TheoremComplete

Bonnet-Myers Theorem

The Bonnet-Myers theorem shows that a complete Riemannian manifold with positive Ricci curvature lower bound must be compact with bounded diameter. It is one of the fundamental results connecting curvature to global topology.


Statement

Theorem8.6Bonnet-Myers theorem

Let (Mn,g)(M^n, g) be a complete Riemannian manifold with Ricci curvature satisfying Ric≥(n−1)κg\mathrm{Ric} \geq (n-1)\kappa g for some constant κ>0\kappa > 0. Then:

  1. The diameter of MM is bounded: diam(M)≤πκ\mathrm{diam}(M) \leq \frac{\pi}{\sqrt{\kappa}}.
  2. MM is compact.
  3. The fundamental group π1(M)\pi_1(M) is finite.

Proof

Proof

Step 1: Diameter bound. Let p,q∈Mp, q \in M with d(p,q)=Ld(p,q) = L. By Hopf-Rinow (completeness), there exists a minimizing unit-speed geodesic γ:[0,L]→M\gamma: [0, L] \to M from pp to qq.

Choose parallel orthonormal fields E2(t),…,En(t)E_2(t), \ldots, E_n(t) along γ\gamma perpendicular to γ′(t)\gamma'(t). Consider the variation vector fields Vi(t)=sin⁡(πtL)Ei(t)V_i(t) = \sin\left(\frac{\pi t}{L}\right) E_i(t) for i=2,…,ni = 2, \ldots, n. Since γ\gamma is minimizing, the second variation of energy is non-negative:

d2Eds2∣s=0≥0for any fixed-endpoint variation.\frac{d^2 E}{ds^2}\Big|_{s=0} \geq 0 \quad \text{for any fixed-endpoint variation.}

Computing the second variation for ViV_i:

I(Vi,Vi)=∫0L(∣∇γ′Vi∣2−g(R(Vi,γ′)γ′,Vi))dt.I(V_i, V_i) = \int_0^L \left(|\nabla_{\gamma'} V_i|^2 - g(R(V_i, \gamma')\gamma', V_i)\right) dt.

We have ∇γ′Vi=πLcos⁡πtLEi\nabla_{\gamma'} V_i = \frac{\pi}{L}\cos\frac{\pi t}{L} E_i (since EiE_i is parallel), so ∣∇γ′Vi∣2=π2L2cos⁡2πtL|\nabla_{\gamma'} V_i|^2 = \frac{\pi^2}{L^2}\cos^2\frac{\pi t}{L}.

Summing over i=2,…,ni = 2, \ldots, n:

∑i=2nI(Vi,Vi)=∫0Lsin⁡2πtL((n−1)π2L2cos⁡2πtLsin⁡2πtL−Ric(γ′,γ′))dt.\sum_{i=2}^n I(V_i, V_i) = \int_0^L \sin^2\frac{\pi t}{L}\left(\frac{(n-1)\pi^2}{L^2}\frac{\cos^2\frac{\pi t}{L}}{\sin^2\frac{\pi t}{L}} - \mathrm{Ric}(\gamma', \gamma')\right) dt.

Wait -- more carefully: ∑iI(Vi,Vi)=∫0L((n−1)π2L2cos⁡2πtL−sin⁡2πtLRic(γ′,γ′))dt\sum_i I(V_i, V_i) = \int_0^L \left(\frac{(n-1)\pi^2}{L^2}\cos^2\frac{\pi t}{L} - \sin^2\frac{\pi t}{L}\mathrm{Ric}(\gamma', \gamma')\right) dt.

Using Ric(γ′,γ′)≥(n−1)κ\mathrm{Ric}(\gamma', \gamma') \geq (n-1)\kappa:

∑iI(Vi,Vi)≤(n−1)∫0L(π2L2cos⁡2πtL−κsin⁡2πtL)dt=(n−1)2(π2L−κL).\sum_i I(V_i, V_i) \leq (n-1)\int_0^L \left(\frac{\pi^2}{L^2}\cos^2\frac{\pi t}{L} - \kappa\sin^2\frac{\pi t}{L}\right) dt = \frac{(n-1)}{2}\left(\frac{\pi^2}{L} - \kappa L\right).

If L>π/κL > \pi/\sqrt{\kappa}, then ∑iI(Vi,Vi)<0\sum_i I(V_i, V_i) < 0, so at least one I(Vi,Vi)<0I(V_i, V_i) < 0, contradicting the minimality of γ\gamma. Hence L≤π/κL \leq \pi/\sqrt{\kappa}.

Step 2: Compactness. The diameter bound implies M=Bˉπ/κ(p)M = \bar{B}_{\pi/\sqrt{\kappa}}(p) is a closed bounded set. By Hopf-Rinow (completeness implies Heine-Borel), this is compact.

Step 3: Finite fundamental group. The universal cover M~\tilde{M} inherits the Riemannian metric with the same Ricci bound. By Steps 1-2, M~\tilde{M} is compact. The fundamental group π1(M)\pi_1(M) acts on M~\tilde{M} by deck transformations (isometries). Since M~\tilde{M} is compact, the fibers of M~→M\tilde{M} \to M are finite, so π1(M)\pi_1(M) is finite. ■\blacksquare

■

Applications

ExampleDiameter bounds
  1. SnS^n with K=1K = 1: Ric=(n−1)g\mathrm{Ric} = (n-1)g, so κ=1\kappa = 1 and diam≤π\mathrm{diam} \leq \pi. Equality holds.
  2. Sn(r)S^n(r) with K=1/r2K = 1/r^2: Ric=n−1r2g\mathrm{Ric} = \frac{n-1}{r^2}g, so diam≤πr\mathrm{diam} \leq \pi r. Again equality holds.
  3. Any Einstein manifold with Ric=λg\mathrm{Ric} = \lambda g, λ>0\lambda > 0, has diam≤πn−1λ\mathrm{diam} \leq \pi\sqrt{\frac{n-1}{\lambda}} and is compact.
RemarkRigidity: Cheng's theorem

Cheng's maximal diameter theorem states that if Ric≥(n−1)κ>0\mathrm{Ric} \geq (n-1)\kappa > 0 and diam(M)=π/κ\mathrm{diam}(M) = \pi/\sqrt{\kappa} (equality), then MM is isometric to the sphere Sn(1/κ)S^n(1/\sqrt{\kappa}). This is a rigidity result: the sphere is the only manifold achieving the maximal diameter.