$\DeclareMathOperator{\SO}{\mathrm{SO}}\DeclareMathOperator{\Spin}{\mathrm{Spin}}\DeclareMathOperator{\Inn}{\mathrm{Inn}}\DeclareMathOperator{\Out}{\mathrm{Out}}\DeclareMathOperator{\Aut}{\mathrm{Aut}}$ How do we construct a precise map of inner + outer automorphism of special orthogonal group $\SO(n;\mathbb{R})$?

  • $d=2$; We can look at $\SO(2;\mathbb{R})=U(1)$ which is abelian, and we know the inner $$ \Inn(\SO(2;\mathbb{R}))=\SO(2;\mathbb{R})/Z(\SO(2;\mathbb{R}))=1 $$ $$ \Out(\SO(2;\mathbb{R}))=\mathbb{Z}/2 $$ The total $\Aut(\SO(2;\mathbb{R}))=\Inn(\SO(2;\mathbb{R})) \rtimes \Out(\SO(2;\mathbb{R}))=\mathbb{Z}/2$ We have no $\Inn(\SO(2;\mathbb{R}))$ except the identity map. I believe that we can get the $\Out(\SO(2;\mathbb{R}))=\mathbb{Z}/2$ by flipping $t \to -t$ in $$ U(1)=\{\exp(i t) | t \in [0, 2 \pi)\} \to \{\exp(-i t) | t \in [0, 2 \pi)\}. $$ I wish to see explicit answer like the above for my following questions ---

  • other $n$ but $n\neq 2,8$ is discussed in MSE with answer still pending.

  • for $n=8$

Question 1: How do we construct the inner automorphism map explicitly (if my result is correct?)? Let us consider $\Spin(8;\mathbb{R})$, $\SO(8;\mathbb{R})$, $\SO(8;\mathbb{R})/(\mathbb{Z}/2)$.

for $n=8$

$$ \Inn(\Spin(n;\mathbb{R}))=\Spin(n;\mathbb{R})/Z(\Spin(n;\mathbb{R})) = \SO(8;\mathbb{R})/\mathbb{Z}/2 $$ $$ \Inn(\SO(n;\mathbb{R})/\mathbb{Z}/2)=(\SO(n;\mathbb{R})/\mathbb{Z}/2)/Z(\SO(n;\mathbb{R})/(\mathbb{Z}/2)) = \SO(8;\mathbb{R})/\mathbb{Z}/2 $$ $$ \Inn(\SO(n;\mathbb{R}))=\SO(n;\mathbb{R})/Z(\SO(n;\mathbb{R})) = \SO(8;\mathbb{R})/\mathbb{Z}/2 $$

Question 2: How do we construct the outer automorphism map explicitly $$ \Out(Spin(8;\mathbb{R}))=S_3 $$ $$ \Out(\SO(8;\mathbb{R}))=\mathbb{Z}/2 $$ $$ \Out(\SO(8;\mathbb{R})/\mathbb{Z}/2 )=S_3 $$ Given the parametrization of $\SO(n;\mathbb{R})$ how to map to itself via the $\Out$ map?

Question 3: How do we construct the total automorphism map explicitly

$$ \Aut(\Spin(8;\mathbb{R}))=\Inn(\Spin(8;\mathbb{R})) \rtimes \Out(\Spin(8;\mathbb{R})) =(\SO(8;\mathbb{R})/\mathbb{Z}/2 ) \rtimes S_3 ? $$ $$ \Aut(\SO(8;\mathbb{R}))=\Inn(\SO(8;\mathbb{R})) \rtimes \Out(\SO(8;\mathbb{R})) =(\SO(8;\mathbb{R})/\mathbb{Z}/2 ) \rtimes \mathbb{Z}/2 ? $$ $$ \Aut(\SO(8;\mathbb{R})/\mathbb{Z}/2)=\Inn(\SO(8;\mathbb{R})/\mathbb{Z}/2) \rtimes \Out(\SO(8;\mathbb{R})/\mathbb{Z}/2) =(\SO(8;\mathbb{R})/\mathbb{Z}/2 ) \rtimes S_3 ? $$ Given the parametrization of $\Spin(8;\mathbb{R})$, $\SO(8;\mathbb{R})$, $\SO(8;\mathbb{R})/(\mathbb{Z}/2)$, how to map to itself via the $\Aut$ map?

P.S. Possible useful link but with not explicit (not enough) constructions in Automorphism group of real orthogonal Lie groups

  • $\begingroup$ See mathoverflow.net/questions/116666/triality-of-spin8?rq=1 $\endgroup$
    – Q. Zhang
    Sep 28, 2020 at 2:30
  • $\begingroup$ which answer do you suggest to concerns the explicit map of $S_3$? and inner map? thanks! $\endgroup$ Sep 28, 2020 at 2:32
  • $\begingroup$ An automorphism permits the three extremal simple weights; that's the map from the automorphism group (hence from the outer automorphism group, since it's trivial) to $\operatorname S_3$. An automorphism also carries a so called pinning (Borel subgroup, torus, set of simple root vectors) to another pinning, which is conjugate by a unique inner automorphism to the first; that's the map to the inner automorphism group. What does "Given the parametrization of $\operatorname{SO}(n; \mathbb R)$ mean", and what do you want to map to itself? $\endgroup$
    – LSpice
    Sep 28, 2020 at 3:34
  • 3
    $\begingroup$ The assertion that $\mathrm{Out}(\mathrm{SO}(8))$ has order 6 is false. It has order 2, and the full automorphism group is $\mathrm{PO}(8)$. The claim about $S_3$ is correct with $\mathrm{PSO}(8)$ instead. $\endgroup$
    – YCor
    Sep 28, 2020 at 6:30
  • 3
    $\begingroup$ I'm not sure that I understand what you are looking for regarding inner automorphisms. The inner automorphisms all come from conjugation via a matrix of ${\rm SO}(8,\mathbb{R})$ (and clearly the central involution acts trivially), so how much more explicit do you want to be? $\endgroup$ Sep 29, 2020 at 8:38

1 Answer 1


If you just want an explicit realization the of outer automorphisms of $\mathrm{Spin}(8)$, here is one, assuming that you know about the algebra of octonions $\mathbb{O}$, the unique $8$-dimensional (and hence nonassociative) inner product algebra over $\mathbb{R}$ with positive definite inner product.

The subgroup $\mathrm{Spin}(8)\subset{\mathrm{SO}(8)}^3$ is defined as the set of triples $g = (g_1,g_2,g_3)$ that satisfy $$ \mathrm{Re}\bigl(g_1(a_1)g_2(a_2)g_3(a_3)\bigr) = \mathrm{Re}(a_1a_2a_3) $$ for all $a_i\in\mathbb{O}$. (Here, $\mathrm{Re}(a_1a_2a_3) = (a_1a_2a_3)\cdot\mathbf{1}$, where $\mathbf{1}\in\mathbb{O}$ is the multiplicative unit.) The group of outer automorphisms of $\mathrm{Spin}(8)$ is generated the element $\beta:\mathrm{Spin}(8)\to \mathrm{Spin}(8)$ of order $3$ defined by $$ \beta(g_1,g_2,g_3) = (g_2,g_3,g_1) $$ and the element $\alpha:\mathrm{Spin}(8)\to \mathrm{Spin}(8)$ of order $2$ defined by $$ \alpha(g_1,g_2,g_3) = \bigl(\ cg_2c,\ cg_1c,\ cg_3c\ \bigr) $$ where $c:\mathbb{O}\to\mathbb{O}$ is octonionic conjugation, i.e., $c(x) = 2(x{\cdot}\mathbf{1})\,\mathbf{1} - x$. (Note that $c$ belongs to $\mathrm{O}(8)$ but not $\mathrm{SO}(8)$.)

The facts that $\mathrm{Spin}(8)$, as defined as above is a subgroup of $\mathrm{SO}(8)^3$ and that each of the projections $\pi_i:\mathrm{Spin}(8)\to\mathrm{SO}(8)$ defined by $\pi_i(g_1,g_2,g_3) = g_i$ is a nontrivial double cover of $\mathrm{SO}(8)$ and that $\alpha$ and $\beta$ are outer automorphisms of $\mathrm{Spin}(8)$ follow from basic facts about the algebra $\mathbb{O}$.

If the above description is not explicit enough, or a description that does not mention the octonions is preferred, here is Cartan's description at the level of the Lie algebra ${\frak{so}}(8)$, drawn from his paper Le principe de dualité et la théorie des groupes simples et semi-simples (Bull. Sc. Math. 49 (1925), 361–374:

Let indices run from $0$ to $7$ with the understanding that, if a formula gives an index greater than $7$, one subtracts $7$. (Thus, $8=1$, but $7\not=0$.) Then an element $a\in {\frak{so}}(8)$ is a skew-symmetric matrix with entries $a = (a_{i,j})$ where $a_{i,j}=-a_{j,i}$. There are essentially 28 distinct entries, and these break up into $7$ groups of $4$: $$ b_i = \begin{pmatrix}a_{0,i}\\a_{i+1,i+5}\\a_{i+4,i+6}\\ a_{i+2,i+3}\\\end{pmatrix},\qquad i=1,\ldots,7 $$ Let $$ H=\frac12\,\begin{pmatrix} -1&-1&-1&-1\\ \phantom{-}1&\phantom{-}1&-1&-1\\ \phantom{-}1&-1&\phantom{-}1&-1\\ \phantom{-}1&-1&-1&\phantom{-}1\\\end{pmatrix}\quad\text{and}\quad K = \begin{pmatrix}-1&0&0&0\\ 0&1&0&0\\ 0&0&1&0\\ 0&0&0&1\end{pmatrix}. $$ Note that $H^3=K^2=I_4$ and $KHK = H^2$.

Consider the linear mappings $h:{\frak{so}}(8)\to {\frak{so}}(8)$ and $k:{\frak{so}}(8)\to {\frak{so}}(8)$ induced by the transformations $$ b_i\mapsto Hb_i\,\quad\text{and}\quad b_i\mapsto Kb_i, \quad i = 1,\ldots,7. $$ Then $h$ and $k$ are automorphisms of ${\frak{so}}(8)$ that generate a group of order $6$ that maps isomorphically onto the group of outer automorphisms of ${\frak{so}}(8)$.

  • $\begingroup$ Is there a way to show whether or not this action via matrix multiplication exists for defining/spin representations as well? I.e. is there a convenient (matrix) way to perform the triality in other representations, given that in the adjoint rep it may be given by simple matrix multiplication, or am I left with decomposing and shuffling my bases around 'manually'? $\endgroup$
    – Craig
    Feb 25 at 22:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.