betaFeedback

WorksMathematics

Axioms for relative entropy

Abstract

For normal states of arbitrary von Neumann algebras, we characterize positive finite multiples of Araki relative entropy among non-degenerate divergences. The axioms are monotonicity under normal unital completely positive maps, tensor additivity, superadditivity for product second states, and separate lower semicontinuity in the predual norm. Alternatively, a conditional-expectation identity replaces the two tensor assumptions. The proof uses order bounds and asymptotic estimates on tensor powers. An appendix proves the known analytic estimates needed here, using standard modular theory and Haagerup L^p spaces. We also obtain the corresponding characterization in the C* algebraic setting.

Introduction

Relative entropy is a primary information-theoretic notion. It measures the distinguishability of two states of a system. It was introduced by Kullback and Leibler [12] for classical probability distributions. If p=(p1,…,pr)p=(p_1,\ldots,p_r), q=(q1,…,qr)q=(q_1,\ldots,q_r) are finite probability measures, their relative entropy is given by

SCr(p∥q)=∑ipilog⁡piqi.S_{\mathrm{Cr}}(p\mathbin{\|}q)=\sum_i p_i\log\frac{p_i}{q_i}.

Umegaki introduced the relative entropy in the semifinite setting [24]; for density matrices of a matrix algebra AA,

SA(ρ∥σ)=Tr⁡ρ(log⁡ρ−log⁡σ).S_A(\rho\mathbin{\|}\sigma)=\operatorname{Tr}\rho(\log\rho-\log\sigma).

Araki extended it to normal positive functionals on arbitrary von Neumann algebras [1, 2], thus for arbitrary quantum systems. For normal faithful states φ,ψ\varphi,\psi of a von Neumann algebra MM in standard form, Araki’s definition is

SM(φ∥ψ)=−⟨ξφ,log⁡Δψ,φξφ⟩.S_M(\varphi\mathbin{\|}\psi)=-\langle\xi_\varphi,\log\Delta_{\psi,\varphi}\xi_\varphi\rangle.

Here Δψ,φ\Delta_{\psi,\varphi} is the relative modular operator between cyclic and separating vectors ξψ,ξφ\xi_\psi,\xi_\varphi representing ψ,φ\psi,\varphi. The expression is interpreted as an extended spectral integral.

Uhlmann’s construction by interpolation of quadratic forms [23] agrees with Araki’s for normal functionals; see Hiai–Ohya–Tsukada [8 Section 3]. Kosaki’s formula [11] gives another important variational description of the same functional. A recent, related description as ff-divergence is given by the integral representation of Correa da Silva–Fröb–Lechner–Sangaletti [3 Theorem 3.19].

The author’s interest in relative entropy originated from work on von Neumann algebras and quantum black holes [13]. Quantum field theory provides a natural setting for infinite quantum systems, in which relative entropy plays a crucial role, in particular in the study of local algebras, entropy-energy inequalities and black hole entropy. See [9, 14, 15] for examples of recent developments and further references.

The question here is whether relative entropy is determined by a small number of structural properties. Petz [20] and Ohya–Petz [18] gave earlier matrix-algebra characterizations involving conditional expectations. Donald [4] gives a distinct axiomatic construction of a generalized relative entropy and proves its agreement with the standard relative entropy for normal state pairs on injective von Neumann algebras. For nuclear C∗C^*-algebras, Petz [21, Theorem 1.2] states a characterization using the conditional-expectation identity, monotonicity under unital Schwarz maps, a direct-sum rule, and lower semicontinuity.

The asymptotic argument in this paper also has antecedents. Matsumoto [16] characterizes finite-dimensional relative entropy, up to scale, by monotonicity, additivity on tensor powers, and asymptotic lower semicontinuity. Wilming–Gallego–Eisert [25] obtain a characterization for matrix states with a faithful second state from continuity in the first state, monotonicity, additivity, and superadditivity. Their use of superadditivity to pass from regularity on one algebra to an estimate on its tensor powers is the mechanism used in Lemma 2.4, with lower semicontinuity in place of continuity.

In this paper we give a characterization for arbitrary von Neumann algebras, without injectivity or faithfulness assumptions. Starting with a divergence DD, that is a family of maps DMD_M taking values in [0,+∞][0,+\infty] and defined for all normal state pairs on arbitrary von Neumann algebras MM, we show that DD is a positive finite multiple of Araki relative entropy, with the same constant for every algebra and every pair of states, if DD satisfies four axioms: monotonicity, tensor additivity, superadditivity for product second states, and separate norm lower semicontinuity. We assume that the divergence DD is non-degenerate, meaning that it has a positive finite value at one state pair on one algebra. Neither the location of that pair nor its value is prescribed. The formula on finite commutative algebras is derived. Alternatively, the conditional-expectation identity can replace tensor additivity and superadditivity, with monotonicity and separate norm lower semicontinuity retained. An analogous characterization holds in the C∗C^*-algebraic setting. The C∗C^*-algebra proof uses the minimal tensor product and Kaplansky density, without a nuclearity assumption.

The analytic estimates in Proposition 3.1 are known: they follow from Fawzi–Gao–Rahaman [5], with the estimates for separation of tensor powers by positive contractions due to Poutrat–Wang [19]. Appendix § 9 gives a pedagogical proof of the dominated case needed here, using standard modular theory and Haagerup LpL^p spaces [6]. Its role is expository; it supplies established analytic input independently of the function DD being characterized.

Sections 1–4 state the axioms and prove the von Neumann result. § 6 gives the conditional-expectation formulation, and § 7 proves the C∗C^*-algebra result.

Setting, axioms, and plan of the proof

States, norms, and unital completely positive maps

Let MM be a von Neumann algebra, M∗M_* its predual, and Sn(M)S_n(M) its normal state space. Tensor products of von Neumann algebras are denoted by M⊗NM \otimes N. As usual, the support s(φ)s(\varphi) is the least projection e∈Me \in M with φ(e)=1\varphi(e)=1. Faithfulness means s(φ)=1s(\varphi)=1. We call MM σ\sigma-finite if it has a faithful normal state.

We consider a family of maps D=(DM)MD=(D_M)_M

DM:Sn(M)×Sn(M)⟶[0,+∞],D_M:S_n(M)\times S_n(M)\longrightarrow[0,+\infty],

indexed by all von Neumann algebras MM. No symmetry, bilinearity, or continuity is included in this definition. The value +∞+\infty is allowed and is important below. We call the family DD a divergence, and we say that the divergence DD is non-degenerate if there exist a von Neumann algebra M0M_0 and normal states φ0,ψ0\varphi_0,\psi_0 such that

0<DM0(φ0∥ψ0)<+∞.(1)0 < D_{M_0}(\varphi_0 \mathbin\Vert\psi_0) < +\infty. \tag*{(1)}

Non-degeneracy is a condition on the family: such a pair need not exist on each algebra. In particular, no positive value is required on C\mathbb{C}, which has only one state. Vanishing on equal states is not part of the definition; it will follow from non-degeneracy and the axioms.

A normal unital completely positive map (UCP) Φ:N→M\Phi:N\to M induces the map φ↦φ∘Φ\varphi\mapsto\varphi\circ\Phi from Sn(M)S_{\mathrm n}(M) to Sn(N)S_{\mathrm n}(N). For finite commutative algebras we often describe this induced map directly on probability vectors. If Tji≥0T_{ji}\ge0 and ∑jTji=1\sum_j T_{ji}=1, its action is p↦Tpp\mapsto Tp, and its adjoint is

ΦT:CJ→CI,(ΦTf)i=∑jTjifj.\Phi_T:\mathbb{C}^{J}\to\mathbb{C}^{I},\qquad(\Phi_Tf)_i=\sum_j T_{ji}f_j.

This adjoint is unital and completely positive. Consequently, the monotonicity axiom applies to every stochastic map used in the proof.

The entropy to be characterized

We use the nonnegative convention for Araki relative entropy SM(φ∥ψ)S_M(\varphi\mathbin\Vert\psi) [1, 2, 18]. In standard form, let ξφ\xi_\varphi be the natural-cone vector representing φ\varphi, and let Δψ,φ\Delta_{\psi,\varphi} be the relative modular operator in this order. Then

SM(φ∥ψ)=−⟨ξφ,log⁡Δψ,φ ξφ⟩.(2)S_M(\varphi\mathbin\Vert\psi)=-\langle\xi_\varphi,\log\Delta_{\psi,\varphi}\,\xi_\varphi\rangle. \tag*{(2)}

The expression is interpreted as an extended spectral integral and is well-defined.

Theorem 1.1 (Axiomatic characterization). Let D=(DM)MD=(D_M)_M be a non-degenerate divergence on all von Neumann algebras and all their normal state pairs. Assume the following four conditions.

(A1) Monotonicity. For every normal unital completely positive map Φ:N→M\Phi:N\to M,

DN(φ∘Φ∥ψ∘Φ)≤DM(φ∥ψ).D_N(\varphi\circ\Phi\mathbin\Vert\psi\circ\Phi)\le D_M(\varphi\mathbin\Vert\psi).

(A2) Tensor additivity. For φi,ψi∈Sn(Mi)\varphi_i,\psi_i\in S_{\mathrm n}(M_i),

DM1⊗M2(φ1⊗φ2∥ψ1⊗ψ2)=DM1(φ1∥ψ1)+DM2(φ2∥ψ2).D_{M_1\otimes M_2}(\varphi_1\otimes\varphi_2\mathbin\Vert\psi_1\otimes\psi_2)=D_{M_1}(\varphi_1\mathbin\Vert\psi_1)+D_{M_2}(\varphi_2\mathbin\Vert\psi_2).

(A3) Superadditivity for product second states. For ω∈Sn(M⊗N)\omega\in S_{\mathrm n}(M\otimes N), denote its marginal states by ωM(x)=ω(x⊗1)\omega_M(x)=\omega(x\otimes1) and ωN(y)=ω(1⊗y)\omega_N(y)=\omega(1\otimes y). Then

DM⊗N(ω∥ψM⊗ψN)≥DM(ωM∥ψM)+DN(ωN∥ψN).D_{M\otimes N}(\omega\mathbin\Vert\psi_M\otimes\psi_N)\ge D_M(\omega_M\mathbin\Vert\psi_M)+D_N(\omega_N\mathbin\Vert\psi_N).

(A4) Separate lower semicontinuity in norm. With either state fixed, the function is lower semicontinuous in the other state for the predual norm. Explicitly,

∥φk−φ∥→0⟹DM(φ∥ψ)≤lim inf⁡kDM(φk∥ψ),\lVert\varphi_k-\varphi\rVert\to0\quad\Longrightarrow\quad D_M(\varphi\mathbin\Vert\psi)\le\liminf_k D_M(\varphi_k\mathbin\Vert\psi),
∥ψk−ψ∥→0⟹DM(φ∥ψ)≤lim inf⁡kDM(φ∥ψk).\lVert\psi_k-\psi\rVert\to0\quad\Longrightarrow\quad D_M(\varphi\mathbin\Vert\psi)\le\liminf_k D_M(\varphi\mathbin\Vert\psi_k).

Then there is a constant 0<λ<∞0 < \lambda< \infty, independent of MM and of the states, such that

DM(φ∥ψ)=λSM(φ∥ψ)D_M(\varphi\mathbin\Vert\psi)=\lambda S_M(\varphi\mathbin\Vert\psi)

for every normal state pair on every von Neumann algebra. More precisely, for δ=(1,0)\delta=(1,0) and u=(12,12)u=\left(\frac{1}{2},\frac{1}{2}\right),

References

  1. [1]H. Araki, Relative entropy of states of von Neumann algebras, Publ. Res. Inst. Math. Sci. 11 (1976), 809–833. doi:10.2977/prims/1195191148.DOI
  2. [2]H. Araki, Relative entropy for states of von Neumann algebras II, Publ. Res. Inst. Math. Sci. 13 (1977), 173–192.DOI
  3. [3]R. Correa da Silva, M. B. Fröb, G. Lechner and L. Sangaletti, Integral representations of f-divergences for general von Neumann algebras (2026), arXiv:2607.05195.arxiv.org/abs/2607.05195
  4. [4]M. J. Donald, On the relative entropy, Commun. Math. Phys. 105 (1986), 13–34. doi:10.1007/BF01212339.DOI
  5. [5]O. Fawzi, L. Gao and M. Rahaman, Asymptotic equipartition theorems in von Neumann algebras, Ann. Henri Poincaré 27 (2026), 95–141 (published online in 2025). doi:10.1007/s00023-025-01545-3. arXiv:2212.14700v3. Numbered citations to this work use the arXiv version.DOI
  6. [6]U. Haagerup, L_p-spaces associated with an arbitrary von Neumann algebra, in Algèbres d’opérateurs et leurs applications en physique mathématique (Marseille, 1977), Colloques Internationaux du CNRS, vol. 274, CNRS, Paris, 1979, pp. 175–184. Available scan.
  7. [7]F. Hiai, Quantum f-divergences in von Neumann algebras I. Standard f-divergences (2018), arXiv:1805.02050.DOI
  8. [8]F. Hiai, M. Ohya and M. Tsukada, Sufficiency and relative entropy in *-algebras with applications in quantum systems, Pacific J. Math. 107 (1983), 117–143. Full text.DOI
  9. [9]S. Hollands and R. Longo, A new proof of the QNEC, Commun. Math. Phys. 406 (2025), 269. Volume dedicated to H. Araki. doi:10.1007/s00220-025-05450-y.DOI
  10. [10]V. Jakšić, Y. Ogata, C.-A. Pillet and R. Seiringer, Quantum hypothesis testing and non-equilibrium statistical mechanics, Rev. Math. Phys. 24 (2012), 1230002. doi:10.1142/S0129055X12300026. arXiv:1109.3804.DOI
  11. [11]H. Kosaki, Relative entropy of states: A variational expression, J. Operator Theory 16 (1986), 335–348. Full text.
  12. [12]S. Kullback and R. A. Leibler, On information and sufficiency, Ann. Math. Statist. 22 (1951), 79–86. doi:10.1214/aoms/1177729694.DOI
  13. [13]R. Longo, An analogue of the Kac–Wakimoto formula and black hole conditional entropy, Commun. Math. Phys. 186 (1997), 451–479. doi:10.1007/s002200050116.DOI
  14. [14]R. Longo, A Bekenstein-type bound in QFT, Commun. Math. Phys. 406 (2025), 95. doi:10.1007/s00220-025-05261-1.DOI
  15. [15]R. Longo and F. Xu, Relative entropy in CFT, Adv. Math. 337 (2018), 139–170. doi:10.1016/j.aim.2018.08.015.DOI
  16. [16]K. Matsumoto, Reverse test and characterization of quantum relative entropy (2010), arXiv:1010.1030.arxiv.org/abs/1010.1030
  17. [17]Y. Ogata, A generalization of the inequality of Audenaert et al. (2010), arXiv:1011.1340.arxiv.org/abs/1011.1340
  18. [18]M. Ohya and D. Petz, Quantum Entropy and Its Use, Springer, Berlin, 1993; corrected printing, 2004. doi:10.1007/978-3-642-57997-4.DOI
  19. [19]Y. Pautrat and S. Wang, Ke Li’s lemma for quantum hypothesis testing in general von Neumann algebras, Ann. Henri Poincaré 24 (2023), 2323–2339. arXiv:2010.02177.
  20. [20]D. Petz, Characterization of the relative entropy property of states of matrix algebras, Acta Math. Hungar. 59 (1992), 449–455. doi:10.1007/BF00050907.DOI
  21. [21]D. Petz, Relative entropy in quantum information theory, in O. E. Barndorff-Nielsen and K. Mølmer (eds.), Workshop on Stochastics and Quantum Physics, MaPhySto Miscellanea no. 16, University of Aarhus, 1999, pp. 84–88, especially Theorem 1.2. Workshop proceedings.
  22. [22]D. Petz, Quantum entropy, in Encyclopedia of Mathematical Physics, vol. 3, Elsevier, 2006, pp. 177–182.DOI
  23. [23]A. Uhlmann, Relative entropy and the Wigner–Yanase–Dyson–Lieb concavity in an interpolation theory, Commun. Math. Phys. 54 (1977), 21–32. doi:10.1007/BF01609834.DOI
  24. [24]H. Umegaki, Conditional expectation in an operator algebra, IV (Entropy and information), Kodai Math. Sem. Rep. 14 (1962), 59–85. doi.org/10.2996/kmj/1138844604.
  25. [25]H. Wilming, R. Gallego and J. Eisert, Axiomatic characterization of the quantum relative entropy and free energy, Entropy 19 (2017), 241. doi:10.3390/e19060241. arXiv:1702.08473.DOI

Paper details

vlnc:W_y4435edv
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

ChatGPT (OpenAI)

View statement
Looking for related papers…
Contents