Tema 8: Boundary at infinity of Hadamard manifolds
In this topic, we provide some (equivalent) ways to compactify a Hadamard manifold M.
Asymptotic rays
A ray is a geodesic γ:[0,∞)→M.
Asymptotic rays: Two rays γ1,γ2:[0,∞)→M are asymptotic if the function t∈[0,∞)↦d(γ1(t),γ2(t)) is bounded. In this case, we write γ1∼γ2.
It is easy to check that ∼ is an equivalence relation.
Boundary at infinity M(∞): The boundary at infinity of M is the set of equivalence relations of ∼, M(∞)={[γ]:γ is a ray}.
Every p∈M defines a map fp:SMp→M(∞) by fp(v)=[γv].
Lemma 1. The map fp is a bijection.
Proof. We start proving injectivity. If fp(v)=fp(w) then γv∼γw and so d(t)=d(γv(t),γw(t)) is uniformly bounded for t≥0. By convexity, d(t)=const, and so d(t)=d(0)=0. This implies that v=w. Now we prove that fp is surjective. Fix a ray γ. For each n, let γvn= ray starting at p and passing through γ(n), say γvn(tn)=γ(n). Passing to a subsequence, we can assume that vn→v. We claim that γv∼γ, with d(γv(t),γ(t))≤d(p,γ(0)) for all t≥0. To see that, start observing that d(γv(t),γ(t))=limn→∞d(γvn(t),γ(t)). By convexity, the function d(γvn(t),γ(t)) is non-increasing in [0,tn]. \footnote{The function fn(t)=d(γvn(t),γ(t)) satisfies f″n(t)≥0 and f′n(tn)=0. If f′n(t∗)>0 for some 0≤t∗≤tn, then f′n(t)>0 for all t≥t∗, a contradiction.} Hence, for tn>t we have d(γvn(t),γ(t))≤d(p,γ(0)).
Lemma 2. For every p,q∈M, the composition f−1q∘fp:SMp→SMq is an homeomorphism.
Sphere topology on M(∞): The sphere topology on M(∞) is the topology that makes any fp an homeomorphism.
By Lemma 2, the definition does not depend on p. In particular, M(∞) is homeomorphic to a sphere of Rn. There is another way to understand this topology, that actually characterizes the topology of the union ¯M=M∪M(∞). For p∈M, let B1(p)⊂TMp be the closed unit ball centered at the origin. Define the map φ=φp:B1(p)→¯M by φ(v)={expp(‖v‖1−‖v‖v), if ‖v‖<1,fp(v), if ‖v‖=1.
Theorem 1. (Eberlein-O'Neil) The map φ is an homeomorphism.
Hence ¯M is homeomorphic to a closed ball of Rn. The topology of ¯M is called the cone topology.
Discussion