Entanglement Swapping

Entanglement swapping answers one of the most startling questions quantum theory allows: can two particles that have never met, never interacted, never shared a source, nonetheless end up entangled? The answer is yes, and the recipe is disarmingly economical. Take two independent Einstein–Podolsky–Rosen pairs, $(A,B)$ from one source and $(C,D)$ from another. Send $A$ and $D$ to distant parties and bring the inner partners $B$ and $C$ together for a single joint measurement in the Bell basis. That one measurement severs the original links and welds a new one: $A$ and $D$ , which share no past, emerge maximally entangled. No entanglement was created from nothing — it was rerouted, teleported from where it was to where it is wanted. The mechanism is a corollary of quantum teleportation seen from an unusual angle. Teleportation uses a Bell measurement to move an unknown state onto a distant qubit; in swapping the teleported qubit is itself half of an entangled pair, so what moves is not a state but a correlation — $D$ inherits the entanglement $B$ had with $A$ . Two features organize everything that follows. First, the protocol is probabilistic in which Bell outcome occurs — all four equally likely at $1/4$ — yet deterministic in success, since every outcome leaves $A$ – $D$ maximally entangled, related to a fixed target by one local Pauli correction once the outcome is announced. Second, nothing travels faster than light: before that announcement the marginal state of $A,D$ is maximally mixed. Swapping is the beating heart of the quantum repeater, the primitive that turns short, high-fidelity links into continent-spanning entanglement.

Key takeaways

A joint Bell-state measurement on the inner partners $B,C$ of two independent pairs $(A,B)$ and $(C,D)$ projects the outer particles $A,D$ into an entangled state — although they never interacted.
It is teleportation of one half of an entangled pair: the Bell measurement transfers not a single-qubit state but the entanglement that qubit carried, leaving $A$ entangled with $D$ .
The four Bell outcomes each occur with probability $1/4$ ; every outcome leaves $A$ – $D$ in a definite Bell state, deterministic up to a local Pauli correction conditioned on the announced result.
No information moves superluminally: without the classical outcome, tracing out $B,C$ leaves $A,D$ in the maximally mixed state $I/4$ — separable and featureless.
Cascaded swaps are the elementary step of a quantum repeater, but imperfections compound as $v_{\text{out}}=v^{N}$ , so real networks interleave entanglement purification; the primitive underpins device-independent QKD, loophole-free Bell tests, and the quantum internet.
Key idea: Re-express the product of two Bell states in the Bell basis of the inner particles — a Bell measurement on those inner particles factorizes the four-qubit state so that each of the four equiprobable outcomes leaves the outer particles in a corresponding Bell state.

The core insight

Re-express the product of two Bell states in the Bell basis of the inner particles: a Bell measurement on those inner particles factorizes the four-qubit state so each of the four equiprobable outcomes leaves the outer particles in a corresponding Bell state, related to a fixed target by one local Pauli correction.

Where this lives today

Entanglement swapping has moved from proof-of-principle to infrastructure. At Delft, nitrogen-vacancy centres in diamond emit entangled photons whose mid-point Bell measurement swaps entanglement onto the distant spins; this heralded link — extended to a three-node chain in 2021 — is the first rudimentary quantum network, its rate set by the visibility decay of the second figure. The same primitive drives DLCZ memory-based and "all-photonic" cluster-state repeaters. Swapping's heralding is what made the 2015 loophole-free Bell tests possible: a successful inner measurement announces that a clean entangled pair exists, closing the detection and locality loopholes at once. And entanglement-based QKD such as E91 and its device-independent descendants use swapping to distribute pairs across repeater nodes without any node holding a copy of the key. Open frontiers: multiplexed swapping at metropolitan and satellite scales, and deterministic Bell analyzers that beat the $50\%$ linear-optics ceiling.

The physical picture

Picture two short chains of entanglement end to end: $A$ tied to $B$ , $C$ tied to $D$ , the chains entirely separate. Bring the inner links $B$ and $C$ into one apparatus and fuse them with a Bell-state measurement. The fusion destroys the old $A$ – $B$ and $C$ – $D$ correlations (those qubits are consumed) and forges a brand-new $A$ – $D$ correlation across the whole span. Entanglement is conserved bookkeeping, not conjured substance — one ebit before, one ebit after, relocated onto a longer baseline. The first figure lays out this "weld two links into one" picture with the identity that makes it work.

Figure

Two independent sources each emit a Φ⁺ pair; qubits A and D fly to distant parties while the inner qubits B and C meet at a Bell-state measurement that fuses the two short links into one long A–D link, with the swap identity Φ⁺(AB) ⊗ Φ⁺(CD) = ½ Σ β_k(BC) ⊗ β_k(AD) giving each of the four Bell outcomes probability ¼

A second, sharper reading sees the Bell measurement as teleporting $B$ onto $D$ : because $B$ carries no definite state — it is entangled with $A$ — its "message" is the correlation with $A$ itself, which linear teleportation transplants intact onto $A$ – $D$ (Derivation 2). The classical outcome is the wiring label naming which of four Pauli-related Bell states the far parties hold, and it must be transmitted before the pair is usable — averaged over the four unknown outcomes, $A$ and $D$ sit in $I/4$ , indistinguishable from two never-entangled qubits, nothing changing locally at the distant measurement. So the quantum repeater is limited not by causality but by how fast imperfections accumulate down a chain — the subject of the second figure.

Figure

Cascaded entanglement swapping degrades a Werner state's visibility multiplicatively as v_out = vᴺ over N segments — a near-ideal chain at v = 0.99 (green) still holds 0.818 after 20 swaps, a v = 0.95 chain (blue) falls to 0.599 by ten swaps, and a v = 0.90 chain (red) drops below the separability threshold v = 1/3 by about the eleventh swap, which is why real repeaters interleave entanglement purification

The core equations

The starting resource is a maximally entangled Bell state, one from each source; taking $|\Phi^+\rangle$ as the standard is only a convention — the algebra runs identically for any Bell input.

|\Phi^+\rangle = \tfrac{1}{\sqrt{2}}\big(|00\rangle + |11\rangle\big) \qquad (1)

where: $|\Phi^+\rangle$ — the canonical maximally entangled Bell state from each source; $|00\rangle,|11\rangle$ — computational-basis product states. Reading: perfectly correlated in both $Z$ and $X$ bases, the signature of one ebit. The protocol begins with two such pairs, produced independently and combined as a tensor product. The labelling matters: $A,D$ are the outer qubits that fly apart, $B,C$ the inner qubits to be jointly measured.

|\psi\rangle_{ABCD} = |\Phi^+\rangle_{AB}\otimes|\Phi^+\rangle_{CD} \qquad (2)

where: $A,B$ — source 1 ( $A$ outer, $B$ inner); $C,D$ — source 2 ( $D$ outer, $C$ inner). Reading: a pure product with all correlation within each pair, none between them. The Bell states form an orthonormal basis of the two-qubit space. The inner measurement projects $B,C$ onto one of these four vectors; expanding $|\psi\rangle_{ABCD}$ in this basis reveals the swap.

\{|\Phi^\pm\rangle,|\Psi^\pm\rangle\}=\Big\{\tfrac{|00\rangle\pm|11\rangle}{\sqrt2},\ \tfrac{|01\rangle\pm|10\rangle}{\sqrt2}\Big\} \qquad (3)

where: $|\Phi^\pm\rangle$ — even-parity Bell states; $|\Psi^\pm\rangle$ — odd-parity Bell states. Reading: four orthogonal maximally entangled states spanning the two-qubit space. The central identity re-expresses the four-qubit product as a sum over the inner Bell basis, each term tensored with an outer Bell state (Derivation 1). The content: measuring $B,C$ sorts the state into four equally weighted branches.

|\psi\rangle_{ABCD} = \tfrac12\sum_{k} |\beta_k\rangle_{BC}\otimes\big(\sigma_k\otimes I\big)|\Phi^+\rangle_{AD} \qquad (4)

where: $|\beta_k\rangle_{BC}$ — the $k$ -th inner Bell state; $\sigma_k\in\{I,Z,X,ZX\}$ — the Pauli correction for outcome $k$ , on $A$ ; $|\Phi^+\rangle_{AD}$ — the fixed target. Reading: outcome $k$ leaves $A,D$ in $(\sigma_k\otimes I)|\Phi^+\rangle$ . The Pauli operators relate the four Bell states, which is why one one-sided correction suffices: applying $\sigma_k^{-1}$ on $A$ maps any outcome back to the standard target.

|\Phi^-\rangle = (Z\otimes I)|\Phi^+\rangle,\quad |\Psi^+\rangle = (X\otimes I)|\Phi^+\rangle,\quad |\Psi^-\rangle = (ZX\otimes I)|\Phi^+\rangle \qquad (5)

where: $X,Z$ — Pauli flip and phase operators; $\otimes I$ — identity on the partner. Reading: one-sided Pauli action moves between all four Bell states, so learning $k$ suffices to correct the pair. Because all four branches in equation (4) carry amplitude $1/2$ , the Born rule assigns equal probability — probabilistic in which Bell state appears, certain that some maximally entangled state always does.

P(k) = \big|\tfrac12\big|^2 = \tfrac14,\qquad k\in\{\Phi^+,\Phi^-,\Psi^+,\Psi^-\} \qquad (6)

where: $P(k)$ — probability of the $k$ -th outcome on $B,C$ . Reading: a flat distribution over the four outcomes. Averaging over an unrecorded outcome collapses this to $\rho_{AD}^{\text{no comm}}=\sum_k P(k)\,|\beta_k\rangle_{AD}\langle\beta_k|=I_{AD}/4=(I_A/2)\otimes(I_D/2)$ , since the four Bell projectors sum to the identity. This maximally mixed, separable state is the algebraic statement that swapping cannot signal — derived in full below.

Derivation AThe swap identity by expansion in the inner Bell basis

The goal is to prove equation (4) from scratch — that the product of two independent Bell pairs, with the inner qubits re-expressed in their own Bell basis, factorizes into a perfectly correlated sum pairing each inner Bell state one-to-one with an outer Bell state at equal amplitude. This single fact is the whole phenomenon: measuring the inner pair deterministically projects the outer pair into a matching maximally entangled state, fixing outcome probabilities and corrections at once. We expand the product, regroup by inner versus outer qubits, and convert back to the Bell basis.

Step 1. Write the product in the computational basis, ordering $A,B,C,D$ : $|\Phi^+\rangle_{AB}|\Phi^+\rangle_{CD}=\tfrac12(|00\rangle_{AB}+|11\rangle_{AB})(|00\rangle_{CD}+|11\rangle_{CD})=\tfrac12(|0000\rangle+|0011\rangle+|1100\rangle+|1111\rangle)$ . Why: everything downstream is just a regrouping of these four equally weighted terms — no approximation enters.

Step 2. Regroup each term by the inner pair $(B,C)$ and outer pair $(A,D)$ : the four kets become $|00\rangle_{BC}|00\rangle_{AD}$ , $|01\rangle_{BC}|01\rangle_{AD}$ , $|10\rangle_{BC}|10\rangle_{AD}$ , $|11\rangle_{BC}|11\rangle_{AD}$ , so $|\psi\rangle_{ABCD}=\tfrac12\sum_{x,y}|xy\rangle_{BC}|xy\rangle_{AD}$ . Why: inner and outer registers are perfectly correlated in the computational basis — the seed of the correlation the measurement exposes.

Step 3. Convert both registers to the Bell basis via $|00\rangle=\tfrac{|\Phi^+\rangle+|\Phi^-\rangle}{\sqrt2}$ , $|11\rangle=\tfrac{|\Phi^+\rangle-|\Phi^-\rangle}{\sqrt2}$ , $|01\rangle=\tfrac{|\Psi^+\rangle+|\Psi^-\rangle}{\sqrt2}$ , $|10\rangle=\tfrac{|\Psi^+\rangle-|\Psi^-\rangle}{\sqrt2}$ ; cross terms cancel pairwise, leaving $\tfrac12\sum_k|\beta_k\rangle_{BC}|\beta_k\rangle_{AD}$ . Why: the identity $\sum_{x,y}|xy\rangle|xy\rangle=\sum_k|\beta_k\rangle|\beta_k\rangle$ is the mathematical core — it holds because the Bell basis is a real orthogonal rotation of the computational basis.

Step 4. Apply the Pauli relations of equation (5) to write each outer Bell state as $(\sigma_k\otimes I)|\Phi^+\rangle_{AD}$ , recovering equation (4). Projecting $B,C$ onto $|\beta_k\rangle$ leaves $A,D$ in $(\sigma_k\otimes I)|\Phi^+\rangle$ with amplitude $1/2$ , hence probability $1/4$ . Why: this is the swap — a Bell measurement on qubits that never touched $A$ or $D$ nonetheless dictates, outcome by outcome, the exact entangled state of the outer pair. *Each branch amplitude is $1/2$ , so each norm is $|1/2|^2=1/4$ and the four norms sum to $1$ , confirming the state is correctly normalised.*

Go deeper · Derivation B · 1 minSwapping as teleportation of half an entangled pair

Derivation BSwapping as teleportation of half an entangled pair

The goal is to re-derive the outer state without expanding the four-qubit product, by recognizing swapping as ordinary teleportation applied to an entangled input. This explains why swapping's classical communication and Pauli corrections match teleportation's, and makes the conservation of entanglement transparent: teleportation transports whatever the input carried, and here that is a correlation rather than a definite state. We start from the teleportation projection, promote the input to half of a Bell pair, and push it through by linearity.

Step 1. Recall the teleportation primitive: projecting $B,C$ onto $|\Phi^+\rangle$ transfers an arbitrary single-qubit state from $B$ to $D$ , $\langle\Phi^+|_{BC}\big(|\phi\rangle_B|\Phi^+\rangle_{CD}\big)=\tfrac12|\phi\rangle_D$ . Why: a Bell projection copies $B$ 's amplitudes onto $D$ , up to the normalization $1/2$ that reappears as probability $1/4$ .

Step 2. Make the input entangled instead of definite: $|\Phi^+\rangle_{AB}=\tfrac{1}{\sqrt2}\sum_x|x\rangle_A|x\rangle_B$ , so the "state of $B$ " is a superposition correlated with $A$ . Why: this is the only structural change from teleportation to swapping — the teleported qubit is half of a Bell pair, so its partner $A$ rides along.

Step 3. Apply the same $BC$ projection and use linearity to pull the $A$ sum outside: $\langle\Phi^+|_{BC}|\Phi^+\rangle_{AB}|\Phi^+\rangle_{CD}=\tfrac{1}{\sqrt2}\sum_x|x\rangle_A\,\langle\Phi^+|_{BC}|x\rangle_B|\Phi^+\rangle_{CD}$ . Why: teleportation is linear, so it acts term by term, teleporting each $|x\rangle_B$ to $|x\rangle_D$ while $A$ is untouched.

Step 4. Each inner projection sends $|x\rangle_B\to\tfrac12|x\rangle_D$ , giving $\tfrac{1}{\sqrt2}\sum_x|x\rangle_A\cdot\tfrac12|x\rangle_D=\tfrac12|\Phi^+\rangle_{AD}$ : the correlation that bound $A$ to $B$ now binds $A$ to $D$ . Why: the entanglement was transplanted, not created; other outcomes give the Pauli-related Bell states of equation (5). *As a consistency check, $\big\|\tfrac12|\Phi^+\rangle_{AD}\big\|^2=1/4=P(\Phi^+)$ , in agreement with the equiprobable outcome law.*

Go deeper · Derivation C · 1 minNo-signalling from the unconditioned reduced state

Derivation CNo-signalling from the unconditioned reduced state

The goal is to prove that swapping cannot transmit information faster than light, by showing that without the classical outcome the outer pair is in the maximally mixed state $I/4$ — statistically indistinguishable from two never-entangled qubits. This reconciles the "spooky" language of swapping with relativity: the distant measurement changes no local statistic at $A$ or $D$ . We average over outcomes and use completeness of the Bell basis.

Step 1. From the swap identity, outcome $k$ leaves $A,D$ in the pure Bell state $|\beta_k\rangle_{AD}$ with probability $P(k)=1/4$ . Why: this is the conditional description — correct only for an observer who knows $k$ .

Step 2. An experimenter who has not received the classical result must average over the unknown outcome, giving the incoherent mixture $\rho_{AD}=\tfrac14\sum_k|\beta_k\rangle\langle\beta_k|$ . Why: ignorance of a genuinely random outcome is a classical mixture, not a superposition — the proper state is the ensemble average.

Step 3. The four Bell projectors resolve the identity, $\sum_k|\beta_k\rangle\langle\beta_k|=I_{AD}$ , so $\rho_{AD}=I_{AD}/4$ ; equivalently, tracing the original product, $\mathrm{Tr}_{BC}|\psi\rangle\langle\psi|=(I_A/2)\otimes(I_D/2)$ . Why: both routes agree because the Bell basis is complete — the averaged state carries no imprint of the measurement.

Step 4. Since $I/4=(I_A/2)\otimes(I_D/2)$ is separable and equals two independent maximally mixed qubits, no local measurement at $A$ or $D$ can detect whether — or when — the Bell measurement happened. Why: a change with no local consequence transmits no information; the entanglement is real but certified only once the classical bits arrive. *As a quick check: $\mathrm{Tr}\,\rho_{AD}=1$ and $\rho_{AD}^2=I/16=\tfrac14\rho_{AD}$ , confirming that the purity is $1/4$ — the hallmark of a maximally mixed state.*

Limiting cases

Case	Result	Meaning
Ideal pair, ideal measurement	$A$ – $D$ in $\|\beta_k\rangle$ , $P(k)=\tfrac14$	perfect swap: each outcome yields one ebit across the long baseline
No classical communication	$\rho_{AD}=\mathrm{Tr}_{BC}[\cdots]=\tfrac{I}{4}$	outer pair maximally mixed and separable — no superluminal signalling
Linear-optics Bell analyzer	$p_{\text{success}}=\tfrac12$	only $\|\Psi^+\rangle,\|\Psi^-\rangle$ are distinguishable with beam splitters and detectors, capping heralded swaps at $50\%$
Imperfect Werner inputs	$v_{\text{out}}=v_1 v_2$	swapping two visibility- $v$ pairs multiplies visibilities; degradation compounds
Cascaded repeater chain	$v_{\text{out}}=v^{N}$	$N$ imperfect swaps decay visibility exponentially — separable once $v^N\le\tfrac13$ , forcing purification

Common trapthe misconception that swapping creates usable entanglement instantly and at a distance

The dramatic phrasing — "particles that never met become entangled the moment $B$ and $C$ are measured" — invites a false inference: that a usable, detectable correlation springs into being at $A$ and $D$ instantaneously, which would let you signal. It does not. Before the outcome is broadcast, tracing out $B,C$ leaves $A,D$ in the maximally mixed state $I/4$ , a separable state identical to two never-entangled qubits; every local statistic at $A$ or $D$ is unchanged whether or not the distant measurement happened. The entanglement becomes usable only once the outcome $k$ is communicated classically — at or below light speed — and the Pauli correction applied. Swapping is not communication-free creation of entanglement at a distance; it is a heralded re-routing whose herald travels classically — which is why it powers quantum networks without threatening relativity.

Worked examples

1 · A single run of the protocol, outcome by outcome (fully worked)

Two sources emit $|\Phi^+\rangle_{AB}$ and $|\Phi^+\rangle_{CD}$ . Alice holds $A$ , Bob holds $D$ ; a middle station Bell-measures $B,C$ and announces $|\Psi^+\rangle$ . What state do $A,D$ share, what correction restores $|\Phi^+\rangle$ , and what would Alice have seen before the announcement?

Step 1. Read the swapped state off the swap identity, equation (4): outcome $|\Psi^+\rangle_{BC}$ pairs with the outer state $(X\otimes I)|\Phi^+\rangle_{AD}=|\Psi^+\rangle_{AD}$ . Alice and Bob now share $\mathbf{|\Psi^+\rangle_{AD}=\tfrac{1}{\sqrt2}(|01\rangle+|10\rangle)}$ — maximally entangled, though $A$ and $D$ came from different sources and never interacted.

Step 2. The outcome probability is $P(\Psi^+)=|1/2|^2=\mathbf{1/4}$ , and by symmetry each outcome is equally likely; the swap succeeds every time, since some Bell state always results.

Step 3. To standardize, Alice applies $\sigma^{-1}=X$ to $A$ : since $|\Psi^+\rangle=(X\otimes I)|\Phi^+\rangle$ , applying $X$ again gives $|\Phi^+\rangle_{AD}$ . Two classical bits naming the outcome plus one Pauli flip convert any run into the target $|\Phi^+\rangle$ .

Step 4. Before the announcement, Alice's marginal is $\rho_A=\mathrm{Tr}_D|\Psi^+\rangle\langle\Psi^+|=I/2$ (likewise $\rho_D=I/2$ ) — completely random, even averaged over the four unannounced outcomes. She learns nothing about whether the swap happened until the classical result reaches her — no-signalling in action.

2 · Why a repeater is unavoidable, and why it needs purification (fully worked)

Compare direct entanglement distribution over $L=1000\ \mathrm{km}$ of fibre (attenuation length $L_{\text{att}}=22\ \mathrm{km}$ ) against a chain of $N=10$ segments joined by swaps. Then, treating each segment's pair as a Werner state of visibility $v=0.95$ , find the end-to-end visibility and whether the final pair is still entangled.

Step 1. Direct transmission succeeds with probability $\eta=e^{-L/L_{\text{att}}}=e^{-1000/22}\approx\mathbf{1.8\times10^{-20}}$ . At a source rate of $10^{10}/\mathrm{s}$ , that is one delivered pair every $\approx5.5\times10^{9}\ \mathrm{s}\approx\mathbf{175\ years}$ — utterly hopeless.

Step 2. Split the path into $N=10$ segments of $100\ \mathrm{km}$ . Each now has transmittance $e^{-100/22}\approx\mathbf{1.06\times10^{-2}}$ , about $1\%$ — eighteen orders of magnitude better per link. Distributing entanglement on each short segment and swapping at the nine nodes converts an astronomically small probability into a workable one — the entire rationale for the quantum repeater.

Step 3. But swapping is not free of noise. Modelling each segment as a Werner pair of visibility $v=0.95$ and using $v_{\text{out}}=v^{N}$ , the end-to-end visibility after ten swaps is $0.95^{10}=\mathbf{0.599}$ .

Step 4. A Werner state is entangled precisely when $v>1/3$ . Here $0.599>1/3$ , so the final pair is still entangled but badly degraded, from $0.95$ to $0.60$ . Had the segments been $v=0.90$ , the chain would give $0.90^{10}=0.349$ , and one more swap ( $0.90^{11}=0.314$ ) would push it below $1/3$ into separability. This exponential decay is why real repeaters interleave purification between swaps — the message of the second figure.

Physics of the portrait

Entanglement swapping is a young idea with a clear lineage: proposed as a thought experiment to make Bell tests "event-ready," then turned into a defining experiment of the quantum-information era by the group that did most to bring multi-particle entanglement into the laboratory.

Origins · 1 minAnton Zeilinger and how this was discovered

Anton Zeilinger

born 1945 · experimental quantum information, multiphoton entanglement, and entanglement swapping

Anton Zeilinger is an Austrian physicist whose experiments turned entanglement from a paper curiosity into a working laboratory resource, and who shared the 2022 Nobel Prize in Physics with John Clauser and Alain Aspect for pioneering tests of Bell inequalities and quantum information science. In 1993, with Marek Żukowski, Michael Horne and Artur Ekert, he co-authored the proposal that named entanglement swapping — an "event-ready-detectors" scheme in which a Bell measurement on two independent pairs both entangles the remaining particles and heralds that a clean entangled pair exists. Five years later his group, with Jian-Wei Pan, Dik Bouwmeester and Harald Weinfurter, performed the first experimental swap, using down-conversion photons and a beam-splitter Bell analyzer to entangle photons that had never shared a source. The same laboratory achieved the first photon teleportation (1997) and later the Micius satellite links. His enduring contribution was insisting that the strangest predictions of quantum mechanics — nonlocal correlations between systems with no common past — are not merely consistent but demonstrable and useful, laying the foundation for quantum repeaters, entanglement-based cryptography, and the nascent quantum internet.

History

1935

— Einstein, Podolsky and Rosen expose the entanglement of separated systems, and Schrödinger names it the characteristic trait of quantum mechanics

1964

— Bell shows entangled correlations violate any local-hidden-variable bound, making entanglement empirically testable

1993

— Żukowski, Zeilinger, Horne and Ekert propose entanglement swapping as an event-ready Bell experiment

1998

— Pan, Bouwmeester, Weinfurter and Zeilinger demonstrate entanglement swapping with down-converted photons

2015

— Hensen and co-workers use heralded swapping between distant NV centres for a loophole-free Bell test

Going further

Texts. Nielsen and Chuang, Quantum Computation and Quantum Information, section 1.3.7 and Chapter 12 · Preskill, Lecture Notes on Quantum Computation, Chapter 4 · Wilde, Quantum Information Theory, 2nd ed., Chapter 6 · Watrous, The Theory of Quantum Information, Chapter 6 · Kitaev, Shen and Vyalyi, Classical and Quantum Computation, Chapter 11. The four-qubit factorization of equation (4), the equiprobable $1/4$ outcomes, the Pauli relations of equation (5), and the maximally mixed $\rho_{AD}=I/4$ are all standard results. Key numerical checks: $\mathrm{Tr}_{BC}[\cdots]=I/4$ ; $e^{-1000/22}\approx1.8\times10^{-20}$ ; $0.95^{10}=0.599$ . The $50\%$ linear-optics ceiling follows Lütkenhaus et al. (1999); the repeater scaling and visibility multiplication $v_{\text{out}}=v^N$ follow Briegel, Dür, Cirac and Zoller, Phys. Rev. Lett. 81, 5932 (1998). The founding papers are Żukowski, Zeilinger, Horne and Ekert, Phys. Rev. Lett. 71, 4287 (1993) — the proposal — and Pan, Bouwmeester, Weinfurter and Zeilinger, Phys. Rev. Lett. 80, 3891 (1998) — the first demonstration.

Prerequisites. Bell-State Measurement · Quantum Teleportation · Superdense Coding. Leads to. E91 Protocol · The No-Cloning Theorem in Cryptography.

Ask the tutor: want to trace the swap identity through a $|\Psi^-\rangle$ outcome and its correction, or work out how many purification rounds a 500 km chain needs to stay above the entanglement threshold?

How can two particles that never interacted become entangled?

They become entangled only relative to a measurement record. Each outer particle was previously entangled with an inner partner, and the Bell measurement on the two inner partners projects the joint state onto a branch in which the outer particles are maximally entangled. No influence travels from $A$ to $D$ ; the correlation is inherited through the shared measurement, not any interaction. The precise statement is the swap identity — written in the inner Bell basis, the product of two Bell pairs pairs each inner Bell state with an outer one, so measuring the inner pair selects the outer state. It is entanglement re-routed, not created from a mutual past.

Does entanglement swapping let you signal faster than light?

No. Before the outcome is communicated, the reduced state of the outer pair is exactly $I/4$ — separable and identical to two never-entangled qubits. Every local statistic at $A$ or $D$ is unchanged by the distant measurement, so nothing done locally reveals whether or when it happened. Usable entanglement appears only after the outcome is transmitted classically (at or below light speed) and the matching Pauli correction applied. Swapping heralds and re-routes entanglement but transmits no information on its own, fully consistent with relativity.

Why does each Bell outcome occur with probability exactly one quarter?

Because the swap identity writes the initial product as an equal-amplitude sum $\tfrac12\sum_k|\beta_k\rangle_{BC}|\beta_k\rangle_{AD}$ , every branch carrying coefficient $1/2$ ; the Born rule assigns $|1/2|^2=1/4$ to each of the four outcomes. The flatness follows from starting with maximally entangled inputs — two $|\Phi^+\rangle$ pairs have no bias toward any Bell sector, so all four appear equally often. Partially entangled inputs would generally give unequal probabilities.

How is entanglement swapping different from quantum teleportation?

They are the same operation on different inputs. Teleportation transfers an unknown single-qubit state onto the far half of a resource pair. In swapping the "data qubit" is itself half of an entangled pair, so what transfers is not a state but the entanglement it carried — the far particle inherits the correlation. The machinery is identical: a Bell measurement, four equiprobable outcomes, one classical message, a local Pauli correction. Swapping is teleportation with a pre-entangled input, which is why it moves entanglement from one baseline to another.

Why can standard linear-optics implementations succeed only half the time?

With only beam splitters, phase shifters and photon counters — no nonlinearity or ancilla — a photonic Bell analyzer cannot distinguish all four Bell states. The odd-parity $|\Psi^+\rangle$ and $|\Psi^-\rangle$ give distinct coincidence patterns, but the even-parity $|\Phi^+\rangle$ and $|\Phi^-\rangle$ share a detector signature. Only two of four outcomes are cleanly heralded, so an unambiguous swap succeeds at most $50\%$ of the time. Beating this ceiling needs added resources — auxiliary photons, hyperentanglement, or nonlinear light–matter interactions realizing a complete deterministic Bell measurement.

Why does a repeater chain need purification if each swap is exact?

Because no real pair or measurement is exact. Model each segment by a visibility $v<1$ ; swapping multiplies visibilities, so $N$ cascaded swaps give $v_{\text{out}}=v^{N}$ , an exponential decay. A Werner state is entangled only while $v>1/3$ , so the end-to-end state eventually crosses into separability — for $v=0.90$ , by about the eleventh swap. Purification counteracts this by consuming several noisy pairs to distil fewer higher-visibility ones, resetting $v$ before it collapses. Nested schemes alternating swapping and purification make the distance scaling polynomial rather than exponential — the founding insight of the quantum repeater.