0
$\begingroup$

any isometry of Euclidean vector space $\mathbb{R}^3$ has the form

$$\mathbb{R}^3\ni x\mapsto A\cdot x + b\in\mathbb{R}^3,$$

where $A\in O(3)$ is an orthonormal matrix, and $b\in\mathbb{R}^3$. However, geometrically there are only a few types of isometries, namely: translations (iff $A$ is the unit matrix), rotations, reflections, and compositions of those three types of geometric maps.

I am wondering, how to classify those geometric mappings in terms of $A$ and $b$? Are there any properties of $A$ and $b$, such that one can classify the isometries of Euclidean space? I looked in many books, but I could not find anything satisfactory.

I would appreciate your help!

|cite|improve this question
$\endgroup$
  • $\begingroup$ A translation corresponds to $A=I$, the identity matrix, and an arbitrary translation vector $b$, etc. $\endgroup$ – Dietrich Burde Sep 7 at 14:52
  • $\begingroup$ Thank you! The case of translations is not the problem here. I am interested in the "etc." Do you know a reference, where the other isometries are classified similarly to the translations (A=I) $\endgroup$ – Oliver Watt Sep 7 at 14:54
  • $\begingroup$ For rotation around the origin, we have $b=0$ and and an arbitrary $A\in SO_3(\Bbb R)$, see 3D rotation group. In general, we have the group $O(3)\ltimes \Bbb R^3$, so the semidirect product, see the wikipedia article for the corresponding geometric types. $\endgroup$ – Dietrich Burde Sep 7 at 14:57
  • $\begingroup$ Why is that the case? Cant be a map with $b\neq 0$ become a rotation? $\endgroup$ – Oliver Watt Sep 7 at 14:59
  • $\begingroup$ @OliverWatt No. Not around the origin. If $T$ is a rotation around $b$, however, then $Tx = A(x-b)+b = Ax + (b-Ab)$, where $A$ is a rotation around the origin. $\endgroup$ – amsmath Sep 7 at 16:03
0
$\begingroup$

A reflection is an inverse isometry, hence $\det A=-1$, whereas $\det A=1$ for a rotation.

In addition, a rotation has an invariant line, while a reflection has an invariant plane. It follows that, for a rotation, $A$ has only one unit eigenvalue, while for a reflection $A$ has two unit eigenvalues. The corresponding eigenvectors span the rotation axis or the reflection plane.

|cite|improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.