Exercise: The Correlation Function of the Bell States

Introduction

This article is one of a series of worked exercises that illustrate quantum physics in two parallel presentations: the standard Hilbert-space formulation, and the biquaternion formulation developed in the companion articles. The goal is not to derive new physics, but to give the reader a concrete, computationally explicit demonstration of the correspondence between the two formalisms.

Each exercise in the series is solved twice: first in the standard way, then in the biquaternion way. The results agree, as they must: the two formulations are the same mathematics in different notation. To keep the two solutions strictly parallel, each is presented in the same steps, and the final result is stated in the same form.

The earlier exercises treated a single qubit (Exercises: Measuring Spin Along an Arbitrary Direction, Spin Precession in a Magnetic Field and Successive Measurements of Spin), an entangled pair of qubits in the singlet state (Exercise: Two Spins in the Singlet State), and the reduced state of one subsystem (Exercise: The Reduced State of an Entangled Subsystem). This exercise treats all four Bell states — the four maximally entangled two-qubit states — and the spin correlation function, the quantity that encodes the full dependence of joint measurement outcomes on the two measurement directions.

The correlation function is the object measured in Bell-inequality experiments. It is also the natural bilinear pairing on the tensor-product algebra, and it reveals a clean structural fact: the four Bell states correspond to the four sign patterns $\epsilon_j = \pm 1$ with $\epsilon_1\epsilon_2\epsilon_3 = +1$, and their correlation functions are $E(\vec u_x, \vec u_y) = -\sum_j \epsilon_j\,a_j b_j$.

The conventions are those of the companion articles. The biquaternion algebra is $\mathbb{B} = \mathbb{C}\otimes_\mathbb{R}\mathbb{H}$, with the Hermitian subspace $\mathbb{B}_{\Psi}$ and the anti-Hermitian subspace $\mathbb{B}_{\Phi}$. The quaternion units are $e_t = 1, \vec e_x, \vec e_y, \vec e_z$ with $\vec e_\eta^2 = -e_t$, and $i$ is the scalar imaginary. Twice the scalar part of an element of $\mathbb{B}_{\Psi}$ is the matrix trace of its image: $2\,\mathrm{Sc}(\tilde Q)=\operatorname{tr}\mathsf M_2(\tilde Q)$. The two-qubit state space is the tensor product $\mathbb{B}\otimes\mathbb{B}$, isomorphic to $M_4(\mathbb{C})$, with the trace $\mathrm{Tr}\bigl(\mathsf M_2(x)\otimes\mathsf M_2(y)\bigr) = \mathrm{Tr}_\mathbb{B}\mathsf M_2(x)\cdot\mathrm{Tr}_\mathbb{B}\mathsf M_2(y)$.

The Exercise

Consider the four Bell states of two spin-$\tfrac{1}{2}$ particles:

$$ |Ξ^{\pm}\rangle = \frac{1}{\sqrt{2}}\left(|\uparrow\uparrow\rangle \pm |\downarrow\downarrow\rangle\right), \qquad |\tilde{Ξ}^{\pm}\rangle = \frac{1}{\sqrt{2}}\left(|\uparrow\downarrow\rangle \pm |\downarrow\uparrow\rangle\right). $$

For two measurement directions $\vec u_x$ and $\vec u_y$, define the spin correlation function

$$ E(\vec u_x, \vec u_y) = \langle \left(\vec u_x\cdot\boldsymbol{\sigma}_1\right)\left(\vec u_y\cdot\boldsymbol{\sigma}_2\right)\rangle, $$

where $\boldsymbol{\sigma}_1 = \boldsymbol{\sigma}\otimes I$ acts on particle 1, and $\boldsymbol{\sigma}_2 = I\otimes\boldsymbol{\sigma}$ acts on particle 2.

Compute $E(\vec u_x, \vec u_y)$ for each of the four Bell states. Express the result in the form

$$ E(\vec u_x, \vec u_y) = \sum_{j,\eta=1}^{3} a_j\,T_{j\eta}\,b_\eta = \vec u_x^T T\, \vec u_y, $$

and identify the $3\times 3$ matrix $T$ for each Bell state.

Solution in the Standard Formulation

Step 1: The four Bell states

The four Bell states are the maximally entangled two-qubit states

$$ |Ξ^{\pm}\rangle = \frac{1}{\sqrt{2}}\left(|\uparrow\uparrow\rangle \pm |\downarrow\downarrow\rangle\right), \qquad |\tilde{Ξ}^{\pm}\rangle = \frac{1}{\sqrt{2}}\left(|\uparrow\downarrow\rangle \pm |\downarrow\uparrow\rangle\right). $$

They form an orthonormal basis of the two-qubit Hilbert space $\mathbb{C}^2\otimes\mathbb{C}^2$. The singlet $|\tilde{Ξ}^-\rangle$ was treated in Exercise: Two Spins in the Singlet State; the other three are its companions.

Each Bell state is a rank-one projector. Using the identity

$$ |Ξ^{\pm}\rangle\langleΞ^{\pm}| = \tfrac{1}{4}\left(I\otimes I \pm \sigma_x\otimes\sigma_x \mp \sigma_y\otimes\sigma_y + \sigma_z\otimes\sigma_z\right), $$

$$ |\tilde{Ξ}^{\pm}\rangle\langle\tilde{Ξ}^{\pm}| = \tfrac{1}{4}\left(I\otimes I \pm \sigma_x\otimes\sigma_x \pm \sigma_y\otimes\sigma_y - \sigma_z\otimes\sigma_z\right), $$

we may expand each Bell projector in the Pauli-string basis $\{I\otimes I, \sigma_j\otimes\sigma_\eta\}$. The Pauli strings are trace-orthogonal:

$$ \mathrm{Tr}\!\left[(\sigma_j\otimes\sigma_\eta)(\sigma_l\otimes\sigma_m)\right] = 4\,\delta_{jl}\,\delta_{\eta m}. $$

Step 2: The correlation matrix

The correlation function is the expectation of the tensor-product observable $\vec u_x\cdot\boldsymbol{\sigma}_1\otimes\vec u_y\cdot\boldsymbol{\sigma}_2$:

$$ E(\vec u_x, \vec u_y) = \langle\tilde{Ξ}|\left(\sum_j a_j\sigma_j\otimes I\right)\left(\sum_\eta b_\eta I\otimes\sigma_\eta\right)|\tilde{Ξ}\rangle = \sum_{j,\eta} a_j b_\eta\,\langle\tilde{Ξ}|\sigma_j\otimes\sigma_\eta|\tilde{Ξ}\rangle. $$

So the correlation function is determined by the $3\times 3$ matrix

$$ T_{j\eta} = \langle\tilde{Ξ}|\sigma_j\otimes\sigma_\eta|\tilde{Ξ}\rangle = \mathrm{Tr}\!\left(|\tilde{Ξ}\rangle\langle\tilde{Ξ}|\cdot\sigma_j\otimes\sigma_\eta\right). $$

Because the Pauli strings are trace-orthogonal, the projector admits the expansion

$$ |\tilde{Ξ}\rangle\langle\tilde{Ξ}| = \tfrac{1}{4}\sum_{j,\eta} T_{j\eta}\,\sigma_j\otimes\sigma_\eta, $$

and $T_{j\eta}$ can be read off directly as the coefficient of $\sigma_j\otimes\sigma_\eta$ inside the bracket. The off-diagonal entries vanish because the Bell projectors contain only the diagonal Pauli strings $\sigma_x\otimes\sigma_x$, $\sigma_y\otimes\sigma_y$, $\sigma_z\otimes\sigma_z$.

Step 3: Computing $T$ for each Bell state

Singlet $|\tilde{Ξ}^-\rangle$. The projector is

$$ |\tilde{Ξ}^-\rangle\langle\tilde{Ξ}^-| = \tfrac{1}{4}\left(I\otimes I - \sigma_x\otimes\sigma_x - \sigma_y\otimes\sigma_y - \sigma_z\otimes\sigma_z\right). $$

Reading off the coefficients:

$$ T_{11} = T_{22} = T_{33} = -1, \qquad T_{j\eta} = 0 \text{ for } j \neq k. $$

So $T_{\mathrm{sp}^-} = -\mathrm{diag}(1,1,1)$, and $E = -\vec u_x\cdot\vec u_y$. This reproduces the result of Exercise: Two Spins in the Singlet State.

State $|Ξ^{+}\rangle$. The projector is

$$ |Ξ^{+}\rangle\langleΞ^{+}| = \tfrac{1}{4}\left(I\otimes I + \sigma_x\otimes\sigma_x - \sigma_y\otimes\sigma_y + \sigma_z\otimes\sigma_z\right). $$

Reading off the coefficients:

$$ T_{11} = +1, \quad T_{22} = -1, \quad T_{33} = +1, \qquad T_{j\eta} = 0 \text{ for } j \neq k. $$

So $T_{Ξ^+} = \mathrm{diag}(1,-1,1)$, and $E = a_1 b_1 - a_2 b_2 + a_3 b_3$.

State $|Ξ^{-}\rangle$. The projector is

$$ |Ξ^{-}\rangle\langleΞ^{-}| = \tfrac{1}{4}\left(I\otimes I - \sigma_x\otimes\sigma_x + \sigma_y\otimes\sigma_y + \sigma_z\otimes\sigma_z\right). $$

Reading off the coefficients:

$$ T_{11} = -1, \quad T_{22} = +1, \quad T_{33} = +1. $$

So $T_{Ξ^-} = \mathrm{diag}(-1,1,1)$, and $E = -a_1 b_1 + a_2 b_2 + a_3 b_3$.

State $|\tilde{Ξ}^+\rangle$. The projector is

$$ |\tilde{Ξ}^+\rangle\langle\tilde{Ξ}^+| = \tfrac{1}{4}\left(I\otimes I + \sigma_x\otimes\sigma_x + \sigma_y\otimes\sigma_y - \sigma_z\otimes\sigma_z\right). $$

Reading off the coefficients:

$$ T_{11} = +1, \quad T_{22} = +1, \quad T_{33} = -1. $$

So $T_{\mathrm{sp}^+} = \mathrm{diag}(1,1,-1)$, and $E = a_1 b_1 + a_2 b_2 - a_3 b_3$.

Step 4: The result

The four Bell states give the four diagonal correlation matrices:

Bell state $T = \mathrm{diag}(T_{11}, T_{22}, T_{33})$ $E(\vec u_x, \vec u_y)$
$|\tilde{Ξ}^-\rangle$ $\mathrm{diag}(-1,-1,-1)$ $-\vec u_x\cdot\vec u_y$
$|Ξ^{+}\rangle$ $\mathrm{diag}(+1,-1,+1)$ $a_1 b_1 - a_2 b_2 + a_3 b_3$
$|Ξ^{-}\rangle$ $\mathrm{diag}(-1,+1,+1)$ $-a_1 b_1 + a_2 b_2 + a_3 b_3$
$|\tilde{Ξ}^+\rangle$ $\mathrm{diag}(+1,+1,-1)$ $a_1 b_1 + a_2 b_2 - a_3 b_3$

Each diagonal entry is $\pm 1$, and the product $T_{11}T_{22}T_{33}$ equals $-1$ in every case. Equivalently, writing $T_{jj} = -\epsilon_j$, the signs satisfy $\epsilon_1\epsilon_2\epsilon_3 = +1$. Only the singlet has $T = -I$, and hence the rotationally invariant correlation $E = -\vec u_x\cdot\vec u_y$.

Solution in the Biquaternion Formulation

Step 1: The four Bell idempotents

In the biquaternion framework, a single-qubit pure state is an idempotent of $\mathbb{B}_{\Psi}$, of the form $\tilde\Pi(\vec u) = \tfrac{1}{2}(e_t + i\vec u)$. A two-qubit pure state is an idempotent of $\mathbb{B}_{\Psi}\otimes\mathbb{B}_{\Psi} \subset \mathbb{B}\otimes\mathbb{B}$.

The four Bell states correspond to the four Bell idempotents

$$ \tilde\Pi_{\epsilon} = \frac{1}{4}\left(e_t\otimes e_t + \epsilon_1\,\vec e_x\otimes \vec e_x + \epsilon_2\,\vec e_y\otimes \vec e_y + \epsilon_3\,\vec e_z\otimes \vec e_z\right), $$

where $\epsilon = (\epsilon_1, \epsilon_2, \epsilon_3)$ runs over the four sign patterns with $\epsilon_1\epsilon_2\epsilon_3 = +1$:

State $\epsilon = (\epsilon_1, \epsilon_2, \epsilon_3)$ Biquaternion idempotent
$|\tilde{Ξ}^-\rangle$ $(+1,+1,+1)$ $\tfrac{1}{4}(e_t\otimes e_t + \vec e_x\otimes \vec e_x + \vec e_y\otimes \vec e_y + \vec e_z\otimes \vec e_z)$
$|Ξ^{+}\rangle$ $(-1,+1,-1)$ $\tfrac{1}{4}(e_t\otimes e_t - \vec e_x\otimes \vec e_x + \vec e_y\otimes \vec e_y - \vec e_z\otimes \vec e_z)$
$|Ξ^{-}\rangle$ $(+1,-1,-1)$ $\tfrac{1}{4}(e_t\otimes e_t + \vec e_x\otimes \vec e_x - \vec e_y\otimes \vec e_y - \vec e_z\otimes \vec e_z)$
$|\tilde{Ξ}^+\rangle$ $(-1,-1,+1)$ $\tfrac{1}{4}(e_t\otimes e_t - \vec e_x\otimes \vec e_x - \vec e_y\otimes \vec e_y + \vec e_z\otimes \vec e_z)$

Idempotency. The square of $\tilde\Pi_\epsilon$ is

$$ \tilde\Pi_\epsilon^2 = \frac{1}{16}\sum_{\mu,\mu_2\in\{t,x,y,z\}}\epsilon_\mu\epsilon_{\mu_2}\,(e_\mu e_{\mu_2})\otimes(e_\mu e_{\mu_2}), $$

with $\epsilon_t \equiv 1$. We group the sixteen pairs $(m,m_2)$ into three classes.

Diagonal pairs $m = m_2$. For $m = 0$, $e_t e_t = e_t$; for $m = 1, 2, 3$, $e_m e_m = -e_t$. In all four cases the tensor product $(e_m e_m)\otimes(e_m e_m)$ equals $e_t\otimes e_t$, with coefficient $\epsilon_m^2 = 1$. Contribution: $4\,e_t\otimes e_t$.

Off-diagonal pairs involving $e_t$. The pairs $(t,j)$ and $(j,t)$ with $j \in \{x,y,z\}$ both give $\vec e_j\otimes \vec e_j$, with coefficient $\epsilon_j$. Contribution: $2\epsilon_j\,\vec e_j\otimes \vec e_j$ for each $j$.

Off-diagonal pairs among $\{x,y,z\}$. For $j \neq k$ in $\{x,y,z\}$, let $l$ be the remaining index. Then $\vec e_j \vec e_\eta = \epsilon_{j\eta l}\vec e_l$ and $\vec e_\eta \vec e_j = -\epsilon_{j\eta l}\vec e_l$, and the two tensor products are both $\vec e_l\otimes \vec e_l$. The coefficients $\epsilon_j\epsilon_k$ and $\epsilon_k\epsilon_j$ are equal. Contribution: $2\epsilon_j\epsilon_k\,\vec e_l\otimes \vec e_l$.

Collecting, the coefficient of $\vec e_l\otimes \vec e_l$ in $16\,\tilde\Pi_\epsilon^2$ is $2\epsilon_l + 2\epsilon_j\epsilon_k$, where $\{j,k,l\} = \{x,y,z\}$. Since

$$ 16\,\tilde\Pi_\epsilon = 4\,e_t\otimes e_t + 4\sum_{l\in\{x,r_y,r_z\}}\epsilon_l\,\vec e_l\otimes \vec e_l, $$

the requirement $\tilde\Pi_\epsilon^2 = \tilde\Pi_\epsilon$ reduces to

$$ \epsilon_j\epsilon_k = \epsilon_l \qquad \text{for each cyclic triple } (j,k,l). $$

Multiplying the three equations and using $\epsilon_j^2 = 1$ gives $\epsilon_1\epsilon_2\epsilon_3 = +1$. Conversely, if $\epsilon_1\epsilon_2\epsilon_3 = +1$, multiplying by $\epsilon_1$ gives $\epsilon_2\epsilon_3 = \epsilon_1$, and similarly for the other two cyclic triples. So idempotency holds if and only if $\epsilon_1\epsilon_2\epsilon_3 = +1$.

Normalisation. Four times the scalar part is

$$ 4\,\mathrm{Sc}(\tilde\Pi_\epsilon) = \frac{1}{4}\sum_{j\in\{t,x,y,z\}}\bigl(2\,\mathrm{Sc}(\vec e_j)\bigr)^2 = \frac{1}{4}\cdot 4 = 1, $$

because $2\,\mathrm{Sc}(e_t) = 2$ and $2\,\mathrm{Sc}(\vec e_j) = 0$ for $j \in \{x,y,z\}$.

Step 2: The correlation observable

The correlation function is the expectation of the tensor-product Hermitian element

$$ \tilde{H}_{\Psi,ab} = (i\vec u_x)\otimes(i\vec u_y), $$

where $\vec u_x = \sum_j a_j \vec e_j$ and $\vec u_y = \sum_\eta b_\eta \vec e_\eta$ are pure real quaternions. The expectation is given by the tensor-product pairing formula:

$$ E(\vec u_x, \vec u_y) = 4\,\mathrm{Sc}\!\left(\tilde\Pi_\epsilon\tilde{H}_{\Psi,ab}\right). $$

Step 3: Computing the pairing for each Bell state

We compute the pairing term by term. Multiplying $\tilde\Pi_\epsilon$ by $\tilde{H}_{\Psi,ab}$,

$$ \tilde\Pi_\epsilon\tilde{H}_{\Psi,ab} = \frac{1}{4}\sum_{\mu\in\{t,x,y,z\}}\epsilon_\mu\,(e_\mu \cdot i\vec u_x)\otimes(e_\mu \cdot i\vec u_y), $$

with $\epsilon_t \equiv 1$.

Term $m = t$. The pairing is $4\,\mathrm{Sc}(i\vec u_x)\,\mathrm{Sc}(i\vec u_y) = 0$, because the scalar part of $i\vec u_x$ vanishes.

Terms $m \in \{x,y,z\}$. We have

$$ e_\mu\cdot i\vec u_x = i\,e_\mu \vec u_x = i\left(-a_\mu e_t + \text{vector}\right), $$

using $e_\mu \vec u_x = -a_\mu e_t + \sum_{\eta \neq \mu} a_\eta e_\mu \vec e_\eta$. The scalar part is $-i a_\mu$, and similarly $\mathrm{Sc}(e_\mu\cdot i\vec u_y) = -i b_\mu$. The product of the two scalar parts is

$$ (-i a_\mu)(-i b_\mu) = i^2\, a_\mu b_\mu = -a_\mu b_\mu. $$

Assembly. Substituting, the factor $\frac14$ carried by the product and the factor $4$ carried by the pairing cancel,

$$ E(\vec u_x, \vec u_y) = \sum_{\mu\in\{x,y,z\}}\epsilon_\mu\,(-a_\mu b_\mu) = -\sum_{\mu\in\{x,y,z\}}\epsilon_\mu\,a_\mu b_\mu. $$

So the correlation function is

$$ \boxed{\;E(\vec u_x, \vec u_y) = -\epsilon_1\,a_1 b_1 - \epsilon_2\,a_2 b_2 - \epsilon_3\,a_3 b_3\;}. $$

Step 4: The result

Applying this formula to the four sign patterns:

Bell state $\epsilon = (\epsilon_1,\epsilon_2,\epsilon_3)$ $E(\vec u_x, \vec u_y)$
$|\tilde{Ξ}^-\rangle$ $(+1,+1,+1)$ $-a_1 b_1 - a_2 b_2 - a_3 b_3 = -\vec u_x\cdot\vec u_y$
$|Ξ^{+}\rangle$ $(-1,+1,-1)$ $+a_1 b_1 - a_2 b_2 + a_3 b_3$
$|Ξ^{-}\rangle$ $(+1,-1,-1)$ $-a_1 b_1 + a_2 b_2 + a_3 b_3$
$|\tilde{Ξ}^+\rangle$ $(-1,-1,+1)$ $+a_1 b_1 + a_2 b_2 - a_3 b_3$

These match the standard results exactly.

Limiting Cases

Both formulations give the same results, and both give the same limits. The cleanest way to read the table is by the direction of the measurements.

Parallel measurements along a coordinate axis. Set $\vec u_x = \vec u_y = \vec u_\eta$; the correlation is the diagonal entry $T_{\eta\eta}$:

Axis $\tilde{Ξ}^-$ $Ξ^{+}$ $Ξ^{-}$ $\tilde{Ξ}^+$
$k=x$ $-1$ $+1$ $-1$ $+1$
$k=y$ $-1$ $-1$ $+1$ $+1$
$k=z$ $-1$ $+1$ $+1$ $-1$

The singlet is the unique Bell state for which the parallel correlation is $-1$ along every direction: the singlet is perfectly anti-correlated along any common axis, and its correlation matrix is $-I$. The other three Bell states are perfectly correlated (or anti-correlated) only along the three coordinate axes, with sign patterns that depend on the direction.

Joint probabilities. For any Bell state, the four joint probabilities of the measurement outcomes are

$$ p(+,+) = p(-,-) = \tfrac{1}{4}\left(1 + E\right), \qquad p(+,-) = p(-,+) = \tfrac{1}{4}\left(1 - E\right). $$

This follows from $\langle \sigma_j \otimes I\rangle = \langle I \otimes \sigma_\eta\rangle = 0$ in every Bell state: the single-particle expectations vanish, and the Born probabilities reduce to $\tfrac{1}{4}(1 \pm E)$. Exercise: Two Spins in the Singlet State's probability $p(+,+) = \tfrac{1}{4}(1 - \vec u_x\cdot\vec u_y)$ is the special case $E = -\vec u_x\cdot\vec u_y$ for the singlet.

Consistency with Exercise: Two Spins in the Singlet State. For the singlet, $E(\vec u_x, \vec u_y) = -\vec u_x\cdot\vec u_y$, so $p(+,+) = \tfrac{1}{4}(1 - \vec u_x\cdot\vec u_y)$, recovering Exercise: Two Spins in the Singlet State.

What the Biquaternion Solution Illustrates

1. The four Bell states have a uniform algebraic form. Each is an idempotent

$$ \tilde\Pi_\epsilon = \frac{1}{4}\left(e_t\otimes e_t + \epsilon_1\,\vec e_x\otimes \vec e_x + \epsilon_2\,\vec e_y\otimes \vec e_y + \epsilon_3\,\vec e_z\otimes \vec e_z\right), $$

with a sign pattern satisfying $\epsilon_1\epsilon_2\epsilon_3 = +1$. This is the algebraic characterization of the Bell basis: four idempotents, differing only by the sign pattern of the three vector-vector terms.

2. The correlation function is a bilinear pairing. The correlation function $E(\vec u_x, \vec u_y) = 4\,\mathrm{Sc}(\tilde\Pi_\epsilon\tilde{H}_{\Psi,ab}^{*})$ is the tensor-product Hermitian form between the state idempotent and the observable, read as four times the scalar part on the tensor product $\tilde{H}_{\Psi,ab} = (i\vec u_x)\otimes(i\vec u_y)$. The normalised Hermitian form, read through the trace, is the same algebraic operation that gives the Born rule (Exercise: Measuring Spin Along an Arbitrary Direction) and the expectation value (Exercise: Spin Precession in a Magnetic Field).

3. The sign pattern encodes the correlation matrix. In the standard formulation, the correlation matrix $T$ is computed by expanding the Bell projector in the Pauli-string basis and reading off the coefficients. In the biquaternion formulation, the same coefficients appear directly as the signs $\epsilon_j$ in the idempotent. The matrix $T$ is $-\mathrm{diag}(\epsilon_1,\epsilon_2,\epsilon_3)$.

4. The singlet is the isotropic case. Among the four Bell states, only the singlet has all signs equal, giving $T = -I$ and the rotationally invariant correlation $E = -\vec u_x\cdot\vec u_y$. The other three break rotational symmetry; their correlation matrices have one sign pattern of the form $(+,-,+)$ up to permutation, and their correlations depend on the direction of the measurements. The four sign patterns with $\epsilon_1\epsilon_2\epsilon_3 = +1$ correspond exactly to the four Bell states, and no other patterns are possible because the idempotency condition enforces the product to be $+1$.

5. The result agrees with the standard formulation. This is not surprising, because the two formulations are the same mathematics in different notation: $\mathbb{B}\otimes\mathbb{B} \cong M_4(\mathbb{C})$, and the Bell idempotent $\tilde\Pi_\epsilon$ corresponds exactly to the rank-one projector $|\tilde{Ξ}_{\epsilon}\rangle\langle\tilde{Ξ}_{\epsilon}|$ under the isomorphism. But seeing the same result emerge from the algebra is a useful sanity check, and it illustrates the structural claim of the companion articles: the kinematics of two-qubit entanglement is contained in the tensor-product algebra $\mathbb{B}\otimes\mathbb{B}$, with the Born rule given by the tensor-product trace, and the four Bell states are the four idempotents with sign pattern satisfying $\epsilon_1\epsilon_2\epsilon_3 = +1$.

Comparison with Exercises: Measuring Spin Along an Arbitrary Direction, Spin Precession in a Magnetic Field, Reduced State of an Entangled Subsystem, Successive Measurements of Spin and Two Spins in the Singlet State

Measuring Spin Along an Arbitrary Direction Spin Precession in a Magnetic Field Successive Measurements of Spin Two Spins in the Singlet State Reduced State of an Entangled Subsystem Correlation Function of the Bell States
System One qubit One qubit One qubit Two qubits Two qubits Two qubits
Physical process Single measurement Evolution Successive measurements Singlet measurement Reduced state Correlation function
Structural object Idempotent + idempotent Idempotent + $*$-unitary Sequence of idempotents Idempotent in $\mathbb{B}\otimes\mathbb{B}$ Partial trace Bell idempotents
Rule used Pairing formula Pairing formula + rotor conjugation Product of pairing formulas Tensor-product pairing formula Partial trace + pairing formula Tensor-product trace reading of the Hermitian form
Result $\cos^2(\theta/2)$ $(\frac{\hbar}{2})\cos(\omega_L r'_t/c)$ $\cos^2(\theta_1/2)\cos^2((\theta_2-\theta_1)/2)$ $(1/4)(1 - \vec u_x\cdot\vec u_y)$ $\tilde{Ϛ}_{\Psi,1} = \tfrac{1}{2}e_t$ $E = -\sum_j \epsilon_j a_j b_j$

Summary

We have solved a sixth exercise in quantum physics — the spin correlation function of the four Bell states — in two parallel presentations.

In the standard formulation, the solution proceeds through the four Bell projectors, their expansions in the Pauli-string basis, and the trace-orthogonality of the Pauli strings, giving the four diagonal correlation matrices $T = \mathrm{diag}(\pm 1, \pm 1, \pm 1)$ and the correlation functions $E = \vec u_x^T T\vec u_y$.

In the biquaternion formulation, the solution proceeds through the four Bell idempotents $\tilde\Pi_\epsilon$ in $\mathbb{B}\otimes\mathbb{B}$, the correlation observable $(i\vec u_x)\otimes(i\vec u_y)$, and the tensor-product Hermitian form, read as the Hermitian pairing, giving the general formula $E = -\sum_j \epsilon_j a_j b_j$ for the sign pattern $\epsilon$ with $\epsilon_1\epsilon_2\epsilon_3 = +1$.

The two formulations agree, as they must. The biquaternion formulation makes explicit the algebraic structure of the Bell basis: the four Bell states are the four idempotents with the sign pattern $\epsilon_1\epsilon_2\epsilon_3 = +1$; the correlation function is the tensor-product Hermitian form between the state idempotent and the correlation observable, read as the Hermitian pairing; and the singlet is the unique Bell state with the rotationally invariant correlation $E = -\vec u_x\cdot\vec u_y$.

The correlation function is the object measured in Bell-inequality experiments, and the four correlation matrices exhibited here are the full content of the two-qubit spin correlations in the Bell basis.

Summary of Notation

Symbol Meaning
$\mathbb{B}$ Biquaternion algebra
$\mathbb{B}_{\Psi}$ Hermitian subspace (states, observables)
$\mathbb{B}\otimes\mathbb{B}$ Tensor product, isomorphic to $M_4(\mathbb{C})$
$e_t = 1, \vec e_x, \vec e_y, \vec e_z$ Quaternion basis, $\vec e_\eta^2 = -e_t$
$i$ Scalar imaginary, $i^2 = -1$
$\boldsymbol{\sigma} = (\sigma_1, \sigma_2, \sigma_3)$ Pauli matrices
$\boldsymbol{\sigma}_1 = \boldsymbol{\sigma}\otimes I$ Spin operator on particle 1
$\boldsymbol{\sigma}_2 = I\otimes\boldsymbol{\sigma}$ Spin operator on particle 2
$|Ξ^{\pm}\rangle_{*}, |\tilde{Ξ}^{\pm}\rangle$ The four Bell states
$\tilde\Pi_\epsilon = \tfrac{1}{4}(e_t\otimes e_t + \sum_j \epsilon_j\,\vec e_j\otimes \vec e_j)$ Bell idempotent, with $\epsilon_1\epsilon_2\epsilon_3 = +1$
$\tilde{H}_{\Psi,ab} = (i\vec u_x)\otimes(i\vec u_y)$ Correlation observable
$E(\vec u_x, \vec u_y) = 4\,\mathrm{Sc}(\tilde\Pi_\epsilon\tilde{H}_{\Psi,ab}^{*})$ Correlation function, four times the scalar part
$T_{j\eta} = \langle\tilde{Ξ}|\sigma_j\otimes\sigma_\eta|\tilde{Ξ}\rangle$ Correlation matrix
$\mathrm{Tr}\bigl(\mathsf M_2(x)\otimes\mathsf M_2(y)\bigr) = \mathrm{Tr}_\mathbb{B}\mathsf M_2(x)\cdot\mathrm{Tr}_\mathbb{B}\mathsf M_2(y)$ Trace on the tensor product

Further Reading