Bonnet–Myers Diameter Bounds

• Bonnet–Myers bounds are foundational results that relate lower Ricci curvature conditions to compactness and explicit diameter upper limits in Riemannian geometry.
• Methodologies include ODE comparisons, spectral conditions, and discrete analogues, which yield sharp, quantifiable diameter estimates in both smooth manifolds and graphs.
• Recent innovations extend these bounds to sub-Riemannian, Lorentzian, and generalized curvature frameworks, offering new rigidity results and classification insights.

The Bonnet–Myers diameter bound is a foundational result in Riemannian geometry and its many generalizations, relating lower Ricci curvature control to finiteness of diameter and compactness. These principles have been extended and sharpened across a wide array of geometries, ranging from discrete graphs to sub-Riemannian, Lorentzian, and Finsler manifolds, and to objects equipped with generalized curvature tensors such as the Bakry–Émery or spectral Ricci tensors. This article presents an exhaustive account of Bonnet–Myers diameter bounds, emphasizing the most recent characterizations, diameter estimates, rigidity, and the interplay with discrete and sub-Riemannian analogues.

1. Classical Bonnet–Myers Theorem and Its Variants

The classical Bonnet–Myers theorem states that if a complete Riemannian manifold $(M^n,g)$ satisfies $\mathrm{Ric}_g \ge (n-1)K>0$, then $M$ is compact, $\pi_1(M)$ is finite, and

$\mathrm{diam}(M)\le\frac{\pi}{\sqrt{K}}\,,$

with equality only for the round sphere of radius $K^{-1/2}$. This result is paralleled in Finsler geometry under analogous lower Ricci scalar bounds, yielding the same diameter upper bound and compactness criteria, with arguments based on comparison of the index form along geodesics (Anastasiei, 2014).

Generalizations now encompass:

• Integral and average Ricci conditions, accommodating polynomial or exponential decay of Ricci curvature, with explicit diameter and compactness criteria via ODE/comparison techniques and integral barriers (Li et al., 2 Sep 2025).
• Spectral variants, where the lower bound on $\mathrm{Ric}$ is replaced by a spectral condition involving the first eigenvalue of $-\gamma\Delta+\mathrm{Ric}$, leading to sharp diameter bounds dependent on the oscillation of a positive eigenfunction $u$ (Antonelli et al., 2024).

2. Discrete Bonnet–Myers Bounds: Graphs and Curvature Notions

Discrete analogues for graphs relate positive lower bounds on various discrete Ricci curvature notions (Ollivier, Lin–Lu–Yau, Bakry–Émery, entropic Ricci, CD-condition) to upper bounds on combinatorial diameter.

2.1. Ollivier, Lin–Lu–Yau, and Entropic Ricci Curvature

For a simple connected graph $G=(V,E)$ and Lin–Lu–Yau curvature $\mathrm{Ric}_g \ge (n-1)K>0$0 on edges, the discrete Bonnet–Myers theorem asserts: $\mathrm{Ric}_g \ge (n-1)K>0$1 with sharpness achieved in highly symmetric families such as hypercubes, Johnson graphs, demi-cubes, and Gosset graphs (Kamtue, 2020, Li et al., 2024).

Bonnet–Myers sharp graphs satisfy $\mathrm{Ric}_g \ge (n-1)K>0$2, with every edge along some diameter-realizing path attaining this minimal curvature. In the context of Lin–Lu–Yau curvature, the formula admits an optimal-transport interpretation via minimum-cost blow-up matchings between neighbor sets (Li et al., 2024).

2.2. Bakry–Émery CD-Condition

If $\mathrm{Ric}_g \ge (n-1)K>0$3 satisfies the CD$\mathrm{Ric}_g \ge (n-1)K>0$4 condition and bounded vertex degree $\mathrm{Ric}_g \ge (n-1)K>0$5, then (Liu et al., 2016): $\mathrm{Ric}_g \ge (n-1)K>0$6 This is sharp on hypercubes. For CD$\mathrm{Ric}_g \ge (n-1)K>0$7 with finite $\mathrm{Ric}_g \ge (n-1)K>0$8, even allowing unbounded degree: $\mathrm{Ric}_g \ge (n-1)K>0$9 Comparison to manifold case shows these bounds interpolate between discrete and smooth geometry, with nodal rigidity questions paralleling the classical setting.

2.3. Recent Innovations: Long-Scale and Entropic Approaches

Diametric bounds have been refined using long-scale Ollivier–Ricci curvature, allowing estimates across vertices at graph distances $M$0. For distance-regular or amply regular graphs, such long-scale arguments yield bounds strictly improving previous results, sometimes achieving sharpness in exceptional graphs (e.g., Wells or Golay coset graphs) (Chen et al., 2024).

Entropic Ricci curvature, defined via convexity of discrete entropy or localized Bochner inequalities, also leads to a Bonnet–Myers diameter bound, which is optimal under the arithmetic mean (true CD$M$1), but suboptimal with the default logarithmic mean; only hypercubes attain equality in the sharp bound (Kamtue, 2020).

3. Rigidity and Sharpness

Sharpness of the Bonnet–Myers diameter upper bound is both geometric and combinatorial:

• In the Riemannian context, equality requires constant Ricci curvature and the round sphere model (Antonelli et al., 2024).
• Discrete rigidity theorems assert that, in the regular self-centered case, only certain symmetric graphs are sharp (hypercubes, Johnson, demi-cubes, Gosset), and recent work extends classification to non-regular cases at small diameter and to layered graphs called “symmetrical antitrees” (Kamtue, 2020, Cushing et al., 2024).
• In directed graphs with positive Lin–Lu–Yau Ricci curvature, a Cheng-type theorem characterizes those achieving the diameter bound as spherically suspended graphs, with sharpness uniquely realized by specific direct products and suspensions (Ozawa et al., 2020).

4. Extensions: Sub-Riemannian, Lorentzian, and Generalized Structures

Sub-Riemannian Geometry

Bonnet–Myers type theorems apply to Carnot–Carathéodory geometries (3-Sasakian, quaternionic contact), provided suitable Ricci-type bounds on canonical subbundles are satisfied. The universal diameter bounds take the form $M$2, with sharpness realized in quaternionic Hopf fibrations and canonical structures (Rizzi et al., 2015, Barilari et al., 2017).

Lorentzian Length Spaces

A Bonnet–Myers style diameter bound holds for globally hyperbolic Lorentzian length spaces $M$3 with global timelike curvature bounded below by $M$4: $M$5 attained precisely in anti-de Sitter models. The result hinges on a synthetic (Alexandrov-type) hinge and triangle comparison in Lorentzian settings and a uniform control on timelike triangle degeneracy (Beran et al., 2023).

Kähler and Quaternionic Kähler Manifolds

Improved diameter bounds are attainable under lower bounds on Bakry–Émery type orthogonal (holomorphic or quaternionic) Ricci curvature:

• Gradient and non-gradient versions of the orthogonal Bakry–Émery tensor lead to stronger diameter estimates compared to the Riemannian case.
• The diffusion-type result yields:

$M$6

for Kähler manifolds with appropriate orthogonal curvature and holomorphic sectional curvature lower bounds, matched only by $M$7 with Fubini–Study metric (Gordina et al., 2021).

Beckner–Sobolev inequalities adapted to Kähler geometry allow further strict improvements in the diameter upper bound, taking advantage of complex differential identities (Baudoin et al., 2019).

5. Extrinsic Diameter and Further Generalizations

For hypersurfaces isometrically embedded in Euclidean space, the extrinsic Bonnet–Myers theorem asserts that the Euclidean diameter is strictly less than $M$8 when Ricci curvature is bounded below: $M$9 Sharpness is achieved by a family of “needle-shaped” spheres, whose intrinsic metric degenerates to a segment. Almost rigidity theorems quantify Gromov-Hausdorff closeness to the interval as the extrinsic diameter approaches this upper bound (Li et al., 2024).

Spectral generalizations show that the spectral condition $\pi_1(M)$0 is enough to deduce Bonnet–Myers diameter, volume, and fundamental group finiteness, with rigidity only at the round sphere (Antonelli et al., 2024).

6. Applications, Open Problems, and Outlook

Bonnet–Myers diameter bounds serve as a control mechanism for global geometry across smooth and discrete spaces, impacting compactness, symmetry classification, and spectral geometry. Rigidity at equality ties the global structure to constant-curvature or highly symmetric models. Key open directions include:

• Complete classification of Bonnet–Myers sharp irregular graphs beyond small diameter (Li et al., 2024, Cushing et al., 2024).
• Sharpness and uniqueness of bounds under weaker or “non-standard” curvature lower bounds, including along average or decay profiles (Li et al., 2 Sep 2025).
• Generalizations to singular metric spaces, non-symmetric Laplacians, and diffusion generators.

The unifying theme is that positive lower curvature—Ricci, Bakry–Émery, or synthetic—forces geometric and spectral rigidity, compactness, and explicit quantitative control on distance. Ongoing research actively refines these bounds and uncovers new rigidity phenomena in a range of geometric and combinatorial contexts (Li et al., 2024, Chen et al., 2024, Antonelli et al., 2024, Li et al., 2 Sep 2025).