Analysis of two-player quantum games using geometric algebra

James M. Chappell*,a,b , Azhar Iqbala,c and Derek Abbotta

a
School of Electrical and Electronic Engineering,
University of Adelaide 5005, Australia
b
School of Chemistry and Physics, University of Adelaide, South Australia 5005, Australia
arXiv:1007.1332v2 [quant-ph] 11 Aug 2010

c
Centre for Advanced Mathematics and Physics,
National University of Sciences & Technology,
Peshawar Road, Rawalpindi, Pakistan
∗
Email: james.m.chappell@adelaide.edu.au
(Dated: August 12, 2010)

Abstract
The framework for playing quantum games in an Einstein-Podolsky-Rosen (EPR) type setting
is investigated using the mathematical formalism of geometric algebra (GA). In this setting, the
players’ strategy sets remain identical to the ones in the classical mixed-strategy version of the
game, which is then obtained as proper subset of the corresponding quantum game. As examples,
using GA we analyze the games of Prisoners’ Dilemma and Stag Hunt when played in the EPR
type setting.

PACS numbers: 03.67.Lx

1
I. INTRODUCTION

Although its origins can be traced to earlier works [1–4], the extension of game theory
[5, 6] to the quantum regime [7] was proposed by Meyer [8] and Eisert et al [9] and have since
been investigated by others [10–46]. Game theory is a vast subject but many interesting
games can still be found in simple-to-analyze two-player two-strategy non-cooperative games.
The well known games of Prisoners’ Dilemma (PD) and Stag Hunt [5, 6] are such examples.
The general idea in the quantization scheme proposed by Eisert et al [9] for such games
involves a referee who forwards a two-qubit entangled state to the two players. Players
perform their strategic actions on the state that consist of local unitary transformations to
their respective qubits. The qubits are subsequently returned to the referee for measurement
from which the players’ payoffs are determined. The setup ensures that players sharing a
product initial state corresponds to the mixed-strategy version of the considered classical
game. However, players sharing an entangled state can lead to new Nash equilibria (NE)
[5, 6] consisting of pairs of unitary transformations [7, 9]. At these quantum NE the players
can have higher payoffs relative to what they obtain at the NE in the mixed-strategy version
of the classical game.
This approach to constructing quantum games was subsequently criticized [12] as follows.
The players’ strategic actions in the quantum game are extended operations relative to their
actions in the original mixed-strategy version of the classical game, in which, each player
can perform a strategic action consisting of a probabilistic combination of their two pure
strategies. The mentioned criticism [12] argued that as the quantum players have expanded
strategy sets and can do more than what the classical players can do, it is plausible to
represent the quantum game as an extended classical game that also involves new pure
strategies. The entries in the extended game matrix can then be suitably chosen so to be
representative of the players’ payoffs at the obtained quantum NE. This line of reasoning can
be extended further in stating that quantum games are in fact ‘disguised’ classical games
and to quantize a game is equivalent to replacing the original game by an extended classical
game.
Generally speaking, quantum games are constructed as continuous variable games and
are to be understood as proper extensions of the classical mixed-strategy games. When we
accept this as their basic character, and observe the structure of the payoff relations in mixed-

2
strategy versions of the corresponding extended classical game, we are led to conclude that
this quantum game structure cannot be identical to the payoff structure found in extended
classical games. It therefore, appears to us, that simply adding new classical strategies
cannot be claimed to capture the essence of a quantum game.
As an alternative way to counter the criticism in [12], two-party Einstein-Podolsky-Rosen
(EPR) type experiments [47–54] are recognized to have genuinely quantum features. One
observes that the setting of such experiments can be fruitfully adapted [25, 28, 34, 41, 44]
for playing a quantum version of a two-player two-strategy game, which allows us to avoid
the criticism from another perspective. In particular, with the EPR type setting the players’
strategies can be defined entirely classically—consisting of a probabilistic combination of a
player’s choice between two measurement directions. That is, with this setting, the players’
strategy sets remain identical to ones they have in a standard arrangement for playing a
mixed-strategy version of a classical two-player two-strategy game. As the players’ strategy
sets in the quantum game are not extended relative to the classical game, for this route to
constructing quantum games, the mentioned criticism [12] does not apply.
The usefullness of applying the formalism of geometric algebra (GA) [55–59] in the inves-
tigation of quantum games has recently been shown [45] for the well known quantum penny
flip game [8]. In the present paper, we use GA to analyze two-player two-strategy quantum
games that are constructed using the EPR type setting. The constraints to be imposed on
the parameters of EPR type arrangements are obtained that ensure a faithful embedding of
the mixed-strategy version of the original classical game within the corresponding quantum
game. In particular, we show how using GA we can determine new NE in quantum games
of Stag Hunt and Prisoners’ Dilemma played in the EPR type setting.

II. EPR SETTING FOR PLAYING A QUANTUM GAME

We have the following payoff matrices

Bob Bob
S1′ S2′ S1′ S2′
   
S1 G00 G01 S1 H001 H01
A = Alice  , B = Alice  , (1)
S2 G10 G11 S2 H10 H11

3
giving Alice’s and Bob’s payoffs, respectively. Here Alice’s pure strategies are S1 and S2
and Bob’s pure strategies are S1′ and S2′ . In a run, Alice chooses her strategy to be either
S1 or S2 and likewise, in the same run, Bob chooses his strategy to be either S1′ or S2′ . We
consider games with symmetrical payoffs for which B = AT , where T indicates transpose.
This requires H00 = G00 , H01 = G10 , H10 = G01 , and H11 = G11 .
The EPR setting assumes that players Alice and Bob are spatially-separated participants,
who are located at the two arms of the EPR system. In a run, each player receives one
half of a two-particle system emitted by the same source. We associate Alice’s strategies
S1 , S2 to the directions κ11 , κ12 respectively and similarly, associate Bob’s strategies S1′ , S2′ to
the directions κ21 , κ22 , respectively. On receiving a pair of particles, players Alice and Bob
together choose a pair of directions from the four possible cases (κ11 , κ21 ), (κ11 , κ22 ), (κ12 , κ21 ),
(κ12 , κ22 ) and a quantum measurement is performed along the chosen pair. The outcome of the
measurement at either arm is +1 or −1. Over a large number of runs, a record is maintained
of the players’ choices of directions, representing their strategies, and one of the four possible
outcomes (+1, +1), (+1, −1), (−1, +1), (−1, −1) emerging out of the measurement. Within
each of the brackets, the first entry is reserved for the outcome at Alice’s side and the second
entry for the outcome at Bob’s side. Players’ payoff relations are expressed in terms of the
outcomes of measurements that are recorded for a large number of runs, as the players
sequentially receive, two-particle systems emitted from the source. These payoffs depend on
the strategic choices that each player adapts for his/her two directions over many runs, and
on the dichotomic outcomes of the measurements performed along those directions.

III. GEOMETRIC ALGEBRA

Geometric algebra (GA) [55–59] is an associative non-commutative algebra, that can

provide an equivalent description to the conventional Dirac bra-ket and matrix formalisms
of quantum mechanics. Recently, Christian [60] has used the formalism of GA in exciting
and penetrating investigations of some of the foundational questions in quantum mechanics.
In the area of quantum games, GA has been used by Chappell et al [45] to determine all
possible unitary transformations which enable the player Q to implement a winning strategy
in Meyer’s PQ penny flip quantum game [8].
Given a linear vector space V with elements u, v, . . . we may form [61] the tensor product

4
U ⊗ V of vector spaces U, V , containing elements (bivectors) u ⊗ v and hence construct
the exterior or wedge product u ∧ v = u ⊗ v − v ⊗ u. This may be extended to a vector
space Λ(V ) with elements consisting of multivectors that can be multiplied by means of the
exterior product. The geometric product uv of two vectors u, v is defined by uv = u.v +u∧v,
where u.v is the scalar inner product. The geometric product is in general not commutative
but it is always associative, that is u(vw) = (uv)w.
Denote by {σ i } an orthonormal basis in ℜ3 , then σ i · σ j = δ ij . We also have σ i ∧ σ i = 0
for each i = 1, 2, 3 and so in terms of the geometric product we have σ 2i = σ i σ i = 1, and
σ i σ j = σ i ∧ σ j = −σ j σ i for each i 6= j. Hence the basis vectors anticommute with respect
to the geometric product. If we denote by ι the trivector

ι = σ1 σ2σ3 , (2)

then for distinct basis vectors we have

σ i σ j = δ ij + ιǫijk σ k , (3)

where ǫijk is the Levi-Civita symbol. We find that ι2 = σ 1 σ 2 σ 3 σ 1 σ 2 σ 3 = σ 1 σ 2 σ 1 σ 2 = −1 and

commutes with all other elements and so has identical properties to the conventional complex
√
number i = −1. Thus we have an isomorphism between the basis vectors σ 1 , σ2 , σ 3 and
the Pauli matrices through the use of the geometric product.
In order to express quantum states in GA we use the one-to-one mapping [57–59] defined
as follows
 
a0 + ia3
|ψi = α|0i + β|1i =   ↔ ψ = a0 + a1 ισ 1 + a2 ισ 2 + a3 ισ 3 , (4)
−a2 + ia1

where ai are real scalars.

It can then be shown using the Schmidt decomposition of a general two qubit state, that
a general two-particle state can be represented (see Appendix) in GA as
γ γ
ψ = AB(cos + sin ισ 12 ισ 22 ), (5)
2 2
where γ ∈ [0, π2 ] is a measure of the entanglement and where A, B are single particle rotors
applied to the first and second qubit, respectively. General unitary operations are called
[57] rotors in GA, represented as

R(θ1 , θ2 , θ3 ) = e−θ3 ισ3 /2 e−θ1 ισ2 /2 e−θ2 ισ3 /2 . (6)

5
This rotation, in Euler angle form, can completely explore the available space of a single
qubit, and is equivalent to a general unitary transformation acting on a spinor. So, we have
the rotors for each qubit defined as

A = R(α1 , α2 , α3 ) = e−α3 ισ3 /2 e−α1 ισ2 /2 e−α2 ισ3 /2 , (7)

B = R(β 1 , β 2 , β 3 ) = e−β 3 ισ3 /2 e−β 1 ισ2 /2 e−β 2 ισ3 /2 . (8)

For example, for A = B = 1 and γ = π2 , we find the Bell state, and A = 1 and B = R(π, 0, 0)
π
and γ = 2
we recover the singlet state. This can be checked using Eq. (4), where we note
that −ισ 2 → |1i.
To simulate the process of measurement in GA, we form a separable state φ = RS, where
R and S are single particle rotors, which allow general measurement directions to be speci-
fied, on the first and second qubit respectively. The state to be measured is now projected
onto the separable state φ. In the N-particle case, the probability that the quantum state
ψ returns the separable state φ is given is Ref. [57] as

P (ψ, φ) = 2N −2 hψEψ † φEφ† i0 − hψJψ † φJφ† i0 ,

(9)

where the angle brackets hi0 mean to retain only the scalar part of the expression. We have
the two observables ψJψ † and ψEψ † , which in the two particle case involves [57]

1 1
E = (1 − ισ 13 ισ 23 ), J = (ισ 13 + ισ 23 ). (10)
2 2

The † operator acts the same as complex conjugation, flipping the sign of ι and inverting
the order of terms.

A. Calculating observables

Employing Eq. (9), we firstly calculate

1 γ γ γ γ
ψEψ † = AB(cos + sin ισ 12 ισ 22 )(1 − ισ 13 ισ 23 )(cos + sin ισ 12 ισ 22 )B † A†
2 2 2 2 2
1
= AB 1 − ισ 13 ισ 23 + sin γ(ισ 12 ισ 22 − ισ 11 ισ 21 ) B † A†


2
1
1 − ιAσ 13 A† ιBσ 23 B † + sin γ(ιAσ 12 A† ιBσ 22 B † − ιAσ 11 A† ιBσ 21 B † )


= (11)
2

6
and
1 γ γ γ γ
ψJψ † = AB(cos + sin ισ 12 ισ 22 )(ισ 13 + ισ 23 )(cos + sin ισ 12 ισ 22 )B † A†
2 2 2 2 2
1 γ γ
= AB(cos2 − sin2 )(ισ 13 + ισ 23 )B † A†
2 2 2
1
= cos γ(ιAσ 13 A† + ιBσ 23 B † ). (12)
2
1 σ1 2 σ2
To describe the players measurement directions, we have R = e−ικ 2 and S = e−ικ 2 . For
the quantum game in the EPR setting, κ1 can be either of Alice’s two directions i.e. κ11 or
κ12 . Similarly, in the expression for S the κ2 can be either of Bob’s two directions i.e. κ21 or
κ22 . Hence we obtain

φJφ† = RSJS † R†
1
= (ιRσ 13 R† + ιSσ 23 S † )
2
1 1 1 2 2
= (ισ 13 eικ σ2 + ισ 23 eικ σ2 ), (13)
2
and

φEφ† = RSES † R†
1
= (1 − ιRσ 13 R† ιSσ 23 S † )
2
1 1 1 2 2
= (1 − ισ 13 eικ σ2 ισ 23 eικ σ2 ). (14)
2
Now from Eq. (9), we calculate
1 1 1 2 2
−hψJψ † φJφ† i0 = − hcos γ(ιAσ 13 A† + ιBσ 23 B † )(ισ 13 eικ σ2 + ισ 23 eικ σ2 )i0
4
1
= cos γ[(−)m X(κ1 ) + (−)n Y (κ2 )], (15)
4
where m, n ∈ {0, 1} refers to measuring a |0i or a |1i state, respectively, and using the
results in Appendix, we have

X(κ1 ) = cos α1 cos κ1 + cos α3 sin α1 sin κ1 , (16)

Y (κ2 ) = cos β 1 cos κ2 + cos β 3 sin β 1 sin κ2 . (17)

Also, from Eq. (9) we obtain

hψEψ † φEφ† i0 = h(1 − ιAσ 13 A† ιBσ 23 B † + sin γ(ιAσ 12 A† ιBσ 22 B † − ιAσ 11 A† ιBσ 21 B † ))
1 2
×(1 − ισ 13 ισ 23 eικσ2 eιτ σ2 )i0
1
= [1 + (−)m+n XY − (−)m+n sin γ{U(k 1 )V (k 2 ) − F (k 1 )G(k 2 )}], (18)
4

7
where

F (κ1 ) = cos α2 (cos κ1 sin α1 − cos α3 sin κ1 cos α1 ) + sin κ1 sin α2 sin α3 , (19)
G(κ2 ) = cos β 2 (cos κ2 sin β 1 − cos β 3 sin κ2 cos β 1 ) + sin κ2 sin β 2 sin β 3 (20)

and

U(κ1 ) = − sin α2 (cos κ1 sin α1 − cos α3 sin κ1 cos α1 ) + sin κ1 cos α2 sin α3 , (21)
V (κ2 ) = − sin β 2 (cos κ2 sin β 1 − cos β 3 sin κ2 cos β 1 ) + sin κ2 cos β 2 sin β 3 . (22)

To simplify the equations, we define

Z(κ1 , κ2 ) = F (k 1 )G(k 2 ) − U(k 1 )V (k 2 ) (23)

= cos φ[cos κ1 cos κ2 sin β 1 sin α1 − sin κ1 cos κ2 sin β 1 cos α1 cos α3
+ sin κ1 sin κ2 (cos α1 cos α3 cos β 1 cos β 3 − sin α3 sin β 3 )
− cos κ1 sin κ2 sin α1 cos β 1 cos β 3 ]
+ sin φ[sin κ1 cos κ2 sin α3 sin β 1 + cos κ1 sin κ2 sin α1 sin β 3

+ sin κ1 sin κ2 (cos β 1 cos β 3 sin α3 + cos α1 cos α3 sin β 3 )]. (24)

Now combining Eq. (15) and Eq. (18) we have the probability to observe a particular state

1
Pmn = [1 + cos γ{(−)m Xi + (−)n Yj } + (−)m+n (Xi Yj + sin γZij )]. (25)
4

To simplify notation we have written Zij = Z(κ1i , κ2j ) , Xi = X(κ1i ) and Yj = Y (κ2j ), where
i, j ∈ {1, 2} represent the two possible measurement directions available to each player. We
notice that we were able to make the substitution φ = α2 + β 2 in Eq. (24), which as shown
in Appendix, is as expected from the known redundancy in the rotors. If we put γ = 0, that
is, for no entanglement, we have the probability

1
Pmn = (1 + (−)m Xi + (−)n Yj + (−)m+n Xi Yj )
4
(1 + (−)m Xi )1 (1 + (−)n Yj )2
= , (26)
2 2

which shows a product state incorporating general measurement directions for each qubit.

8
 Writing out the probabilities for the four measurement outcomes we find
1
P00 (κ1i , κ2j ) = [1 + cos γ(Xi + Yj ) + (Xi Yj + sin γZij )], (27)
4
1 2 1
P01 (κi , κj ) = [1 + cos γ(Xi − Yj ) − (Xi Yj + sin γZij )], (28)
4
1 2 1
P10 (κi , κj ) = [1 + cos γ(−Xi + Yj ) − (Xi Yj + sin γZij )], (29)
4
1 2 1
P11 (κi , κj ) = [1 + cos γ(−Xi − Yj ) + (Xi Yj + sin γZij )]. (30)
4

B. Finding the payoff relations

We allow each player the classical probabilistic choice between their two chosen measure-
ment directions for their Stern-Gerlach detectors. The two players, Alice and Bob choose
their first measurement direction with probability x and y respectively, where x, y ∈ [0, 1].
Now, we have the mathematical expectation of Alice’s payoff, where she chooses the direction
κ11 with probability x and the measurement direction κ12 with probability 1 − x, as

ΠA (x, y) = xy[P00 (κ11 , κ21 )G00 + P01 (κ11 , κ21 )G01 + P10 (κ11 , κ21 )G10 + P11 (κ11 , κ21 )G11 ]

+ x(1 − y)[P00 (κ11 , κ22 )G00 + P01 (κ11 , κ22 )G01 + P10 (κ11 , κ22 )G10 + P11 (κ11 , κ22 )G11 ]
+ y(1 − x)[P00 (κ12 , κ21 )G00 + P01 (κ12 , κ21 )G01 + P10 (κ12 , κ21 )G10 + P11 (κ12 , κ21 )G11 ]
+ (1 − x)(1 − y)[P00 (κ12 , κ22 )G00 + P01 (κ12 , κ22 )G01 + P10 (κ12 , κ22 )G10 + P11 (κ12 , κ22 )G11 ],
(31)

using the payoff matrix, defined for Alice, in Eq. (1). We also define

∆1 = G10 − G00 , ∆2 = G11 − G01 , ∆3 = ∆2 − ∆1 , (32)

so that we find the payoff for Alice

1
ΠA (x, y) = [G00 + G10 + G01 + G11
4
+ ∆3 {x((X1 − X2 )Y2 + (Z12 − Z22 ) sin γ) + y((Y1 − Y2 )X2 + (Z21 − Z22 ) sin γ)
+ xy{(X1 − X2 )(Y1 − Y2 ) + sin γ(Z11 + Z22 − Z12 − Z21 )} + X2 Y2 + Z22 sin γ}

− cos γ{(∆1 + ∆2 )((X1 − X2 )x + X2 ) − (G00 − G01 + G10 − G11 )((Y1 − Y2 )y + Y2 )}].

(33)

Bob’s payoff can be easily deduced by symmetry.

9
 C. Solving the general two-player game

We now find the optimal solutions by calculating the Nash equilibrium (NE), that is, the
expected response assuming rational self interest. To find the NE we simply require

ΠA (x∗ , y ∗) ≥ ΠA (x, y ∗ ), ΠB (x∗ , y ∗) ≥ ΠB (x∗ , y), (34)

which is stating that any unilateral movement of a player away from the NE of (x∗ , y ∗), will
result in a lower payoff for that player. We find

ΠA (x∗ , y ∗) − ΠA (x, y ∗)
1
= (x∗ − x)[∆3 {y ∗ ((X1 − X2 )(Y1 − Y2 ) + sin γ(Z11 + Z22 − Z12 − Z21 ))
4
+ (X1 − X2 )Y2 + (Z12 − Z22 ) sin γ} − cos γ(∆1 + ∆2 )(X1 − X2 )] (35)

and for the second player Bob we have similarly

ΠB (x∗ , y ∗ ) − ΠB (x∗ , y)
1
= (y ∗ − y)[∆3 {x∗ ((X1 − X2 )(Y1 − Y2 ) + sin γ(Z11 + Z22 − Z12 − Z21 ))
4
+ (Y1 − Y2 )X2 + (Z21 − Z22 ) sin γ} − cos γ(∆1 + ∆2 )(Y1 − Y2 )]. (36)

D. Embedding the classical game

To embed the classical game, we require at zero entanglement, not only the same pair of
strategies being a NE but also to have the same bilinear structure of the payoff relations.
At a NE of (x∗ , y ∗) = (0, 0), with zero entanglement, we find the payoff from Eq. (33) to be

1
ΠA (x, y) = [G00 (1 + X2 )(1 + Y2 ) + G10 (1 − X2 )(1 + Y2 )
4
+ G01 (1 + X2 )(1 − Y2 ) + G11 (1 − X2 )(1 − Y2 )]. (37)

This result illustrates how we could select any one of the payoff entries we desire with the
appropriate selection of X2 and Y2 , however in order to achieve the classical payoff of G11
for this NE, we can see that we require X2 = −1 and Y2 = −1. If we have a game which
also has a classical NE of (x∗ , y ∗) = (1, 1) then from Eq. (33) at zero entanglement we find

10
the payoff
1
ΠA (x, y) = [G00 (1 + X1 )(1 + Y1 ) + G10 (1 − X1 )(1 + Y1 )
4
+ G01 (1 + X1 )(1 − Y1 ) + G11 (1 − X1 )(1 − Y1 )]. (38)

So, we can see, that we can select the required classical payoff, of G00 , by the selection of
X1 = 1 and Y1 = 1.
Referring to Eq. (17), we then have the conditions

X(κ1 ) = cos α1 cos κ1 + cos α3 sin α1 sin κ1 = ±1, (39)

Y (κ2 ) = cos β 1 cos κ2 + cos β 3 sin β 1 sin κ2 = ±1. (40)

Looking at the equation for Alice, we have two classes of solution: If α3 6= 0, then for the
equations satisfying X2 = Y2 = −1, we have for Alice in the first equation α1 = 0, κ12 = π
or α1 = π, κ12 = 0 and for the equations satisfying X1 = Y1 = +1, we have α1 = κ11 = 0 or
α1 = κ11 = π, which can be combined to give either α1 = 0, κ11 = 0 and κ12 = π or α1 = π,
κ11 = π and κ12 = 0. For the second class with α3 = 0, we have the solution α1 − κ12 = π and
for X1 = Y1 = +1 we have α1 − κ11 = 0. So, in summary, for both cases we have that the two
measurement directions are π out of phase with each other, and for the first case (α3 6= 0)
we can freely vary α2 and α3 , and for the second case (α3 = 0), we can freely vary α1 and α2
to change the initial quantum quantum state without affecting the game NE or payoffs. The
same arguments hold for the equations for Y . Combining these results and substituting into
Eq. (24), we see by inspection for the two cases that F (κ1 ) = G(κ2 ) = U(κ1 ) = V (κ2 ) = 0
and hence, we find that Z22 = Z21 = Z12 = Z11 = 0.
This then reduces the equation governing the NE in Eq. (35) to
1
ΠA (x∗ , y ∗) − ΠA (x, y ∗) = (x∗ − x)[∆3 {2y ∗ − 1} − cos γ(∆1 + ∆2 )] ≥ 0, (41)
2
which now has the new quantum behavior governed solely by the entanglement angle γ. We
have the associated payoffs
1
ΠA (x, y) = [G00 + G11 − cos γ(G00 − G11 ) + 2xy∆3
2
− x{∆3 + cos γ(∆1 + ∆2 )} − y{∆3 − cos γ(G00 − G01 + G10 − G11 )}]. (42)

Setting γ = 0 in Eq. (42) we find the payoff

ΠA (x, y) = G11 + x(G01 − G11 ) + y(G10 − G11 ) + xy(G00 − G01 − G10 + G11 ), (43)

11
which is the required classical payoff structure. Hence we have faithfully embedded the
classical game inside a quantum version of the game, when the entanglement goes to zero.
We also have the probabilities for each state |mi|ni, after measurement from Eq. (25),
for this form of the quantum game as

1
(Pmn )ij = [1 + cos γ((−)m+i+1 + (−)n+j+1 ) + (−)m+n+i+j ], (44)
4

for the two measurement directions i and j.

IV. EXAMPLES

Here we explore the above results for the games of Stag Hunt and Prisoners’ Dilemma.
The quantum versions of these games are discussed in Refs. [9, 11, 19, 20, 24, 43].
A game with two NE is the Stag Hunt game defined by the conditions ∆3 > ∆2 > 0 and
∆1 +∆2 > 0 and ∆3 > ∆1 +∆2 . In the classical mixed-strategy version of this game three NE
appear consisting of (x∗ , y ∗ ) = (0, 0), (x∗ , y ∗) = (1, 1) and (x∗ , y ∗) = ( ∆ 2 ∆2
, ). Checking the
∆3 ∆3

mixed NE for the Stag Hunt game we require from Eq. (41), ∆3 {2y ∗ −1}−cos γ(∆1 +∆2 ) = 0
or
cos γ(∆1 + ∆2 ) + ∆2 − ∆1
y∗ = , (45)
2∆3
∆2
which returns the classical mixed NE of y = ∆3
at zero entanglement. For PD we have
∆1 , ∆2 > 0. For the NE of (x∗ , y ∗) = (0, 0), at maximum entanglement, from Eq. (42) we
have the payoff
1
ΠA (x, y) = ΠB (x, y) = (G00 + G11 ), (46)
2
that gives a significant improvement over the classical payoff of G11 . From Eq. (41), we
notice that the NE of (x∗ , y ∗) = (1, 1) is achievable, provided ∆3 > 0, which is true in a
subset of PD games. We then have a phase transition at ∆3 − cos γ(∆1 + ∆2 ) = 0 or

∆3 ∆2 − ∆1
cos γ = = . (47)
∆1 + ∆2 ∆2 + ∆1

As ∆1 , ∆2 > 0, a phase transition is guaranteed for the case ∆3 > 0. Consider a particular
example of PD by taking G00 = 3 = H00 , G01 = 0 = H10 , G10 = 4 = H01 , and G11 = 2 = H11
in matrices (1). From (32) we find ∆1 = 1, ∆2 = 2 and ∆3 = 1 and we obtain γ = cos−1 (1/3),
for a transition to the new NE.

12
V. CONCLUDING REMARKS

The EPR type setting for playing a quantum version of a two-player two-strategy game is
explored using the formalism of geometric algebra (GA), used for the representation of the
quantum states, and the calculation of observables. It is shown that a GA based analysis
provides a faithful embedding of a symmetric mixed-strategy version of a classical two-player
two-strategy game into its quantum version that is constructed using an EPR type setting.
For a general two-player two-strategy game, we find the governing equation for a strategy
pair forming a NE and the associated payoff relations. We find that at zero entanglement
the quantum game returns the same pair(s) of NE as the classical mixed-strategy game,
while the payoff relations in the quantum game reduce themselves to their bilinear form
corresponding to a mixed-strategy classical game. We find that, within our GA based
analysis, even though the requirement to properly embed a classical game puts constraints
on the possible quantum states allowing this, we still have a degree of freedom, available
with the entanglement angle γ, with which we can generate new NE. As a specific example
the PD was found to have a NE of (x∗ , y ∗) = (1, 1) at high entanglement.

VI. APPENDIX

A. Deriving the general two qubit state representation in GA

The Schmidt decomposition of a general two particle state can be written [59]

1 γ −iφ α1 −iα3 α1 iα3 β −iβ3 β iβ3

|ψi = ρ 2 eiχ [cos e 2 (cos e 2 |0i + sin e 2 |1i) ⊗ (cos 1 e 2 |0i + sin 1 e 2 |1i)
2 2 2 2 2
γ iφ α1 −iα3 α1 iα3 β 1 −iβ3 β 1 iβ3
+ sin e 2 (sin e 2 |0i − cos e 2 |1i) ⊗ (sin e 2 |0i − cos e 2 |1i)]. (48)
2 2 2 2 2

13
We note that the Schmidt decomposition assumes that cos γ ≥ sin γ, hence we need to
enforce the condition γ ∈ [0, π2 ]. Using the mapping defined in Eq. (4), we find that
−iφ α1 −iα3 α1 iα3
e 2 e 2 |0i + sin e 2 |1i)
(cos
2 2
α1 −i(φ+α3 ) α1 i(α3 −φ
= cos e 2 |0i + sin e 2 |1i
2 2
α1 φ + α3 α1 φ + α3 α1 α3 − φ α1 α3 − φ
= (cos cos − i cos sin )|0i + (sin cos + i sin sin )|1i
2 2 2 2 2 2 2 2
α1 φ + α3 α1 φ + α3 α1 α3 − φ α1 α3 − φ
→ cos cos − cos sin ισ 3 − sin cos ισ 2 + sin sin ισ 1
2 2 2 2 2 2 2 2
= e−ια3 σ3 /2 e−ια1 σ2 /2 e−ιφσ3 /2 . (49)

We can then also deduce that

β 1 −iβ3 β iβ3
eiχ (cos e 2 |0i + sin 1 e 2 |1i) → e−ιβ 3 σ3 /2 e−ιβ 1 σ2 /2 eιχσ3 . (50)
2 2
α1 α1
If we then make the substitution 2
→ −( π2 − 2
) we find
iφ α1 −iα3 α1 iα3 π α1
e 2 (sin e 2 |0i − cos e 2 |1i) → −e−ια3 σ3 /2 eι( 2 − 2 )σ2 eιφσ3 /2
2 2
= −e−ια3 σ3 /2 e−ια1 σ2 /2 e−ιφσ3 /2 ισ 12 , (51)

and
β 1 −iβ3 β iβ3
eiχ (sin e 2 |0i − cos 1 e 2 |1i) → −e−ιβ 3 σ3 /2 e−ιβ 1 σ2 /2 ισ 22 eιχσ3 . (52)
2 2
The general state is then represented as
1 γ γ
ψ = ρ 2 e−ια3 σ3 /2 e−ια1 σ2 /2 e−ιφσ3 /2 e−ιβ 3 σ3 /2 e−ιβ 1 σ2 /2 (cos + sin ισ 12 ισ 22 )eιχσ3 . (53)
2 2
We notice the first qubit has an extra degree of freedom available in the applied rotor as
compared to the second qubit, namely, the parameter φ. However, we had a choice where we
associated the complex phase during the derivation. So if we split the phase eiφσ3 /2 between
the two qubits as eiα2 σ3 /2 and eiβ 2 σ3 /2 , we can have symmetrical rotors on each qubit with

A = e−ια3 σ3 /2 e−ια1 σ2 /2 e−ια2 σ3 /2 , B = e−ιβ 3 σ3 /2 e−ιβ 1 σ2 /2 e−ιβ 2 σ3 /2 , (54)

where φ = α2 + β 2 . The general two qubit state can now be written as

1 γ γ
ψ = ρ 2 AB(cos + sin ισ 12 ισ 22 )eιχσ3 E. (55)
2 2
The extra E term on the end of the wave function is included because we have introduced
a second copy of ι for the second particle space, whereas normally the complex number i

14
is common across all particle spaces. We can remove this redundancy with a projection
operator, E = 12 (1 − ισ 13 ισ 23 ). However, when we come to form the observables ψEψ † and
ψJψ † , we notice that because E 2 = E and EJ = JE = J, we see that the inclusion of E, as
well as the global phase term eιχσ3 , will cancel out in both cases and hence is superfluous.
Also for a pure state we have ρ = 1 and so without loss of generality, in regard to the
observables, we can write a general two-particle state in GA, compactly as

γ γ
ψ = AB(cos + sin ισ 12 ισ 22 ). (56)
2 2

We see that with the five degrees of freedom provided with the rotors and also the entan-
glement angle γ, we have defined a two particle quantum state with six degrees of freedom,
as required for a normalized two-particle quantum state ignoring the global phase.

B. Calculating the observables

These three results are useful when calculating observables in GA. We will firstly calculate
the general case with general measurement directions, although we discover, that general
measurement directions, fails to add generality to the equations for NE and payoffs. That
is, in Eq. (58) we only require κ non-zero. We have a rotor defined as

A = e−α3 ισ3 /2 e−α1 ισ2 /2 e−α2 ισ3 /2 , (57)

and for measurement we use

R = e−λισ3 /2 e−κισ2 /2 e−ηισ3 /2 . (58)

We evaluate the quantities ιAσ 1 A† , ιAσ 2 A† , and ιAσ 3 A† as follows.

ιAσ 1 A† = ιe−α3 ισ3 /2 e−α1 ισ2 /2 e−α2 ισ3 /2 σ 1 eα2 ισ3 /2 eα1 ισ2 /2 eα3 ισ3 /2
= e−α3 ισ3 /2 e−α1 ισ2 /2 (cos α2 − sin α2 ισ 3 )e−α1 ισ2 /2 e−α3 ισ3 /2 ισ 1
= e−α3 ισ3 /2 (cos α2 cos α1 − cos α2 sin α1 ισ 2 − sin α2 ισ 3 )e−α3 ισ3 /2 ισ 1
= (cos α1 cos α2 cos α3 − sin α2 sin α3 )ισ 1 − sin α1 cos α2 ισ 3

+ (cos α1 cos α2 sin α3 + sin α2 cos α3 )ισ 2 , (59)

15
 ιAσ 2 A† = ιe−α3 ισ3 /2 e−α1 ισ2 /2 e−α2 ισ3 /2 σ 2 eα2 ισ3 /2 eα1 ισ2 /2 eα3 ισ3 /2
= (cos α2 e−α3 ισ3 − cos α1 sin α2 ισ 3 e−α3 ισ3 − sin α1 sin α2 ισ 1 )ισ 2

= (cos α2 cos α3 − cos α1 sin α2 sin α3 )ισ 2 − (cos α2 sin α3 + cos α1 sin α2 cos α3 )ισ 1
+ sin α1 sin α2 ισ 3 , (60)

ιAσ 3 A† = cos α1 ισ 3 + sin α1 cos α3 ισ 1 + sin α1 sin α3 ισ 2 . (61)

We thus find the following results for completely general measurement directions

hιAσ 3 A† ιRσ 3 R† i0 = − cos α1 cos κ − sin α1 cos α3 sin κ cos λ − sin α1 sin α3 sin κ sin λ
= − cos α1 cos κ − cos(α3 − λ) sin α1 sin κ = −X(κ),
(62)

hιAσ 2 A† ιRσ 3 R† i0 = − cos κ sin α1 sin α2 + sin κ(cos α1 cos(α3 − λ) sin α2 + cos α2 sin(α3 − λ))
= U(κ), (63)

hιAσ 1 A† ιRσ 3 R† i0 = cos α2 (cos κ sin α1 − cos α1 cos(α3 − λ) sin κ1 ) + sin α2 sin κ sin(α3 − λ)
= F (κ). (64)

These expressions being substituted directly to form Eq. (15) and Eq. (18), with λ = 0.

C. When measurement directions are in three dimensions

To give players completely general measurement directions we allow

1 1 1 σ1 1 σ1 2 2 2 σ2 2 σ2
σ3 σ3
R = e−ιλ e−ικ 2 e−ιη 3 , S = e−ιλ e−ικ 2 e−ιη 3 , (65)

which have the form (58) with superscripts 1 and 2 reserved for Alice and Bob, respectively.
We find
1
−hψJψ † φJφ† i0 = − hcos γ(ιAσ 13 A† + ιBσ 23 B † )(ιRσ 13 R† + ιSσ 23 S † )i0
4
1
= cos γ[(−)m X(κ1 ) + (−)n Y (κ2 )], (66)
4

16
where

X(κ1 ) = cos α1 cos κ1 + cos(α3 − λ1 ) sin α1 sin κ1 , (67)

Y (κ2 ) = cos β 1 cos κ2 + cos(β 3 − λ2 ) sin β 1 sin κ2 . (68)

Similarly, we find

hψEψ † φEφ† i0 = h(1 − ιAσ 13 A† ιBσ 23 B † + sin γ(ιAσ 12 A† ιBσ 22 B † − ιAσ 11 A† ιBσ 21 B † ))
× (1 − ιRσ 13 R† ιSσ 23 S † )i0
1
= [1 + (−)m+n XY − sin γ(−)m+n {U(k 1 )V (k 2 ) − F (k 1 )G(k 2 )}], (69)
4
where

F (κ1 ) = cos α2 (cos κ1 sin α1 − cos α1 cos(α3 − λ1 ) sin κ1 ) + sin α2 sin κ sin(α3 − λ1 ),

G(κ2 ) = cos β 2 (cos κ2 sin β 1 − cos β 1 cos(β 3 − λ2 ) sin κ2 ) + sin β 2 sin κ2 sin(β 3 − λ2 )
(70)

and

U(κ1 ) = − sin α2 (cos κ1 sin α1 − sin κ1 cos α1 cos(α3 − λ1 )) + sin κ1 cos α2 sin(α3 − λ1 ),
V (κ2 ) = − sin β 2 (cos κ2 sin β 1 − sin κ2 cos β 1 cos(β 3 − λ2 )) + sin κ2 cos β 2 sin(β 3 − λ2 )
(71)

and Z(κ1 , κ2 ) is then found as before from the Eq. (23). The total probability is calculated
as before from Eq. (25) but now with modified definitions of Xi , Yj and Zij . We now enforce
the classical game with X1 = Y1 = 1 and X2 = Y2 = −1. For the equation for Alice, we have
two classes of solution: If α3 − λ1 6= 0, then for the equations satisfying X2 = Y2 = Z2 = −1
we have for Alice in the first equation α1 = 0, κ12 = π or α1 = π, κ12 = 0 and for the equations
satisfy X1 = Y1 = Z1 = +1 we have α1 = κ11 = 0 or α1 = κ11 = π, which can be combined
to give either α1 = 0, κ11 = 0 and κ12 = π or α1 = π, κ11 = π and κ12 = 0. For the second
class with α3 − λ1 = 0 we have the solution α1 − κ12 = π, and for X1 = Y1 = Z1 = +1
we have α1 − κ11 = 0. Combining these results and substituting into Eq. (23) we find that
Z22 = Z21 = Z12 = Z11 = 0, and so we will recover the same expressions for the NE and
payoffs given in Eq. (41) and Eq. (42), which were calculated for the measurement directions
based on the single parameter κ. This is in accordance with the known result [7] that Bell’s

17
inequalities can still be maximally violated when the allowed directions of measurement,
for the two observers, are located in a single plane as opposed to being defined in three
dimensions.

[1] A. Blaquiere, Wave mechanics as a two-player game, in Dynamical Systems and Microphysics,
33 (1980), Springer-Verlag. Available at the quantum computation archive maintained by
Tom Toffoli: http://pm1.bu.edu/~tt/qcl/pdf/blaquiea198277787a07.pdf
[2] S. Wiesner, Conjugate coding, SIGACT News 15/1, 78 (1983). Avail-
able at the quantum computation archive maintained by Tom Toffoli:
http://pm1.bu.edu/~tt/qcl/pdf/wiesners198316024137.pdf
[3] N.D. Mermin, Am. J. Phys. 58 (1990) 731.
[4] N.D. Mermin, Phys. Rev. Lett. 65 (1990) 1838.
[5] K. Binmore, Game Theory: A Very Short Introduction (Oxford University Press, USA, 2007).
[6] E. Rasmusen, Games & Information: An Introduction to Game Theory (Blackwell Publishers
Ltd., Oxford, 2001) 3rd Edition.
[7] A. Peres, Quantum Theory: Concepts and Methods (Kluwer Academic Publishers, 1995).
[8] D.A. Meyer, Phys. Rev. Lett. 82 (1999) 1052.
[9] J. Eisert, M. Wilkens, and M. Lewenstein, Phy. Rev. Lett. 83 (1999) 3077. See also, J. Eisert
and M. Wilkens, J. Mod. Opt. 47 (2000) 2543.
[10] L. Vaidman, Found.Phys. 29 (1999) 615.
[11] S.C. Benjamin and P. M. Hayden, Phys. Rev. Lett. 87 (2001) 069801.
[12] S.J. van Enk, R. Pike, Phys Rev A 66 (2002), 024306.
[13] N.F. Johnson, Phys. Rev. A 63 (2001) 020302.
[14] L. Marinatto and T. Weber, Phys. Lett. A 272 (2000) 291.
[15] A. Iqbal and A.H. Toor, Phys. Lett. A 280 (2001) 249.
[16] J. Du, H. Li, X. Xu, X. Zhou and R. Han, Phys. Lett. A 302 (2002) 229.
[17] J. Du, H. Li, X. Xu, M. Shi, J. Wu, X. Zhou and R. Han, Phys. Rev. Lett. 88 (2002) 137902.
[18] E.W. Piotrowski and J. Sladkowski, Physica A 312 (2002) 208.
[19] A.P. Flitney and D. Abbott, Proc. Royal Society London A 459 (2003) 2463.
[20] A.P. Flitney and D. Abbott, Fluct. Noise Lett. 2 (2002) R175.

18
[21] A. Iqbal and A.H. Toor, Phys. Rev. A 65 (2002) 022306.
[22] E.W. Piotrowski and J. Sladkowski, Int. J. Theor. Phys. 42 (2003) 1089.
[23] J. Shimamura, S.K. Özdemir, F. Morikoshi, and N. Imoto, Phys. Lett. A 328 (2004) 20.
[24] A.P. Flitney and D. Abbott, J. Phys. A: Math. & Gen. 38 (2005) 449.
[25] A. Iqbal and S. Weigert, J. Phys. A: Math. & Gen. 37 (2004) 5873.
[26] R. Mendes, Quant. Infor. Process., 4 (2005) 1.
[27] T. Cheon and I. Tsutsui, Phys. Lett. A 348 (2006) 147.
[28] A. Iqbal, J. Phys. A: Math. & Gen. 38 (2005) 9551.
[29] A. Nawaz and A.H. Toor, J. Phys. A: Math. & Gen. 37 (2004) 11457.
[30] T. Cheon, in Proceedings of International Conference, ”Quantum Information: Back Action”,
Indian Institute of Technology, Kanpur, Kanpur India, march, 2006, Arxiv: quant-ph/0605134
[31] J. Shimamura, S.K. Özdemir, F. Morikoshi and N. Imoto, Int. J. Quant. Inf. 2 (2004) 79.
[32] T. Ichikawa and I. Tsutsui, Annals of Physics 322 (2007) 531.
[33] S.K. Özdemir, J. Shimamura, and N. Imoto, New J. Phys. 9 (2007) 43.
[34] A. Iqbal and T. Cheon, Phys. Rev. E 76 (2007) 061122.
[35] M. Ramzan and M.K. Khan, J. Phys. A: Math. & Theor. 41 (2008) 435302.
[36] A.P. Flitney and L.C.L. Hollenberg, Phys. Lett. A 363 (2007) 381.
[37] N. Aharon and L. Vaidman, Phys. Rev. A 77 (2008) 052310.
[38] S.A. Bleiler, arXiv:0808.1389 and arXiv: 0808.1391.
[39] H. Guo, J. Zhang and G.J. Koehler, Decision Support Systems 46 (2008) 318.
[40] T. Ichikawa, I. Tsutsui and T. Cheon, J. Phys. A: Math. & Theor. 41 (2008) 135303.
[41] A. Iqbal, T. Cheon and D. Abbott, Phys. Lett. A 372 (2008) 6564.
[42] Q. Li, Y. He and J. Jiang, J. Phys. A: Math. & Theor. 42 (2009) 445303.
[43] A. Iqbal and D. Abbott, Phys. Lett. A 373 (2009) 2537.
[44] A. Iqbal and D. Abbott, J. Phy. Soc. Japan 78 (2009) 014803.
[45] J.M. Chappell, A. Iqbal, M.A. Lohe, L. von Smekal, J. Phys. Soc. Jpn. 78 (2009) 054801.
[46] J.M. Chappell, A. Iqbal and D. Abbott, arXiv:1005.5262
[47] A. Einstein, B. Podolsky and N. Rosen, Phys. Rev. 47 (1935) 777.
[48] D. Bohm, Quantum Theory, (Prentice-Hall, Englewood Cliffs, NJ, 1951), pp. 611–623.
[49] J.S. Bell: Physics 1 (1964) 195.
[50] J.S. Bell, Speakable and Unspeakable in Quantum Mechanics (Cambridge University Press,

19
 Cambridge, 1987).
[51] J.S. Bell, Rev. Mod. Phys. 38 (1966) 447452.
[52] A. Aspect, J. Dalibard, and G. Roger, Phys. Rev. Lett. 49 (1982) 1804.
[53] J.F. Clauser and A. Shimony, Rep. Prog. Phys. 41 (1978) 1881.
[54] J.L. Cereceda, Found. Phys. Lett. 13 (2000) 427.
[55] D. Hestenes, New Foundations for Classical Mechanics, 2nd ed. (Springer Verlag. 1999).
[56] D. Hestenes and G. Sobczyk, Clifford Algebra to Geometric Calculus: A Unified Language for
Mathematics and Physics (Fundamental Theories of Physics). (Springer 1987).
[57] C. Doran and A. Lasenby, Geometric Algebra for Physicists (Cambridge University Press
2003).
[58] V.D. Sabbata and B.K. Datta, Geometric Algebra and Applications to Physics (Taylor &
Francis 2007).
[59] R. Parker and C. Doran, arxiv:quant-ph/0106055.
[60] J. Christian, arXiv:quant-ph/0703179, arXiv:quant-ph/0703244, arXiv:0707.1333,
arXiv:0806.3078, arXiv:0904.4259 and arXiv:1005.4932.
[61] P. Szekeres, A Course in Modern Mathematical Physics: Groups, Hilbert Space and Differential
Geometry (Cambridge University Press, 2004).

20