, then for p = π(Op ) and q = π(Oq ) we have


Oq if and only if Oq and Oq commute,
TOp (Oq ) = (8)
Oq · Op if and only if Oq and Oq anticommute.

The points of W(2N − 1, 2) are N -qubit observables, lines correspond to triplet of mutually
commuting elements whose product is ±IN and form contexts. Planes and higer dimensional
linear subspaces of W(2N −1, 2) can also be of some use to generate contexts. Let us illustrate
this for N = 2 and N = 3 [25, 26].

2.1 N = 2, the Doily W(3, 2)

The symplectic polar space of rank 2 and order 2, known as the doily, encapsulates the
commutation relations within the two-qubit Pauli group. This space features 15 points and
15 lines and form a 153 configuration, i.e. a point-line incidence structure with 3 points per
line and 3 lines per point. Out of 245, 342 non isomorphic 153 -configurations, the doily is
the only one to be triangle free. Using the canonical representatives (s = 1 in (1)), one gets
a labelling of the doily by two-qubit Pauli observables as given in Figure 3.
Peres-Mermin magic squares are subgeometries of W(3, 2). More precisely they are
geometric hyperplanes of W(3, 2), i.e. sets of points such that a line of W(3, 2) is either fully
contained in such set or shares with it a single point. An example of such a set is given in
Figure 4. Starting with this canonical labeling one can form 9 more copies of Peres-Mermin
square by acting by transvections. Each of them have an odd number, namely 1 or 3, of
negative contexts and is thus contextual.
To conclude this subsection it is worth mentioning that the space W(3, 2) of its own is a
contextual configuration, as first proved in [13]. In fact, Cabello shows that the configuration
corresponding to points and lines of W(2N − 1, 2) is contextual for any N ≥ 2.
1 A linear space is said to be totally isotropic if and only if the symplectic form vanishes identically on

the space.

4
 IY

XY ZY

YZ
XI ZI
YX

XZ
IX ZX IZ
YI

XX YY ZZ

Figure 3: The symplectic polar space W(3, 2) labelled by two-qubit Pauli observables. Each
line of the configuration forms a context. The three negative contexts are indicated in blue.

IY

XY ZY

YZ
XI ZI
YX

XZ
IX ZX IZ
YI

XX YY ZZ

Figure 4: A Peres-Mermin square as a subgeometry of the doily. The square is the same
as the one introduced in Figure 1. It is a geometric hyperplane in the sense that a line of
W(3, 2) is either contained in it (six such lines depicted in red) or has one point common
with it (each of the remaining nine lines).

2.2 N = 3, the symplectic polar space W(5, 2)

The symplectic polar space of rank 3, W(5, 2), comprises 63 points, 315 lines and 135
Fano planes. An interesting study of geometric hyperplanes of W(5, 2) and their physical
interpretations in terms of representation theory and invariants can be found in [26]. Fano
planes of W(5, 2) are totally isotropic 2-dimensional spaces over F2 . An example of such
a plane is given in Figure 5. The product of the seven observables in a Fano plane yields
±I8 . If one removes a line, one gets a new type of context – an affine plane of order
two – made of four observables whose product is ±I8 . For instance, in the negative Fano
plane of Figure 5, removing the line ZZI − IZZ − ZIZ, which is positive, leads to the
context XXX − Y Y X − Y XY − XY Y that is exactly the negative line of the Mermin
pentagram of Figure 2. Using three-qubit Pauli operator contexts of this type one can
create altogether 12, 096 distinct Mermin’s pentagrams in W(5, 2), as it was first shown
by computer calculations [27] and later proved also rigorously by sole geometric arguments
[28]. In [29] it was demonstrated that geometric hyperplanes of W(5, 2) defined by quadratic
equations (hyperbolic or elliptic) are also contextual configurations; contextual inequalities
for the hyperbolic case were also tested on a quantum computer [15].
In Section 3 we will introduce another prominent configuration, namely the split Cay-
ley hexagon of order two, which lives in W(5, 2) in two distinct modes featuring different
contextuality behavior.

5
 XXX

YYX Y XY

XY Y

ZZI IZZ ZIZ

Figure 5: An example of the (negative) Fano plane in W(5, 2); the blue line is also negative.

2.3 A simple method for checking contextuality of a configuration

In order to prove that a particular point-line configuration of observables, i.e. a set of ob-
servables {Oj }j∈J grouped into contexts {Ci }i∈I , provides a KS proof, we reformulate the
question as a linear problem [29]. Indeed, the existence of an NCHV theory that accommo-
dates all contexts Ci of a configuration means that there exists a classical function, f , such
that f (Oj ) = ±1 for all observables and for all contexts Ci of the configuration. This means
that f satisfies the following system of equations:

ΠOj ∈Ci f (Oj ) = sign(Ci ), ∀i ∈ I, (9)

where sign(Ci ) is the sign of the context Ci .

Consider now the incidence matrix A of the configuration, i. e. an I × J matrix where
aij = 1 if and only if Oj ∈ Ci and aij = 0 otherwise. Let b ∈ FI2 such that bi = 0 if and
only if sign(Ci ) = 1 and bi = 1 if and only if sign(Ci ) = −1. Then the existence of a classical
function f that satifies the constrains of Eq. (9) boils down to finding a vector x ∈ FJ2 which
is the solution of the following linear system

Ax = b. (10)

For instance, in the case of the Peres-Mermin configuration of Figure 1 the linear system to
be solved over the two-element field is
 
x1
  x2   
1 1 1 0 0 0 0 0 0   0
0 0 0 1 1 1 0 0 0 x3  0
 
0 0 0 0 0 0 1 1 1 x4  0
     
1 0 0 1 0 0 1 0 0 · x5  = 0 . (11)
     
  x6   
0 1 0 0 1 0 0 1 0   0
x7 
0 0 1 0 0 1 0 0 1  
x8  1
x9

One readily sees that this system has no solution as the sum of the first five rows of A
equals the last row, while the sum of the first five coordinates of b does not equal the last
coordinate. The lack of solution means that the corresponding configuration is contextual,
i. e. it furnishes a proof of the KS Theorem. This linear formulation of the problem allows
us to get a relatively fast computer-based check of the contextual nature of larger sets of
contexts and observables. Note that the incidence matrix A encodes the geometry of the
configuration, while the vector b encodes the signs of the contexts; the latter, of course,
depends on the choice of labeling of the points of the symplectic polar space by N -qubit
observables.

6
3 The split Cayley hexagon of order two and its two
non-equivalent symplectic embeddings
A generalized n-gon G of order (k, l) is an point-line incidence structure such that every
line contains k + 1 points, every point is contained in l + 1 lines and G does not contain
any ordinary m-gons for 2 ≤ m < n, but two points, two lines or a point and a line are
always contained in an n-gon [30]. When k = l one says that the order of G is k. The Fano
plane, Figure 5, is the unique example of the generalized triangle of order two and the doily
W(3, 2), Figure 3, is the unique (self-dual) generalized quadrangle of order two. There is
no generalized 5-gon of order two, but there exist two generalized hexagons of order two,
the split Cayley hexagon and its dual [30]. We will focus now on the split Cayley hexagon,
H, which also lives in W(5, 2); this means that one can label the 63 points of H by the 63
non trivial three-qubit observables in such a way that the lines of the hexagon are lines of
W(5, 2), i. e. sets of mutually commuting three-qubit observables.
As already stressed, there are two unequivalent embeddings of H in W(5, 2). The first
one, called classical, was explicitly worked out in [22] and in the context of quantum in-
formation further discussed, for instance, in [20, 27, 32]; an example of such embedding is
portrayed in Figure 6.

XZY

IYX YIZ
ZXY

XZI XIY
IZY XZX XZZ
IZI XII
XIX IIY IZZ
XYY
YZY
YXZ YYY ZYX
ZZZ IYY YIY XXX
YII IYI
ZYY YXY
XIZ
IZX ZIZ
IXX YIX
IYZ YZX XYZ
ZXX
ZXZ
YYZ YYX
ZII IXI YZI
XYI IXY ZXI
ZIY

ZZY ZIX XYX YZZ IXZ XXY

IIX IIZ

XXI
ZZX ZZI XXZ
YYI

ZYZ YXX

YXI ZYI

Figure 6: Three-qubit classical embedding of the split Cayley hexagon of order two as
described in [22]. The illustration of points and lines of the hexagon follows closely that of
Ref. [31]
.

In W(5, 2) there are 120 classically-embedded copies of H; each of them can be obtained
from the copy shown in Figure 6 by the action of the symplectic group Sp(6, 2). The labelling
of Figure 6 was obtained from the observation that the split Cayley hexagon contains a copy
of the Heawood graph, the incidence graph of the Fano plane, and following the procedure
outlined in [33].
Another description of this embedding is the following one [30]. In P G(6, 2), consider

7
the following parabolic quadric Q

x1 x4 + x2 x5 + x3 x6 − x27 = 0. (12)

This quadric contains 63 points. Consider now the lines of Q that satisfy the following
Plücker equations:

p62 = p17 , p13 = p72 , p24 = p37 , p35 = p74 ,

(13)
p46 = p57 , p51 = p76 , p14 + p25 + p36 = 0,

where pij = xi xj − xj xi . There are 63 such lines and the obtained 633 configuration is
isomorphic to H. This configuration can be bijectively projected into P G(5, 2) and W(5, 2).
Indeed the projection

[x1 : · · · : x7 ] 7→ [x1 : · · · : x6 ] (14)

is a bijection over F2 as x7 = x1 x2 + x3 x4 + x5 x6 and the lines of Q that satisfy the Plücker

relations (13) are mapped to totally isotropic lines of P G(5, 2), i. e. the lines of W(5, 2) for
the sympletic form (5). This embedding will be denoted by HC .
Interestingly, there exists another, non-equivalent embedding of H into W(5, 2), as dis-
covered by Coolsaet [23]. Coolsaet considers, in P G(6, 2), the following coordinate map

 : [x1 : · · · : x7 ] 7→ [x1 + x6 + f5 (x) : x2 + x3 + f4 (x) : x3 : x4 : x5 : x6 : x7 ] (15)

with f4 (x) = x3 x5 + x7 x4 and f5 (x) = x4 x6 + x7 x5 , which indeed establishes a different

type of embedding of H into Q, called skew; its projection to W(5, 2) will be denoted by
HS . There are altogether 7560 copies of HS in W(5, 2); one of them is depicted in Figure
7.
XZY

IZZ ZII YIZ

XZI ZYX
YXX XZX XII XYX
IZI IIY
XIX IYX
IXX ZIX
IYY YYY XIY
XXI YXZ YIY IYI ZYI
IZX
ZYY IYZ
YII XIZ
ZIZ
YZY YIX
YZX XYZ
XYI ZXZ
YXY YYZ YZZ IIX
IXI
ZXX ZXY ZIY
IXY ZXI
XYY IXZ
ZZY YZI XZZ YYX IIZ XXY

ZYZ XXZ ZZZ ZZI

YYI

ZZX IZY

YXI XXX

Figure 7: A copy of the split Cayley hexagon of order two that is skew-embedded into
W(5, 2).

8
 We shall conclude this section by a brief description of the main difference between the
two embeddings. In HC , all the three lines passing through any of its points are coplanar,
i. e. lie in the same plane of W(5, 2). This is, however, not the case for HS , where only
15 points exhibit this property; for each of the remaining 48 points, only two of the three
lines are coplanar. To illustrate this, let us take the point XXX of the copy shown in
Figure 7. The three lines passing through it are {XXX, XY Y, IZZ}, {XXX, ZZI, Y Y X}
and {XXX, IY Z, XZY }; clearly, only the first two lie in the same plane. Next, take any
observable and all the 30 other observables that commute with it. These 31 observables will
form a geometric hyperplane in both HC and HS . But the two embeddings differ in that
while this hyperplane is of the same type for each point of HC , an HS features two different
kinds of them.

4 HS is contextual, whereas HC is not!

At this point an interesting question arises: Can the two embeddings be ascribed some
physical distinction? To answer this question (in affirmative), we ran several different ex-
0
periments on W(5, 2), HC s, HS0 s and their complements. By the complement, G, of a con-
figuration G ⊂ W(2N − 1, 2) we mean the set of all line-contexts of W(2N − 1, 2) that are
not in G. We tested by computer the contextual nature of all HC , HS , HC , HS of W(5, 2)
and the results of our analysis are summarized in Table 1.

Geometry Contextual # of Copies

HC No 120
HS No 7560
HC No 120
HS Yes 7560

Table 1: Results on the contextuality properties of differently-embedded hexagons and their

complements. The difference between the two embeddings reveals itself in terms of contex-
tuality when one considers the complement of the configuration.

Our calculations were performed with the software Magma and the computation re-
sources of the supercomputer of the Mésocentre de calcul de Franche-Comté. All our codes
and results are available on the QuantCert GitHub page2 . To implement the incidence struc-
tures of HC and HS we first encoded the description in coordinates of the two embeddings
provided by Coolsaet [23] (Eq. (12), Eq. (13) and Eq. (15)). Then we created all 120 copies
of HC and all 7560 copies of HS by repeatedly acting by transvections. Their complements
were easily obtained from the implementation of W(5, 2). Once all incidence structures were
created we checked their contextuality by using the function IsConsistent that establishes
if a given configuration is contextual by employing the procedure described in Section 2.
The IsConsistent Magma intrinsic (compiled function) checks if the rank of the aug-
mented matrix (A|b) is greater that the one of the coefficient matrix A. This operation is
in O(|O| + |C|) operations; in other words, its execution lasts a time proportional to the
number of observables and contexts of the geometry. Given the fact that this precise system
had 315 − 63 = 252 contexts, the duration of this operation is quite negligible, even repeated
several thousands of times. Building the 7560 copies of HS was more intensive though: to
reach all copies we started with a single copy of HS and ran every possible combination of
4 transvections, including the identity, for a total of 644 = 22 4 ≈ 17E6 total operations.
This computation lasted around 11 hours (10.87 to be more precise), which speaks for the
efficiency of Magma.
2 URL: https://quantcert.github.io/Magma-contextuality/

9
5 Conclusion
We have found a very important physical property regarding the two non-equivalent em-
beddings of a very special subconfiguration of the symplectic polar space W(5, 2), the split
Cayley hexagon of order two. Using the interpretation in terms of three-qubit Pauli ob-
servables we showed that the complement of any skew-embedded hexagon is a contextual
configuration, i. e. provides a proof of the Kochen-Specker Theorem. We tested our find-
ings on all possible embeddings of the split Cayley hexagons and found out that only skew
embeddings enjoy this property. To extend this work one may try to measure the degree
of contextuality of HS . The degree of contextuality indicates how far is a given contextual
configuration to be satisfiable, i.e. non-contextual, if one changes the constrains imposed by
the vector b. In other words, if |J| is the number of contexts and P the maximum number
of constraints that can be satisfied, then the degree of contextuality is d = |J| − P . For a
non-contextual configuration d = 0. The calculation of d boils down to finding the Hamming
distance between b and the image of A, the incidence matrix of the configuration. However,
in the case of the split Cayley hexagon a brute force calculation to compute d is out of reach
of the supercomputer resources we have currently at our disposal. It, therefore, requires a
deeper understanding of the geometry to reduce the calculation cost. The degree of contex-
tuality d is also necessary for calculating the classical bound of the contextual inequalities
given in [13] as well as for a possible testing of these inequalities on a quantum computer
[15].

6 Acknowledgements
This work was supported by the Conseil Régional de Bourgogne Franche-Comté, mobility
grant GeoQuant, the Thomas Jefferson Fundation and by the Slovak VEGA Grant Agency,
Project # 2/0004/20. Part of this work was done during a visit of HdB and a one-year stay of
FH at the Mathematics and Statistics Department of Auburn University. Both authors wish
to acknowledge the people of the department that made our work here possible, in particular
Dr. Luke Oeding for his precious help and friendship as well as Drs. Gaetan Bakalli and
Mucyo Karemera for their hospitality and enthusiastic discussions about mathematics and
life.

References
[1] Howard, M., Wallman, J., Veitch, V., & Emerson, J. (2014). Contextuality supplies the
‘magic’for quantum computation. Nature, 510(7505), 351-355.
[2] Bermejo-Vega, J., Delfosse, N., Browne, D. E., Okay, C., & Raussendorf, R. (2017).
Contextuality as a resource for models of quantum computation with qubits. Physical
review letters, 119(12), 120505.
[3] Lillystone, P., Wallman, J. J., & Emerson, J. (2019). Contextuality and the single-qubit
stabilizer subtheory. Physical review letters, 122(14), 140405.
[4] Shahandeh, F. (2021). Quantum computational advantage implies contextuality. arXiv
preprint arXiv:2112.00024.
[5] Kochen, S., & Specker, E. P. (1975). The problem of hidden variables in quantum me-
chanics. In The logico-algebraic approach to quantum mechanics (pp. 293-328). Springer,
Dordrecht.
[6] Lisoněk, P., Badziag, P., Portillo, J. R., & Cabello, A. (2014). Kochen-Specker set with
seven contexts. Physical Review A, 89(4), 042101.
[7] Cabello, A., Kleinmann, M., & Portillo, J. R. (2016). Quantum state-independent con-
textuality requires 13 rays. Journal of Physics A: Mathematical and Theoretical, 49(38),
38LT01.

10
[8] Bartosik, H., Klepp, J., Schmitzer, C., Sponar, S., Cabello, A., Rauch, H., & Hasegawa,
Y. (2009). Experimental test of quantum contextuality in neutron interferometry. Physical
review letters, 103(4), 040403.
[9] D’Ambrosio, V., Herbauts, I., Amselem, E., Nagali, E., Bourennane, M., Sciarrino, F.,
& Cabello, A. (2013). Experimental implementation of a Kochen-Specker set of quantum
tests. Physical Review X, 3(1), 011012.
[10] Budroni, C., Cabello, A., Gühne, O., Kleinmann, M., & Larsson, J. Ǎ. (2021). Quantum
contextuality. arXiv preprint arXiv:2102.13036.
[11] Mermin, N. D. (1993). Hidden variables and the two theorems of John Bell. Reviews of
Modern Physics, 65(3), 803.
[12] Peres, A. (1990). Incompatible results of quantum measurements. Physics Letters A,
151(3-4), 107-108.
[13] Cabello, A. (2010). Proposed test of macroscopic quantum contextuality. Physical Re-
view A, 82(3), 032110.
[14] Dikme, A., Reichel, N., Laghaout, A., & Björk, G. (2020). Measuring the Mermin-Peres
magic square using an online quantum computer. arXiv preprint arXiv:2009.10751.
[15] Holweck, F. (2021). Testing quantum contextuality of binary symplectic polar spaces on
a Noisy Intermediate Scale Quantum Computer. Quantum Information Processing 7/2021

[16] Saniga, M., & Planat, M. (2007). Multiple Qubits as Symplectic Polar Spaces of Order
Two. Advanced Studies in Theoretical Physics, 1, 1-4.
[17] Thas, K. (2009). The geometry of generalized Pauli operators of N-qudit Hilbert space,
and an application to MUBs. EPL (Europhysics Letters), 86(6), 60005.

[18] Havlicek, H., Odehnal, B., & Saniga, M. (2009). Factor-group-generated polar spaces
and (multi-) qudits. SIGMA. Symmetry, Integrability and Geometry: Methods and Ap-
plications, 5, 096.
[19] Saniga, M., Planat, M., & Rosu, H. (2004). Mutually unbiased bases and finite projec-
tive planes. Journal of Optics B: Quantum and Semiclassical Optics, 6(9), L19.

[20] Lévay, P., Saniga, M., Vrana, P., & Pracna, P. (2009). Black hole entropy and finite
geometry. Physical Review D, 79(8), 084036.
[21] Lévay, P., & Holweck, F. (2019). Finite geometric toy model of spacetime as an error
correcting code. Physical Review D, 99(8), 086015.

[22] Lévay, P., Saniga, M., & Vrana, P. (2008). Three-qubit operators, the split Cayley
hexagon of order two, and black holes. Physical Review D, 78(12), 124022.
[23] Coolsaet, K. (2010). The smallest split Cayley hexagon has two symplectic embeddings.
Finite Fields and Their Applications, 16(5), 380-384.

[24] Holweck, F., & Saniga, M. (2017). Contextuality with a small number of observables.
International Journal of Quantum Information, 15(04), 1750026.
[25] Saniga, M., Planat, M., Pracna, P., & Havlicek, H. (2007). The Veldkamp space of
two-qubits. SIGMA. Symmetry, Integrability and Geometry: Methods and Applications,
3, 075.

[26] Lévay, P., Holweck, F., & Saniga, M. (2017). Magic three-qubit Veldkamp line: A finite
geometric underpinning for form theories of gravity and black hole entropy. Physical
Review D, 96(2), 026018.

11
 [27] Planat, M., Saniga, M., & Holweck, F. (2013). Distinguished three-qubit ‘magicity’via
automorphisms of the split Cayley hexagon. Quantum information processing, 12(7), 2535-
2549.
[28] Lévay, P., Planat, M., & Saniga, M. (2013). Grassmannian connection between three-
and four-qubit observables, Mermin’s contextuality and black holes. Journal of High En-
ergy Physics, 2013(9), 1-35.
[29] de Boutray, H., Holweck, F., Giorgetti, A., & Masson, P. A. (2021). Automated syn-
thesis of contextuality proofs from subspaces of symplectic polar spaces. arXiv preprint
arXiv:2105.13798.

[30] Maldeghem, van H., Generalized Polygons, Monographs in Mathematics 93, Birkhäuser,
Basel, 1998.
[31] Schroth, A. E. (1999). How to draw a hexagon. Discrete Math, 199, 161-171.
[32] Saniga, M., Planat, M., Pracna, P., & Lévay, P. (2012). ’Magic’Configurations of Three-
Qubit Observables and Geometric Hyperplanes of the Smallest split Cayley hexagon.
SIGMA. Symmetry, Integrability and Geometry: Methods and Applications, 8, 083.
[33] Polster, B., Schroth, A. E., & Van Maldeghem, H. (2001). Generalized flatland. The
Mathematical Intelligencer, 23(4), 33-47.

12
View publication stats