2
$\begingroup$

In most places, a Riemannian manifold is defined to be a locally euclidean surface (with other restrictions). When thinking of 2D-surfaces, this raises two questions:

  1. Does locally euclidean mean locally flat (angles of triangle sum to 180 degrees), or just that it can be represented by a 2D coordinate system? Taking the example of a sphere without poles, both the sphere without poles and a tiny patch on it have a coordinate representation in R2. But only the small patch is (almost) flat. So which of the two properties are really locally euclidean? Or does one is a consequence of the other?

  2. In 2D, a regular surface with a metric can be considered a Reimannian manifold. So what property of a regular surface makes it locally euclidean?

  • 0
    Have you actually seen general $2$-manifolds defined as "locally Euclidean", or just "locally modeled on $\mathbf{R}^{2}$"? – 2017-01-21
  • 1
    Yes. For instance this talks about a general n-manifold (http://mathworld.wolfram.com/Manifold.html). – 2017-01-21
  • 1
    General manifolds do not have a notion of curvature and angles. You have to add that separately to make a Riemannian manifold. The "locally euclidean" part of general manifolds is purely topological. – 2017-01-21
  • 0
    @Arthur: Does the question read correctly now? – 2017-01-21

1 Answers 1

3

  1. $\newcommand{\Reals}{\mathbf{R}}$"Just that the manifold can be represented (locally) by an $n$-dimensional coordinate system."

    In the context of the Mathworld definition of smooth manifold, "locally Euclidean" might better be phrased "locally Cartesian": locally, the space is modeled on the smooth structure of $\Reals^{n}$.

    As Arthur says, a smooth manifold comes with no natural definitions of length, angle, or curvature.

    If a differential geometer speaks of a locally-Euclidean manifold, by contrast, it may well mean a flat Riemannian manifold, in which small geodesic triangles have total interior angle $\pi$. It's therefore potentially ambiguous to speak of "locally-Euclidean" in the context of Riemannian geometry.

    Incidentally, a Riemannian manifold is "pointwise Euclidean" in the sense that each tangent space comes equipped with a positive-definite inner product. Flatness, however, is a differential condition on the metric components.

  2. If $U \subset \Reals^{n}$ is a non-empty open set, a regular mapping $\Phi:U \to \Reals^{N}$ is an immersion, and each point $x$ in $U$ has a neighborhood $V$ such that

    • $\Phi(V) \subset \Reals^{N}$ is a smooth $n$-manifold, and

    • $\Phi:V \to \Phi(V)$ is a diffeomorphism.

    The flat (Euclidean) metric on $\Reals^{N}$ pulls back to a Riemannian metric on $U$, but generally this metric is not flat.