| Holevo bound | |
|---|---|
| Name | Holevo bound |
| Field | Quantum information theory |
| Named after | Alexander Holevo |
| Introduced | 1970 |
| Statement | "Upper bound on accessible classical information from quantum states" |
Holevo bound
The Holevo bound is a fundamental inequality in Quantum information theory that limits the amount of classical information that can be retrieved from an ensemble of quantum states. Formulated by Alexander Holevo in 1973 (often cited as 1973–1974 work following earlier notes), it shows that even though a quantum system in a Hilbert space of dimension d can be described by continuous parameters, the amount of reliably extractable classical information is constrained by the ensemble's von Neumann entropy. The result underpins limits on quantum communication capacity and distinguishes quantum measurements from classical information processing.
The Holevo bound provides an upper limit on the accessible information obtainable about a classical random variable X encoded into quantum states {ρ_x} with prior probabilities p_x. For an ensemble {p_x, ρ_x} with average state ρ = Σ_x p_x ρ_x, the bound states that the mutual information I(X:Y) between the classical variable X and the outcome Y of any quantum measurement satisfies I(X:Y) ≤ S(ρ) − Σ_x p_x S(ρ_x), where S(σ) = −Tr(σ log σ) is the von Neumann entropy. This right-hand side is often called the Holevo χ quantity χ( {p_x, ρ_x} ). The inequality is independent of the specific positive-operator valued measure (POVM) used for measurement, and hence sets a fundamental limit for protocols in quantum communication and quantum cryptography.
Formally, let X be a discrete random variable with distribution {p_x} and let ρ_x be density operators on a finite-dimensional Hilbert space H. The sender prepares ρ_x with probability p_x and the receiver performs a general measurement described by a POVM {E_y}. The joint distribution of X and outcome Y yields a classical mutual information I(X:Y). The Holevo χ is defined as χ = S(ρ) − Σ_x p_x S(ρ_x). The proof uses properties of relative entropy (quantum Kullback–Leibler divergence) D(σ||τ) = Tr[σ (log σ − log τ)], monotonicity of relative entropy under completely positive trace-preserving maps (Uhlmann's theorem and Lieb's theorem are often invoked), and the data-processing inequality. A standard proof constructs a cq-state (classical-quantum state) ρ_XQ = Σ_x p_x |x⟩⟨x| ⊗ ρ_x on a composite space and applies subadditivity and monotonicity of quantum relative entropy to relate I(X:Y) to χ. Variants of the proof reference results by Lieb, Ruskai, and the framework of completely positive maps and quantum channels.
Accessible information is defined as the maximum mutual information over all possible measurements: I_acc = max_{POVM} I(X:Y). The Holevo bound implies I_acc ≤ χ. In many ensembles χ is not achievable with any single measurement; equality holds in special cases such as ensembles of mutually commuting states or ensembles of orthogonal pure states. The bound is central in comparing classical mutual information from a measurement with quantum mutual information quantities like the quantum mutual information I(A:B) = S(A)+S(B)−S(AB) for bipartite states. It also connects to the Holevo–Schumacher–Westmoreland theorem (HSW theorem), which uses χ to characterize the product-state classical capacity of a quantum channel.
Holevo's bound constrains protocols in quantum communication, such as classical information transmission over quantum channels and schemes using quantum key distribution (QKD) where an eavesdropper's information gain is limited. In the context of the HSW theorem, χ provides the achievable rate for sending classical messages via product-state encodings over memoryless channels; for entangled inputs the capacity theory involves regularized χ-like quantities and the Holevo capacity concept. Example ensembles include binary pure-state encodings (e.g., coherent states in quantum optics), ensembles of mixed states used in dense coding variants, and ensembles used to study capacities of channels such as the depolarizing channel and amplitude damping channel. Practical implications arise in design of encoding strategies for systems developed at institutions like IBM Research, Google Quantum AI, and experimental groups in quantum optics.
The Holevo bound has been extended and related to many results: the HSW theorem for channel capacity, the Holevo–Jozsa–Schumacher–Westmoreland–Wootters results on ensemble distinguishability, and trade-off relations involving entanglement and classical information. Related inequalities include strong subadditivity of von Neumann entropy, Araki–Lieb inequality, and bounds based on quantum relative entropy and Sandwiched Rényi divergence used in one-shot and finite-blocklength regimes. Generalizations adapt χ to infinite-dimensional systems (e.g., in quantum optics with energy constraints), to continuous-variable encodings, and to scenarios with entangled inputs leading to additivity questions historically investigated by researchers such as Peter Shor, Christopher King, and Michael Nielsen.
Operationally, the Holevo bound quantifies how quantum systems limit classical information extraction and thus shapes the design of quantum communication protocols, error correction schemes, and cryptographic security proofs. It helps distinguish capacities: the classical capacity of quantum channels, the private capacity for secure communication, and entanglement-assisted capacities where superdense coding achieves higher rates by exploiting shared entanglement. The bound is a cornerstone of resource-theoretic analyses comparing classical and quantum resources and is taught in graduate texts such as Nielsen and Chuang's Quantum Computation and Quantum Information and Holevo's own monograph. Its influence spans theory and experiment, guiding work at universities and labs including Caltech, MIT, and national laboratories involved in quantum technologies. Category:Quantum information theory