Gaussian Quantum Information

Christian Weedbrook; Stefano Pirandola; Raul Garcia-Patron; Nicolas J. Cerf; Timothy C. Ralph; Jeffrey H. Shapiro; Seth Lloyd

Source-linked AI summary

Gaussian Quantum Information

Christian Weedbrook, Stefano Pirandola, Raul Garcia-Patron, Nicolas J. Cerf, Timothy C. Ralph, Jeffrey H. Shapiro, Seth Lloyd

arXiv:1110.3234v1Oct 14, 2011quant-ph

TL;DR

Continuous-variable quantum information offers an alternative to qubit-based processing, but practical implementations face limitations such as short-distance quantum key distribution. This review synthesizes Gaussian states, operations, and measurements, finding that their first- and second-order moments simplify analysis while supporting diverse information-processing tasks.

Problem
Continuous-variable QKD implementations remained limited to short distances, partly because Gaussian-modulation reconciliation is inefficient at low signal-to-noise ratios.
Method
The review synthesizes continuous-variable quantum information processes using combinations of Gaussian states, Gaussian operations, and Gaussian measurements.
Results
Characterizing Gaussian states and operations through first- and second-order statistical moments substantially simplifies the mathematical analysis of quantum-information protocols.
Takeaways & Limitations
Gaussian resources support quantum communication, cryptography, computation, teleportation, and state and channel discrimination within a tractable framework.

Abstract

from arXiv · show

The science of quantum information has arisen over the last two decades centered on the manipulation of individual quanta of information, known as quantum bits or qubits. Quantum computers, quantum cryptography and quantum teleportation are among the most celebrated ideas that have emerged from this new field. It was realized later on that using continuous-variable quantum information carriers, instead of qubits, constitutes an extremely powerful alternative approach to quantum information processing. This review focuses on continuous-variable quantum information processes that rely on any combination of Gaussian states, Gaussian operations, and Gaussian measurements. Interestingly, such a restriction to the Gaussian realm comes with various benefits, since on the theoretical side, simple analytical tools are available and, on the experimental side, optical components effecting Gaussian processes are readily available in the laboratory. Yet, Gaussian quantum information processing opens the way to a wide variety of tasks and applications, including quantum communication, quantum cryptography, quantum computation, quantum teleportation, and quantum state and channel discrimination. This review reports on the state of the art in this field, ranging from the basic theoretical tools and landmark experimental realizations to the most recent successful developments.

I. INTRODUCTION … A. Bosonic systems in a nutshell

The review introduces continuous-variable quantum information through Gaussian states and transformations, then develops its theoretical foundations, protocols, channels, cryptography, computation, notation, and bosonic-mode formalism. It emphasizes an accessible, self-contained, largely theoretical treatment centered on optical Gaussian systems while connecting the framework to experiments and related readings.

I. INTRODUCTION: Quantum information uses intrinsically quantum systems such as atoms and photons to register, transform, and transmit information, including discrete qubits and continuous variables.
A. Gaussian quantum information processing: Gaussian states and transformations provide the primary analytical tools for continuous-variable quantum information, with Gaussian transformations mapping Gaussian states to Gaussian states.
A. Gaussian quantum information processing: The review is self-contained and presents foundational facts, detailed derivations, and recent developments for readers from quantum optics, atomic physics, and quantum information.
B. Outline of review: The review mainly develops optical Gaussian protocols theoretically, while also discussing atomic-ensemble protocols and citing relevant optical and atomic experiments.
B. Outline of review: The review covers Gaussian theory, state discrimination, teleportation, cloning, bosonic communication channels, quantum cryptography, and continuous-variable cluster-state computation.
C. Further readings: Further readings include reviews and books on continuous-variable quantum information, Gaussian channels, quantum cryptography, and cluster-state computation.
D. Comment on notation: The review normalizes vacuum-noise variance to 1, corresponding to ℏ=2, while noting common alternatives of 1/4, 1/2, and 1.
A. Bosonic systems in a nutshell: A bosonic continuous-variable system has an infinite-dimensional Hilbert space, with N electromagnetic-field modes described by bosonic operators or canonical quadratures q_k and p_k having continuous spectra.

1. Phase-space representation and Gaussian states · 2. Gaussian unitaries · B. Examples of Gaussian states and Gaussian unitaries

Bosonic quantum states admit a phase-space description through Wigner functions and, for Gaussian states, are completely characterized by their displacement vectors and covariance matrices. Gaussian unitaries preserve Gaussianity and act as affine symplectic transformations on phase space, providing the framework for elementary Gaussian-state examples.

1. Phase-space representation and Gaussian states: A density operator ˆρ completely represents an N-mode bosonic system and is pure when it satisfies ˆρ² = ˆρ.Pure states can be written as ˆρ = |ϕ⟩⟨ϕ|.
1. Phase-space representation and Gaussian states: Any quantum state is equivalently represented by a normalized, generally non-positive Wigner function W(x) over the 2N-dimensional phase space K.The phase-space variables x ∈R2N are eigenvalues of quadrature operators.
1. Phase-space representation and Gaussian states: The first two statistical moments are the displacement vector and covariance matrix V, whose diagonal entries give quadrature variances.V is a real, symmetric 2N × 2N matrix constrained by the uncertainty principle.
1. Phase-space representation and Gaussian states: Gaussian states are bosonic states with Gaussian Wigner representations and are completely characterized by their displacement vector and covariance matrix.For pure states, Gaussianity is equivalent to non-negativity of the Wigner function.
2. Gaussian unitaries: A Gaussian operation maps Gaussian states to Gaussian states, while reversible quantum channels are represented by unitary transformations.Gaussian channels and Gaussian unitaries preserve the Gaussian character of quantum states.
2. Gaussian unitaries: Gaussian unitaries are generated by Hamiltonians that are second-order polynomials in field operators and correspond to linear unitary Bogoliubov transformations.The Hamiltonian form uses α ∈CN and N × N complex matrices F and G.
2. Gaussian unitaries: In quadrature coordinates, a Gaussian unitary is an affine symplectic map (S, d): x →Sx + d, combining a linear symplectic transformation with a phase-space translation.The matrix S is real and 2N × 2N, while d ∈R2N.
2. Gaussian unitaries: The action of a Gaussian unitary on a Gaussian state is completely determined by its transformation of the displacement vector and covariance matrix.This moment-based description is summarized by Eq. (25).

1. Vacuum states and thermal states … 6. Beam splitter

The section develops the basic one- and two-mode Gaussian states and transformations, from vacuum and thermal states through displacement, squeezing, phase rotation, general Gaussian decomposition, and beam splitters.

1. Vacuum states and thermal states: The vacuum state |0⟩ has zero photons, is annihilated by â, and reaches the minimum symmetric quadrature variance, V(q̂) = V(p̂) = 1.Its covariance matrix is the identity, and this variance is called vacuum noise or quantum shot-noise.
1. Vacuum states and thermal states: Thermal states maximize von Neumann entropy at fixed mean photon number n̄ and have Gaussian Wigner functions with covariance V = (2n̄ + 1)I.Every Gaussian state can be decomposed into thermal states.
2. Displacement and coherent states: Displacement translates quadratures by dα and transforms the vacuum into coherent states with unchanged covariance V = I but mean x̄ = dα.The displacement operator is generated by a linear Hamiltonian, with α = (q + ip)/2.
3. One-mode squeezing and squeezed states: One-mode squeezing transforms annihilation and quadrature operators linearly, producing squeezed vacuum states with one quadrature variance below shot-noise and the other above it.The squeezing parameter is r, and the covariance matrix is V = S(r)S(r)^T = S(2r).
4. Phase rotation: Phase rotation is defined relative to a local oscillator and applies the unitary R(θ) = exp(−iθâ†â), rotating the quadratures by angle θ.In the Heisenberg picture, â transforms as â → e^iθâ.
5. General one-mode Gaussian states: Any one-mode symplectic matrix decomposes as S = R(θ)S(r)R(φ), so any one-mode Gaussian unitary consists of displacement, rotations, and squeezing.Applying this unitary to a thermal state yields the most general one-mode Gaussian state.
5. General one-mode Gaussian states: Setting n̄ = 0 gives the most general one-mode pure Gaussian state, a rotated and displaced squeezed state |α, θ, r⟩ = D(α)R(θ)S(r)|0⟩.The general one-mode construction extends to arbitrary N bosonic modes.
6. Beam splitter: For two modes, the beam splitter is the simplest interferometer; its transmissivity is τ = cos^2θ, with τ = 1/2 defining a balanced beam splitter.It transforms both annihilation and quadrature operators through linear unitary and symplectic maps.

7. Two-mode squeezing and EPR states … 3. Two-mode Gaussian states

The paper develops two-mode squeezing and EPR states, then provides symplectic and thermal tools for analyzing arbitrary multimode Gaussian states. It concludes with compact analytical descriptions of two-mode Gaussian states, including their covariance structure and symplectic spectrum.

7. Two-mode squeezing and EPR states: Two-mode squeezing generates signal-idler photon pairs through a nonlinear crystal, with the bilinear interaction term a†b† defining the corresponding Gaussian unitary.The squeezing strength is quantified by r.
7. Two-mode squeezing and EPR states: Applying the two-mode squeezing operator to two vacua produces a two-mode squeezed vacuum, also called an Einstein-Podolsky-Rosen state.This state is Gaussian and has zero mean.
7. Two-mode squeezing and EPR states: For r > 0, EPR quadrature correlations beat quantum shot-noise and imply bipartite entanglement, approaching perfect correlations as r →∞.At r = 0, the state consists of two vacua with variances equal to 1.
C. Symplectic analysis for multimode Gaussian States: Symplectic analysis studies multimode Gaussian states through second-order statistical moments, centered on Williamson’s theorem and the Euler decomposition.Williamson’s theorem diagonalizes positive-definite even-dimensional real matrices, including covariance matrices.
1. Thermal decomposition of Gaussian states: The symplectic spectrum consists of positive quantities νk obtained from a covariance matrix and determines fundamental properties of the associated Gaussian state.A quantum covariance matrix must have symplectic eigenvalues satisfying νk ≥1.
1. Thermal decomposition of Gaussian states: Every zero-mean Gaussian state is unitarily equivalent to a tensor product of one-mode thermal states whose photon numbers are specified by the symplectic spectrum.Thermal states can be purified by EPR states, yielding a pure Gaussian purification after adding a reference system.
2. Euler decomposition of canonical unitaries: Euler decomposition expresses every canonical unitary as a passive interferometer, parallel one-mode squeezers, and a second passive transformation, enabling preparation of arbitrary multimode Gaussian states.The preparation uses thermal states, interferometers, squeezers, and final displacements.
3. Two-mode Gaussian states: Two-mode Gaussian states provide simple analytical formulas for entanglement studies and are represented by a block covariance matrix with symmetric local blocks A and B and correlation block C.Their Williamson form is V⊕= (ν−I) ⊕(ν+I), with symplectic eigenvalues {ν−, ν+}.

D. Entanglement in bipartite Gaussian states … 1. Homodyne detection

The review characterizes bipartite Gaussian entanglement through separability tests, entanglement measures, and experimentally accessible Gaussian measurements. It emphasizes partial transposition, the limitations of mixed-state measures and Gaussian distillation, and homodyne detection as the standard measurement technique.

D. Entanglement in bipartite Gaussian states: Entanglement is defined by correlations that cannot be generated through local operations and classical communication, motivating tests for presence and quantification.Separable states are convex combinations of product states, whereas entangled states are not separable.
1. Separability: Partial transposition tests Gaussian separability by transforming one subsystem’s quadratures and checking whether the resulting covariance matrix satisfies the uncertainty principle.For Gaussian states, this is equivalent to requiring the smallest partially transposed symplectic eigenvalue ˜ν−≥1.
1. Separability: The partial-transpose criterion is sufficient for 1 × M and bisymmetric Gaussian states but is not generally equivalent to separability, including for 2 × 2 states.Nonlinear maps and semidefinite programming provide alternative complete characterization techniques.
2. Entanglement measures: For pure bipartite Gaussian states, entanglement is quantified by the entropy of entanglement, equal to the von Neumann entropy of either reduced state.Local Gaussian unitaries map any pure bipartite Gaussian state to a tensor product of EPR states.
2. Entanglement measures: Mixed Gaussian states lack a unique entanglement measure: entanglement of formation is generally difficult to optimize and is solved in continuous variables only for two-mode symmetric states.For those states, the entanglement of formation is a function of ˜ν−.
2. Entanglement measures: Distillable entanglement equals entanglement of formation for pure states, but Gaussian LOCC cannot distill mixed Gaussian states into pure Gaussian states.Such distillation can be achieved using non-Gaussian operations that map Gaussian states into Gaussian states.
2. Entanglement measures: Logarithmic negativity is an easily computable entanglement monotone that quantifies partial-transpose non-positivity and upper-bounds distillable entanglement.It has been characterized for 1 × M and N × M bisymmetric Gaussian states.
1. Homodyne detection: Homodyne detection measures a bosonic mode’s q or p quadrature, with outcome distributions obtained by marginalizing the Wigner function over the conjugate quadrature.Experimentally, the mode and local oscillator are combined at a balanced beam splitter, and two photodetector signals are subtracted.

2. Heterodyne detection and Gaussian POVMs … 2. Quantum Chernoff bound

The paper develops Gaussian measurement models, including heterodyne detection and partial measurements, then surveys photon-counting measurements and Gaussian-state distinguishability. It introduces Helstrom and Chernoff bounds, including a multimode Gaussian formula based on first moments and symplectic spectra.

2. Heterodyne detection and Gaussian POVMs: Heterodyne detection is a Gaussian POVM projecting onto coherent states and implemented by mixing the signal with vacuum before homodyne detection.General pure-Gaussian POVMs decompose into Gaussian unitaries, vacuum ancillas, and homodyne measurements; noisy POVMs additionally trace out output modes.
3. Partial Gaussian measurement: Partial Gaussian measurement updates the retained subsystem through ancillary modes, Gaussian unitaries, homodyne detection, discarding selected outputs, and retaining the remaining modes.The resulting covariance matrix does not depend on the specific measurement result.
4. Counting and detecting photons: Photon counting uses non-Gaussian number-state measurements or avalanche photodiodes that distinguish vacuum from one or more photons.These measurements support protocols including Gaussian-state discrimination, entanglement distillation, and universal quantum computation.
III. DISTINGUISHABILITY OF GAUSSIAN STATES: Non-orthogonal quantum states cannot generally be perfectly distinguished, limiting Gaussian protocols such as quantum cloning and cryptographic security.The distinguishability section introduces the Helstrom bound, quantum Chernoff bound, and quantum fidelity, then considers minimum-error discrimination of optical coherent states.
2. Quantum Chernoff bound: The quantum Chernoff bound provides an upper bound pe,min ≤ pQC when the trace distance lacks an accessible analytical expression.It requires minimizing over s ∈ [0, 1].
2. Quantum Chernoff bound: For Gaussian states, the quantum Chernoff bound is computable from the first two statistical moments and, in the multimode case, relates to symplectic spectra.The general formula applies to N-mode Gaussian states and also yields the quantum Bhattacharyya bound through the corresponding Cs term.
2. Quantum Chernoff bound: The Gaussian Chernoff formula is easy to evaluate for pure states but can require non-trivial calculations of symplectic matrices in general cases.When those matrices are difficult to derive, weaker bounds depending only on symplectic spectra, such as the Minkowski bound, are available.

3. Quantum fidelity · 4. Multicopy discrimination · B. Distinguishing optical coherent states

The paper develops fidelity-based bounds for Gaussian-state discrimination, extends Helstrom discrimination to multiple copies with asymptotically tight Chernoff behavior, and applies minimum-error discrimination to binary optical coherent states. It describes coherent-state encodings, their transformations, Helstrom bounds, and the cat-state measurement basis achieving the optimum.

3. Quantum fidelity: Quantum fidelity compares input and output states and ranges from zero for orthogonal states to one for identical states.It is used in contexts including quantum teleportation and cloning.
3. Quantum fidelity: For two single-mode Gaussian states, fidelity is expressed through their covariance matrices, displacement difference, and determinant-dependent quantities.The relevant quantities are Δ := det(V0 + V1), δ := (det V0 − 1)(det V1 − 1), and d := x̄1 − x̄0.
3. Quantum fidelity: The resulting fidelity bounds provide further estimates for the minimum error probability and satisfy a chain of inequalities.These bounds are attributed to Fuchs and de Graaf.
4. Multicopy discrimination: With M equiprobable copies of an unknown state, optimal discrimination uses a collective measurement projecting onto the positive part of the M-copy Helstrom matrix.The two hypotheses remain ρ0 and ρ1, each occurring with the same probability.
4. Multicopy discrimination: For M ≫ 1, the quantum Chernoff bound is exponentially tight, with the minimum-error and Chernoff probabilities sharing the same error-rate exponent.The exponent is characterized by the quantum Chernoff information κ, with ϑ ≤ υ.
4. Multicopy discrimination: The M-copy quantum Bhattacharyya bound is easier to compute but is not exponentially tight in the general case.It is presented as another measure of distinguishability.
B. Distinguishing optical coherent states: Minimum-error discrimination of weak, largely overlapping binary coherent states is a fundamental optical-communication task, with information encoded through amplitude or phase modulation.Alice selects one of two known coherent states with known prior probabilities, while Bob must identify the received state.
B. Distinguishing optical coherent states: Amplitude- and phase-shift keyed coherent-state encodings can be interconverted by displacement, and Helstrom bounds apply to both schemes.The optimal measurement for two coherent states is a Schrödinger cat-state basis, a superposition of coherent states whose weights depend on the displacement.

1. Kennedy receiver … A. Quantum teleportation and variants

The reviewed receivers range from the Kennedy scheme to the Helstrom-optimal Dolinar receiver, with optimized displacement outperforming Kennedy and homodyne detection. Continuous-variable teleportation transfers coherent states through Gaussian EPR correlations, surpasses the classical fidelity threshold when entanglement is present, and extends to several related protocols.

1. Kennedy receiver: The Kennedy receiver displaces |α⟩ and |−α⟩ by α, then uses direct photon counting to decide between vacuum and |2α⟩.It always identifies the vacuum correctly ideally; errors arise from vacuum fluctuations in |2α⟩, yielding an error probability twice the Helstrom bound.
2. Dolinar receiver: The Dolinar receiver saturates the Helstrom bound, the lowest possible error for distinguishing two pure coherent states.It combines photon counting with real-time feedback through a causally adjusted local oscillator and decides from the final photon-count parity.
3. Homodyne receiver: The homodyne receiver is the simplest Gaussian-only setup and is optimal among all Gaussian measurements.For weak coherent states with amplitudes |α|^2 < 0.4, it is near-optimal and has lower error probability than the Kennedy receiver.
4. Optimized displacement receiver: The optimized displacement receiver displaces |α⟩ and |−α⟩ by an optimized real value β and minimizes the sum of both decision-error terms.Unlike Kennedy detection, which minimizes only one error term, the optimized scheme jointly optimizes both probabilities.
4. Optimized displacement receiver: The optimized displacement receiver outperforms both homodyne and Kennedy receivers for all values of α.It has applications in quantum cryptography, where it can increase secret-key rates, and squeezing can further improve performance.
4. Optimized displacement receiver: The receiver-performance hierarchy is Dolinar, optimized displacement, Kennedy, then homodyne, with only Dolinar being optimal.Kennedy nevertheless has lower error probability than homodyne for most amplitude values, and optimized displacement has been demonstrated experimentally.
A. Quantum teleportation and variants: Continuous-variable teleportation transfers coherent states between distant parties using EPR correlations shared through a two-mode Gaussian state acting as a virtual channel.Alice and Bob’s modes are prepared in a zero-mean Gaussian state with covariance matrix in (A, B, C)-block form, while the input state may be arbitrary.
A. Quantum teleportation and variants: Entanglement in the virtual channel is generally necessary for fidelity above the classical threshold, and EPR correlations with r > 0 give F > 1/2 for coherent states.Continuous-variable teleportation has also been extended to number-phase and all-optical teleportation, networks, single-photon states, entanglement swapping, and transfer to an atomic ensemble of about 1012 Caesium atoms.

B. Quantum cloning … B. One-mode Gaussian channels

Gaussian quantum cloning cannot produce two perfect copies of an arbitrary state, but an optical Gaussian cloner generates two coherent-state clones with optimal added noise and fidelity 2/3. Bosonic Gaussian channels provide a standard noise model and, in the one-mode case, admit a complete canonical classification determined by three invariants.

B. Quantum cloning: Perfect cloning of an arbitrary quantum state is impossible; it is possible only when the input belongs to a set of orthogonal states.A von Neumann measurement can then perfectly discriminate the orthogonal inputs.
B. Quantum cloning: The Gaussian cloning machine produces two noisy copies of an arbitrary coherent state, with imperfection quantified by each clone’s excess noise variance.The added-noise variances obey a generalized uncertainty relation, preventing simultaneous vanishing noise in conjugate quadratures.
B. Quantum cloning: Each coherent-state clone has one unit of additional shot noise and fidelity F = 2/3, while the anti-clone has two units and fidelity F = 1/2.The optical implementation uses a phase-insensitive amplifier of intensity gain two followed by a balanced beam splitter.
V. BOSONIC GAUSSIAN CHANNELS: Gaussian channels are bosonic channels that model noise in quantum communication by describing interactions between information-carrying bosonic systems and decohering environments.They are defined by transforming Gaussian states into Gaussian states and can be represented through Gaussian dilations.
A. General formalism: A multimode bosonic channel is a completely positive, trace-preserving linear map, with a Stinespring dilation that provides both system and complementary environmental outputs.Degradable channels allow the environmental output to be obtained from the system output by another completely positive, trace-preserving map.
B. One-mode Gaussian channels: Every one-mode Gaussian channel decomposes into input and output Gaussian unitaries surrounding a canonical form with zero displacement and diagonal matrices Tc and Nc.The canonical form is identified by generalized transmissivity τ, rank r, and thermal number n̄; τ and r determine its class.
B. One-mode Gaussian channels: All one-mode Gaussian-channel forms with transmissivity τ ≤1/2 are antidegradable, including classes A1, A2, D and lossy channels in part of class C.Degradability and antidegradability are unchanged by Gaussian unitary equivalence.

C. Classical capacity of Gaussian channels

Classical capacity quantifies optimal reliable communication rates, with quantum channels governed by the Holevo bound and bosonic channels constrained by input energy. For pure-loss bosonic channels, the classical capacity is known exactly, while entanglement-enhanced capacity remains open for general one-mode Gaussian channels.

Classical Gaussian channels: Classical Gaussian channels model telephone channels and satellite links through b = τa + ξ, where ξ is Gaussian noise with variance V.The input signal has variance P, and τ denotes channel transmissivity.
Classical Gaussian channels: The classical Gaussian-channel formula predicts infinite communication rate when V = 0, reflecting the absence of measurement-accuracy limits in classical physics.Quantum treatment removes this divergence because information is encoded in quantized electromagnetic-field modes.
Quantum channel capacity: Quantum channel capacity is obtained by maximizing the Holevo information over input sources, with bosonic channels subject to a bounded-energy constraint.For memoryless channels, collective measurements achieve the Holevo information asymptotically; multi-letter entangled inputs define the full capacity.
Quantum channel capacity: For one-mode bosonic Gaussian channels, whether entangled inputs enhance classical capacity remains an open question.Hastings proved that such enhancement occurs for some channels, but the issue remains unresolved for one-mode bosonic Gaussian channels.
Pure-loss channels: C(Lp) = g(τµ + 1 −τ) is the classical capacity of a pure-loss channel with transmissivity τ and mean input photon number parameter µ := 2 ¯m + 1.The result follows from matching upper and lower bounds based on von Neumann entropy sub-additivity.

1. Bosonic minimum output entropy conjecture · D. Quantum capacity of Gaussian channels · E. Quantum dense coding and entanglement-assisted classical capacity

The section presents unresolved and established capacity results for Gaussian channels, from the unproved minimum-output-entropy conjecture to tight quantum-capacity bounds and entanglement-assisted classical communication advantages.

1. Bosonic minimum output entropy conjecture: Gaussian encodings provide lower bounds for the classical capacity of one-mode Gaussian channels, although calculating these capacities remains difficult.For a lossy channel, coherent-state encoding yields an explicit lower bound.
1. Bosonic minimum output entropy conjecture: The lower-bound capacity is conjectured to be tight because the bosonic minimum output entropy conjecture identifies vacuum input as minimizing lossy-channel output entropy.The conjecture remains unproved despite its intuitive appeal.
D. Quantum capacity of Gaussian channels: Quantum capacity requires regularization because coherent information is non-additive and may involve inputs entangled across multiple channel uses.For bosonic channels, coherent information remains finite even with infinite input energy, so no energy constraint is needed in the capacity definition.
D. Quantum capacity of Gaussian channels: For one-mode Gaussian channels, restricting to one use and pure Gaussian inputs yields a lower bound applicable to canonical classes A1, A2, C, and D.The bound is tight for degradable channels through additivity and Gaussian-state extremality.
E. Quantum dense coding and entanglement-assisted classical capacity: Quantum dense coding increases a channel’s classical capacity when sender and receiver share entanglement, while ignoring the cost of distributing that entanglement.The continuous-variable protocol uses an EPR state, dual-quadrature modulation, and Bell detection.
E. Quantum dense coding and entanglement-assisted classical capacity: The dense-coding rate can exceed the identity channel’s classical capacity at the same fixed average photon number for a considerable range of squeezing and detector efficiency.The relevant parameters are Vsq and η.
E. Quantum dense coding and entanglement-assisted classical capacity: Entanglement-assisted classical capacity is the maximum asymptotic reliable bit-transmission rate with unlimited pre-shared entanglement, and for the identity channel it equals twice the classical capacity.For one-mode Gaussian channels, the capacity is evaluated under an input-energy constraint and achieved by a Gaussian state.

F. Entanglement distribution and secret-key capacities … 3. No-switching protocol (heterodyne detection)

The paper develops Gaussian quantum-information applications from channel capacities and discrimination to continuous-variable QKD. It emphasizes practical Gaussian protocols, including coherent-state encoding and no-switching heterodyne detection, alongside security and implementation foundations.

F. Entanglement distribution and secret-key capacities: Entanglement-distribution capacity equals quantum capacity, E(M) = Q(M), while secret-key capacity quantifies secure bits distributed per channel use.Reverse coherent information is additive and gives a lower bound on reverse capacities.
G. Gaussian channel discrimination and applications: Gaussian channel discrimination minimizes the output error probability perr for distinguishing two possible channels, with optimal fixed-energy input states remaining an open problem.Quantum illumination uses this framework to detect low-reflectivity objects in bright thermal noise.
G. Gaussian channel discrimination and applications: Quantum illumination with EPR states yields pEPR(M) ≃ exp(−Mκ ¯m/¯n)/2, while coherent states yield pcoh(M) ≃ exp(−Mκ ¯m/4¯n)/2, a 6 dB disadvantage.The advantage persists despite the transmitted signal being entangled with an idler that does not probe the target.
VI. QUANTUM CRYPTOGRAPHY USING CONTINUOUS VARIABLES: Quantum key distribution lets Alice and Bob generate secret keys with theoretically unconditional security guaranteed by quantum mechanics, using Gaussian modulation and homodyne or heterodyne measurements.Continuous-variable QKD is presented as the Gaussian-state version of QKD.
A. Continuous-variable QKD protocols: Continuous-variable QKD protocols are organized as prepare-and-measure schemes, with an entanglement-based representation obtained by measuring a suitable entangled source.Alice prepares signal-state ensembles using a random number generator.
1. A generic protocol: QKD consists of quantum communication followed by classical post-processing, including sifting, parameter estimation, error correction, and privacy amplification over an authenticated public channel.Post-processing maps Alice’s and Bob’s raw keys into a shared secret key.
2. Coherent-state protocol (homodyne detection): Coherent states are sufficient for secret-key distribution and are easier to generate experimentally than other Gaussian states, enabling demonstrations and field implementations.The first Gaussian-modulated coherent-state protocol used direct reconciliation.
3. No-switching protocol (heterodyne detection): No-switching heterodyne detection measures q and p simultaneously, removes random basis switching, simplifies experiments, and produces higher secret-key rates.It generates two correlated data strings and is compatible with all known continuous-variable QKD protocols, despite an uncertainty-principle noise penalty.

4. Squeezed-state protocols … 4. Full characterization of collective Gaussian attacks

The review develops Gaussian continuous-variable QKD protocols from squeezed-state and fully-Gaussian constructions to security analyses, finite-size effects, and collective Gaussian attack characterizations. It emphasizes entanglement-based representations, practical imperfections, and Gaussian attacks as central tools for assessing asymptotic security.

4. Squeezed-state protocols: Squeezed-state protocols began with Gaussian modulation of squeezed states and Gaussian measurements, later incorporating heterodyne detection in reverse reconciliation.The heterodyne-based protocol is described as a noisy version of squeezed-state homodyne detection.
5. Fully-Gaussian protocols and entanglement-based representation: Fully-Gaussian protocols combine Gaussian-modulated states and Gaussian measurements, forming eight direct- or reverse-reconciliation variants assessable against collective Gaussian attacks.Their entanglement-based representation uses an EPR state, beam splitters, and homodyne or heterodyne detection to generate squeezed- or coherent-state protocols.
9. Thermal state QKD: Thermal-state QKD analyzes noisy coherent states to account for impurity arising from experimental imperfections in Alice’s initial states.This practical issue was first investigated by Filip (2008) and Usenko and Filip (2010).
B. Security analysis: The strongest security definition requires that, except with probability ǫ, Alice and Bob’s generated key is identical to an ideal secret key, otherwise they abort.Security is expressed using the trace distance between the actual joint state and the ideal secret-key state.
1. Main eavesdropping attacks: In coherent attacks, Eve interacts globally with all signals and later performs an optimal joint measurement, whereas collective attacks use independently prepared ancillas interacting with individual signals.The quantum de Finetti theorem enables unconditional asymptotic security against coherent attacks in the continuous-variable setting.
2. Finite-size analysis: Finite-size analysis replaces the infinite-signal idealization with corrections depending on exchanged signals, key-generation data, reconciliation efficiency, parameter estimation, and privacy amplification.Although collective and coherent attacks are equally powerful asymptotically, the finite-regime correction can be improved, and proposed alternatives remain partial.
3. Optimality of collective Gaussian attacks: For fully-Gaussian protocols, permutation symmetry implies that the secret key is minimized by the Gaussian state sharing the same covariance matrix as the relevant tripartite state.Thus, collective Gaussian attacks provide the fundamental benchmark for asymptotic security tests of Gaussian-modulated continuous-variable QKD.
4. Full characterization of collective Gaussian attacks: A collective Gaussian attack is characterized by reducing a one-mode Gaussian channel to canonical form using transmissivity τ, rank r, and thermal number n̄, then describing its canonical dilation.In the asymptotic regime, ancillary details can be ignored, leaving the dilation {L(τ, r), |ν⟩} and Gaussian unitaries {U, W}; canonical attacks set U = W = I.

5. Secret-key rates … VII. CONTINUOUS-VARIABLE QUANTUM COMPUTATION USING GAUSSIAN CLUSTER STATES

The paper analyzes continuous-variable QKD secret-key rates under collective Gaussian attacks, postselection, discrete modulation, and two-way communication, while identifying reconciliation, finite-size effects, and Gaussian entanglement distillation as important challenges. It then introduces continuous-variable quantum computation through circuit, cluster-state, and GKP approaches, emphasizing finite squeezing and fault tolerance.

5. Secret-key rates: Secret-key rates are derived for fully Gaussian protocols under collective entangling-cloner attacks characterized by channel transmission τ and excess noise χ.The protocols differ in Alice’s and Bob’s measurements and reconciliation, while Eve’s information is obtained through symplectic-eigenvalue calculations.
b. Postselection: Postselection complicates security analysis because filtering makes the description non-Gaussian, requiring tighter bounds and numerical evaluation of the overall secret-key rate.The effective binary channel contributes zero whenever βI(a : b) − S_a,b(E : x) < 0.
C. Future directions: Continuous-variable QKD still requires unconditional security in realistic conditions, improved reconciliation, and finite-size analysis to extend operation toward distances comparable to about 100 km.Field implementations had reached only 27 km, largely because Gaussian-modulation reconciliation is inefficient at low signal-to-noise ratios.
c. Discrete modulation of Gaussian states: Nearly an order of magnitude higher secret-key rates from four coherent states enable secret-key distribution over distances of about 50 km despite finite-size effects.The protocol uses eight-dimensional-sphere modulation and Bob’s heterodyne-based no-switching detection.
d. Two-way quantum communication: Two-way coherent-state QKD activates security in the ON configuration, whose excess-noise threshold exceeds the OFF threshold for every τ ∈(0, 1).Thus, channels with excess noise intolerable in OFF can remain tolerable in ON.
C. Future directions: Continuous-variable repeaters face a serious limitation because Gaussian operations cannot distill Gaussian entanglement, despite repeaters’ potential to distribute long-distance entanglement for secret-key extraction.The proposed repeater approach combines entanglement distillation, swapping, and quantum memories.
VII. CONTINUOUS-VARIABLE QUANTUM COMPUTATION USING GAUSSIAN CLUSTER STATES: Continuous-variable computation includes circuit and cluster-state models, while GKP computation encodes finite-dimensional qubits into the harmonic oscillator to support fault tolerance.Cluster states are highly entangled multimode states, ideally generated with infinitely squeezed states but practically approximated using finite squeezing.
VII. CONTINUOUS-VARIABLE QUANTUM COMPUTATION USING GAUSSIAN CLUSTER STATES: The computation section develops continuous-variable gates and universality, one-way computation through teleportation, graph states, nullifiers, and finite-squeezing Gaussian errors.The models must be combined with fault-tolerant, error-correctable systems in which continuous variables are eventually discretized.

A. Continuous-variable quantum gates · 1. Universal set of quantum gates · B. One-way quantum computation using continuous variables

Continuous-variable quantum computation is built from Gaussian single- and two-mode gates, but universality additionally requires a non-Gaussian nonlinear gate. One-way computation instead realizes arbitrary Hamiltonians through adaptive single-mode measurements on entangled cluster states.

A. Continuous-variable quantum gates: Gaussian continuous-variable computation uses displacement, beam-splitter, single-mode squeezing, and two-mode squeezing gates as core operations.These gates are introduced as important Gaussian gates for continuous-variable quantum computation.
A. Continuous-variable quantum gates: The Fourier gate is the Gaussian analogue of the qubit Hadamard gate and implements a π/2 phase-space rotation between quadratures.It relates position and momentum basis states through a Fourier transform.
A. Continuous-variable quantum gates: The phase gate performs a shearing operation, while CPHASE is a two-mode Gaussian gate that transforms momentum quadratures and leaves position quadratures unchanged.The phase gate combines rotations and squeezers; CPHASE acts nontrivially on momenta but does nothing to positions.
A. Continuous-variable quantum gates: The circuit model represents displacement, squeezing, Fourier, and phase gates as single-mode operations, and CPHASE, beam splitter, and two-mode squeezing as two-mode operations.The review distinguishes the graphical representations of single-mode and two-mode Gaussian gates.
1. Universal set of quantum gates: Universal continuous-variable computation requires all Gaussian operations plus a non-Gaussian gate exp[itq^n] with n≥3.The Lloyd-Braunstein criterion specifies Gaussian gates Z(t), P(η), F, and arbitrary multimode Gaussian gates, together with a nonlinear polynomial transformation.
1. Universal set of quantum gates: Gaussian gates alone cannot synthesize arbitrary Hamiltonians, and Gaussian processing from an initial Gaussian state can be efficiently simulated classically.This limitation includes Gaussian measurements and Gaussian operations.
B. One-way quantum computation using continuous variables: One-way continuous-variable quantum computation performs algorithms through single-mode measurements on an entangled cluster state, with measurement outcomes selecting later bases.Quantum gates are not required because arbitrary Hamiltonians are simulated through measurements; adaptiveness governs the measurement order.
B. One-way quantum computation using continuous variables: Cluster-state computation initializes highly squeezed vacuum qumodes, applies CZ gates to entangle them, and then measures the relevant qumodes.When only Gaussian gates are implemented, measurement order becomes irrelevant through parallelism rather than adaptiveness.

1. Understanding one-way computation via teleportation … 2. Stabilizers and nullifiers

The paper explains one-way continuous-variable computation through teleportation, showing that measurements can simulate gates and enable universal transformations. It then represents cluster states graphically and characterizes them through commuting stabilizers and nullifiers.

1. Understanding one-way computation via teleportation: Gate teleportation transforms an arbitrary input into X(m1)F|ψ⟩, with corrections recovering the original state.The correction relation is F†X†(m1)|ψ′⟩=|ψ⟩.
1. Understanding one-way computation via teleportation: Measurement-based computation replaces explicit gate implementation by measuring in a basis determined by the desired gate.The gate U can be absorbed into the measurement process, enabling its effect through basis choice.
2. Implementing gates using measurements: Measurements implement the universal Hamiltonians q, q^2, and q^3, while the Gaussian two-mode gate is supplied during cluster construction.The cubic Hamiltonian is used because it is optically implementable.
2. Implementing gates using measurements: The corresponding transformations realize displacement Z(t), phase P(t), and cubic phase V(t) through momentum or rotated-quadrature measurements.The first transformation adds t to the p-measurement result; the second uses homodyne detection in a rotated quadrature.
1. Graph states: A continuous-variable graph state is specified by vertices for squeezed momentum eigenstates and edges for CZ operations between qumodes.This graph construction provides a convenient depiction of cluster states.
1. Graph states: For a two-mode cluster, initialization creates two vertices, CZ adds their connecting edge, and a p-quadrature measurement is shown on the first node.The graph formalism is presented as equivalent to the teleportation-circuit description.
2. Stabilizers and nullifiers: A continuous-variable stabilizer is an operator with eigenvalue +1 on the state, exemplified by X(s)|0⟩p=|0⟩p for all s.The stabilizer formalism defines and analyzes cluster or graph states.
2. Stabilizers and nullifiers: Nullifiers satisfy Hi|φ⟩=0, commute pairwise, and are related to stabilizers by Ki(s)=e^−isHi; graph neighborhoods determine their form.For a three-node linear cluster, the nullifiers are p1−q2, p2−q1−q3, and p3−q2.

3. Shaping clusters: removing nodes and shortening wires … 1. Creating the cubic phase state

Gaussian cluster states can be reshaped through quadrature measurements, implemented through several optical architectures, and used for universal computation when supplemented by non-Gaussian resources. Finite squeezing introduces Gaussian distortions, while cubic-phase-state synthesis uses measurement-dependent corrections and squeezing operations.

3. Shaping clusters: removing nodes and shortening wires: Quadrature measurements remove graph-state nodes, while momentum measurements preserve correlations between neighboring nodes, enabling cluster-topology shaping.A Gaussian cluster can be reshaped into a required algorithmic topology; four-mode shaping was experimentally demonstrated with homodyne detection and feedforward.
D. Gaussian errors from finite squeezing: Finite squeezing causes Gaussian distortion with zero mean and variance 1/VS, equivalently convolution in momentum space by a Gaussian of variance VS.The same distortion mechanism applies to gate teleportation, where the output is MX(m1)FU |ψ⟩.
E. Optical implementations of Gaussian cluster states: Optical continuous-variable cluster-state generation is deterministic, and homodyne detection alone implements any multimode Gaussian transformation once the cluster is established.The optical approach uses readily available Gaussian elements, but practical implementations must account for errors from finite squeezing.
1. Canonical method: The canonical method prepares momentum-squeezed vacua and applies CZ gates, whose mutual commutation removes dependence on application order.Each CZ gate is implemented optically with two beam splitters and two online squeezers.
2. Linear-optics method: The linear-optics method creates cluster states using only offline squeezed states and a beam-splitter network, moving difficult online squeezing offline.Antisqueezing is suppressed, smaller input squeezing is required, and experiments demonstrated four-mode cluster geometries and simple one-way computations.
3. Single-OPO method: The single-OPO method generates an ultra-compact, scalable, universal N-mode cluster state in one top-down step using O(N^2) resources.An appropriately constructed multi-frequency pump enables generation of any continuous-variable cluster state with a bipartite graph.
4. Single-QND-gate method: The single-QND-gate method revisits CZ-based generation with a compact scheme for arbitrarily large cluster states, using temporally encoded qumodes repeatedly fed through one gate.This contrasts with the canonical method’s need for O(N^2) low-noise CZ gates to establish the initial cluster.
5. Temporal-mode-linear-optics method; F. Universal quantum computation; 1. Creating the cubic phase state: The temporal-mode-linear-optics method combines earlier approaches, while universal computation additionally requires a non-Gaussian element such as the cubic phase state.The cubic phase state is synthesized after a measurement-dependent cubic strength γ′(n), using two squeezing gates and a cubic phase gate for correction.

2. Implementing the cubic phase gate … VIII. CONCLUSION AND PERSPECTIVES

The review presents cubic-phase-gate implementations, continuous-variable error-correction constraints, and continuous-variable algorithms within cluster-state quantum computation. It concludes that Gaussian quantum information is mathematically simple and technologically versatile, while several important protocols require non-Gaussian resources and fault-tolerance remains a central future direction.

2. Implementing the cubic phase gate: Embedding a cubic phase state into a cluster enables cubic-gate teleportation using only Gaussian homodyne measurements.The finite-squeezing cluster becomes non-Gaussian, and the gate is teleported onto the input state up to known Gaussian corrections.
2. Implementing the cubic phase gate: Starting from a universal Gaussian cluster instead requires both Gaussian and non-Gaussian measurements, online squeezing corrections, and sequential measurement dependence.Photon counting introduces probabilistic outcomes, invalidating the usual parallelism of the measurement procedure.
G. Quantum error correction: Continuous-variable fault tolerance is necessary because uncorrected physical errors can propagate during large-scale quantum processing and make computation useless.The review frames fault-tolerant error correction as necessary for establishing computational capability in an imperfect physical system.
G. Quantum error correction: Gaussian operations alone cannot error-correct Gaussian noise on Gaussian states, whereas non-Gaussian resources or operations provide routes around this limitation.Examples include non-Gaussian encodings corrected with Gaussian operations and non-Gaussian operations for Gaussian-state entanglement distillation.
H. Continuous-variable algorithms: Continuous-variable analogs of Grover’s search and Deutsch-Jozsa algorithms have been developed using the quantum circuit model.Both algorithms originated in discrete-variable systems before continuous-variable analogs were formulated.
I. Future directions: Future work should develop fault tolerance for continuous-variable cluster computation and incorporate existing and new continuous-variable algorithms into the cluster-state model.The review identifies these as important avenues for advancing continuous-variable quantum computation.
VIII. CONCLUSION AND PERSPECTIVES: Gaussian quantum information gains analytical simplicity from first- and second-order moments and supports protocols involving Gaussian states, operations, and measurements.The review also covers distinguishability bounds and Gaussian bosonic communication-channel capacity and statistical discrimination.
VIII. CONCLUSION AND PERSPECTIVES: Universal computation, entanglement distillation, and error correction are impossible within a purely Gaussian framework unless supplemented by a non-Gaussian state, operation, or measurement.The review nevertheless anticipates a key role for Gaussian quantum information because its protocols are simple, versatile, and technologically available.