A Dynkin condition for manifolds with boundary

Abstract: We propose a Dynkin-type condition for smooth Riemannian manifolds with boundary. We show that this condition implies bi-Lipschitz equivalence with a Bakry-Émery weighted Riemannian manifold obtained via a time change. As a consequence, we obtain various results, including a local doubling property as well as lower bounds on the Neumann spectral gap and logarithmic Sobolev constant. The local doubling property also yields a new precompactness theorem for manifolds with boundary.

• The paper introduces a Neumann–Dynkin condition that integrates negative parts of Ricci curvature and the second fundamental form for manifolds with boundary.
• Using a novel time change method, it establishes Bakry–Émery curvature-dimension conditions and obtains explicit lower bounds for the spectral gap and logarithmic Sobolev constant.
• The framework derives local doubling and Poincaré inequalities, leading to robust convergence and compactness results under weak, integral curvature control.

A Dynkin Condition for Manifolds with Boundary

Introduction and Context

The paper "A Dynkin condition for manifolds with boundary" (2605.31223) establishes a novel analytic framework extending integral Ricci curvature lower bounds, specifically the Dynkin and Kato-type conditions, to the setting of smooth Riemannian manifolds with boundary. While pointwise Ricci curvature lower bounds and their integral relaxations (Kato/Dynkin) drive much of modern geometric analysis and geometric convergence theory (notably for geodesically complete, boundaryless manifolds), the role of such analytic control in the presence of boundary is nontrivial. The authors formulate a Neumann--Dynkin condition, reflecting both interior (negative part of Ricci curvature) and boundary (negative part of the second fundamental form) contributions, and derive sharp geometric and analytic consequences.

Neumann--Dynkin Condition: Formulation and Technical Innovation

The core contribution is the introduction of the Neumann--Dynkin condition, which employs the Neumann heat kernel $p^N(t, x, y)$ and a measure $m = \mathrm{Ric}_- \mu + \mathrm{II}_- \sigma$, combining contributions from the interior and the boundary. For some $T > 0$ and $\gamma \in [0, 1/(n-2))$, the condition reads

$$k_T(m) := \sup_{x\in M} \int_0^T \int_{M} p^N(s,x,y) \, dm(y) ds \le \gamma.$$

This condition, critically, is non-pointwise—distinct from classical lower curvature bounds—and only involves integrated information on the negative parts of Ricci curvature and the second fundamental form.

The analytic framework integrates recent developments on Kato, Dynkin, and $L^p$-Ricci curvature bounds for closed manifolds (see, e.g., [RoseStollmann2], [CarronRose], [CMT], [RoseWei]), adapting them to the case of boundary by coupling the interior and boundary geometry via the Neumann process. The development of this framework requires careful measure-theoretic and functional-analytic control, particularly in constructing associated Schrödinger operators with interior and (Robin-type) boundary potentials.

Main Results

Bi-Lipschitz Equivalence and Bakry-Émery Reduction via Time Change

A centerpiece of the manuscript is the proof that a Neumann--Dynkin condition implies the existence of a time change (i.e., a conformal/weighted metric transformation) resulting in a Bakry-Émery weighted manifold with convex boundary and explicit lower bounds for the curvature-dimension condition over a controlled scale. Precisely, for $k_T(m) < 1/(n-2)$, there exists a smooth, positive function $h$ such that $(M, e^{2h}g, e^{2h}\mu)$ is bi-Lipschitz to $(M,g)$ and solves the synthetic curvature condition $m = \mathrm{Ric}_- \mu + \mathrm{II}_- \sigma$0, where $m = \mathrm{Ric}_- \mu + \mathrm{II}_- \sigma$1 and $m = \mathrm{Ric}_- \mu + \mathrm{II}_- \sigma$2 are controlled in terms of $m = \mathrm{Ric}_- \mu + \mathrm{II}_- \sigma$3 and $m = \mathrm{Ric}_- \mu + \mathrm{II}_- \sigma$4. This is achieved by constructing $m = \mathrm{Ric}_- \mu + \mathrm{II}_- \sigma$5 as a solution to a Schrödinger equation with interior and boundary potentials reflecting the (negative) Ricci and second fundamental form, generalizing previous Kato-based conformal deformations to the context with boundary.

The time change argument leverages the perturbation theory for Dirichlet forms and a detailed analysis of regularity theory for elliptic equations with Robin-type boundary conditions. The method provides a robust analytic pathway from integrated curvature bounds to geometric regularity and control.

Local Doubling, Poincaré Inequality, and Precompactness

As an immediate corollary of the time-change reduction and synthetic curvature control, the authors establish that the Neumann--Dynkin condition ensures:

• Local Doubling: For any ball $m = \mathrm{Ric}_- \mu + \mathrm{II}_- \sigma$6 of radius much less than $m = \mathrm{Ric}_- \mu + \mathrm{II}_- \sigma$7,

$m = \mathrm{Ric}_- \mu + \mathrm{II}_- \sigma$8

where $m = \mathrm{Ric}_- \mu + \mathrm{II}_- \sigma$9 depends only on $T > 0$0.

• Local $T > 0$1 Poincaré Inequality: For $T > 0$2 and $T > 0$3,

$T > 0$4

These lead by standard arguments to precompactness results (in the pointed Gromov–Hausdorff topology) for the class of smooth complete manifolds with boundary satisfying such a Neumann--Dynkin condition [Gromov, CMT].

Spectral and Functional Estimates

Employing connections to time-changed geometry and referencing established bounds for Bakry-Émery spaces with convex boundary, the paper derives explicit lower bounds for the Neumann spectral gap. That is, under the Neumann--Dynkin condition and a diameter bound, the first nontrivial Neumann eigenvalue $T > 0$5 admits a computable lower bound, generalizing results based on classical pointwise bounds (e.g., [Li-Yau, Cheeger, Wang]) to the integral, "Dynkin-type" regime. This extends, and in some situations strictly improves on, previous eigenvalue lower bounds under less flexible Kato/Dynkin hypotheses [RoseWei]. Additionally, bounds on the logarithmic Sobolev constant are deduced, with explicit estimates in terms of $T > 0$6, and the diameter.

Weakening and Generalization of Curvature Control

The analytic Neumann--Dynkin condition can be much weaker than assumptions appearing in most earlier compactness and convergence theorems for manifolds with boundary, all the more since it does not impose pointwise or uniform Ricci/second fundamental form bounds, nor requires strong geometric controls such as uniform injectivity radius or rolling ball conditions.

Techniques and Proof Strategy

The authors develop a sophisticated framework based on Dirichlet forms and the Neumann heat semigroup, extending the machinery of Stollmann–Voigt and Guneysu for interior Kato/Dynkin bounds. The central technical device is a simultaneous time change and weight transformation, constructed via solutions to boundary-value Schrödinger equations incorporating both negative Ricci and boundary curvature (as negative potential terms). Regularity theory (elliptic PDE with Robin boundary) is crucial to global representation and uniform estimates.

Limiting theories (Gromov–Hausdorff and Mosco convergence) are controlled via the established volume and Poincaré inequalities, following the tradition of Cheeger-Gromov-Kodani–Wong–Perales in the boundary case, but under strictly weaker analytic hypotheses on the sequence.

Consequences and Implications

Practically, the Neumann--Dynkin condition allows for a broad class of manifolds with boundary to be included in compactness and convergence frameworks, including those with small singular sets, regions of negative curvature, or thin boundary strips where only integral control (rather than sectional or Ricci lower bounds) is available. The results suggest the possibility of extending synthetic Ricci and Bakry-Émery curvature-dimension theory to the boundary setting under integral curvature bounds, with implications for heat kernel estimates, convergence of spectral data, and limit space regularity.

Theoretically, the approach bridges potential theory, stochastic analysis (involving reflected Brownian motion through heat kernel representations), and geometric analysis under non-uniform curvature control. This opens new directions for studying non-smooth limits of manifolds with boundary, stability of analytic and geometric invariants, and extensions of RCD/Kato geometry in the presence of boundary.

Conclusion

This work significantly advances the analytic theory of integral curvature bounds as applied to manifolds with boundary, providing a unified approach that yields both precise geometric regularity statements and effective analytic inequalities (doubling, Poincaré, spectral gap, log-Sobolev constant) under rigorously quantified control conditions. The Neumann--Dynkin framework is robust, strictly weaker than existing compactness assumptions, and likely to yield further insight in geometric analysis and the study of synthetic lower curvature bounds for spaces with boundary.