8

I have to prove that this matrix is a rotation-matrix $$\begin{pmatrix} \frac12 & 0 & \frac{\sqrt{3}}{2} \\ 0 & 1 & 0 \\ \frac{\sqrt{3}}{2} & 0 & \frac12 \end{pmatrix}$$ How do I do this? My idea is to multiplicate it with $\begin{pmatrix} x \\ y \\ z\end{pmatrix}$ and show that one component will remain unchanged . Is this enough? Do non-rotational transformations exist, which leave one component unchanged ?

Christian
  • 1,781
  • Certainly projection onto one component leaves that component unchanged, but (in dimensions greater than $1$) projections onto a component are singular and so are not orthogonal. – Travis Willse Nov 15 '14 at 11:04
  • 2
    How do you define rotation matrix? – Git Gud Nov 15 '14 at 11:16
  • No, simply multiplying by (x, y ,z) and showing that one component does not change is not enough. This property would be true for several non-rotational matrices as well. One examples are scaling (diagonal) matrices for the other two components (i.e. all non main-diagonal entries are zero, one main diagonal entry is one, the other ones are arbitrary). – Dreamer Nov 15 '14 at 17:08
  • 1
    Also, only matrices for the rotation around a coordinate axis leave one vector component unchanged. But matrices for the rotation around an arbitrary axis (e.g. the axis y=x) do not. – Dreamer Nov 15 '14 at 17:11

2 Answers2

24

The following characterization of rotational matrices can be helpful, especially for matrix size $n > 2$.

$M$ is a rotational matrix if and only if $M$ is orthogonal, i.e. $MM^T = M^TM = I$, and $\det(M) = 1$.

Git Gud
  • 31,356
Empiricist
  • 7,933
  • 4
    Actually, if you define rotation as 'rotation about an axis,' this is false for $n>3$. The matrix $$\left[\begin{array}{cccc} \frac{\sqrt{2}}{2} & -\frac{\sqrt{2}}{2} & 0 & 0\ \frac{\sqrt{2}}{2} & \frac{\sqrt{2}}{2} & 0 & 0\ 0 & 0 & \frac{\sqrt{3}}{2} & -\frac{1}{2}\ 0 & 0 & \frac{1}{2} & \frac{\sqrt{3}}{2}\end{array}\right]$$ satisfies all of your conditions, but has no real eigenvalues (and hence, no axis of rotation). –  Nov 16 '14 at 01:33
  • I don't think that the check det(M) = 1 is relevant to test if a matrix has rotation unless it is to test that it is a pure rotation. A det of 1 means, in 3 dimensions, that the cube formed by the axes given by the matrix as an area of 1 cubic unit. Consequently, this also means that the matrix does not contain scale. It is possible to have a rotation matrix with a det of 1 (eg. 2 flipped axis). – brita_ Apr 06 '18 at 11:55
  • A rotation matrix $M$ does not need to satisfy $\det(M)=1$. This is only true if $M$ describes a proper rotation; otherwise it describes an improper rotation, and $\det(M)=-1$. – HelloGoodbye Feb 17 '21 at 22:09
8

I think there is a minus sign missing. As it is, the determinant is not $1$. After fixing, this specific case is easy.

$$\begin{pmatrix} \frac12 & 0 & -\frac{\sqrt{3}}{2} \\ 0 & 1 & 0 \\ \frac{\sqrt{3}}{2} & 0 & \frac12 \end{pmatrix} = \begin{pmatrix} \cos \frac{\pi}{3} & 0 & -\sin \frac{\pi}{3} \\ 0 & 1 & 0 \\ \sin \frac{\pi}{3} & 0 & \cos \frac{\pi}{3} \end{pmatrix}$$

It is a rotation of $\pi/3$ around the $y$-axis.

Ivo Terek
  • 77,665