Bonnet–Myers Type Theorems

• Bonnet–Myers type theorems are results that guarantee compactness and provide diameter bounds for spaces with a positive lower curvature bound.
• They extend classical Riemannian methods using index form computations, Laplacian and mean curvature comparisons, and are adapted to settings such as Bakry–Émery, sub-Riemannian, and discrete geometries.
• These theorems have practical implications in diverse areas including weighted manifolds, foliated spaces, Kähler and Lorentzian geometries, and even non-smooth and graph structures.

A Bonnet–Myers type theorem is a compactness and diameter estimate for a geometric or analytic structure under suitable lower curvature bounds or their analogues. Classically, Bonnet–Myers establishes that any complete Riemannian manifold with Ricci curvature bounded below by a positive multiple of the metric is compact and has finite fundamental group, with an explicit upper bound on the diameter. The landscape of Bonnet–Myers type theorems now encompasses a diverse array of frameworks—weighted manifolds, sub-Riemannian and foliated geometries, discrete structures such as graphs, Kähler and quaternionic geometries, as well as analytic generalizations involving spectral and integral curvature conditions. This entry surveys the principal formulations, structural strategies, key extensions, and methodological paradigms of Bonnet–Myers type theorems across these settings.

1. Classical Bonnet–Myers Theorem: Core Statement, Method, and Optimality

The foundational statement for a complete $n$-dimensional Riemannian manifold $(M, g)$ is: $\Ric_g \ge (n-1)K>0 \implies \mathrm{diam}(M) \le \frac{\pi}{\sqrt{K}}$ with compactness and finiteness of the fundamental group; equality holds if and only if $(M, g)$ is isometric to the round sphere of radius $1/\sqrt{K}$.

The key proof techniques are (a) index form computations along minimizing geodesics, exploiting the relation between the second variation and presence of conjugate points, and (b) Laplacian comparison for the distance function using the Bochner–Weitzenböck and matrix Riccati equations. Extensions and refinements, including those allowing a limited amount of negative Ricci curvature or incorporating potential terms, exhibit fundamental tension between pointwise curvature hypotheses, oscillation theory of Jacobi equations, and the control of volume growth (Wu, 2017).

2. Bakry–Émery Ricci and Weighted Variants

The Bakry–Émery Ricci tensor generalizes Ricci to the weighted or drift setting. For a smooth vector field $X$ or function $f$, these are

$\Ric_X = \Ric + \tfrac12 \mathcal{L}_X g, \qquad \Ric_f = \Ric + \Hess f$

Bonnet–Myers type results hold if

$\Ric_X \ge (n-1)K g, \qquad |X| \le A,$

with resulting sharp diameter bound (Wu, 2017): $$\mathrm{diam}(M) \le \frac{2A}{(n-1)K} + \frac{\pi}{\sqrt{K}}$$ For $(M, g)$0, we recover the classical bound. The general methodology replaces the classical second variation with mean curvature comparison applied to the excess function, relying on weighted Laplacian analysis and comparison with the model sphere. This approach achieves bounds sharper than preceding ones, and directly connects to work using the index form and second variation methods, such as those by Limoncu and Tadano.

Further, polynomial and exponential-type decay assumptions on $(M, g)$1 or $(M, g)$2 suffice for compactness and diameter bounds (Qiu, 2019, Li et al., 2 Sep 2025). Under appropriate growth of the weight function (e.g., bounded or of controlled polynomial growth), explicit compactness results follow.

3. Generalized Geometric Settings: Kähler, Quaternionic, and Finsler Structures

In Kähler and quaternionic Kähler geometry, the "orthogonal Bakry–Émery Ricci tensor" provides a more refined localization of curvature, splitting the Ricci tensor and including extension via a potential or drift (Gordina et al., 2021). For a Kähler manifold $(M, g)$3 with holomorphic sectional curvature $(M, g)$4, the orthogonal Ricci is

$(M, g)$5

and the Bakry–Émery extension is

$(M, g)$6

Diameter bounds, in both gradient and non-gradient Bakry–Émery cases, are strictly sharper than the classical case. For instance: $(M, g)$7 For Fubini–Study geometry, these are sharp and attain half the classical Riemannian bound.

In Finsler geometry, Bonnet–Myers type theorems operate via the index form along geodesics and averaged Ricci curvature conditions. Integral Ricci assumptions and upper Ricci bounds plus average control yield compactness; pointwise lower bounds recover familiar diameter bounds (Anastasiei, 2014).

4. Metric Measure, Synthetic, and Non-Smooth Theories

Synthetic and analytic extensions circumvent pointwise curvature in favor of integral, spectral, or Kato-type smallness, extending applicability to settings with negative Ricci curvature on large sets or decaying curvature at infinity:

• Integral criteria: Compactness or diameter bounds are deduced when an integral of the form $(M, g)$8 is satisfied, where $(M, g)$9 computes the minimal normalized Ricci along rays (Li et al., 2 Sep 2025).
• Kato-type conditions: If the Kato constant for the negative part of the Ricci curvature at level $\Ric_g \ge (n-1)K>0 \implies \mathrm{diam}(M) \le \frac{\pi}{\sqrt{K}}$0 is strictly less than 1, then the manifold is compact, and explicit volume and diameter estimates are available (Rose, 2019).
• Spectral criteria: If the lowest eigenvalue $\Ric_g \ge (n-1)K>0 \implies \mathrm{diam}(M) \le \frac{\pi}{\sqrt{K}}$1, then the manifold is compact and $\Ric_g \ge (n-1)K>0 \implies \mathrm{diam}(M) \le \frac{\pi}{\sqrt{K}}$2; equality is achieved only by the sphere (Antonelli et al., 2024). These spectral formulations are sharp and control finiteness of the fundamental group.

5. Sub-Riemannian, Foliated, and Non-Riemannian Structures

Bonnet–Myers type compactness theorems extend to sub-Riemannian manifolds, sub-Laplacian contexts, and spaces with horizontal/vertical structures.

• Sub-Riemannian Bonnet–Myers: For bracket-generating H-type (or Carnot-type) manifolds, if the horizontal Ricci tensor minus torsion corrections satisfies a lower bound $\Ric_g \ge (n-1)K>0 \implies \mathrm{diam}(M) \le \frac{\pi}{\sqrt{K}}$3, then compactness and diameter bounds of the form $\Ric_g \ge (n-1)K>0 \implies \mathrm{diam}(M) \le \frac{\pi}{\sqrt{K}}$4 hold (Baudoin et al., 2019, Barilari et al., 2017, Barilari et al., 2014).
• Foliations with minimal leaves: For Riemannian foliations with bracket-generating horizontal bundle $\Ric_g \ge (n-1)K>0 \implies \mathrm{diam}(M) \le \frac{\pi}{\sqrt{K}}$5, positive lower bounds on the "horizontal Ricci-type" tensor $\Ric_g \ge (n-1)K>0 \implies \mathrm{diam}(M) \le \frac{\pi}{\sqrt{K}}$6 yield

$\Ric_g \ge (n-1)K>0 \implies \mathrm{diam}(M) \le \frac{\pi}{\sqrt{K}}$7

where $\Ric_g \ge (n-1)K>0 \implies \mathrm{diam}(M) \le \frac{\pi}{\sqrt{K}}$8 (Baudoin, 16 Sep 2025).

• Sub-Laplacian comparison and generalizations to quaternionic and Sasakian structures produce sharp and unified diameter estimates directly generalizing Riemannian results.

6. Discrete, Graph, and Noncommutative Variants

Bonnet–Myers type theorems for finite and infinite graphs use analogues of Ricci curvature—Ollivier, Lin–Lu–Yau (LLY), or Bakry–Émery. Key outcomes include:

• For a connected D-regular graph with Ollivier Ricci curvature $\Ric_g \ge (n-1)K>0 \implies \mathrm{diam}(M) \le \frac{\pi}{\sqrt{K}}$9 on all edges, $(M, g)$0 (Kamtue, 2020, Li et al., 2024).
• Bonnet–Myers sharpness (equality cases) drive structural classifications: e.g., regular graphs achieving the bound at diameter 3 are classified; hypo/hypercube and Johnson graphs are known examples.
• Matching-cost formulas relate edge curvature to optimal transport costs between blown-up neighbor sets, providing powerful combinatorial criteria for sharpness, irregular cases, and forbidden subgraph constraints (Li et al., 2024).
• Bakry–Émery and entropic Ricci discrete curvature yield gradient bounds on Markov semigroups and corresponding sharp diameter estimates. For arithmetic mean (Bakry–Émery) curvature, equality is realized by Hamming cubes and sharpness is attained (Kamtue, 2020).

7. Extensions: Extrinsic, Lorentzian, and Noncommutative Geometries

Broader generalizations include:

• Extrinsic Bonnet–Myers: For a compact Riemannian manifold isometrically immersed in Euclidean space with Ricci lower bound $(M, g)$1, there is an upper bound on the extrinsic diameter $(M, g)$2, strictly less than the intrinsic (geodesic) diameter (Li et al., 2024).
• Lorentzian pre-length spaces: In the synthetic setting of globally hyperbolic Lorentzian spaces with global lower timelike curvature bound $(M, g)$3, finite time-separation diameter is bounded by $(M, g)$4 (Beran et al., 2023).
• Spectral gap and compactness: Using spectral gap lower estimates for the Laplace–Beltrami operator, mean Ricci bounds, and controlled negative part of Ricci, generalized Bonnet–Myers compactness and volume bounds are established for both weighted and unweighted manifolds (Bonnefont et al., 2021).

---

Summary Table: Core Bonnet–Myers Type Theorems

Context	Curvature Hypothesis	Main Conclusion	Reference
Riemannian	$(M, g)$5	$(M, g)$6	(Wu, 2017)
Bakry–Émery	$(M, g)$7, $(M, g)$8	$(M, g)$9	(Wu, 2017)
Integral Ricci (Ray)	$1/\sqrt{K}$0	$1/\sqrt{K}$1 is compact	(Li et al., 2 Sep 2025)
Sub-Riemannian	Horizontal Ricci–torsion bound (*)	$1/\sqrt{K}$2	(Baudoin et al., 2019)
Kähler/Quaternionic	Orthogonal Bakry–Émery Ricci, $1/\sqrt{K}$3	Sharper diameter bounds, e.g. $1/\sqrt{K}$4	(Gordina et al., 2021)
Discrete Graph	$1/\sqrt{K}$5	$1/\sqrt{K}$6	(Kamtue, 2020)
Spectral	$1/\sqrt{K}$7	$1/\sqrt{K}$8	(Antonelli et al., 2024)

The synthesis of geometric, analytic, and combinatorial methods in Bonnet–Myers type theorems unifies broad compactness, finiteness, and rigidity phenomena across several geometric frameworks. The invariants used, and the resulting diameter/compactness criteria, are now recognized as central structure theorems within metric, sub-Riemannian, discrete, and even Lorentzian geometry, providing sharp thresholds and illuminative model examples for the appearance of compactness and upper diameter control.