Macroscopic Bell States

• Macroscopic Bell states are entangled quantum states of large systems with high-dimensional Hilbert spaces, including high-spin singlets, squeezed states, and cat-state superpositions.
• They enable testing quantum nonlocality when local observables have macroscopic spectra or are coarse-grained, leading to violations of generalized Bell-type inequalities.
• Experimental platforms range from photonic and mechanical systems to hybrid setups, addressing challenges like loss tolerance, decoherence, and measurement precision.

Searching arXiv for relevant papers on macroscopic Bell states and closely related macroscopic nonlocality constructions. arXiv search query: "macroscopic Bell states" Macroscopic Bell states are entangled states of two or more subsystems each possessing many degrees of freedom or a large Hilbert-space dimension. In practice, the term covers high-spin singlets, continuous-variable two-mode squeezed states, and cat-state superpositions of macroscopically distinct coherent states. The central question is whether Bell nonlocality survives when local observables have macroscopic spectra or when measurements are deliberately coarse-grained to discriminate only macroscopically distinct outcomes. A particularly direct formulation is the non-dichotomic Bell-type inequality for macroscopic photon numbers in two spatially separated beams derived for macroscopic squeezed vacuum, with predicted violation by experimentally realized macroscopic Bell states (Teh et al., 2021, Chekhova et al., 2012).

1. Formal definitions and representative state families

A useful general definition is that a macroscopic Bell state is an entangled state of two subsystems in which each subsystem is itself macroscopic, either because it carries a large total spin, a large excitation number, or a continuous-variable superposition with macroscopically separated components. Three representative constructions recur across the literature: higher-spin singlets, two-mode squeezed states, and entangled cat states (Teh et al., 2021).

For spin systems, a standard example is the zero-total-spin generalization of the singlet,

$$|\Psi\rangle = \frac{1}{\sqrt{2J+1}} \sum_{m=-J}^{J} (-1)^{J-m} |m\rangle_A \otimes |-m\rangle_B,$$

where $J\gg \tfrac12$ is the macroscopic spin magnitude. For continuous variables, the paradigmatic state is the two-mode squeezed vacuum

$$|TMSS(r)\rangle = \frac{1}{\cosh r} \sum_{n=0}^{\infty} (\tanh r)^n |n\rangle_A |n\rangle_B,$$

which approaches the original EPR state as $r\to\infty$ (Teh et al., 2021).

For coherent-state encodings, a two-mode cat-Bell state is commonly written as

$$|\Psi_{-}\rangle = N_{-}\bigl(|\alpha\rangle_A|-\alpha\rangle_B-|-\alpha\rangle_A|\alpha\rangle_B\bigr),$$

with

$N_{-}=[2(1-e^{-4|\alpha|^2})]^{-1/2}.$

In the large-$|\alpha|$ limit, $\langle \alpha|-\alpha\rangle=e^{-2|\alpha|^2}\to0$, so the logical identification $|+\rangle\equiv|\alpha\rangle$, $|-\rangle\equiv|-\alpha\rangle$ turns the state into an effective Bell pair of macroscopic qubits (Thenabadu et al., 2020).

The high-gain photonic literature introduces a distinct but related four-mode family, the macroscopic analogues of the two-photon Bell basis,

$J\gg \tfrac12$0

$J\gg \tfrac12$1

$J\gg \tfrac12$2

where $J\gg \tfrac12$3 is the parametric-gain parameter (Iskhakov et al., 2011).

Class	Representative form	Macroscopicity parameter
Spin Bell state	$J\gg \tfrac12$4	$J\gg \tfrac12$5
CV Bell state	$J\gg \tfrac12$6	$J\gg \tfrac12$7
Cat-Bell state	$J\gg \tfrac12$8	$J\gg \tfrac12$9
High-gain photonic Bell state	$$|TMSS(r)\rangle = \frac{1}{\cosh r} \sum_{n=0}^{\infty} (\tanh r)^n |n\rangle_A |n\rangle_B,$$0, $$|TMSS(r)\rangle = \frac{1}{\cosh r} \sum_{n=0}^{\infty} (\tanh r)^n |n\rangle_A |n\rangle_B,$$1	$$|TMSS(r)\rangle = \frac{1}{\cosh r} \sum_{n=0}^{\infty} (\tanh r)^n |n\rangle_A |n\rangle_B,$$2

2. Bright photonic macroscopic Bell states

In high-gain parametric down-conversion, two orthogonal polarization modes in two frequency channels are coupled to produce bright four-mode squeezed vacua that are direct macroscopic analogues of the Bell basis. Their Fock expansions make explicit the perfect photon-number correlations between polarization-frequency modes, with mean photon number per mode $$|TMSS(r)\rangle = \frac{1}{\cosh r} \sum_{n=0}^{\infty} (\tanh r)^n |n\rangle_A |n\rangle_B,$$3 (Iskhakov et al., 2011).

A salient feature is first-order unpolarization. Using the Stokes operators

$$|TMSS(r)\rangle = \frac{1}{\cosh r} \sum_{n=0}^{\infty} (\tanh r)^n |n\rangle_A |n\rangle_B,$$4

all four macroscopic Bell states satisfy $$|TMSS(r)\rangle = \frac{1}{\cosh r} \sum_{n=0}^{\infty} (\tanh r)^n |n\rangle_A |n\rangle_B,$$5 for $$|TMSS(r)\rangle = \frac{1}{\cosh r} \sum_{n=0}^{\infty} (\tanh r)^n |n\rangle_A |n\rangle_B,$$6, so the first-order degree of polarization is zero. Higher-order moments, however, distinguish the states. The operational second-order degree of polarization is defined as

$$|TMSS(r)\rangle = \frac{1}{\cosh r} \sum_{n=0}^{\infty} (\tanh r)^n |n\rangle_A |n\rangle_B,$$7

and extends to arbitrary order through

$$|TMSS(r)\rangle = \frac{1}{\cosh r} \sum_{n=0}^{\infty} (\tanh r)^n |n\rangle_A |n\rangle_B,$$8

Triplet states exhibit hidden polarization in second and higher orders, while the singlet analogue is nearly unpolarized up to the fourth order (Iskhakov et al., 2011).

Experimentally reported values illustrate the separation between first-order unpolarization and higher-order structure:

State	$$|TMSS(r)\rangle = \frac{1}{\cosh r} \sum_{n=0}^{\infty} (\tanh r)^n |n\rangle_A |n\rangle_B,$$9	$r\to\infty$0	$r\to\infty$1
$r\to\infty$2	$r\to\infty$3	$r\to\infty$4	$r\to\infty$5
$r\to\infty$6	$r\to\infty$7	$r\to\infty$8	$r\to\infty$9

Three-dimensional quantum polarization tomography further reconstructs the polarization quasiprobability distribution by a three-dimensional inverse Radon transform of measured Stokes marginals. The reconstructed distributions are centered at the origin, confirming first-order unpolarization, but their shapes differ: the triplet-state distribution is an ellipsoid with squeezing in one Stokes component, whereas the singlet-like case is approximately spherical with simultaneous suppression of noise in all three Stokes observables (Kanseri et al., 2011).

The naming of singlet and triplet sectors is not presented uniformly across summaries. One summary identifies $$|\Psi_{-}\rangle = N_{-}\bigl(|\alpha\rangle_A|-\alpha\rangle_B-|-\alpha\rangle_A|\alpha\rangle_B\bigr),$$0 as the singlet analogue, while another tomography summary associates the singlet or “polarization scalar” behavior with $$|\Psi_{-}\rangle = N_{-}\bigl(|\alpha\rangle_A|-\alpha\rangle_B-|-\alpha\rangle_A|\alpha\rangle_B\bigr),$$1. The noncontroversial content is the existence of four high-gain Bell analogues, their vanishing first-order Stokes vector, and the use of higher-order polarization observables to separate hidden-polarization triplet-type behavior from singlet-like isotropy (Iskhakov et al., 2011, Kanseri et al., 2011).

3. Cat-state and nonlinear-dynamics constructions

A major line of work realizes macroscopic Bell states as coherent-state superpositions. In one construction, single-mode cats are produced from an initial coherent state $$|\Psi_{-}\rangle = N_{-}\bigl(|\alpha\rangle_A|-\alpha\rangle_B-|-\alpha\rangle_A|\alpha\rangle_B\bigr),$$2 by Kerr-type nonlinear evolution under

$$|\Psi_{-}\rangle = N_{-}\bigl(|\alpha\rangle_A|-\alpha\rangle_B-|-\alpha\rangle_A|\alpha\rangle_B\bigr),$$3

with even $$|\Psi_{-}\rangle = N_{-}\bigl(|\alpha\rangle_A|-\alpha\rangle_B-|-\alpha\rangle_A|\alpha\rangle_B\bigr),$$4, commonly $$|\Psi_{-}\rangle = N_{-}\bigl(|\alpha\rangle_A|-\alpha\rangle_B-|-\alpha\rangle_A|\alpha\rangle_B\bigr),$$5 or $$|\Psi_{-}\rangle = N_{-}\bigl(|\alpha\rangle_A|-\alpha\rangle_B-|-\alpha\rangle_A|\alpha\rangle_B\bigr),$$6. At special interaction times, $$|\Psi_{-}\rangle = N_{-}\bigl(|\alpha\rangle_A|-\alpha\rangle_B-|-\alpha\rangle_A|\alpha\rangle_B\bigr),$$7 generates superpositions $$|\Psi_{-}\rangle = N_{-}\bigl(|\alpha\rangle_A|-\alpha\rangle_B-|-\alpha\rangle_A|\alpha\rangle_B\bigr),$$8, which can then be entangled across two modes by balanced beam splitting, postselection, or cross-Kerr interaction (Thenabadu et al., 2020).

A particularly explicit qubit mapping uses the coarse-grained observable

$$|\Psi_{-}\rangle = N_{-}\bigl(|\alpha\rangle_A|-\alpha\rangle_B-|-\alpha\rangle_A|\alpha\rangle_B\bigr),$$9

so that only the sign of a homodyne outcome is retained. Local basis changes are implemented by

$N_{-}=[2(1-e^{-4|\alpha|^2})]^{-1/2}.$0

For $N_{-}=[2(1-e^{-4|\alpha|^2})]^{-1/2}.$1, one has in the $N_{-}=[2(1-e^{-4|\alpha|^2})]^{-1/2}.$2 basis

$N_{-}=[2(1-e^{-4|\alpha|^2})]^{-1/2}.$3

$N_{-}=[2(1-e^{-4|\alpha|^2})]^{-1/2}.$4

In the orthogonal limit, the predicted correlation is exactly the spin-$N_{-}=[2(1-e^{-4|\alpha|^2})]^{-1/2}.$5 form

$N_{-}=[2(1-e^{-4|\alpha|^2})]^{-1/2}.$6

and the CHSH choice $N_{-}=[2(1-e^{-4|\alpha|^2})]^{-1/2}.$7, $N_{-}=[2(1-e^{-4|\alpha|^2})]^{-1/2}.$8, $N_{-}=[2(1-e^{-4|\alpha|^2})]^{-1/2}.$9, $|\alpha|$0 yields $|\alpha|$1. For finite $|\alpha|$2, the violation persists for $|\alpha|$3 (Thenabadu et al., 2020).

A related time-domain formulation replaces the conventional choice of analyzer angle with a choice of local evolution time under a nonlinear Hamiltonian. In the $|\alpha|$4-boson case, Josephson–Kerr dynamics

$|\alpha|$5

implements an effective nonlinear beam splitter, while in the coherent-state case a single-mode Kerr Hamiltonian

$|\alpha|$6

generates quadrature-separated cats. For the coherent-state scheme, the large-$|\alpha|$7 CHSH sum tends to

$|\alpha|$8

with $|\alpha|$9 already for $\langle \alpha|-\alpha\rangle=e^{-2|\alpha|^2}\to0$0. In the discrete $\langle \alpha|-\alpha\rangle=e^{-2|\alpha|^2}\to0$1-boson scheme, numerical solutions show a Clauser–Horne ratio $\langle \alpha|-\alpha\rangle=e^{-2|\alpha|^2}\to0$2 for, for example, $\langle \alpha|-\alpha\rangle=e^{-2|\alpha|^2}\to0$3, and the violation persists up to $\langle \alpha|-\alpha\rangle=e^{-2|\alpha|^2}\to0$4 with carefully chosen $\langle \alpha|-\alpha\rangle=e^{-2|\alpha|^2}\to0$5 (Thenabadu et al., 2019).

Phase-entangled coherent states provide a propagation-oriented variant,

$\langle \alpha|-\alpha\rangle=e^{-2|\alpha|^2}\to0$6

with correlation function $\langle \alpha|-\alpha\rangle=e^{-2|\alpha|^2}\to0$7 and $\langle \alpha|-\alpha\rangle=e^{-2|\alpha|^2}\to0$8. For $\langle \alpha|-\alpha\rangle=e^{-2|\alpha|^2}\to0$9, $|+\rangle\equiv|\alpha\rangle$0, and fiber loss $|+\rangle\equiv|\alpha\rangle$1, Bell violation remains possible over approximately $|+\rangle\equiv|\alpha\rangle$2 in commercially available optical fibers (Kirby et al., 2013).

4. Bell tests, coarse-graining, and loss tolerance

The measurement theory of macroscopic Bell states departs from standard qubit CHSH tests in two distinct ways: local outcomes are often multivalued or unbounded, and practical readout is typically coarse-grained. For spin-$|+\rangle\equiv|\alpha\rangle$3 systems, a generalized CHSH construction uses

$|+\rangle\equiv|\alpha\rangle$4

with the usual local-hidden-variable bound $|+\rangle\equiv|\alpha\rangle$5. If one coarse-grains to the sign operator $|+\rangle\equiv|\alpha\rangle$6 for $|+\rangle\equiv|\alpha\rangle$7 and $|+\rangle\equiv|\alpha\rangle$8 otherwise, Bell violation can still survive, but only when measurement directions are adjusted with precision finer than $|+\rangle\equiv|\alpha\rangle$9 (Teh et al., 2021).

For macroscopic squeezed vacuum, a different route is to abandon dichotomic assumptions altogether: the 2012 analysis derives a non-dichotomic Bell-type inequality directly for macroscopic photon numbers in two spatially separated beams and shows that the experimentally realized macroscopic Bell states can violate it (Chekhova et al., 2012).

Coarse-graining has also been studied as a controllable quantum-to-classical transition. For the macroscopic entangled state

$|-\rangle\equiv|-\alpha\rangle$0

with fuzzy dichotomic observables $|-\rangle\equiv|-\alpha\rangle$1 and coarsened references $|-\rangle\equiv|-\alpha\rangle$2, the family of symmetric Bell functions

$|-\rangle\equiv|-\alpha\rangle$3

shows an odd-even dichotomy: for even $|-\rangle\equiv|-\alpha\rangle$4, increasing the number of settings reduces robustness to coarse-graining, whereas for odd $|-\rangle\equiv|-\alpha\rangle$5, increasing the number of settings enhances robustness. By contrast, the linear steering inequality

$|-\rangle\equiv|-\alpha\rangle$6

has monotonic robustness improvement with $|-\rangle\equiv|-\alpha\rangle$7 for both odd and even numbers of settings (Naik et al., 2023).

Loss tolerance depends strongly on where the loss enters. For the multi-photon Bell-type state

$|-\rangle\equiv|-\alpha\rangle$8

loss after the local unitary rotations can be counteracted by increasing $|-\rangle\equiv|-\alpha\rangle$9. The optimal CHSH expression obeys

$J\gg \tfrac12$00

and the critical efficiencies read $J\gg \tfrac12$01 for $J\gg \tfrac12$02, approximately $J\gg \tfrac12$03 for $J\gg \tfrac12$04, approximately $J\gg \tfrac12$05 for $J\gg \tfrac12$06, and $J\gg \tfrac12$07 for $J\gg \tfrac12$08. Loss before the unitary transformations, however, produces decoherence that cannot be compensated by increasing macroscopicity (Lim et al., 2011).

This distinction is consistent with negative results in some micro-macro optical tests. In amplified singlet schemes with thresholded polarization measurements, a CHSH violation requires a distinguishability parameter $J\gg \tfrac12$09, but for large photon number one finds $J\gg \tfrac12$10, precluding violation. The result is that simple preselection can help only in the low-photon regime, not in the asymptotically macroscopic one (Stobińska et al., 2011).

5. Implementations in photonic, mechanical, and hybrid platforms

Photonic systems remain the most mature setting. High-gain optical parametric amplification produces bright macroscopic Bell states of light, and homodyne- or parity-based continuous-variable Bell tests can in principle approach the Tsirelson bound. For two-mode squeezed vacuum measured via displaced parity, the CHSH parameter satisfies $J\gg \tfrac12$11 as $J\gg \tfrac12$12 (Teh et al., 2021).

Mechanical platforms extend the concept to genuinely massive oscillators. An entanglement-swapping protocol based on projective Bell-state measurements on superconducting qubits generates four bipartite phononic Bell cat states, including

$J\gg \tfrac12$13

Using displaced-parity pseudospin measurements with $J\gg \tfrac12$14 and optimal displacements $J\gg \tfrac12$15, $J\gg \tfrac12$16, the reported CHSH value is $J\gg \tfrac12$17, above the classical bound $J\gg \tfrac12$18 (Nongthombam et al., 2024).

Reservoir-engineered optomechanics provides a complementary route. With two-tone driving, a cavity can prepare two mechanical resonators in an entangled Gaussian steady state. For parity detection, Bell violation is predicted only in a transient regime with extremely small cavity losses and mechanical damping; a representative peak is $J\gg \tfrac12$19 at times $J\gg \tfrac12$20, but it vanishes if $J\gg \tfrac12$21 or $J\gg \tfrac12$22. For on-off detection, the predicted steady-state value is $J\gg \tfrac12$23, with tolerance to $J\gg \tfrac12$24 up to $J\gg \tfrac12$25, provided $J\gg \tfrac12$26 (Li et al., 2017).

Hybrid spin-superconducting devices demonstrate a different notion of macroscopicity, based on atom number rather than oscillator separation. A millimeter-sized YIG sphere containing $J\gg \tfrac12$27 spins was entangled with a superconducting qubit through a microwave cavity, yielding the Bell state

$J\gg \tfrac12$28

with a reconstructed fidelity

$J\gg \tfrac12$29

This is described as the largest system, in the sense of atom number, capable of generating the maximally entangled quantum state (Xu et al., 2023).

The experimental landscape remains uneven. The 2021 review notes that EPR steering has already been observed in mesoscopic atomic and optomechanical regimes, while Bell nonlocality is rarer; it cites the first optomechanical Bell violation as a heralded phonon-phonon entanglement test on two nanomechanical resonators with $J\gg \tfrac12$30 (Teh et al., 2021).

6. Classical analogues, simulations, and interpretive cautions

Not every macroscopic Bell-state construction is intended as a quantum resource. A planar optical or microwave resonator with a weak parabolic transverse potential can realize a classical analogue of two qubits, using the $J\gg \tfrac12$31 and $J\gg \tfrac12$32 transverse directions as independent subsystems and Gaussian/Hermite modes as computational basis states. In that setting, the Bell-type wave function

$J\gg \tfrac12$33

leads to a CHSH value $J\gg \tfrac12$34 under Pauli-like mode measurements. The same source stresses that this is entirely classical, that there is no spatial nonlocality, and that what remains uniquely quantum is the combination of particle discreteness, true nonlocal measurements, and entanglement beyond the spatial-dimension bound (Snoke, 2014).

A different classical simulation employs multilayer gray-painted spheres as macroscopic hidden-variable analogues of spin-$J\gg \tfrac12$35 systems. By assigning measurement-direction-dependent partitions of the hidden-variable space, the model reproduces the quantum singlet correlation

$J\gg \tfrac12$36

and the CHSH value $J\gg \tfrac12$37. The point of the construction is not that Bell nonlocality disappears, but that Bell’s derivation relies on a single joint probability space for incompatible settings; the model abandons that requirement (Slavnov, 2011).

These analogues correct a common misconception. A CHSH value above $J\gg \tfrac12$38 is not, by itself, sufficient to establish the full physical content of macroscopic quantum nonlocality. Spatial separation, measurement independence, the absence of postselection artifacts, and physically quantum state preparation remain decisive. This suggests a useful distinction between mathematical Bell-state structure and experimentally compelling macroscopic Bell nonlocality.

7. Foundations, limits, and open problems

Macroscopic Bell states were developed partly to test whether quantum nonlocality survives confrontation with macroscopic realism. In coherent-state Bell and Leggett–Garg analyses, a sharp distinction is drawn between deterministic macroscopic realism and weak macroscopic realism. Deterministic macroscopic realism assumes that, prior to any choice of angle $J\gg \tfrac12$39, the field already carries definite values for the possible coarse-grained observables; CHSH violations negate this assumption. Weak macroscopic realism assigns a definite hidden outcome only for the pointer basis actually set by the local unitary dynamics, and can be made consistent with the observed violations (Thenabadu et al., 2020).

That compatibility is not the end of the matter. The same analysis argues that if, after the local unitary, the system is in one or the other of two macroscopic states $J\gg \tfrac12$40 associated with $J\gg \tfrac12$41, then EPR reasoning promotes them to elements of reality; yet interference fringes in the conjugate quadrature or entanglement-inferred squeezing imply that the actual superposition can satisfy a relation of the form $J\gg \tfrac12$42, so such weak-macrorealist elements of reality cannot correspond to single quantum states. This is presented as a macroscopic EPR-type paradox (Thenabadu et al., 2020).

From the broader review perspective, macroscopic Bell violations “bring into question the validity of certain forms of macroscopic realism” and undermine naive appeals to the correspondence principle. At the same time, the technical obstacles remain severe: decoherence rates grow rapidly for large cats, measurement-setting precision can shrink as $J\gg \tfrac12$43 or $J\gg \tfrac12$44, detection efficiency remains critical, and many atomic or optomechanical tests are not yet space-like separated (Teh et al., 2021).

Several results underscore that macroscopicity alone does not guarantee Bell violation. In a four-mode BEC formulation with large local spin outcomes, the generalized CHSH-type Bell parameter satisfies

$J\gg \tfrac12$45

and numerical optimization over several candidate states found no violation beyond the trivial microscopic case $J\gg \tfrac12$46 (Dalton, 2018). For mixed macroscopic entangled states, coarse-grained Bell and steering thresholds can already diverge at two settings, with $J\gg \tfrac12$47 for the Werner-type family $J\gg \tfrac12$48 (Naik et al., 2023).

Open problems are correspondingly concrete: loophole-free Bell tests with truly macroscopic degrees of freedom, strong mechanical cat states suitable for CHSH tests, quantum nonlocality for masses far beyond current optomechanical demonstrations, multipartite coarse-grained Bell schemes, and robust protection of macroscopic entanglement and steering against environmental decoherence (Teh et al., 2021). A plausible implication is that future progress will depend less on a single universal Bell inequality than on platform-specific combinations of state engineering, coarse-grained measurement design, and loss-aware certification strategies.