[Paper Review] An H1-BMO duality theory for semigroups of operators

Tao Mei|arXiv (Cornell University)|Apr 23, 2012

Advanced Operator Algebra Research15 references3 citations

TL;DR

This paper establishes an abstract H₁-BMO duality theory for semigroups of operators on general sigma-finite measure spaces, defining Hardy space H₁ and BMO space via square functions of P. A. Meyer's gradient form and semigroup actions. The key contribution is a duality inequality with absolute constants, valid under the Γ₂ ≥ 0 condition and additional regularity assumptions, extending to noncommutative settings like von Neumann algebras without relying on geometric or kernel-based structures.

ABSTRACT

Let (M,μ) be a sigma-finite measure space. Let (T_t) be a semigroup of positive preserving maps on (M,μ) with standard assumptions. We prove a H_1-BMO duality theory with assumptions only on T_t. The BMO is defined as spaces of functions f such that the L_\infty norm of sup_tT_t|f-T_tf|^2 is finite. The H1 is defined by square functions of P. A. Meyer's gradient form. Our argument does not rely on any geometric/metric structure of M nor on the kernel of the semigroups of operators. This abstract argument allows to extend our main results to the noncommutative setting, e.g. the case where L_\infty(M,μ) is replaced by von Neuman algebras with a semifinite trace. We also prove a Carleson embedding theorem for semigroups of operators.

Motivation & Objective

• To develop an H₁-BMO duality theory for semigroups of operators that relies only on the semigroup's intrinsic properties, not on geometric or metric assumptions of the underlying space.
• To define H₁ and BMO spaces using square functions of P. A. Meyer’s gradient form and semigroup actions, avoiding dependence on heat kernel bounds or specific function spaces.
• To establish a duality between H₁ and BMO with absolute constants, generalizing Stein’s universal H^p theory to the H₁-BMO case.
• To extend the theory to noncommutative settings, such as von Neumann algebras with semifinite traces, by removing reliance on the geometric structure of the measure space.
• To prove a Carleson embedding theorem for semigroups, supporting the duality results and enabling applications in noncommutative harmonic analysis.

Proposed method

• Define BMO space via the seminorm ‖f‖_BMO(T) = sup_t ‖T_t |f - T_t f|²‖_∞^{1/2}, and bmo space via ‖f‖_bmo(T) = sup_t ‖T_t |f|² - |T_t f|²‖_∞^{1/2}.
• Define H₁^S(T) and H₁^G(T) using square functions S_Γ(f) = (∫₀^∞ T_s Γ(T_s f) ds)^{1/2} and G_Γ(f) = (∫₀^∞ Γ(T_s f) ds)^{1/2}, where Γ is Meyer’s carré du champ.
• Use the Γ₂ ≥ 0 condition to control the gradient form and ensure positivity and regularity in the semigroup framework.
• Apply interpolation theory between bmo(T) and L₁(M), and use the boundedness of maximal operators M_t to control time-averaged norms.
• Prove a Carleson embedding theorem for semigroups by estimating tent space norms via time-integrated semigroup actions and dyadic decomposition.
• Establish duality via a key inequality: τ(fg) ≤ c₁‖f‖_{H₁^S}^{1/2}‖f‖_{H₁^G}^{1/2}‖g‖_{bmo} ≤ c₂‖f‖_{H₁^S}‖g‖_{bmo}, with absolute constants c₁, c₂.

Experimental results

Research questions

• RQ1 Under what conditions does the duality (H₁^S(T))⁰ = bmo(T) hold for semigroups of operators?
• RQ2 When are the two BMO norms ‖·‖_BMO(T) and ‖·‖_bmo(T) equivalent?
• RQ3 When does H₁^S(T) = H₁^G(T) hold, i.e., when are the two Hardy space definitions equivalent?
• RQ4 Can the H₁-BMO duality be extended to noncommutative settings such as von Neumann algebras without geometric assumptions?
• RQ5 What are the minimal regularity conditions on the semigroup ensuring the duality and equivalence results?

Key findings

• The duality inequality τ(fg) ≤ c₂‖f‖_{H₁^S(T)}‖g‖_{bmo(T)} holds with absolute constants c₂ independent of the semigroup or underlying space, under the Γ₂ ≥ 0 condition.
• If the semigroup satisfies additional regularity conditions—specifically, Hölder continuity in time and a uniform L¹-estimate on time-averaged semigroups—then (H₁^S(T))⁰ = bmo(T) holds with equivalent norms.
• Under the same regularity assumptions, bmo(T) = BMO(T) and H₁^S(T) = H₁^G(T), establishing norm equivalence between the two BMO and two Hardy space definitions.
• The constants in the norm equivalences depend only on the regularity parameters c₃, c₄, and r, not on the geometry of the underlying measure space.
• The theory extends to noncommutative settings, such as von Neumann algebras with semifinite traces, due to the abstract formulation relying only on semigroup structure and the Γ₂ condition.
• A Carleson embedding theorem is proven for semigroups, showing that the tent space norm of a measure is controlled by the L¹ norm of the function, with uniform constants.