---
title: Bell-Diagonal States in Quantum Information
url: https://www.emergentmind.com/topics/bell-diagonal-states
type: topic
---

# Bell-Diagonal States in Quantum Information

Bell-diagonal states are bipartite density operators that are diagonal in a Bell basis of maximally entangled states. In the two-qubit setting they form a three-parameter family with maximally mixed marginals, exact formulas for several correlation measures, and a state space represented by a tetrahedron in correlation coordinates. In higher dimensions they generalize to convex mixtures of generalized Bell projectors; for the Weyl–Heisenberg construction this family becomes the “magic simplex” \(M_d\), whose symmetry strongly constrains positivity, separability, twirling, and noise models. Because of this combination of algebraic symmetry and geometric tractability, Bell-diagonal states are a standard laboratory for entanglement theory, steering, discord, coherence dynamics, error correction, state estimation, and experimental benchmarking [2307.10727][1912.06105].

## 1. Definitions and canonical parameterizations

For two qubits, the Bell basis is
\[
\begin{aligned}
|\Phi^{\pm}\rangle &= \frac{|00\rangle \pm |11\rangle}{\sqrt{2}},\\
|\Psi^{\pm}\rangle &= \frac{|01\rangle \pm |10\rangle}{\sqrt{2}}.
\end{aligned}
\]
A two-qubit Bell-diagonal state is any density operator diagonal in this basis,
\[
\rho_{\mathrm{BD}}=\sum_{i=1}^4 p_i\,|\beta_i\rangle\langle\beta_i|,\qquad p_i\ge 0,\qquad \sum_i p_i=1.
\]
Equivalently, Bell-diagonal states have maximally mixed marginals and admit the Pauli-correlator form
\[
\rho_{\mathrm{BD}}=\frac{1}{4}\Big(I\otimes I+\sum_{k=1}^3 c_k\,\sigma_k\otimes\sigma_k\Big),
\qquad \mathrm{Tr}_A\,\rho=\mathrm{Tr}_B\,\rho=I/2.
\]
The Bell weights and correlation coefficients are related by
\[
\begin{aligned}
c_1 &= p_{\Phi^+}-p_{\Phi^-}+p_{\Psi^+}-p_{\Psi^-},\\
c_2 &= -p_{\Phi^+}+p_{\Phi^-}+p_{\Psi^+}-p_{\Psi^-},\\
c_3 &= p_{\Phi^+}+p_{\Phi^-}-p_{\Psi^+}-p_{\Psi^-},
\end{aligned}
\]
and conversely
\[
\begin{aligned}
p_{\Phi^+} &= \tfrac{1}{4}(1+c_1-c_2+c_3),\qquad
p_{\Phi^-} = \tfrac{1}{4}(1-c_1+c_2+c_3),\\
p_{\Psi^+} &= \tfrac{1}{4}(1+c_1+c_2-c_3),\qquad
p_{\Psi^-} = \tfrac{1}{4}(1-c_1-c_2-c_3).
\end{aligned}
\]
These relations define the standard three-parameter description used throughout the two-qubit literature [1912.06105][1006.2775].

In \(d\times d\) systems, a standard seed maximally entangled state is
\[
|\Phi_d\rangle \equiv \frac{1}{\sqrt d}\sum_{k=0}^{d-1} |k\rangle\otimes|k\rangle.
\]
With Weyl–Heisenberg operators
\[
X\,|k\rangle = |k{+}1\ \mathrm{mod}\ d\rangle,\qquad
Z\,|k\rangle = \omega^k |k\rangle,\qquad \omega=e^{2\pi i/d},
\]
and \(U_{mn}\equiv X^m Z^n\), one defines generalized Bell states
\[
|B_{mn}\rangle \equiv (\mathbb{I}\otimes U_{mn})\,|\Phi_d\rangle
= \frac{1}{\sqrt d}\sum_{k=0}^{d-1} \omega^{\,nk}\,|k\rangle\otimes|k{+}m\rangle.
\]
A \(d\times d\) Bell-diagonal state is then
\[
\rho = \sum_{m,n=0}^{d-1} p_{mn}\,|B_{mn}\rangle\langle B_{mn}|,\qquad
p_{mn}\ge 0,\qquad \sum_{m,n}p_{mn}=1.
\]
In the notation of the Weyl–Heisenberg analysis, the set of such states is
\[
M_d \equiv \Bigl\{\rho=\sum_{k,l} c_{k,l}\,P_{k,l}\ \big|\ \sum_{k,l}c_{k,l}=1,\ c_{k,l}\ge 0\Bigr\},
\]
the “magic simplex” [2307.10727][1004.1655].

The notion has also been generalized to unequal local dimensions. If \(d_A\neq d_B\), one may choose
\[
|\phi^{00}\rangle = \frac{1}{\sqrt{d_A}}\sum_{i=0}^{d_A-1} |i\rangle_A\otimes |i\rangle_B,
\]
and define
\[
|\phi^{\alpha\beta}\rangle = (Z_A^\alpha\otimes X_B^\beta)\,|\phi^{00}\rangle,
\]
leading to generalized Bell-diagonal states
\[
\rho_P = \sum_{\alpha=0}^{d_A-1}\sum_{\beta=0}^{d_B-1} p_{\alpha\beta}\,|\phi^{\alpha\beta}\rangle\langle\phi^{\alpha\beta}|,
\]
which remain diagonal in a correlated Weyl-like operator family even though a complete orthonormal maximally entangled basis does not exist for unequal local dimensions [2308.10607].

## 2. State-space geometry

For two qubits, positivity is equivalent to nonnegativity of the four Bell probabilities. In correlation coordinates \((c_1,c_2,c_3)\), this gives the tetrahedron \(\mathcal{T}\) defined by
\[
1 \pm c_1 \pm c_2 \pm c_3 \ge 0
\]
for the four sign patterns corresponding to the Bell vertices. The vertices are
\[
|\Phi^+\rangle\leftrightarrow (1,-1,1),\quad
|\Psi^+\rangle\leftrightarrow (1,1,-1),\quad
|\Phi^-\rangle\leftrightarrow (-1,1,1),\quad
|\Psi^-\rangle\leftrightarrow (-1,-1,-1).
\]
Inside this tetrahedron, the separable set is the inscribed octahedron
\[
|c_1|+|c_2|+|c_3|\le 1,
\]
which also coincides with the PPT region in \(2\times 2\) systems [1912.06105][1006.2775].

| Property | Condition in \((c_1,c_2,c_3)\)-space | Interpretation |
|---|---|---|
| Physicality | \(1 \pm c_1 \pm c_2 \pm c_3 \ge 0\) | Bell tetrahedron \(\mathcal{T}\) |
| Separability/PPT | \(|c_1|+|c_2|+|c_3|\le 1\) | Inscribed octahedron \(\mathcal{O}\) |
| CHSH nonlocality | \(c_{(1)}^2+c_{(2)}^2>1\) | Outside the CHSH-local cylinders |
| 2-setting steering | \(c_{(1)}^2+c_{(2)}^2>1\) | Coincides with CHSH for Bell-diagonal states |
| 3-setting steering | \(c_1^2+c_2^2+c_3^2>1\) | Sufficient criterion |
| Zero discord | \((c_1,0,0)\), \((0,c_2,0)\), \((0,0,c_3)\) | Three coordinate axes |

The geometric description extends several resource measures. The CHSH-local set is the intersection of three unit cylinders,
\[
c_1^2+c_2^2\le 1,\qquad
c_1^2+c_3^2\le 1,\qquad
c_2^2+c_3^2\le 1,
\]
which strictly contains the separable octahedron. The zero-discord states are exactly the three coordinate axes, while discord level sets form the familiar “tubes” around those axes [1006.2775][1601.00113].

In the Weyl–Heisenberg \(d\times d\) setting, the geometry changes from a tetrahedron to a probability simplex with \(d^2\) vertices \(\{P_{k,l}\}\). The Weyl–Heisenberg relations induce entanglement-class-preserving symmetries and a linear algebra that makes faces, slices, and invariant subfamilies tractable; the analysis in \(d=3,4\) shows that these symmetry-induced convex structures enable unusually strong analytic and numerical control of entanglement within \(M_d\) [2307.10727].

A more restrictive higher-dimensional family arises from maximal abelian symmetry. For
\[
\rho \;=\; \sum_{n=1}^{d-1} \lambda_n\,\Pi_n \;+\; \lambda_d\,P,
\]
the paper on maximal abelian symmetry gives necessary and sufficient conditions inside this subclass:
\[
\rho^{T_B}\ge 0 \quad\Longleftrightarrow\quad \lambda_n\,\lambda_{d-n} \;\ge\; \lambda_d^2,
\]
and
\[
\rho\text{ is separable}\quad\Longleftrightarrow\quad \lambda_n \;\ge\; \lambda_d\quad\text{for all }n=1,\dots,d-1.
\]
This provides an explicitly solvable Bell-diagonal sector beyond the qubit tetrahedron [1011.0515].

## 3. Entanglement, PPT structure, and witnesses

In two-qubit Bell-diagonal states, entanglement is completely characterized by the largest Bell weight. The concurrence is
\[
C(\rho)=\max\Big\{0,\frac{|c_1|+|c_2|+|c_3|-1}{2}\Big\}=\max\{0,\,2p_{\max}-1\},
\]
with \(p_{\max}=\max_i p_i\). Entanglement of formation follows Wootters’ formula,
\[
E(\rho) = h\!\left(\frac{1 + \sqrt{1 - C^2}}{2}\right),
\qquad
h(x)=-x\log_2 x-(1-x)\log_2(1-x).
\]
Thus the octahedral boundary is also the exact separability boundary for the two-qubit family [1912.06105].

The situation changes qualitatively in higher dimensions. Partial transpose acts as
\[
(\ketbra{i}{j}\otimes\ketbra{k}{l})^\Gamma=\ketbra{i}{j}\otimes\ketbra{l}{k},
\]
and while NPT implies entanglement in all dimensions, PPT implies separability only in \(2\times 2\) and \(2\times 3\). In \(d\ge 3\), Bell-diagonal states can be PPT-entangled. Within the Weyl–Heisenberg qutrit simplex \(M_3\), the relative volume of PPT states is about \(60\%\), and the paper reports \(r\mathrm{PPT}=60.0\%\), \(\mathrm{E2/PPT}=10.4\%\), \(\mathrm{E3/PPT}=2.7\%\), and \((\mathrm{E2}\cap \mathrm{E3})/\mathrm{PPT}=1.8\%\) for the standard basis [2307.10727].

Analytic entanglement tests remain effective in Bell-diagonal families. The realignment criterion detects entanglement when
\[
\|\rho_R\|_1>1.
\]
The quasipure concurrence lower bound is
\[
C_{qp}(\rho)=\max(0,S_0-\sum_{i>0}S_i),
\]
and in \(M_d\) the singular values \(S_{k,l}\) admit an explicit formula in terms of the Bell coefficients \(c_{k,l}\) [2307.10727].

Higher-dimensional Bell-diagonal states also furnish explicit bound-entangled constructions. In \(3\times 3\), the family
\[
\rho_\epsilon = N_\epsilon (\Pi_0 + \epsilon \Pi_1 + \epsilon^{-1} \Pi_2),\qquad \epsilon>0,
\]
is PPT for all \(\epsilon\), and separable if and only if \(\epsilon=1\). Hence \(\rho_\epsilon\) is PPT-entangled for \(\epsilon\neq 1\). More generally, the paper constructs diagonal-in-magic-basis entanglement witnesses
\[
W = \sum_{m,n} w_{mn} P_{mn},
\]
and uses them to detect PPT-entangled Bell-diagonal families in \(d\ge 3\) [1004.1655].

For unequal local dimensions, Bell-diagonal entanglement persists and can evade the usual CCNR/realignment and de Vicente criteria. Extending the Sarbicki–Scala–Chruściński criteria to non-Hermitian operator bases yields witnesses strong enough to detect explicit PPT-entangled generalized Bell-diagonal states in \(4\times 6\) that are not detected by those standard criteria [2308.10607].

## 4. Weyl–Heisenberg Bell bases and why the standard basis is special

A central structural result is that not every complete orthonormal Bell basis is equivalent from the perspective of Bell-diagonal entanglement theory. In the Weyl–Heisenberg basis, define the twirl operators
\[
T_{i,j} \equiv W_{i,j}\otimes W_{i,j}^\ast.
\]
They form an abelian group,
\[
T_{i_1,j_1}T_{i_2,j_2}=T_{i_1+i_2,j_1+j_2},\qquad
T_{i,j}^{-1}=T_{-i,-j},
\]
and every Bell state is a simultaneous eigenvector:
\[
T_{i,j}|k,l\rangle = \omega^{\,jk-il}|k,l\rangle.
\]
This stabilizer property gives the standard Weyl–Heisenberg Bell basis a distinguished algebraic status [2307.10727].

One consequence is an exact channel equivalence. The projection channel onto the Bell-diagonal simplex,
\[
\mathcal{P}(\rho)=\sum_{k,l} P_{k,l}\rho P_{k,l}
=\sum_{k,l}\langle k,l|\rho|k,l\rangle P_{k,l},
\]
coincides with the finite Weyl–twirl channel
\[
\mathcal{T}(\rho)=\frac{1}{d^2}\sum_{i,j}(W_{i,j}\otimes W_{i,j}^\ast)\,\rho\,(W_{i,j}\otimes W_{i,j}^\ast)^\dagger.
\]
Theorem 1 states that
\[
\mathcal{P}(\rho)\equiv \mathcal{T}(\rho)
\]
for all bipartite \(\rho\). Through the Choi–Jamiołkowski isomorphism, this identifies generalized Pauli channels with Bell-diagonal Choi states and explains why twirling and Pauli noise collapse to the same operation in the standard basis [2307.10727].

This equivalence has immediate entanglement-theoretic consequences. For a separable state \(\rho_s\),
\[
\mathcal{P}(\rho_s)=\mathcal{T}(\rho_s)\in M_d\cap SEP.
\]
Moreover, if a Bell-diagonal entanglement witness \(K\) is optimal on some separable \(\rho_0\), then it remains optimal after projection:
\[
\mathrm{Tr}(K\rho_0)=0 \;\Rightarrow\; \mathrm{Tr}(K\mathcal{P}(\rho_0))=0,
\]
with \(\mathcal{P}(\rho_0)\in M_d\cap SEP\). This allows witness optimization directly on the magic simplex [2307.10727].

The nontrivial point is that these properties fail for generic complete orthonormal Bell bases. For a phase matrix \(\alpha=(\alpha_{s,l})\), \(|\alpha_{s,l}|=1\), define
\[
|\Phi^\alpha_{k,l}\rangle \equiv \frac{1}{\sqrt d}\sum_{s=0}^{d-1}\omega^{\,k(s-l)}\,\alpha_{s,l}\,|s-l\rangle\otimes|s\rangle,
\]
and the associated simplex
\[
M_d^\alpha \equiv \Bigl\{\rho=\sum_{k,l} c_{k,l}\,P^\alpha_{k,l}\ \big|\ \sum_{k,l}c_{k,l}=1,\ c_{k,l}\ge 0\Bigr\}.
\]
For these generalized bases, the paper finds:

\[
V^\alpha_{k_1,l_1}V^\alpha_{k_2,l_2}\not\propto V^\alpha_{k_1+k_2,l_1+l_2},
\]
\[
(V^\alpha_{i,j}\otimes V^{\alpha\ast}_{i,j})|\Phi^\alpha_{k,l}\rangle\neq \omega^\phi|\Phi^\alpha_{k,l}\rangle,
\]
\[
\mathcal{P}^\alpha\not\equiv \mathcal{T}^\alpha,
\]
and separability need not be preserved by \(\mathcal{P}^\alpha\) [2307.10727].

The entanglement structure changes accordingly. In \(d=3\), across \(1{,}000\) random choices of \(\alpha\), the fraction of PPT states in \(M_3^\alpha\) ranges from \(49\%\) to \(59\%\), with mean \(51.6\%\), whereas the standard Weyl–Heisenberg basis gives \(60.0\%\). Detectability of PPT entanglement also drops:
\[
\mathrm{E2/PPT}: \text{ min }2.1\%,\ \text{max }9.6\%,\ \text{mean }3.6\%,
\]
\[
\mathrm{E3/PPT}: \text{ min }0.6\%,\ \text{max }2.3\%,\ \text{mean }1.0\%,
\]
\[
(\mathrm{E2}\cap \mathrm{E3})/\mathrm{PPT}: \text{ min }0.1\%,\ \text{max }1.4\%,\ \text{mean }0.4\%.
\]
The corresponding correlations with \(r\mathrm{PPT}\) are \(0.99\), \(0.95\), and \(0.96\), respectively. This directly refutes the common assumption that “Bell-diagonal” means basis-independent entanglement geometry [2307.10727].

## 5. Steering, discord, coherence, and resource-theoretic structure

Bell-diagonal states are a rare setting where several nonclassicality measures are exactly computable. For CHSH nonlocality, if \(T=\mathrm{diag}(c_1,c_2,c_3)\), the Horodecki criterion gives
\[
\mathcal{B}_{\max}=2\sqrt{u_{(1)}+u_{(2)}},
\]
where \(u_i=c_i^2\), and violation occurs iff
\[
u_{(1)}+u_{(2)}>1.
\]
In the simplest nontrivial steering scenario, Bell-diagonal states are steerable by two projective measurements if and only if they violate CHSH. This equivalence is specific to Bell-diagonal states and contrasts with the generic strict hierarchy between steering and Bell nonlocality [1912.06105][1601.00113].

For three projective measurements, a simple sufficient steering criterion is
\[
\|T\|_F=\sqrt{c_1^2+c_2^2+c_3^2}>1.
\]
The steering ellipsoid then becomes the natural geometric object: its semiaxes are \(|c_1|,|c_2|,|c_3|\), and its normalized volume is
\[
V=|c_1c_2c_3|.
\]
For generalized POVMs, a constructive local hidden state model exists if
\[
R_T \ge \frac{6}{5},
\qquad
R_T = 2\pi N_T |\det(T)|,
\]
where
\[
N_T^{-1} = \int_{S^2} \mathrm{d}S(\vec{n})\, \frac{1}{[\vec{n}^{\,T} T^{-2} \vec{n}]^2}.
\]
This gives a measurement-set-sensitive distinction between projective steerability and POVM-unsteerability [1909.03963].

Discord is also closed form. Since \(\rho_A=\rho_B=I/2\), one has
\[
I(\rho)=2-H(\{p_i\}),
\]
and the classical correlation is
\[
C_{\mathrm{class}}(\rho)=1-h\!\left(\frac{1+c_*}{2}\right),
\qquad
c_*=\max\{|c_1|,|c_2|,|c_3|\}.
\]
Hence
\[
D(\rho)=1-H(\{p_i\})+h\!\left(\frac{1+c_*}{2}\right).
\]
For Bell-diagonal states, the paper on IBM implementation proves the general inequality
\[
D(\rho)\le D^{(\mathrm{rel,asym})}(\rho),
\]
and shows that equality holds for all Bell-diagonal states:
\[
D(\rho)=D^{(\mathrm{rel,asym})}(\rho).
\]
Geometrically, the zero-discord set is exactly the union of the three coordinate axes [1912.06105][1704.02180].

Coherence depends on the computational basis rather than the Bell basis. In that basis,
\[
\rho_{\mathrm{BD}}=\frac{1}{4}
\begin{pmatrix}
1+c_3 & 0 & 0 & c_1-c_2\\
0 & 1-c_3 & c_1+c_2 & 0\\
0 & c_1+c_2 & 1-c_3 & 0\\
c_1-c_2 & 0 & 0 & 1+c_3
\end{pmatrix},
\]
so
\[
C_{\ell_1}(\rho_{\mathrm{BD}})=\frac{1}{2}\big(|c_1-c_2|+|c_1+c_2|\big),
\]
and
\[
C_{\mathrm{rel}}(\rho_{\mathrm{BD}})=S(\rho_{\mathrm{diag}})-S(\rho).
\]
Coherence vanishes iff \(c_1=c_2=0\), i.e. along the \(c_3\)-axis. Under Markovian noise, Bell-diagonal states remain Bell-diagonal for BF, PF, BPF, DEP, and GAD channels, allowing explicit coherence decay laws. In particular, frozen \(l_1\)-coherence occurs under BPF whenever \(|c_1|\le |c_2|\), and under BF whenever \(|c_2|\le |c_1|\). The later work on \(n\)-th decay rates generalizes this to repeated channels and shows that, for fixed \(n\), Bell-diagonal states can be completely incoherent under GAD, DEP, and PF, but not generically under BF and BPF [1902.02962][2009.01503].

A further subtlety concerns weak measurements. Super-quantum discord satisfies \(D_w\ge D\), but the weak one-way deficit \(\Delta_w^\rightarrow\) can be smaller than the standard one-way deficit, and for Werner states the paper finds the strict ordering
\[
\Delta_w^\rightarrow < \Delta^\rightarrow = D < D_w.
\]
This shows that weak measurement does not necessarily capture more quantumness across all correlation measures [1302.4039].

## 6. Protocols, estimation, and experimental realizations

Bell-diagonal states are central to protocol design precisely because their structure is both rich and manageable. In the Weyl–Heisenberg basis, the stabilizer property enables non-demolition syndrome extraction for maximally entangled qudits. With an ancilla initialized in \(|0\rangle\), applying \(F\), controlled-\(T_{i,j}\), and \(F^\dagger\) yields
\[
(F^\dagger\otimes\mathbb{I})\,CT_{i,j}\,(F\otimes\mathbb{I})\bigl(|0\rangle\otimes|k,l\rangle\bigr)
= |\Phi\rangle\otimes|k,l\rangle,
\]
where
\[
\Phi\equiv jk-il.
\]
The Bell state is left undisturbed, and the measured ancilla identifies the phase/shift error; recovery is performed with \(W_{k,l}^\dagger\). The same Weyl–Heisenberg structure makes twirling to Bell-diagonal form benign, because \(\mathcal{P}\equiv\mathcal{T}\) and separability is preserved [2307.10727].

Bell-diagonal correlations also support more counterintuitive tasks. In the entanglement-distribution-via-separable-states protocol for Bell-diagonal inputs,
\[
\rho = \tfrac{1}{4}\Big(I\otimes I + t_x\,\sigma_x\otimes\sigma_x + t_y\,\sigma_y\otimes\sigma_y + t_z\,\sigma_z\otimes\sigma_z\Big),
\]
the necessary and sufficient resource condition within the separable Bell-diagonal family is \(t_y\neq 0\). If \(t_y=0\), the protocol fails; if \(t_y\neq 0\), separable Bell-diagonal states can distribute distillable entanglement by transmitting a separable qubit [1204.0366].

Several experimental constructions exploit the Bell-diagonal parametrization directly. The IBM implementation papers give circuits that prepare the full two-qubit Bell-diagonal family and evaluate entanglement of formation, CHSH non-locality, steering, discord, and related observables across the full tetrahedron. One paper reports noiseless simulation validation on \(340\) random Bell-diagonal states and \(100\) Werner states with mean fidelity \(99.5\%\pm 0.5\%\), while hardware runs reproduce the expected hierarchy \(N \Rightarrow S \Rightarrow E \Rightarrow D\) but with resource degradation near the Bell corners [1912.06105][1808.10533].

State characterization can also be done under LOCC restrictions. For Bell-diagonal states, local Pauli parity checks satisfy
\[
\Pr(s_i=+1)=\tfrac{1+c_i}{2},\qquad \Pr(s_i=-1)=\tfrac{1-c_i}{2},
\]
so the likelihood factorizes across axes. In this setting, Bell-basis measurements achieve the QCRB, whereas ordered Pauli parity checks incur a factor-of-3 penalty in direct-inversion risk at fixed \(N\). The 2025 estimation study shows that Bayesian mean estimation has the lowest average risk among the estimators tested and remains full rank, while ordered parity checks are the best LOCC measurement design among those compared [2503.11454].

Finally, Bell-diagonal states admit genuinely nondestructive identification. In the probe-based protocol, an unknown Bell-diagonal state is interrogated through auxiliary qubits rather than direct measurement of the Bell-diagonal pair. Three probe settings yield linear relations \(M^1,M^2,M^3\) from which the Bell weights are reconstructed, while repeated applications of the same interaction recover the original Bell-diagonal ensemble after each step. The proposal also gives a cavity-QED realization in which effective \(\sigma_x\sigma_x\), \(\sigma_y\sigma_y\), and \(\sigma_z\sigma_z\) couplings implement the required operations [1111.4038].

Bell-diagonal states therefore occupy a distinctive position in quantum information theory. They are simple enough to admit exact geometry, exact correlation formulas, and experimentally practical workflows, yet structured enough to expose genuinely nontrivial phenomena: PPT entanglement in \(d\ge 3\), basis-sensitive entanglement geometry, coincidence of two-setting steering with CHSH, discord without entanglement, coherence freezing under selected channels, and protocol advantages under twirling, LOCC estimation, and separable-state entanglement distribution [2307.10727][1006.2775].

Source: https://www.emergentmind.com/topics/bell-diagonal-states