1
$\begingroup$

I am trying to understand the quaternion multiplication geometrically.

Following this answer, on What is a good geometric interpretation of quaternion multiplication?, I am trying to see how:

  1. Every non-real quaternion generates a subalgebra isomorphic to $\mathbb{C}$
  2. How $\mathbb{H}$ splits as this subalgebra, and it's orthogonal compliment

So far, I have been unable to construct an isomorphism needed by (1); I have been hoping that this isomorphism was a simple multiplication by an element in $\mathbb{H}$, that depends on the generating-element.

$\endgroup$
  • $\begingroup$ @LordSharktheUnknown Perhaps I misunderstood the original answer. By non-real (or unreal), I thought this meant it looked like $ai + bj +ck + r$ where not all $(a, b, c) \neq (0, 0, 0)$ (i.e: it was not synonymous with "pure" quaternion). In this case of unit quaternions, this would mean it is not 1. $\endgroup$ – Abel May 26 '18 at 18:59
4
$\begingroup$

The quaternions form an algebra over $\mathbb{R}$. If $q$ is a nonreal quaternion, then the subalgebra it generates is $\mathbb{R}[q]$, which is commutative. Now notice that $q+q^*$ and $qq^*=q^*q$ are both real and that $q$ is a root of the polynomial $x^2-(q+q^*)x+qq^*$.

Therefore $\mathbb{R}[q]$ is a $2$-dimensional algebra over $\mathbb{R}$. Since it has no zero divisors, it is a field and thus it it isomorphic to $\mathbb{C}$.

$\endgroup$
  • $\begingroup$ Thanks for the response. Unfortunately, my algebra is a bit rusty, can you elaborate on the 2nd and 3rd step? (q root of real-coefficient polynomial implies 2-dim, and non-existence of zero divisors) $\endgroup$ – Abel May 27 '18 at 21:29
2
$\begingroup$

Alternatively, writing $q$ in polar form as $r\exp(\theta\mathbf{u})=r\cos(\theta)+r\sin(\theta)\mathbf{u}$, in the $\mathbb{R}$-algebra generated by $q$, we can obtain $\mathbf{u}$ from $q$ and vice-versa, so $\mathbb{R}[q]=\mathbb{R}[\mathbf{u}]$. Since $\mathbf{u}$ is a square root of negative one, it functions just as $i$ in $\mathbb{C}$ does, so there is an isomorphism $\mathbb{R}[\mathbf{u}]\cong\mathbb{C}$ given by sending $a+b\mathbf{u}\mapsto a+bi$.

$\endgroup$

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.