Monogamy of Entanglement

Classical correlations are promiscuous. A perfectly correlated classical bit — the value on three carbon copies of the same telegram — can be shared among arbitrarily many parties, each of whom learns it in full. Quantum entanglement refuses to behave this way. If two qubits $A$ and $B$ sit in a maximally entangled Bell state, their joint state is pure, and a pure bipartite state is necessarily a product with the rest of the universe. It follows, with no further assumption, that $A$ can share no entanglement — indeed no correlation of any kind — with any third party $C$ . Strong entanglement on one bond forbids it on every other. This trade-off is what "monogamy of entanglement" names, and it is one of the sharpest lines separating the quantum world from the classical one. Coffman, Kundu, and Wootters turned the slogan into a law in 2000. Using the concurrence $C$ and its square, the tangle $\tau=C^2$ , they proved that for three qubits $\tau_{A(BC)}\geq\tau_{AB}+\tau_{AC}$ : the entanglement $A$ shares with the pair $BC$ upper-bounds the sum of the entanglements it shares with $B$ and with $C$ separately. The deficit is not waste — it is a genuine tripartite invariant, the residual three-tangle $\tau_{ABC}$ , which measures GHZ-type entanglement that lives in all three parties at once and in no pair. Osborne and Verstraete generalized the bound to $n$ qubits in 2006. Operationally it is indispensable: it is why an eavesdropper cannot share the secret an entanglement-based key-distribution protocol establishes, why three parties cannot all violate a Bell inequality maximally, and why holographic states have the rigid entanglement geometry a smooth spacetime dual demands.

Key takeaways

Monogamy of entanglement is the principle that quantum correlations cannot be freely shared: if $A$ is maximally entangled with $B$ , it cannot be entangled at all with a third party $C$ .
The Coffman–Kundu–Wootters (CKW) inequality τ_A(BC) ≥ τ_AB + τ_AC makes this quantitative for three qubits, using the tangle (squared concurrence) as the currency.
The deficit defines the residual three-tangle τ_ABC ≥ 0, a genuine tripartite invariant equal to Cayley's hyperdeterminant of the amplitude array.
The canonical extremes are the GHZ state (all entanglement tripartite, τ_ABC = 1, no pairwise tangle) and the W state (all entanglement pairwise, τ_AB = τ_AC = 4/9, τ_ABC = 0).
Osborne and Verstraete extended CKW to n qubits; monogamy underlies quantum-key-distribution security, the bound S²_AB + S²_AC ≤ 8 on shared Bell violation, and holographic entanglement inequalities.
Monogamy is a property of the right measures: the tangle is monogamous, but bare concurrence and entanglement of formation are not.
Key idea: A pure bipartite state across a cut is uncorrelated with everything else, so strong entanglement on one bond forces weak entanglement on all others; the CKW inequality τ_A(BC) ≥ τ_AB + τ_AC quantifies the trade-off, and the residual three-tangle measures the genuine tripartite part.

The core insight

A pure bipartite state across a cut is uncorrelated with everything else; therefore strong entanglement on one bond forces weak entanglement on all others. The CKW inequality $\tau_{A(BC)} \geq \tau_{AB} + \tau_{AC}$ quantifies the trade-off, and the residual three-tangle $\tau_{ABC}$ measures the genuine tripartite entanglement that no pair can hold.

Where this lives today

Monogamy stopped being a curiosity of three-qubit algebra the moment it became a security proof. Device-independent quantum key distribution rests on it directly: the 2022 loophole-free demonstrations — trapped-ion and neutral-atom experiments that certified secret keys from a measured Bell violation alone — are monogamy in the laboratory, using $S_{AB}^2+S_{AC}^2\leq8$ to guarantee that no eavesdropper shares the correlation Alice and Bob just tested. In many-body physics, the Osborne–Verstraete inequality is why a central spin cannot entangle strongly with all its neighbours at once; this "entanglement frustration" underlies area laws and tensor-network ground states. And in quantum gravity, monogamy is a diagnostic: holographic states dual to a smooth spacetime obey the monogamy of mutual information, $I(A:BC)\geq I(A:B)+I(A:C)$ , which generic quantum states violate — a rigidity, proven for the Ryu–Takayanagi entropy, now read as a fingerprint of emergent geometry.

The physical picture

Picture entanglement as a finite, conserved "attention" a single qubit can pay. If $A$ pours all of it into a Bell pair with $B$ , none is left for $C$ . This is not a metaphor to discard once the equations arrive — it is exactly what the mathematics says. A maximally entangled $\rho_{AB}$ is pure; purity across the $AB\,|\,C$ cut forces the global state to factorize, $\rho_{ABC}=\rho_{AB}\otimes\rho_C$ ; a factorized state carries zero $A$ – $C$ entanglement. The classical world has no such conservation law: three coins glued to show the same face are perfectly correlated in every pair at once, because correlation is a resource you can broadcast. The first figure sets the two worlds side by side and states the budget that governs the quantum one.

Figure

The classical world lets one bit correlate with everyone at once; the quantum world does not — pour entanglement into the A–B bond and the A–C bond starves, subject to the CKW budget τ_AB + τ_AC ≤ τ_A(BC) ≤ 1

Two special three-qubit states show the two ways the budget can be spent, and they are the fixed stars of the subject. The GHZ state $\tfrac{1}{\sqrt2}(|000\rangle+|111\rangle)$ hides all its entanglement in the triple: trace out any one qubit and the remaining pair is a classically correlated mixture with zero tangle, yet $\tau_{ABC}=1$ . The W state $\tfrac{1}{\sqrt3}(|001\rangle+|010\rangle+|100\rangle)$ does the opposite, spreading entanglement as thinly and evenly as monogamy permits across the three pairs, $\tau_{AB}=\tau_{AC}=\tfrac49$ , and keeping nothing back, $\tau_{ABC}=0$ . The second figure draws the CKW budget as a stack of colours, one bar per canonical state — the same unit of entanglement partitioned three different ways.

Figure

The CKW budget τ_A(BC) = τ_AB + τ_AC + τ_ABC partitions one qubit's entanglement three ways — the W state spends 4/9 + 4/9 on pairwise bonds with zero residual, GHZ locks all of it into the three-tangle τ_ABC = 1, and a Bell pair tensored with C puts the whole unit into τ_AB

Why the squared concurrence and not the concurrence itself? Concurrence behaves like a length; the tangle behaves like an area. The additivity monogamy expresses is an additivity of these quadratic contributions — the same reason variances add while standard deviations do not. Feed the bare concurrence into " $E_{A(BC)}\geq E_{AB}+E_{AC}$ " and the statement is simply false; the correct exponent is what makes the budget close. This is the first hint that monogamy is a property of particular, well-chosen measures rather than a universal theorem about correlation.

The core equations

The central statement is the CKW inequality: the entanglement one qubit holds with the rest of a three-qubit system is at least the sum of the entanglements it holds with the two others individually, with the tangle $\tau=C^2$ as the bookkeeping quantity throughout.

\tau_{A(BC)} \;\geq\; \tau_{AB} + \tau_{AC} \qquad (1)

where: $\tau_{A(BC)}$ — the tangle of $A$ against the joint system $BC$ , equal to $C^2_{A|BC}$ ; $\tau_{AB}=C(\rho_{AB})^2$ — the pairwise tangle of $\rho_{AB}$ ; $\tau_{AC}=C(\rho_{AC})^2$ — the pairwise tangle of $\rho_{AC}$ . Reading: the "one against the rest" tangle is a ceiling the two pairwise tangles must fit under together. The left side has a clean closed form. For a pure three-qubit state the tangle across the $A\,|\,BC$ cut is fixed entirely by the single-qubit reduced state $\rho_A$ — through its purity, or equivalently its determinant.

\tau_{A(BC)} = 2\bigl(1 - \mathrm{Tr}\,\rho_A^2\bigr) = 4\det\rho_A \qquad (2)

where: $\rho_A=\mathrm{Tr}_{BC}\,|\psi\rangle\langle\psi|$ — the reduced state of qubit $A$ ; $\mathrm{Tr}\,\rho_A^2$ — its purity, equal to $1$ exactly when $A$ is unentangled with $BC$ ; $\det\rho_A=\lambda_1\lambda_2$ — the product of $\rho_A$ 's eigenvalues. Reading: a qubit maximally mixed on the cut ( $\lambda_1=\lambda_2=\tfrac12$ ) gives $\tau_{A(BC)}=1$ , a pure qubit gives $0$ . (Note the coefficient: $2(1-\mathrm{Tr}\,\rho_A^2)=4\det\rho_A$ , not $2\det\rho_A$ .) The gap in the CKW inequality is itself a physical quantity — a genuine tripartite entanglement monotone, invariant under permutations of the qubits and under local invertible operations.

\tau_{ABC} \;=\; \tau_{A(BC)} - \tau_{AB} - \tau_{AC} \;\geq\; 0 \qquad (3)

where: $\tau_{ABC}$ — the residual three-tangle, the entanglement shared by all three qubits and by no pair. Reading: $\tau_{ABC}=1$ for GHZ and $0$ for W and every biseparable state; it detects exactly the GHZ-type entanglement invisible to any two-qubit measure. Each pairwise tangle comes from Wootters' concurrence, the one closed-form entanglement measure available for an arbitrary mixed two-qubit state.

C(\rho_{AB}) = \max\bigl\{0,\ \sqrt{\mu_1} - \sqrt{\mu_2} - \sqrt{\mu_3} - \sqrt{\mu_4}\bigr\} \qquad (4)

where: $\mu_1\geq\mu_2\geq\mu_3\geq\mu_4$ — the eigenvalues of $\rho_{AB}\,(\sigma_y\otimes\sigma_y)\,\rho_{AB}^{*}\,(\sigma_y\otimes\sigma_y)$ ; $\tau_{AB}=C(\rho_{AB})^2$ — the tangle entering (1). Reading: the spin-flip $\sigma_y\otimes\sigma_y$ and the conjugate encode the antilinear structure of two-qubit entanglement; a separable $\rho_{AB}$ gives $C=0$ . Osborne and Verstraete showed the constraint is not special to three parties: for any number of qubits, the tangle of a focus qubit against everything else bounds the sum of its pairwise tangles.

\tau_{A(B_1 B_2 \cdots B_{n-1})} \;\geq\; \sum_{i=1}^{n-1} \tau_{A B_i} \qquad (5)

where: $A$ — the focus qubit; $B_i$ — the $i$ -th remaining qubit; $\tau_{AB_i}$ — the pairwise tangle of $\rho_{AB_i}$ . Reading: since the left side never exceeds $1$ for a qubit, $A$ cannot be strongly entangled with many partners at once — entanglement is distributed, never duplicated. Monogamy propagates beyond entanglement into nonlocality. If Alice runs the CHSH game with Bob and, using the same settings, simultaneously with Charlie, the two Bell parameters obey a Pythagorean bound.

S_{AB}^2 + S_{AC}^2 \;\leq\; 8 \qquad (6)

where: $S_{AB}, S_{AC}$ — the CHSH values Alice shares with Bob and with Charlie; $8=(2\sqrt2)^2$ — twice the square of Tsirelson's bound. Reading: a near-maximal violation $S_{AB}\to2\sqrt2$ drives $S_{AC}\to0$ — quantum nonlocality cannot be shared, and this bound holds for any state.

Derivation AThe one-versus-rest tangle is four times the determinant of the reduced qubit

The goal is to establish equation (2) from the ground up: to show that for a pure three-qubit state the entanglement across the cut separating $A$ from $BC$ is completely determined by the single-qubit reduced state $\rho_A$ , and specifically equals $2(1-\mathrm{Tr}\,\rho_A^2)=4\det\rho_A$ . This is the object on the left of the CKW inequality, so getting its normalization right is what makes the whole budget close numerically — and it is where a careless factor of two most often creeps in.

Step 1. Any pure bipartite state has a Schmidt decomposition, $|\psi\rangle_{A|BC}=\sqrt{\lambda_1}\,|1\rangle_A|u_1\rangle_{BC}+\sqrt{\lambda_2}\,|2\rangle_A|u_2\rangle_{BC}$ , and because $A$ is a single qubit there are at most two Schmidt terms. The reduced state is then $\rho_A=\mathrm{diag}(\lambda_1,\lambda_2)$ with $\lambda_1+\lambda_2=1$ . Why: the Schmidt rank across the cut is bounded by the smaller dimension, which is $2$ for a qubit, so the entire cut is captured by two numbers.

Step 2. The entanglement across the cut is the linear-entropy concurrence $C^2_{A|BC}=2(1-\mathrm{Tr}\,\rho_A^2)$ , the standard pure-state definition that reduces to the tangle. Compute the purity: $\mathrm{Tr}\,\rho_A^2=\lambda_1^2+\lambda_2^2$ . Why: the purity is the one invariant of a single-qubit density matrix that survives all local unitaries, so any legitimate one-qubit entanglement measure must be a function of it alone.

Step 3. Use the normalization $\lambda_1+\lambda_2=1$ , which gives $(\lambda_1+\lambda_2)^2=\lambda_1^2+2\lambda_1\lambda_2+\lambda_2^2=1$ , hence $\lambda_1^2+\lambda_2^2=1-2\lambda_1\lambda_2$ . Substitute: $1-\mathrm{Tr}\,\rho_A^2=1-(1-2\lambda_1\lambda_2)=2\lambda_1\lambda_2$ . Why: this is the algebraic heart of the identity — the purity and the determinant are two encodings of the same single eigenvalue product $\lambda_1\lambda_2$ .

Step 4. Multiply by the factor of two in the definition: $\tau_{A(BC)}=2\cdot 2\lambda_1\lambda_2=4\lambda_1\lambda_2=4\det\rho_A$ , since $\det\rho_A=\lambda_1\lambda_2$ . The result vanishes iff one $\lambda$ is zero — a product state — and maxes at $\lambda_1=\lambda_2=\tfrac12$ , giving $\tau_{A(BC)}=1$ . Why: fixing the coefficient at $4$ (not $2$ ) is exactly what makes the W-state budget $8/9=4/9+4/9$ and the GHZ budget $1=1$ come out right in the worked examples. *As a concrete check: for the W state $\rho_A=\mathrm{diag}(\tfrac23,\tfrac13)$ , the formula gives $4\cdot\tfrac23\cdot\tfrac13=4\cdot\tfrac29=\tfrac89$ , confirming that $2(1-\mathrm{Tr}\,\rho_A^2)=4\det\rho_A$ holds both symbolically and numerically.*

Go deeper · Derivation B · 1 minExtremal monogamy from the purity of a Bell pair

Derivation BExtremal monogamy from the purity of a Bell pair

The goal is to prove the strongest monogamy statement without any inequality at all: if $A$ and $B$ are exactly maximally entangled inside some larger — possibly mixed — state $\rho_{ABC}$ , then $A$ is not merely weakly entangled with $C$ but completely uncorrelated with it, and the whole system factorizes as a Bell pair tensored with an arbitrary $\rho_C$ . This is the boundary the CKW inequality must reduce to, and it exposes why purity, not any numerical measure, is the true engine of monogamy.

Step 1. Purify the state: introduce a reference system $E$ so that $\rho_{ABC}=\mathrm{Tr}_E\,|\Psi\rangle\langle\Psi|_{ABCE}$ for some global pure $|\Psi\rangle$ . Every mixed state admits such a purification. Why: monogamy arguments are cleanest for pure states, and purification lets us import the pure-state factorization theorem to a mixed $\rho_{ABC}$ .

Step 2. By assumption $\rho_{AB}=\mathrm{Tr}_{CE}\,|\Psi\rangle\langle\Psi|$ is a pure, maximally entangled Bell state $|\Phi^+\rangle\langle\Phi^+|$ . A reduced state of a global pure state is itself pure only when that subsystem factorizes from its complement — equivalently, the Schmidt rank of $|\Psi\rangle$ across the $AB\,|\,CE$ cut is one. Why: mixedness of a marginal is the unique signature of entanglement with the outside; a pure marginal means there is no such entanglement to begin with.

Step 3. Schmidt rank one across $AB\,|\,CE$ means the global state literally splits, $|\Psi\rangle_{ABCE}=|\Phi^+\rangle_{AB}\otimes|\chi\rangle_{CE}$ , with $|\chi\rangle$ some pure state of the remaining systems. Why: a rank-one Schmidt decomposition is by definition a tensor product, so the Bell pair sits in its own sealed sector.

Step 4. Trace out $E$ to recover the physical state: $\rho_{ABC}=|\Phi^+\rangle\langle\Phi^+|_{AB}\otimes\rho_C$ , where $\rho_C=\mathrm{Tr}_E\,|\chi\rangle\langle\chi|$ . Hence $\rho_{AC}=\rho_A\otimes\rho_C$ is a product, so $\tau_{AC}=0$ , and the CKW budget reads $1=1+0$ , saturated with no residual. Why: this is monogamy in its most uncompromising form — a perfect bond leaves literally nothing, not even classical correlation, for any third party. *This boundary is the sharpest case of CKW: $\tau_{AB}=1$ forces $\tau_{AC}=0$ and $\tau_{ABC}=0$ , and no state can do better.*

Go deeper · Derivation C · 1 minThe residual three-tangle and Cayley's hyperdeterminant

Derivation CThe residual three-tangle and Cayley's hyperdeterminant

The goal is to give the gap $\tau_{ABC}=\tau_{A(BC)}-\tau_{AB}-\tau_{AC}$ an independent meaning: to show it is not a leftover of one calculation but a symmetric polynomial invariant of the three-qubit amplitudes, equal to Cayley's hyperdeterminant, and therefore invariant under permuting the qubits and under local invertible (SL(2)) operations. That symmetry is what promotes the three-tangle from an accounting deficit to a genuine measure of tripartite entanglement.

Step 1. Write the pure state in the computational basis, $|\psi\rangle=\sum_{i,j,k\in\{0,1\}}a_{ijk}\,|ijk\rangle$ , so that the eight amplitudes $a_{ijk}$ form a $2\times2\times2$ array. The three-tangle is $\tau_{ABC}=4\,|\mathrm{Hdet}(a)|$ , the modulus of the unique degree-four invariant of that array under SL(2)×SL(2)×SL(2). Why: any genuine tripartite measure must be blind to local basis changes, and the hyperdeterminant is the lowest-degree quantity with exactly that invariance.

Step 2. Expanding the hyperdeterminant collects the amplitudes into three groups, $\tau_{ABC}=4\,|d_1-2d_2+4d_3|$ , with $d_1=a_{000}^2a_{111}^2+a_{001}^2a_{110}^2+a_{010}^2a_{101}^2+a_{100}^2a_{011}^2$ , $d_2$ the six-term product sum $a_{000}a_{111}a_{011}a_{100}+\cdots$ , and $d_3=a_{000}a_{110}a_{101}a_{011}+a_{111}a_{001}a_{010}a_{100}$ . The prefactor $4$ normalizes GHZ to $\tau_{ABC}=1$ . Why: writing the invariant out makes its permutation symmetry manifest and lets one evaluate it on any state directly, without computing reduced density matrices.

Step 3. Evaluate the extremes. For GHZ only $a_{000}=a_{111}=1/\sqrt2$ are nonzero, so $d_1=\tfrac14$ , $d_2=d_3=0$ , and $\tau_{ABC}=4\cdot\tfrac14=1$ . For W the nonzero amplitudes are $a_{001}=a_{010}=a_{100}=1/\sqrt3$ , none of which populate the paired index patterns in $d_1,d_2,d_3$ , so every group vanishes and $\tau_{ABC}=0$ . Why: the two canonical states are precisely the two SL(2)-inequivalent classes of genuine three-qubit entanglement, and the hyperdeterminant separates them cleanly — nonzero for GHZ, zero for W. *The CKW gap $\tau_{A(BC)}-\tau_{AB}-\tau_{AC}$ equals $4|\mathrm{Hdet}(a)|$ on both canonical states: GHZ gives $1-0-0=1$ and W gives $8/9-4/9-4/9=0$ .*

Limiting cases

The whole subject is anchored by how the CKW budget partitions among a few canonical states.

Case	Result	Meaning
GHZ state	$\tau_{AB}=\tau_{AC}=0,\ \tau_{A(BC)}=1,\ \tau_{ABC}=1$	all entanglement is genuinely tripartite; tracing out any qubit leaves a separable pair.
W state	$\tau_{AB}=\tau_{AC}=\tfrac49,\ \tau_{A(BC)}=\tfrac89,\ \tau_{ABC}=0$	pairwise entanglement maximized subject to monogamy, with no residual three-tangle.
Bell pair with spectator	$\tau_{AB}=1$ saturates $\tau_{A(BC)}=1$	forces $\tau_{AC}=0,\ \tau_{ABC}=0$ : extremal monogamy, no room for $A$ to entangle with $C$ .
Fully separable	every tangle $=0$	CKW holds trivially as $0\geq0+0$ ; no entanglement to distribute.
Central qubit, $n-1$ neighbours	$\sum_i \tau_{AB_i}\leq \tau_{A(B_1\cdots B_{n-1})}\leq 1$	Osborne–Verstraete: a spin cannot bond strongly to many neighbours at once — entanglement frustration.

Common trapMonogamy is a law about the right measure, not about all correlations

Two traps hide inside the word "monogamy." First, it is a property of specific measures, not a universal theorem. The squared concurrence (tangle) is monogamous for qubits, but the bare concurrence is not, and neither is the entanglement of formation in raw form — substitute the wrong measure into " $E_{A(BC)}\geq E_{AB}+E_{AC}$ " and you manufacture false statements. Second, monogamy constrains entanglement, not correlation of every kind. Classical correlations remain perfectly shareable, so monogamy does not forbid $A$ from being classically correlated with a whole crowd; it only forbids the quantum part from being copied. Confusing "cannot share entanglement" with "cannot share information" is the most common error in the subject.

Worked examples

1 · The W state saturates the budget (fully worked)

Take $|W\rangle=\tfrac{1}{\sqrt3}(|001\rangle+|010\rangle+|100\rangle)$ . Compute $\tau_{A(BC)}$ , the pairwise tangles, and the residual three-tangle, and confirm CKW.

Step 1. Reduce to qubit $A$ . Across the three branches $A$ reads $0,0,1$ , so $\rho_A=\mathrm{diag}(\tfrac23,\tfrac13)$ with purity $\mathrm{Tr}\,\rho_A^2=\tfrac49+\tfrac19=\tfrac59$ .

Step 2. The one-versus-rest tangle follows from equation (2): $\tau_{A(BC)}=2(1-\tfrac59)=\mathbf{\tfrac89}$ , and as a cross-check $4\det\rho_A=4\cdot\tfrac23\cdot\tfrac13=\tfrac89$ .

Step 3. The pairwise state $\rho_{AB}=\mathrm{Tr}_C|W\rangle\langle W|$ has support on $|00\rangle,|01\rangle,|10\rangle$ ; Wootters' formula (4) gives $C(\rho_{AB})=\tfrac23$ , so $\tau_{AB}=(\tfrac23)^2=\mathbf{\tfrac49}$ , and by permutation symmetry $\tau_{AC}=\tfrac49$ .

Step 4. The residual is $\tau_{ABC}=\tfrac89-\tfrac49-\tfrac49=\mathbf{0}$ . The W state saturates CKW — it packs the maximum pairwise entanglement compatible with the budget and holds nothing back as tripartite entanglement. The budget reads 8/9 = 4/9 + 4/9 + 0.

2 · Monogamy of Bell violation (fully worked)

Alice shares a qubit with Bob and, using the same settings, another with Charlie. She measures a near-Tsirelson $S_{AB}=2.7$ with Bob. What is the most she can violate CHSH with Charlie?

Step 1. The Toner–Verstraete relation (6) bounds the two violations jointly: $S_{AB}^2+S_{AC}^2\leq8$ , holding for any quantum state whatever Alice, Bob, and Charlie share.

Step 2. Insert the measured value: $S_{AC}^2\leq8-(2.7)^2=0.71$ , so $S_{AC}\leq\sqrt{0.71}=\mathbf{0.84}$ .

Step 3. Since $0.84<2$ , Alice's correlation with Charlie cannot even reach the classical CHSH bound of $2$ — it is provably non-violating. As $S_{AB}\to2\sqrt2$ , the ceiling collapses to $S_{AC}\to0$ . A strong violation with Bob forces $S_{AC}\leq0.84$ , below the classical bound — nonlocality is monogamous.

3 · The eavesdropper's dilemma in QKD

Alice and Bob verify that their shared qubits are close to a Bell pair, with tangle $\tau_{AB}\approx0.98$ . Bound the tangle Eve can hold with Alice's qubit.

Step 1. Model Eve as a third party $E$ , so that $A$ , $B$ , $E$ form a tripartite state. CKW applied to focus qubit $A$ gives $\tau_{A(BE)}\geq\tau_{AB}+\tau_{AE}$ , with $\tau_{A(BE)}\leq1$ .

Step 2. Rearrange: $\tau_{AE}\leq\tau_{A(BE)}-\tau_{AB}\leq1-0.98=\mathbf{0.02}$ . The tighter the certified Alice–Bob bond, the less Eve can know — monogamy converts a measured correlation between the legitimate parties directly into a cap on any eavesdropper's, with no assumption about Eve's hardware. A certified $\tau_{AB}\approx0.98$ caps Eve at $\tau_{AE}\leq0.02$ .

Physics of the portrait

Monogamy of entanglement is inseparable from one physicist. William Wootters spent his career finding the exact invariants that make entanglement computable — the no-cloning theorem, the concurrence, and the monogamy law that binds them.

Origins · 1 minWilliam Wootters and how this was discovered

William Wootters

born 1956, American physicist · entanglement measures and quantum information

William Kent Wootters has a rare gift for turning slippery quantum intuitions into exact formulas. As a graduate student he and Wojciech Zurek proved the no-cloning theorem in 1982 — that an unknown quantum state cannot be copied — which is a first cousin of monogamy: if you cannot clone a state, you cannot broadcast its entanglement either. In 1998, with Scott Hill, he gave a closed-form expression for the entanglement of formation of an arbitrary two-qubit state, built on a quantity he named the concurrence — the formula, with its eerie spin-flip and complex conjugation, that makes pairwise entanglement computable and feeds every tangle in this article. Then in 2000, with Valerie Coffman and Joydip Kundu, he proved that these numbers obey a conservation law — the CKW inequality — and defined the residual three-tangle as the genuine tripartite invariant lurking in the deficit. The thread through all three results is the same: entanglement is a resource with a strict accounting, and Wootters found the ledger.

History

1982

— Wootters and Zurek prove the no-cloning theorem

1998

— Hill and Wootters give the closed-form concurrence for two-qubit entanglement of formation

2000

— Coffman, Kundu, and Wootters prove the CKW monogamy inequality and define the three-tangle

2006

— Osborne and Verstraete extend monogamy to n qubits

2006

— Toner and Verstraete establish the monogamy of CHSH (Bell) violation

Going further

Texts. Coffman, Kundu, Wootters, "Distributed entanglement," Phys. Rev. A 61, 052306 (2000) · Osborne, Verstraete, "General monogamy inequality for bipartite qubit entanglement," Phys. Rev. Lett. 96, 220503 (2006) · Toner, Verstraete, "Monogamy of Bell correlations and Tsirelson's bound," arXiv:quant-ph/0611001 · Nielsen & Chuang, Quantum Computation and Quantum Information, §12.5 · Preskill, Lecture Notes on Quantum Computation, Ch. 4 · Wilde, Quantum Information Theory, Ch. 24 · Horodecki, Horodecki, Horodecki, Horodecki, "Quantum entanglement," Rev. Mod. Phys. 81, 865 (2009). Key numerical anchors: the one-versus-rest tangle is $\tau_{A(BC)}=2(1-\mathrm{Tr}\,\rho_A^2)=4\det\rho_A$ (note the coefficient $4$ , not $2$ ); the W-state budget reads $\tau_{A(BC)}=8/9$ , $\tau_{AB}=\tau_{AC}=4/9$ , $\tau_{ABC}=0$ ; and the GHZ budget reads $\tau_{A(BC)}=1$ , pairwise tangles $0$ , $\tau_{ABC}=1$ .

Prerequisites. Concurrence and Wootters' Formula · Separable versus Entangled States · Schmidt Decomposition · Bell States and EPR Pairs. Leads to. CHSH Inequality and Its Quantum Violation · Entanglement Entropy of Pure States · Nonlocality and the No-Signalling Principle.

Ask the tutor: Ask me to walk through the CKW budget for any three-qubit state of your choice — pick the amplitudes and we will compute every tangle and the residual together.

Is monogamy of entanglement a fundamental law or a property of one measure?

Both, depending on what you demand of it. The physical statement — a pure bipartite state factorizes from the rest of the universe, so a perfect Bell pair leaves nothing for a third party — is a theorem, exact and universal. What is measure-dependent is the quantitative version. Turning "cannot share strongly" into an inequality " $E_{A(BC)}\geq E_{AB}+E_{AC}$ " requires the right $E$ : the squared concurrence works for qubits, but the bare concurrence and the raw entanglement of formation do not. So monogamy is a robust physical fact whose sharp numerical form only certain, well-chosen invariants express.

Why the tangle and not the concurrence — what breaks?

The additivity monogamy expresses is an additivity of quadratic quantities, the way variances add while standard deviations do not. Concretely, for the W state the pairwise concurrences are each $2/3$ , so $C_{AB}+C_{AC}=4/3$ , which exceeds the one-versus-rest concurrence $C_{A(BC)}=\sqrt{8/9}\approx0.94$ — the bare-concurrence "budget" is violated. Square everything and it closes exactly: $4/9+4/9=8/9$ . The squaring is not cosmetic; it is what makes the conserved quantity conserved.

What is the residual three-tangle actually measuring?

It measures entanglement that lives in all three qubits jointly and in no pair — the irreducibly tripartite part. A GHZ state is the pure example: every two-qubit marginal is a separable mixture with zero tangle, yet the three qubits are maximally correlated as a unit, giving $\tau_{ABC}=1$ . The three-tangle is a genuine entanglement monotone, invariant under permuting the qubits and under local invertible operations, and it equals the modulus of Cayley's hyperdeterminant of the amplitude array. It is zero for the entire W class, which is precisely why GHZ and W are the two inequivalent species of three-qubit entanglement.

Does monogamy forbid a qubit from correlating with many parties?

Only its entanglement, never its classical correlation. A qubit can be classically correlated with a whole crowd — a broadcast, a shared random key — and monogamy says nothing against it. What is rationed is the quantum resource: the sum of a qubit's pairwise tangles is capped by its total, so strong entanglement with one partner necessarily weakens it with all others. The clean way to hold the distinction is that classical correlation can be copied and entanglement cannot, and monogamy is the quantitative face of that no-cloning fact.

How does monogamy make quantum key distribution secure?

In an entanglement-based protocol, Alice and Bob certify via a Bell test that their qubits are close to a maximally entangled pair. Monogamy then bounds any eavesdropper Eve: if Alice's tangle with Bob is near one, her tangle with Eve is forced near zero, since the total budget is fixed. Device-independent proofs run this on the nonlocal correlations directly, using $S_{AB}^2+S_{AC}^2\leq8$ , so the guarantee holds even if Eve built the devices — the measured violation alone caps her information.

Does monogamy extend beyond qubits and beyond three parties?

Yes, with caveats. Osborne and Verstraete proved the tangle inequality for any number of qubits, $\tau_{A(B_1\cdots B_{n-1})}\geq\sum_i\tau_{AB_i}$ , so a central qubit's entanglement is always a distributed, capped budget. Higher-dimensional systems (qudits) are subtler: the squared concurrence is not automatically monogamous there, and one often needs different measures or higher exponents $\alpha\geq2$ . The general lesson is that monogamy is real in every dimension, but the measure that expresses it cleanly must fit the system.