Distance-preserving maps are compositions of rotations and translations

  • Thread starter Fredrik
  • Start date
  • Tags
    Rotations
In summary, in order to prove that a bijection in ##\mathbb R^n## that preserves distances can be composed of a rotation and a translation, one can define a map ##g## that is the composition of the original bijection ##f## and a translation ##f(0)##, and then prove that ##g## is a rotation using the Mazur-Ulam theorem. Another approach is to use a structure theorem and show that all orthogonal matrices have the structure of a block matrix made up of sines, cosines, and possibly reflections. Additionally, one can use the fact that the metric is generated by an inner-product to show that isometries must preserve the inner-product, but this does not necessarily mean that they
  • #1
Fredrik
Staff Emeritus
Science Advisor
Gold Member
10,877
423
How do you prove that a bijection ##f:\mathbb R^n\to\mathbb R^n## such that ##d(f(x),f(y))=d(x,y)## for all ##x,y\in\mathbb R^n## is a composition of a rotation and a translation? (d=Euclidean metric).

My first thought is to define ##g=f-f(0)##, prove that g is a bijection that preserves distances too, and then try to prove that g is a rotation. I recently learned how to prove that maps that take straight lines to straight lines are linear, so I'm guessing that the strategy involves proving that g takes straight lines to straight lines.

Maybe we can use that g preserves circles around 0 in the sense that if C is such a circle and ##x\in C##, then ##g(x)\in C##. I imagine that we would take an arbitrary line L, and then look at points where it intersects some circle around 0, but I don't see how to turn this idea into a proof. I'm not sure if I should try to prove that g takes straight lines to straight lines first, or if there's a more direct approach.
 
Physics news on Phys.org
  • #2
You may want to start by actually defining what a rotation is. If you define rotations as elements of [itex]O_n[/itex], the orthogonal group, then what you say is known as the Mazur-Ulam theorem. See http://en.wikipedia.org/wiki/Mazur-Ulam_theorem The proof of the Mazur-Ulam theorem is not very difficult and can be found on a pdf in the wiki page.
 
  • #3
Excellent. That's exactly what I wanted. I thought it would be much harder. Thanks.
 
  • #4
the usual method is to first determine all motions preserving a given point, show these are all in your group, and then show that also your motions are transitive. I.e. since translations are transitive, it suffices to show that all isometries preserving the origin are rotations, but of course they are not, since you have included only orientation preserving motions as rotations and translations, so you need also reflections. Of course you can define rotations as all isometries preserving the origin as micromass seems to suggest, but that is not the usual definition. Rather I think they are usually defined as the component SO (special orthogonal group) of the identity, in the full orthogonal group.

perhaps more revealing than simply defining the difficulty away, is to use a structure theorem such as that in herstein, that shows every orthogonal matrix has in some orthogonal basis the structure of a block matrix made up of sines and cosines, i.e. actual plane rotations, and a bunch of -1's, i.e. some 180 rotations and possibly one reflection. the theorem micromass refers to, is something i have been assuming as obvious, i.e. distance preserving maps preserving the origin are linear, essentially by the parallelogram law. maybe i should be more rigorous on that score. I.e. there are two steps 1) distance preserving maps are affine, 2) orientation preserving linear maps are composed of plane rotations in orthogonal planes.
 
Last edited:
  • #5
I think you can use the fact that the metric is generated by an inner-product; then isometries must preserve the inner-product. you can then show that the only maps that preserve the real inner-product are rotations and translations. Of course this would not work on spaces where the inner-product does not generate the metric.
 
  • #6
Bacle2 said:
I think you can use the fact that the metric is generated by an inner-product; then isometries must preserve the inner-product. you can then show that the only maps that preserve the real inner-product are rotations and translations.

It is true that a map that preserves an inner product must preserve the norm induced by that product, and so the metric induced by the norm. But the converse is not the case. For example, consider [itex]f : \mathbb{R}^n \to\mathbb{R}^n : x \mapsto x + a[/itex].

Then [itex]f(x) \cdot f(y) = x \cdot y + x \cdot a + y \cdot a + a \cdot a[/itex] which is not identically equal to [itex]x \cdot y[/itex] unless [itex]a = 0[/itex]. But [itex]\|f(x) - f(y)\| = \|x + a - y - a\| = \|x - y\|[/itex].

Also: Let [itex]e_i[/itex], [itex]1 \leq i \leq n[/itex], be the standard basis on [itex]\mathbb{R}^n[/itex]. For [itex]\sigma \in S_n[/itex], define a linear map [itex]A_\sigma : \mathbb{R}^n \to \mathbb{R}^n[/itex] by [itex]e_i \mapsto e_{\sigma(i)}[/itex]. Then [itex]A_\sigma[/itex] preserves inner products since
[tex]e_i \cdot e_j = \delta_{ij} = \delta_{\sigma(i)\sigma(j)} = e_{\sigma(i)} \cdot e_{\sigma(j)} = A_\sigma e_i \cdot A_\sigma e_j[/tex]
because [itex]i = j[/itex] if and only if [itex]\sigma(i) = \sigma(j)[/itex].

But there are maps [itex]A_\sigma[/itex] which do not correspond to rotations: swapping [itex]e_1[/itex] and [itex]e_2[/itex], for example, which is a reflection in the surface [itex]x_1 = x_2[/itex].
 

FAQ: Distance-preserving maps are compositions of rotations and translations

1. What are distance-preserving maps?

Distance-preserving maps, also known as isometries, are functions that preserve the distances between points in a given space. This means that the distance between any two points on the map will be the same as the distance between those two points in the original space.

2. What is the significance of distance-preserving maps?

Distance-preserving maps are important in mathematics and physics because they help us understand and analyze geometric structures and objects. These maps also have practical applications in fields such as computer graphics, where they are used to create 3D models and animations.

3. How are distance-preserving maps related to rotations and translations?

It has been proven that any distance-preserving map can be expressed as a composition of rotations and translations. This means that any isometry can be broken down into a sequence of rotations and translations, making them easier to understand and analyze.

4. Can distance-preserving maps be applied to any space?

Yes, distance-preserving maps can be applied to any metric space, which is a mathematical space where the distance between any two points is defined. This includes Euclidean spaces, as well as non-Euclidean spaces such as the surface of a sphere.

5. How are distance-preserving maps different from other types of transformations?

Distance-preserving maps are different from other types of transformations because they preserve the distances between points, while other transformations may change the shape or size of an object. Isometries are also different from rigid transformations, which only preserve the shape of an object but not necessarily its size or orientation.

Back
Top