The algebraic Haag theorem without the split property, and Weiner's Conjecture 1.1 on standard half-sided modular inclusions

Weiner (2011) proved an algebraic version of Haag's theorem: under the standard Haag-Kastler axioms together with the Bisognano-Wichmann property and the split property, if a single unitary connects the local von Neumann algebras of two algebraic quantum field theories on all double cones with bases on one half of a fixed spacelike hyperplane, then the two models are fully equivalent — the vacuum vector and the Poincaré representation are forced by the algebra, with no assumption that the connecting unitary preserves them. The proof, however, uses the split property only to extract a convergent subsequence of implementing unitaries, and the author leaves open whether the theorem survives without it, stating Conjecture 1.1: the unitary equivalence class of a standard half-sided modular inclusion of factors already determines its standard vector up to normalization. No proof or refutation of either statement appears in the literature, and the problem remains open.

Background

The classical theorem of Rudolf Haag says, roughly, that within the Wightman axioms a quantum field theory whose fields at one time are unitarily equivalent to those of a free theory is itself free; in particular the interaction picture cannot be made rigorous. In algebraic quantum field theory (AQFT) a model is described not by fields but by a Haag-Kastler net: to each open spacetime region $O$ one assigns a von Neumann algebra $\mathcal{A}(O)$ (a weakly closed algebra of bounded operators, physically the observables measurable in $O$) acting on a fixed Hilbert space. Weiner's minimal axioms are: (1) isotony, $\mathcal{A}(O_1)\subset\mathcal{A}(O_2)$ for $O_1\subset O_2$; (2) locality, spacelike separated algebras commute; (3) covariance under a strongly continuous unitary representation $U$ of the connected Poincaré group; (4) positivity of energy; (5) existence of a unique (up to phase) translation-invariant vacuum vector $\Omega$, cyclic for the algebra generated by all local algebras; (6) the Reeh-Schlieder property, i.e. $\Omega$ is cyclic for $\mathcal{A}(O)$ for every nonempty open region; and (7) the Bisognano-Wichmann property: the modular objects of a wedge algebra $\mathcal{A}(W)$ with respect to $\Omega$ (the modular operator and conjugation furnished by Tomita-Takesaki modular theory, since $\Omega$ is cyclic and separating for $\mathcal{A}(W)$) act geometrically as the Lorentz boosts associated with the wedge $W$ (Bisognano and Wichmann 1975). An eighth assumption is the split property: for double cones $O_1$ with closure contained in $O_2$ the inclusion $\mathcal{A}(O_1)\subset\mathcal{A}(O_2)$ is a split inclusion, meaning there is an intermediate type I factor $\mathcal{N}$ (a von Neumann algebra isomorphic to the algebra $B(\mathcal{K})$ of all bounded operators on some Hilbert space) with $\mathcal{A}(O_1)\subset\mathcal{N}\subset\mathcal{A}(O_2)$ (Doplicher and Longo 1984); physically it expresses a mild independence of regions separated by a finite distance. Non-split nets do exist (e.g. infinite tensor products), so split is a genuine restriction, not a theorem. Weiner's algebraic Haag theorem states: if two nets $\mathcal{A},\tilde{\mathcal{A}}$ satisfy axioms (1)-(8) and there is a single unitary $V$ with $V\mathcal{A}(K^{\diamond})V^{*}=\tilde{\mathcal{A}}(K^{\diamond})$ for every double cone $K^{\diamond}$ with base $K$ contained in one half $H^+$ of a fixed spacelike hyperplane, then the models are equivalent in the strongest sense: there is a unitary $W$ with $W\mathcal{A}(O)W^{*}=\tilde{\mathcal{A}}(O)$ for all double cones $O$ and $WU(g)W^{*}=\tilde{U}(g)$ for all Poincaré transformations $g$ in the connected Poincaré group — and, since the vacuum is the unique translation-invariant vector up to phase, $W$ also maps $\Omega$ to $\tilde{\Omega}$ up to a phase that can be absorbed by rephasing $W$. Thus agreement of the observables on a single time slice already fixes the whole theory — an algebraic counterpart of the classical statement that the interaction picture trivializes. The proof arranges implementing unitaries $W_n$ for larger and larger subalgebras and is completed by passing to the strong limit; the author writes that although he is convinced that this limit exists, he "could not show" it, so the split property is used instead to obtain a compactness condition by which at least a convergent subsequence can be extracted. Weiner's Theorem 4.2 establishes the general $(d+1)$-dimensional statement; in the chiral setting — Möbius covariant nets, i.e. nets of von Neumann algebras on the intervals of the circle $S^1$ covariant under a unitary representation of the Möbius group $PSL(2,\mathbb{R})$ of circle transformations, the framework of one-dimensional conformal field theory — Theorem 5.1 and Corollary 5.3 give the analogous uniqueness statement for strongly additive Möbius covariant nets, where strongly additive means $\mathcal{A}(I_1\cup I_2)=\mathcal{A}(I_1)\vee\mathcal{A}(I_2)$ for adjacent intervals (additivity with no gap). The chiral proof runs through standard half-sided modular inclusions (HSMIs): a triple $(\Omega,\mathcal{N}\subset\mathcal{M})$ where $\mathcal{N}\subset\mathcal{M}$ is an inclusion of von Neumann factors on a Hilbert space, $\Omega$ is a standard vector for the inclusion — i.e. cyclic and separating for both $\mathcal{N}$, $\mathcal{M}$ and the relative commutant $\mathcal{N}'\cap\mathcal{M}$ — and the modular group $\sigma^{\mathcal{M}}_t$ of $\mathcal{M}$ (the one-parameter group of automorphisms furnished by Tomita-Takesaki modular theory) maps $\mathcal{N}$ into itself for $t\leq0$ (half-sided modular inclusion in the sense of Doplicher-Longo and Wiesbrock); such inclusions generate chiral nets and encode Borchers triples (a von Neumann algebra together with a cyclic separating vector and a positive-energy translation group — the raw data equivalent to a two-dimensional model). The same split assumption appears there, and Weiner conjectures (Conjecture 1.1) that it is removable because, at the level of inclusions, the unitary equivalence class of the inclusion itself should already determine the standard vector. The open problem is precisely the removal of the split hypothesis, in both its net-theoretic and inclusion-theoretic formulations, together with the associated question of whether perturbing only the standard vector of a fixed HSMI (Araki perturbation, i.e. replacing $\Omega$ by a perturbed vector $\Omega_X$ while keeping the algebras) can ever generate a genuinely inequivalent net.

Problem Statement

The common objective is to decide whether the local algebraic structure of a quantum field theory, by itself, determines the theory — in particular whether the split property, the one axiom Weiner uses purely technically, can be dropped from the algebraic Haag theorem. The problem as stated in the source comprises three related open sub-problems — the general split-free algebraic Haag theorem (1), Weiner's Conjecture 1.1 on standard half-sided modular inclusions (2), and the perturbation facet (3) that tests (2). Each is an independently refutable claim, and resolving any one of them (positively, by proof, or negatively, by explicit counterexample) constitutes a solution to the task: (1) Split-free algebraic Haag theorem, general case. Prove, or refute by explicit counterexample: if two algebraic QFT models on $(d+1)$-dimensional Minkowski space satisfy Weiner's axioms (1)-(7) — isotony, locality, Poincaré covariance, positivity of energy, unique cyclic vacuum, the Reeh-Schlieder property, and the Bisognano-Wichmann property — but are not assumed to satisfy the split property, and if there exists a unitary $V$ such that $V\mathcal{A}(K^{\diamond})V^{*}=\tilde{\mathcal{A}}(K^{\diamond})$ for every double cone $K^{\diamond}$ with base on a fixed spacelike half-hyperplane, then the two models are equivalent, i.e. there exists a unitary $W$ with $W\mathcal{A}(O)W^{*}=\tilde{\mathcal{A}}(O)$ for all double cones, $WU(g)W^{*}=\tilde{U}(g)$ for all $g$ in the connected Poincaré group, and $W\Omega=\tilde{\Omega}$ (the vacuum condition holds up to a phase by uniqueness of the translation-invariant vacuum, and the phase can be absorbed by rephasing $W$). A refutation consists of two models meeting axioms (1)-(7), non-split, with unitarily equivalent hyperplane double-cone algebras but no such equivalence $W$. (2) Conjecture 1.1, chiral case. Prove or refute: for standard half-sided modular inclusions of factors $(\Omega,\mathcal{N}\subset\mathcal{M})$, the unitary equivalence class of the inclusion determines the standard vector up to normalization — i.e. if $(\tilde{\Omega},\tilde{\mathcal{N}}\subset\tilde{\mathcal{M}})$ is another such inclusion with $\|\tilde{\Omega}\|=\|\Omega\|$ and some unitary $V$ satisfies $V\mathcal{N}V^{*}=\tilde{\mathcal{N}}$ and $V\mathcal{M}V^{*}=\tilde{\mathcal{M}}$, then there exists a unitary $W$ with $W\mathcal{N}W^{*}=\tilde{\mathcal{N}}$, $W\mathcal{M}W^{*}=\tilde{\mathcal{M}}$, and $W\Omega=\tilde{\Omega}$. Together with Weiner's Corollary 5.3 this would remove the split hypothesis from the chiral Theorem 5.1 (equivalence of strongly additive Möbius covariant nets), and its refutation would give a counterexample to the chiral ($S^1$) counterpart of (1) — the split-free uniqueness statement for strongly additive Möbius covariant nets — rather than to the general $(d+1)$-dimensional theorem in (1) itself. (3) Perturbation facet. As a direct test of (2): determine whether, for a fixed inclusion $\mathcal{N}\subset\mathcal{M}$ of factors admitting a standard half-sided modular vector $\Omega$, an Araki-type perturbation of the vector alone (changing the 'dynamics' while keeping the same algebras) can produce a standard half-sided modular inclusion whose generated strongly additive Möbius covariant net is not equivalent to the original one — possible a priori only outside the split, strongly additive setting, since Weiner's theorem rules it out when split holds. The admitted perturbation class is fixed as follows: for a self-adjoint $X\in\mathcal{M}$ with $e^{iXt}\mathcal{N}e^{-iXt}\subset\mathcal{N}$ for all $t\leq0$ (in particular any self-adjoint $X\in\mathcal{N}$), $\Omega_X$ is the vector in the natural cone of $(\Omega,\mathcal{M})$ satisfying $J_{\Omega_X}=J_\Omega$ and $\ln(\Delta_{\Omega_X})=\ln(\Delta_\Omega)+X+JXJ$, and the perturbation is admitted when $(\Omega_X,\mathcal{N}\subset\mathcal{M})$ is again a standard half-sided modular inclusion.

References

1. Mihály Weiner, An Algebraic Version of Haag's Theorem, Communications in Mathematical Physics 305 (2011) 469-485, DOI 10.1007/s00220-011-1236-7, arXiv:1006.4726, https://doi.org/10.1007/s00220-011-1236-7
2. S. Doplicher and R. Longo, Standard and split inclusions of von Neumann algebras, Inventiones Mathematicae 75 (1984) 493-536, DOI 10.1007/BF01388641, https://doi.org/10.1007/BF01388641
3. Joseph J. Bisognano and Eyvind H. Wichmann, On the duality condition for a Hermitian scalar field, Journal of Mathematical Physics 16 (1975) 985-1007, DOI 10.1063/1.522605, https://doi.org/10.1063/1.522605
4. Huzihiro Araki and László Zsidó, Extension of the structure theorem of Borchers and its application to half-sided modular inclusions, Reviews in Mathematical Physics 17 (2005) 491-543, DOI 10.1142/S0129055X05002388, https://doi.org/10.1142/S0129055X05002388
5. Paolo Camassa, Roberto Longo, Yoh Tanimoto and Mihály Weiner, Thermal States in Conformal QFT. I, Communications in Mathematical Physics 309 (2011) 703-735, DOI 10.1007/s00220-011-1337-3, arXiv:1101.2865, https://doi.org/10.1007/s00220-011-1337-3
6. Yoh Tanimoto, Construction of two-dimensional quantum field models through Longo-Witten endomorphisms, Forum of Mathematics, Sigma 2 (2014) e1, DOI 10.1017/fms.2014.3, arXiv:1301.6090, https://doi.org/10.1017/fms.2014.3
7. Luca Giorgetti and Karl-Henning Rehren, Braided categories of endomorphisms as invariants for local quantum field theories, Communications in Mathematical Physics 357 (2017) 3-41, DOI 10.1007/s00220-017-2937-3, arXiv:1512.01995, https://doi.org/10.1007/s00220-017-2937-3
8. Sebastiano Carpi, Yasuyuki Kawahigashi, Roberto Longo and Mihály Weiner, From vertex operator algebras to conformal nets and back, Memoirs of the American Mathematical Society 254 (2018), no. 1213, DOI 10.1090/memo/1213, arXiv:1503.01260, https://doi.org/10.1090/memo/1213
9. Lutz Klaczynski, Haag's theorem in renormalised quantum field theories, PhD thesis, Humboldt-Universität zu Berlin (2016), arXiv:1602.00662, DOI 10.18452/17448, https://arxiv.org/abs/1602.00662
10. Christian D. Jäkel and Jens Mund, The Haag-Kastler Axioms for the $\mathscr{P}(\varphi)_2$ Model on the De Sitter Space, Annales Henri Poincaré 19 (2018) 959-977, DOI 10.1007/s00023-018-0647-9, https://doi.org/10.1007/s00023-018-0647-9
11. Vincenzo Morinelli and Karl-Henning Rehren, Spacelike deformations: higher-helicity fields from scalar fields, Letters in Mathematical Physics 110 (2020) 2019-2038, DOI 10.1007/s11005-020-01294-w, arXiv:1905.08714, https://doi.org/10.1007/s11005-020-01294-w
12. Roberto Longo, Yoh Tanimoto and Yoshimichi Ueda, Free products in AQFT, Annales de l'Institut Fourier 69 (2019) 1229-1258, DOI 10.5802/aif.3269, arXiv:1706.06070, https://doi.org/10.5802/aif.3269
13. Gandalf Lechner and Charley Scotford, Deformations of Half-Sided Modular Inclusions and Non-Local Chiral Field Theories, Communications in Mathematical Physics 391 (2022) 269-291, DOI 10.1007/s00220-022-04324-x, arXiv:2111.03172, https://doi.org/10.1007/s00220-022-04324-x
14. Lauritz van Luijk, Alexander Stottmeister, Reinhard F. Werner and Henrik Wilming, Embezzlement of entanglement, quantum fields, and the classification of von Neumann algebras, preprint (2024), arXiv:2401.07299, https://arxiv.org/abs/2401.07299
15. Ian Koot, Relative Positions of Half-Sided Modular Inclusions, Letters in Mathematical Physics 115 (2025), article 118, DOI 10.1007/s11005-025-02008-w, arXiv:2503.18036, https://doi.org/10.1007/s11005-025-02008-w