6

Let $H$ be the Hamiltonian quaternions, $\mathbb C$ the complex numbers and $\mathbb R$ the real numbers. Identify $H\otimes_{\mathbb R} \mathbb C$ in familiar terms.

This is an exercise of Modern Algebra of Garrett Birkhoff. I think it may have something to do with the Octonions. Any ideas?

any ideas?

zacarias
  • 3,178
  • 1
    Well, the octonions aren't associative, so it seems a little unlikely that they'll show up here, as your algebra has dimension $8$ over the reals. – pjs36 Mar 30 '15 at 20:46
  • @pjs36 Since octonions also have dimension 8 over the reals I'm not sure why you're citing that as a reason they're unlikely to show up. The octonions being nonassociative is justification though, since $\mathbb{H}\otimes\mathbb{C}$ is associative. – anon Jan 05 '17 at 03:04
  • 1
    Hi @arctictern I wish I could provide insight into why me-from-two-years-ago thought the second part of that sentence was a good idea, but unfortunately, I am as clueless as you are :) – pjs36 Jan 05 '17 at 04:22

4 Answers4

8

More concretely, this is just $M_2(\mathbb{C})$.

Not sure exactly how Birkhoff wants you to see this, but here's how I did it, using the judo of central simple algebras. $\mathbb{H}$ is a central division algebra (so a CSA) over $\mathbb{R}$. A commonly used fact is that base-extending a CSA gives a CSA, so $\mathbb{H} \mathop{\otimes}\limits_{\mathbb{R}} \mathbb{C}$, viewed as a $\mathbb{C}$ algebra, is a CSA, of the same dimension (4). Another common result is that any CSA over an algebraically closed field is just a matrix algebra over that field, so this is $M_2(\mathbb{C})$.

Eric Auld
  • 28,997
8

Since $i$ and $j$ both satisfy $x^2+1$, we should look for matrices that also satisfy this polynomial. Take $I$ to be the Jordan canonical form and $J$ to be the rational canonical form for $x^2+1$: $$ I = \begin{pmatrix} \sqrt{-1} & 0\\ 0 & -\sqrt{-1} \end{pmatrix} \qquad J = \begin{pmatrix} 0 & -1\\ 1 & 0 \end{pmatrix} $$ Then $JI = -IJ$ as well, so the map \begin{align*} \mathbb{H} \otimes_\mathbb{R} \mathbb{C} &\to M_2(\mathbb{C})\\ i \otimes 1, j \otimes 1 &\mapsto I,J \end{align*} is a homomorphism, and in fact an isomorphism.

Viktor Vaughn
  • 20,897
5

The quaternions $\mathbb{H}$ are a right vector space over $\mathbb{C}$.

Then there is an algebra homomorphism $\phi:\mathbb{H}\otimes\mathbb{C}\to\mathrm{End}_{\mathbb{C}}(\mathbb{H})$ via

$$\phi(p\otimes z)q=pqz. $$

Using $\{1,\mathbf{j}\}$ as a basis for $\mathbb{H}$ over $\mathbb{C}$ this allows us to identify basis elements of $\mathbb{H}\otimes\mathbb{C}$ with matrices inside $M_2(\mathbb{C})$ in order to check injectivity of $\phi$ hence that it's an isomorphism by dimensions.

anon
  • 155,259
3

$\mathbb{C}\otimes\mathbb{H}$ is isomorphic to the biquaternions or complex quaternions that are similar to $\mathbb{H}$ with complex numbers as scalars. They are a specific feature of Clifford algebra very much useful to reformulate succinctly general relativity.

user26857
  • 53,190
marwalix
  • 17,045