0

I am trying to show that every isometry of $(\mathbb{R}^n, g)$, regarded as a Riemannian manifold with the canonical metric $g = \delta_{ij}dx^idx^j$, is a composition of a rotation and a translation.

Here the definition of isometry is as follows:

Let $(M, g)$ and $(N, h)$ be Riemannian manifolds. An isometry from $(M, g)$ to $(N, h)$ is a diffeomorphism $\phi: M \rightarrow N$ such that $\phi^*h = g$.

My attempt:

Let $\phi: \mathbb{R}^n \rightarrow \mathbb{R}^n$ be an isometry. Write $y = \phi(x) = (\phi^1(x), ..., \phi^n(x))$. Let $h = \delta_{kl}dy^kdy^l$. Then $$\phi^*h = (\delta_{kl}\circ\phi)\phi^*(dy^k)\otimes \phi^*(dy^l)=(\delta_{kl}\circ\phi)\frac{\partial\phi^k}{\partial x^i}\frac{\partial\phi^l}{\partial x^j}dx^idx^j = \frac{\partial\phi^k}{\partial x^i}\frac{\partial\phi^k}{\partial x^j}dx^idx^j.$$

The idea is then to equate this with $g = \delta_{ij}dx^idx^j$, which will give me quadratic equations in $\frac{\partial\phi^k}{\partial x^i}, i, k = 1, ..., n.$ After that I don't know how to proceed. It would be nice if I know that $\phi$ must be affine, but I don't know how to show this either.

I just began learning the Riemannian metric, so some of the above may not make sense. Any help is greatly appreciated.

Joseph
  • 560
  • In $\mathbb{R}^2$ can one express the negation map $v \mapsto -v$ as a composition of a rotation and a translation? – coffeemath Feb 15 '21 at 18:57
  • 2
    Up to translation, you may assume that the isometry maps the origin to itself. Then try to show that it is actually a linear map, hence given by an orthogonal matrix. Finally use a standard result in linear algebra to conclude. – WhatsUp Feb 15 '21 at 18:57
  • 1
    @coffeemath It is a rotation of $\pi$ around the origin. Maybe $\Bbb R^3$ is the example you want to mention, as there is orientation problem. – WhatsUp Feb 15 '21 at 18:58
  • @WhatsUp Thank you! Mimicking the proof here: https://math.stackexchange.com/a/3972686/471269, I can show that $\phi_*$ is linear, but I am not sure about $\phi$. – Joseph Feb 15 '21 at 19:09
  • Here's the sketch: An isometry preserves the distance between two points. If you compose the isometry with a translation, you can assume that it maps the origin to the origin. Now show it preserves lengths of vectors. There is a formula for the dot product in terms of lengths of vectors. That implies that the isometry preserves the dot product. The last step is proving that the map is in fact linear. You do that by choosing an orthonormal basis and observing that the image of this basis is also an orthonormal basis. Lots of details to work out. – Deane Feb 15 '21 at 19:49
  • You forgot reflections. – Moishe Kohan Feb 16 '21 at 15:12

0 Answers0