In JOhn Baez's article The Octonions, he writes the lie algebra isomorphism $\mathfrak sl_2(\mathbb {O}) \space \tilde = \space \mathfrak so(9, 1) $. Why is the former a Lie algbra, even though $SL_2(\mathbb O)$ is not a Lie group, because it is clearly non-associative?

  • 3
    $\begingroup$ Check the definitions carefully; they don't mean the obvious things. $\endgroup$ – Qiaochu Yuan Oct 8 '17 at 8:53
  • $\begingroup$ If there is such a thing, it is not likely to be associative. $\endgroup$ – Lord Shark the Unknown Oct 8 '17 at 9:30
  • $\begingroup$ @LordSharktheUnknown It is. $\endgroup$ – anon Oct 11 '17 at 13:45

$\mathfrak{sl}(2,\mathbb{O})$ is the lie subalgebra of $\mathfrak{gl}(\mathbb{O}^2)$ (treating $\mathbb{O}^2\cong\mathbb{R}^{16}$ as real vector spaces) generated by the linear operators expressible as multiplication by a traceless octonionic $2\times 2$ matrix. And then the Lie group $\mathrm{SL}(2,\mathbb{O})$ within $\mathrm{GL}(16,\mathbb{R})$ is generated by the image of $\mathfrak{sl}(2,\mathbb{O})$ under $\exp$. This is more or less said on page $28$ of Baez's article.

Note $\mathrm{SL}(2,\mathbb{O})$ is associative because composition is associative; it just usually doesn't match up with $2\times2$ octonionic matrix multiplication. If $A$ and $B$ are $2\times 2$ octonionic matrices and $L_A$ is used to denote the left multiplication map $L_A(X)=AX$, then $L_A\circ L_B=L_{AB}$ iff $A,B$ are both $2\times2$ real matrices. If you instead represent $A,B$ by $16\times16$ real matrices $C,D$, then the $16\times16$ real matrix associated with the composition $L_A\circ L_B$ will be the matrix product $CD$.


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.