Distance from a Finsler Submanifold to its Cut Locus and the Existence of a Tubular Neighborhood

Aritra Bhowmick, Sachchidanand Prasad · 2024

Read Distance from a Finsler Submanifold to its Cut Locus and the Existence of a Tubular Neighborhood by Aritra Bhowmick, Sachchidanand Prasad (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2411-01185.txt.

Description

Abstract . In this article we prove that for a closed, not necessarily compact, submanifold N N of a possibly non-complete Finsler manifold ( M , F ) (M,F) , the cut time map is always positive. As a consequence, we prove the existence of a tubular neighborhood of such a submanifold.

Catalog record

CREATORAritra Bhowmick, Sachchidanand Prasad
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE72 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2411-01185

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Distance from a Finsler Submanifold to its Cut Locus and the Existence of a Tubular Neighborhood Aritra Bhowmick, Sachchidanand Prasad Abstract . In this article we prove that for a closed, not necessarily compact, submanifold N N of a possibly non-complete Finsler manifold ( M , F ) (M,F) , the cut time map is always positive. As a consequence, we prove the existence of a tubular neighborhood of such a submanifold. When N N is compact, it then follows that there exists an ϵ > 0 \epsilon>0 such that the distance between N N and its cut locus Cu ⁡ ( N ) \mathrm{Cu}(N) is at least ϵ \epsilon . This was originally proved by B. Alves and M. A. Javaloyes ( Proc. Amer. Math. Soc. 2019). We have given an alternative, rather geometric proof of the same, which is novel even in the Riemannian setup. We also obtain easier proofs of some results from N. Innami et al. ( Trans. Amer. Math. Soc. , 2019), under weaker hypothesis. 2020 Mathematics Subject Classification Primary: 53C22, 53B40; Secondary: 53C60 1. Introduction In the field of Riemannian geometry, geodesics are curves that are locally distance minimizing. If a connected Riemannian manifold (M,g) is assumed to be complete, then between any two points, say, p,q\in M , there always exists a geodesic which is globally distance minimizing, which we simply say is a minimizer . This leads to the notion of the cut locus of a point p , originally introduced by Henri Poincaré [ Poi05 ] . The cut locus \mathrm{Cu}(p) of a point p is the set consisting of all points q such that there exists a minimizer joining p to q , any extension of which fails to be globally distance minimizing. Cut locus has been extensively studied in the literature [ Kob67 , Buc77 , Wol79 , Sak96 ] . In particular, it is well-known that \mathrm{Cu}(p) is the closure of the collection of those points q for which there are at least two distinct minimizers joining p to q . Now, the classical Whitehead convexity theorem [ WHI32 , CE75 ] states that around every point p\in M , there exists a sufficiently small metric ball, say, B such that for any q\in B there exists a unique minimizer joining p to q . As an immediate consequence, one concludes that p\not\in\mathrm{Cu}(p) . The notion of cut locus readily generalizes to submanifolds N\subset M . Given any point q\in M , an N -segment is a minimizer joining a point in N to q , with length equal to the distance from N to q . The cut locus of N , denoted \mathrm{Cu}(N) , consists of those points q such that there is an N -segment joining N to q , any extension of which fails to be distance minimizing from N . It requires some effort to prove that \mathrm{Cu}(N) is disjoint from N . The standard proof involves looking at the normal exponential map associated to N , which is the usual exponential map restricted to the normal bundle of N . One first proves that it is an immersion near the 0 -section, and then, one shows that it is an embedding in some sufficiently small neighborhood [ Lee18 ] . Such a neighborhood is known as a tubular neighborhood . As a consequence, it follows that N\cap\mathrm{Cu}(N)=\emptyset . In [ O’N83 ] , the author proves this fact for semi -Riemannian manifolds. We also note that in the context of sub-Riemannian geometry, there are examples where p\in\mathrm{Cu}(p) for a point p in a sub-Riemannian manifold [ Mon02 ] . Finsler geometry is a natural generalization of the Riemannian one, which was first studied by P. Finsler in his dissertation [ Fin51 ] . A Finsler metric F on a manifold M is a parametrized collection of Minkowski norms on each tangent space, which, unlike a Riemannian norm, may not be induced by an inner product. In this generality one needs to overcome certain challenges, nevertheless, most of the standard notions from Riemannian geometry find their natural counterparts in the Finsler setting, see [ AP94 , BCS00 , She01a , Oht21 , BP24a ] for a survey of results. In particular, geodesics and the notion of cut locus are readily defined. Furthermore, Whitehead convexity theorem still holds [ BCS00 ] , and one immediately obtains that the cut locus of a point in a Finsler manifold is disjoint from the point itself. One of the primary difficulties in dealing with a submanifold N of a Finsler manifold (M,F) is that, in the absence of an inner product, the natural replacement of a normal bundle of N is no longer a vector bundle; it is only a cone bundle, and it may have singularities at the zero section. Consequently, if one attempts to prove that N\cap\mathrm{Cu}(N)=\emptyset , the proof technique from the Riemannian geometry immediately breaks down, since the normal exponential map is not even differentiable near the zero section. The primary goal of this article is to prove Theorem 3.8 , and as an immediate corollary we obtain the following. Theorem A ( 3.9 ) . Given a closed (not necessarily compact) submanifold N of a Finsler manifold (M,F) , we have N\cap\mathrm{Cu}(N)=\emptyset . In fact, when N is compact, in 3.13 we show that d(N,\mathrm{Cu}(N))>0 . This follows immediately from the next theorem. Note that the notion of a tubular neighborhood of a submanifold can be naturally extended to the Finsler context ( 3.1 ). Theorem B ( Theorem 3.11 ) . Given a closed submanifold N of a Finsler manifold (M,F) , there exists a tubular neighborhood of N . If N is compact, then we can get a tubular neighborhood whose image under the normal exponential map is \left\{x\in M\;\middle|\;d(N,x)<\epsilon\right\} for some \epsilon>0 , where d(N,\_) is the distance function from N . We would like to point out that most of the results in this article, including the ones mentioned above, were originally proved in [ AJ19 ] , albeit with a different approach. See subsection 4.1 for a further discussion. Conventions Given a bundle E over a manifold, \Gamma E denotes the sheaf of sections of E , and in particular, X\in\Gamma E denotes a local section of E over an unspecified open set in M . Boldface letters, e.g., \mathbf{v},\mathbf{n} etc. will always denote tangent vectors. Unless stated explicitly, we have not assumed any completeness assumption on the Finsler metric. As such, whenever we have considered a geodesic, it is tacitly assumed to be defined in the maximal possible domain. All manifold are connected, and without boundary. By a closed submanifold, we shall mean topologically closed, but not necessarily compact, embedded submanifold. Organization of the paper In section 2 , we recall some basic definitions and results from Finsler geometry, while deferring the proof of a technical result ( 2.15 ) to the Appendix A . Then, in section 3 we obtain the main results of this article. Finally, in section 4 we discuss how this article relates to [ AJ19 ] and [ IINS19 ] . 2. Preliminaries on Finsler Geometry This section recalls some necessary

Read the complete text