From a textbook on graphics:

To represent a rotation of $\theta$ degrees about a unit length axis $\mathbf{\hat{k}}$, we use the quaternion

$$ \begin{bmatrix} \cos(\theta / 2) \\ \sin(\theta / 2) \mathbf{\hat{k}} \end{bmatrix} $$

A rotation of $\theta + 4\pi$ degrees about an axis $\mathbf{\hat{k}}$ also gives us the same quaternion...Oddly, a rotation of $\theta + 2 \pi$ degrees about an axis $\mathbf{\hat{k}}$, which in fact is the same rotation, gives us the negated quaternion

$$ \begin{bmatrix} -\cos(\theta / 2) \\ -\sin(\theta / 2) \mathbf{\hat{k}} \end{bmatrix} $$

Question: This quote is from a textbook on graphics that lacks rigor; hence this statement is asserted without proof. So I understand conceptually that $\theta + 2\pi$ represents the same rotation as $\theta$ (a trivial fact). But why do we use this second (negated) matrix to represent this equivalent rotation instead of the original one? What is this textbook getting at?

EDIT: Here is an attempt to get further with the problem:

If $\mathbf{\hat{p}}$ represents a vector in $\mathbb{R}^3$ and $q \in H$ represents the quaternion displayed in the first matrix above, then it seems the claim made above would be true if

$$ q \mathbf{\hat{p}} q^{-1} = (-q) \mathbf{\hat{p}} (-q)^{-1} $$


Yes, your last equation is true. That is essentially why the unit quaternions are a "double cover" of the rotation group, because each unit quaternion $u$ and its additive inverse $-u$ represent the same rotation.

The justification is trivial algebra:

First of all $(-q)(- q^{-1})=(-1)^2 q q^{-1}=1$, so $(-q)^{-1}=-(q^{-1})$

Using that,



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.