Skip to the Main Content

Note:These pages make extensive use of the latest XHTML and CSS Standards. They ought to look great in any standards-compliant modern browser. Unfortunately, they will probably look horrible in older browsers, like Netscape 4.x and IE 4.x. Moreover, many posts use MathML, which is, currently only supported in Mozilla. My best suggestion (and you will thank me when surfing an ever-increasing number of sites on the web which have been crafted to use the new standards) is to upgrade to the latest version of your browser. If that's not possible, consider moving to the Standards-compliant and open-source Mozilla browser.

September 13, 2025

A Shadow of Triality?

Posted by John Baez

It’s well known that you can construct the octonions using triality. One statement of triality is that Spin(8)Spin(8) has nontrivial outer automorphisms of order 3. On the other hand, the octonions have nontrivial inner automorphisms of order 3. My question: can we deduce one of these facts from the other?

The second fact is perhaps not very well known. It may even be hard to understand what it means. Though the octonions are nonassociative, for any nonzero octonion gg the map

f: 𝕆 𝕆 x gxg 1 \begin{array}{rccl} f \colon & \mathbb{O} &\to& \mathbb{O} \\ & x & \mapsto & g x g^{-1} \end{array}

is well-defined, since (gx)g 1=g(xg 1)(g x)g^{-1} = g(x g^{-1}), which one can show using the fact that the octonions are alternative. More surprisingly, whenever g 6=1g^6 = 1, this map ff is an automorphism of the octonions:

f(x+y)=f(x)+f(y),f(xy)=f(x)f(y)x,y𝕆 f(x+y) = f(x) + f(y) , \qquad f(x y) = f(x) f(y) \qquad \forall x,y \in \mathbb{O}

and ff has order 3:

f(f(f(x)))=xx𝕆 f(f(f(x))) = x \qquad \forall x \in \mathbb{O}

To understand this latter fact, we can look at

Theorem 2.1 here implies that an octonion gg with |g|=1{|g|} = 1 defines an inner automorphism f:xgxg 1f \colon x \mapsto g x g^{-1} if and only if gg has order 6.

However, the result is stated differently there. Paraphrasing somewhat, Lamont’s theorem says that any g𝕆g \in \mathbb{O} that is not a real multiple of 1𝕆1 \in \mathbb{O} defines an inner automorphism f:xgxg 1f \colon x \to g x g^{-1} if and only if gg obeys

4Re(g) 2=|g| 2 4 \mathrm{Re}(g)^2 = {|g|}^2

This equation is equivalent to Re(g)=±12|g|\operatorname{Re}(g) = \pm \frac{1}{2} {|g|}, which is equivalent to gg lying at either a 60 60^\circ angle or a 120 120^\circ angle from the octonion 11.

Nonzero octonions on the real line clearly define the identity inner automorphism. Thus, a nonzero octonion gg defines an inner automorphism if and only if its angle from 11 is 0 0^\circ, 60 60^\circ, 120 120^\circ or 180 180^\circ. In this case we can normalize gg without changing the inner automorphism it defines, and then we have g 6=1g^6 = 1. Note also that gg and g-g define the same inner automorphism.

It follows that an octonion gg on the unit sphere defines an inner automorphism iff g 6=1g^6 = 1, and that every nontrivial inner automorphism of 𝕆\mathbb{O} has order 3.

However, if you look at Lamont’s proof, you’ll see the equation 4Re(g) 2=|g| 24 \operatorname{Re}(g)^2 = {|g|}^2 plays no direct role! Instead, he really uses the assumption that g 3g^3 is a real multiple of 11, which is implied by this equation (as easily shown using what we’ve just seen).

From Lamont’s work, one can see the Moufang identities and the characteristic equation for octonions are what force all inner automorphisms of the octonions to have order 3.

Thus, an argument giving a positive answer to my question might involve a link between triality and the Moufang identities. Conway and Smith seem to link them in On Quaternions and Octonions. But I haven’t figured out how to get from the outer automorphisms of Spin(8)\text{Spin}(8) to the inner automorphisms of 𝕆\mathbb{O}, or vice versa!

I asked about this on MathOverflow, but I thought some people here would also be interested.

Posted at September 13, 2025 12:06 PM UTC

TrackBack URL for this Entry:   https://golem.ph.utexas.edu/cgi-bin/MT-3.0/dxy-tb.fcgi/3616

14 Comments & 0 Trackbacks

Re: A Shadow of Triality?

Toby Bartels: if you read this, could you remind me how to get the spacing around absolute value signs to look less wretched?

(Or anyone else who knows.)

Posted by: John Baez on September 13, 2025 12:22 PM | Permalink | Reply to this

Re: A Shadow of Triality?

I got ahold of Toby through other channels and he answered:

Curly braces around the absolute value: ${|x|}$.
(Actually, $|x|$ by itself shouldn't cause any 
problems, but $|x|+x$ or $2|x|$ will, and ${|x|}+x$ 
and $2{|x|}$ will fix these.)  I'm in the habit of 
including these braces whenever I write any 
absolute value or similar expressions in itex.
Incidentally, you're not really supposed to use
 | for absolute value at all, but instead the 
directed delimiters \lvert and \rvert (or \lVert 
and \rVert instead of \| for ‖), so $\lvert 
x\rvert$.  Using | instead can cause a completely 
different spacing issue that is much rarer, 
subtler (smaller spaces), and not itex's fault (it 
will also happen in LaTeX or plain TeX); you can 
see this with $|-x|$ for example.  (A quick hack 
to fix this is $|{-x}|$, with braces *inside* the 
absolute value, but $\lvert-x\rvert$ is the really 
proper way to do it.)
Or you could always use \left and \right.  By 
design, \left| is equivalent to \left\lvert etc.  
Furthermore, this also fixes the itex bug that the 
first paragraph addresses (unlike \lvert and 
\rvert), for reasons that I don't understand.  (I 
only discovered this just now, because I was about 
to tell you that it doesn't make any difference, 
but first I tested it by previewing a fake comment 
on the n-Category Café, and it turns out that 
it does make a difference!)
Posted by: John Baez on September 13, 2025 6:13 PM | Permalink | Reply to this

Re: A Shadow of Triality?

Ha, I never noticed, while I was at the Café testing for my reply to you on Mastodon, that I could have replied to you right here on the Café!

Posted by: Toby Bartels on September 13, 2025 6:19 PM | Permalink | Reply to this

Re: A Shadow of Triality?

This is a fascinating question!

I can’t answer it but it got me looking at octonions again this afternoon.

I enjoyed trying to visualize this inner automorphism. Without loss of generality, we can take gg to belong to the complex subalgebra of 𝕆\mathbb{O} spanned by 1,i 01, i_0 and to relax the constraint on order or angle, I’ll write g=cos(ϕ/2)+i 0sin(ϕ/2)g = \cos(\phi/2) + i_0 \sin(\phi/2).

It seems that α:xgxg 1\alpha: x \mapsto g x g^{-1} defines three intersecting quaternion subalgebras of 𝕆\mathbb{O}. We can cycle the three quaternion subalgebras that gg defines using i ni 2ni_n \mapsto i_{2n} (provided that i 0i_0 is the imaginary unit used to build gg).

These three quaternion subalgebras define three planes orthogonal to their intersection (one spanned by i 1,i 3i_1,i_3, another by i 2,i 6i_2,i_6, and another by i 4,i 5i_4,i_5) and α\alpha rotates each of these planes by the same angle ϕ\phi. The constraint that fixes the permissible rotation angle ϕ\phi is that multiplication between quaternions in two of these three planes has its product in the third, and that these three planes must rotate together under the action of α\alpha.

Playing a bit with this today I wondered if these three quaternion subalgebras defined by gg (or rather by i 0i_0) and the map i ni 2ni_n \mapsto i_{2n} might help with your question somehow.

Posted by: Ben on September 13, 2025 5:34 PM | Permalink | Reply to this

Re: A Shadow of Triality?

Ben wrote:

It seems that α:xgxg 1\alpha: x \mapsto g x g^{-1} defines three intersecting quaternion subalgebras of 𝕆\mathbb{O}.

How does it do that?

I think I get most of what you’re saying, but I don’t see how the map α\alpha defines three quaternion subalgebras.

Posted by: John Baez on September 13, 2025 10:09 PM | Permalink | Reply to this

Re: A Shadow of Triality?

I’ll try to explain. Any two octonions x,yx,y define an associative subalgebra of 𝕆\mathbb{O}, which we can write (x,y)\mathbb{R}(x,y) and which is the subalgebra of 𝕆\mathbb{O} generated by all products and \mathbb{R}-linear combinations of 1,x,y1,x,y. If x,yx,y do not belong to a common complex subalgebra of 𝕆\mathbb{O} (if they don’t commute) then that subalgebra is a quaternion subalgebra: (x,y)\mathbb{R}(x,y)\cong \mathbb{H}.

Again, I’ll write α:xgxg 1\alpha: x \mapsto g x g^{-1} with g=cos(ϕ/2)+i 0sin(ϕ/2)g = cos(\phi/2) + i_0 sin(\phi/2) acting on xx in 𝕆\mathbb{O}. The imaginary unit octonion i 0i_0 here is chosen without loss of generality since Aut(𝕆)\mathrm{Aut}(\mathbb{O}) is transitive on octonion imaginary units. We can choose a familiar basis for the octonions (the same used in Conway and Smith’s book), {i ttPL(7)}\{i_t \mid t \in \mathrm{PL}(7)\}.

With this notation, we can see that the map α\alpha looks just like the formula for rotating the quaternion imaginary vectors about the quaternion imaginary unit i 0i_0 by the angle ϕ\phi. Also, every octonion is contained in some quaternion subalgebra (Prop 1.6.4 in Springer and Veldkamp’s book, Octonions, Jordan Algebras, and Exceptional Groups). We can therefore inspect the action of the map α\alpha on the standard octonion basis and identify three quaternion subalgebras closed under α\alpha and intersecting in (i 0)\mathbb{R}(i_0)\cong\mathbb{C}.

First, α(1)=1\alpha(1) = 1, α(i 0)=i 0\alpha(i_0) = i_0, since g,1,i 0g, 1, i_0 belong to the complex subalgebra (i 0)\mathbb{R}(i_0) \cong \mathbb{C} and therefore commute, which makes α(x)\alpha(x) the identity map for any xx that commutes with gg.

Next, i 0,i 1,i 3i_0, i_1, i_3 behave like the quaternion units i,j,ki,j,k, with i 0i 1i 3=1i_0 i_1 i_3 = -1. The quaternion subalgebra (i 0,i 1)\mathbb{R}(i_0, i_1) \cong \mathbb{H} contains (i 0)\mathbb{R}(i_0)\cong\mathbb{C} and is closed under the action of α\alpha since we know that α\alpha rotates the i 1i_1-i 3i_3 plane by angle ϕ\phi. That is, we have α(i 1)=cos(ϕ)i 1+sin(ϕ)i 3\alpha(i_1) = cos(\phi) i_1 + sin(\phi) i_{3} and α(i 3)=cos(ϕ)i 3sin(ϕ)i 1\alpha(i_3) = cos(\phi) i_3 - sin(\phi) i_1. By the octonion automorphism i ni 2ni_n \mapsto i_{2n} we also have α(i 2)=cos(ϕ)i 2+sin(ϕ)i 6\alpha(i_2) = cos(\phi) i_2 + sin(\phi) i_{6}, α(i 6)=cos(ϕ)i 6sin(ϕ)i 2\alpha(i_6) = cos(\phi) i_6 - sin(\phi) i_{2}, α(i 4)=cos(ϕ)i 4+sin(ϕ)i 5\alpha(i_4) = cos(\phi) i_4 + sin(\phi) i_{5}, and α(i 5)=cos(ϕ)i 5sin(ϕ)i 4\alpha(i_5) = cos(\phi) i_5 - sin(\phi) i_{4}.

Taken together, α\alpha fixes (i 0)\mathbb{R}(i_0)\cong\mathbb{C} and acts on the three quaternion subalgebras (i 0,i 1)(i 0,i 2)(i 0,i 4)\mathbb{R}(i_0, i_1) \cong \mathbb{R}(i_0, i_2) \cong \mathbb{R}(i_0, i_4) \cong \mathbb{H} by simultaneously rotating the i ni_n-i 3ni_{3n} (n=1,2,4n=1,2,4) planes orthogonal to their intersection (i 0)\mathbb{R}(i_0)\cong\mathbb{C} by angle ϕ\phi.

That’s what I mean when I say that α\alpha defines three intersecting quaternion subalgebras of 𝕆\mathbb{O}. We know that α\alpha is an automorphism of (i 0,i n)\mathbb{R}(i_0, i_n) \cong \mathbb{H} (for n=1,2,4n=1,2,4), but in order for α\alpha to be an automorphism of 𝕆\mathbb{O} it must also behave well for multiplication between elements in different i ni_n-i 3ni_{3n} planes (i.e. not sharing the same quaternion subalgebra). Playing with these coordinates yesterday it seemed that a couple examples constrain the value of ϕ\phi to the expected 0,120,2400,120, 240 degrees.

Posted by: Ben on September 14, 2025 8:04 AM | Permalink | Reply to this

Re: A Shadow of Triality?

This isn’t an answer, but I’ll contribute what potential clues I know.

G_2 := Aut(octonions) has exactly two conjugacy classes of order 3; pick elements b,c of them. The big difference is that the centralizer of b is isomorphic to U(2) (or at least has that Lie algebra) and the centralizer of c is SU(3). It is this conjugacy class {c} that you’re producing as “inner automorphisms”.

G_2 acts on the pure imaginary unit octonions, a sphere S^6 inside R^7. Your angle condition finds a couple of other S^6s inside the unit octonions S^7, each invariant under G _2. Conjugation by one of your g preserves 1 and g, so acts on the complementary R^6. On that R^6 we can use the imaginary part of g to define a complex structure, so our stabilizer sits inside the U(3) preserving that. By dimension count it’s only SU(3).

Anyway if you want some kind of “correspondence” between your two different kinds of order 3 automorphisms, we should figure what what the spaces are of each. There’s an S^6 worth of inner auts of O.

Meanwhile, if we fix one specific non-inner automorphism tau of Spin(8) having order 3 (which it’s not obvious we can do, because Out(G) is not a subgroup of Aut(G) but rather a subquotient; anyway we indeed can) then the other ones are of the form tau * h where h is in Spin(8), and (tau * h)^3 = 1.

I don’t know what that space {h} looks like. It obviously contains the order 3 elements of the fixed points of tau, at which point we’re back to order 3 elements of G_2. But maybe there’s more? I guess I’ve at least managed to show one containment, for a specific interpretation of your question.

Posted by: Allen Knutson on September 13, 2025 9:51 PM | Permalink | Reply to this

Re: A Shadow of Triality?

It may be worth remarking explicitly that the miracle that lets us realise Out(G)\operatorname{Out}(G) as a subgroup of Aut(G)\operatorname{Aut}(G) is that our group GG is not a random group but a semisimple compact Lie group (or perhaps a complex Lie group, or … whatever generality we decide to work in). In this setting, the choice of a pinning, i.e., one non-00 vector X αX_\alpha in the α\alpha-root subspace of 𝔤 \mathfrak{g}_\mathbb{C} for each α\alpha in a fixed system Δ\Delta of simple roots, is precisely the information needed to determine the splitting Out(G)Aut(G)\operatorname{Out}(G) \to \operatorname{Aut}(G), where we identify Out(G)\operatorname{Out}(G) with the group of diagram automorphisms of GG, then lift σ¯Out(G)\bar{\sigma} \in \operatorname{Out}(G) to the unique automorphism of GG that preserves the set {X α:αΔ}\{X_\alpha : \alpha \in \Delta\} and permutes it according to σ¯\bar{\sigma}.

Posted by: L Spice on September 16, 2025 1:00 AM | Permalink | Reply to this

Re: A Shadow of Triality?

I’m thinking it’s somewhere in the Laves graph.

Posted by: Sambucus Holleraway on September 14, 2025 6:27 PM | Permalink | Reply to this

Re: A Shadow of Triality?

I got some help from Vít Tuček on MathOverflow. He pointed out that Yokota gave a proof that all inner automorphisms of 𝕆\mathbb{O} have order 3 using the Moufang identity and two ideas connected to triality. I’d like to spell it out here.

First, Yokota proves these:

Theorem 1.14.2. For any α 3SO(8)\alpha_3 \in SO(8) there are α 1,α 2SO(8)\alpha_1, \alpha_2 \in SO(8) such that

α 1(x)α 2(y)=α 3(xy) \alpha_1(x) \, \alpha_2(y) = \alpha_3(x y)

for all x,y𝕆x, y \in \mathbb{O}. Moreover α 1,α 2\alpha_1, \alpha_2 are unique up to a common choice of sign.

Lemma 1.14.3. If α 1,α 2,α 3SO(8)\alpha_1, \alpha_2, \alpha_3 \in SO(8), then

α 1(x)α 2(y)=α 3(xy¯)¯ \alpha_1(x) \alpha_2(y) = \overline{\alpha_3(\overline{x y})}

for all octonions x,y𝕆x,y \in \mathbb{O} implies that

α 2(x)α 3(y)=α 1(xy¯)¯ \alpha_2(x) \alpha_3(y) = \overline{\alpha_1(\overline{x y})}

for all x,y𝕆x, y \in \mathbb{O}.

He calls the theorem the “principle of triality for SO(8)”. It’s a step toward proving the fact that Spin(8) has an outer automorphism of order 3 relating its three 8-dimensional representations. He proves the lemma simply by equation-juggling using basic properties of normed division algebras.

Now suppose conjugation by a𝕆a \in \mathbb{O} defines an inner automorphism of the octonions. We can normalize a𝕆a \in \mathbb{O} without changing the inner automorphism it defines, and then we have aa¯=1 a \overline{a} = 1, so a 1=a¯a^{-1} = \overline{a}. Thus we have

(axa¯)(aya¯)=a(xy)a¯ (a x \overline{a})(a y \overline{a}) = a (x y) \overline{a}

Next we use a cool fact about the octonions called the Moufang identity:

(bc)(db)=b(cd)b (b c)(d b) = b(c d)b

where the right-hand side is actually well-defined regardless of how you parenthesize it. Take b=a¯b = \overline{a} and get

(a¯c)(da¯)=a¯(cd)a¯ (\overline{a} c)(d \overline{a}) = \overline{a}(c d) \overline{a}

We can rewrite this equation if we take

α 1(x)=a¯x,α 2(x)=xa¯,α 3(x)=axa \alpha_1(x) = \overline{a} x , \; \alpha_2(x) = x \overline{a} , \; \alpha_3(x) = a x a

Namely, we can rewrite it as

α 1(c)α 2(d)=α 3(cd¯)¯ \alpha_1(c) \alpha_2(d) = \overline{\alpha_3(\overline{c d})}

Since α 1,α 2,α 3SO(8)\alpha_1, \alpha_2, \alpha_3 \in \text{SO}(8), Lemma 1.14.3 jumps in and tells us

α 2(c)α 3(d)=α 1(cd¯)¯ \alpha_2(c) \alpha_3(d) = \overline{\alpha_1(\overline{c d})}

Concretely this means

(ca¯)(ada)=a¯(cd¯)¯ (c \overline a)(a d a) = \overline{\overline{a} (\overline{c d})}

or in other words

(ca¯)(ada)=(cd)a (c \overline a)(a d a) = (c d) a

Now, cleverly taking c=axc = a x, d=yad = y a, we get

(axa¯)(aya 2)=a(xy)a 2 (a x \overline{a})(a y a^2) = a (x y) a^2

But remember we have assumed that conjugation by aa is an inner automorphism and a 1=a¯a^{-1} = \overline{a}. This gives a very similar equation:

(axa¯)(aya¯)=a(xy)a¯ (a x \overline{a})(a y \overline{a}) = a (x y) \overline{a}

Then the uniqueness part of Theorem 1.14.2 implies we must have

a¯=±a 2 \overline{a} = \pm a^2

or

a 3=±1 a^3 = \pm 1

so a 6=1a^6 = 1 and aa defines an inner automorphism of order 3.

This argument is pretty convoluted, so it’s hard to see how the threeness of triality forces inner automorphisms of the octonions to have order 3. However, we can also turn this argument around and show that a 6=1a^6 = 1 implies that aa defines an inner automorphism.

Posted by: John Baez on September 16, 2025 6:51 PM | Permalink | Reply to this

Re: A Shadow of Triality?

I think the answer (or something close to it) turns out to be in Conway and Smith’s On Quaternions and Octonions in their discussion of companions to an orthogonal transformation.

They define an inverse loop (p. 83) and show that any inverse loop will satisfy xyz=yzx=zxy=1x y z=y z x=z x y=1. They also introduce the isotopies of an inverse loop, namely triples of invertible maps (α,β,γ)(\alpha,\beta,\gamma) such that xyz=1x y z =1 implies that x αy βz γ=1x^\alpha y^\beta z^\gamma = 1 (p. 84). The principle of triality is related to the fact that (α,β,γ)(\alpha,\beta,\gamma) being an isotopy also implies that (β,γ,α)(\beta,\gamma,\alpha) and (γ,α,β)(\gamma,\alpha,\beta) are isotopies.

The connection that we are looking for seems to come from the concept of companions to any member of an isotopy (p. 86). If we take the map γ\gamma then there will be a pair of elements bb and aa (in the orbit of the identity) such that (xy) γ=(x γb)(ay γ)(x y)^\gamma = (x^\gamma b)(a y^\gamma). Suppose that we take the map to be γ:xa 1xa\gamma: x \mapsto a^{-1} x a (shown as T a\text{T}_a on p. 98). Provided that diassociativity holds (which I suspect it must for a Moufang loop), then the companions of γ\gamma are a 3a^{-3} and a 3a^3. That is, we have a 1(xy)a=((a 1xa)a 3)(a 3(a 1ya))a^{-1}(x y)a = ((a^{-1} x a)a^{-3})(a^3 (a^{-1}y a)).

I might be missing an assumption somewhere but I think we can conclude something close to this: In any Moufang loop, the map γ:xa 1xa\gamma: x \mapsto a^{-1} x a has companions a 3a^{-3} and a 3a^3. Therefore the map γ\gamma is an automorphism only when these companions cancel out due to (1) the loop being associative or (2) the element a 3a^3 belonging to the nucleus of the loop.

Posted by: Ben on September 24, 2025 3:04 PM | Permalink | Reply to this

Re: A Shadow of Triality?

Wow, this could be the really beautiful explanation I’d been hoping for! Thanks! I’d been trying to read Conway and Smith, hoping I’d find clues there, but I had trouble penetrating this section. I don’t fully understand what you’re saying, but it sounds good.

I will get back to this subject soon, since I have reasons to think that triality and these inner automorphism interact in a nice way in the exceptional Jordan algebra 𝔥 3(𝕆)\mathfrak{h}_3(\mathbb{O}).

Posted by: John Baez on September 25, 2025 12:59 PM | Permalink | Reply to this

Re: A Shadow of Triality?

Ben wrote:

They define an inverse loop (p. 83) and show that any inverse loop will satisfy xyz=yzx=zxy=1x y z=y z x=z x y=1.

I was confused by this at first. In fact they define an inverse loop to be a set with a binary multiplication, a unit 1, and an inverse operation obeying

x 1(xy)=y=(yx)x 1 x^{-1}(x y) = y = (y x) x^{-1}

and

(x 1) 1=x (x^{-1})^{-1} = x

What you’re trying to say here is that in an inverse loop, if three elements obey xyz=1x y z = 1 then we can cycle this relation and get

xyz=yzx=zxy=1x y z = y z x = z x y = 1

and invert these and get

z 1y 1x 1=y 1x 1z 1=x 1z 1y 1=1z^{-1} y^{-1} x^{-1} = y^{-1} x^{-1} z^{-1} = x^{-1} z^{-1} y^{-1} = 1

But anyway, you’re right: in Theorem 14, Conway and Smith use triality, and these ideas, to show that conjugation is by an invertible octonion aa is an automorphism iff a 3a^3 is a real multiple of 11. This is what I wanted!

Posted by: John Baez on September 26, 2025 9:21 AM | Permalink | Reply to this

Re: A Shadow of Triality?

Here are some more details on how Conway and Smith get from triality to this fact: if cc is an octonion with c 3=±1c^3 = \pm 1, then conjugation by cc is an automorphism of the octonions. The argument is essentially the same as the argument by Yokota which I presented here, but it has a different feel. I will skip a lot of steps.

Using triality, Conway and Smith show in Theorem 11 that if γmaps𝕆𝕆\gamma \maps \mathbb{O} \to \mathbb{O} is any rotation of the octonions, then for all x,y𝕆x,y \in \mathbb{O} we have

γ(xy)=(γ(x)a)(bγ(x)) \gamma(x y) = (\gamma(x)a) (b\gamma(x))

for some unit octonions a,ba,b called ‘companions’ of γ\gamma, which are unique up to sign.

Suppose cc is an octonion with c 3=±1c^3 = \pm 1.

The algebra generated by any two octonions is associative, and c 1c^{-1} is in the algebra generated by cc. Given these facts, it follows that (cx)c 1=c(xc 1)(c x)c^{-1} = c(x c^{-1}), so conjugation

xcxc 1 x \mapsto c x c^{-1}

is unambiguous.

Left or right multiplication by an octonion with norm 1 is a rotation of the octonions, and |c|=1{|c|} = 1 implies |c 1|=1{|c^{-1}|} = 1. So xcxc 1 x \mapsto c x c^{-1} is a rotation of the octonions.

In the proof of Theorem 14, Conway and Smith show that this rotation has companions c 3c^3 and c 3c^{-3}. This means

c(xy)c 1=((cxc 1)c 3)(c 3(cxc 1)) c(x y)c^{-1} = ((c x c^{-1}) c^3)(c^{-3} (c x c^{-1}))

But now suppose c 3=±1c^3 = \pm 1. Then we can simplify and get

c(xy)c 1=(cxc 1)(cyc 1) c(x y)c^{-1} = (c x c^{-1}) (c y c^{-1})

which says conjugation by cc is an automorphism.

But how do Conway and Smith show that the map xcxc 1 x \mapsto c x c^{-1} has companions c 3c^3 and c 3c^{-3}? They show it using two simpler results, which are two parts of their Theorem 13:

  1. The map xcx x \mapsto c x has companions cc and c 2c^{-2}

  2. The map xxcx \mapsto x c has companions c 2c^{-2} and cc.

The proofs of these are symmetrical so if we can do the first, we can do the second. The first one claims that

c(xy)=((cx)c)(c 2(cy)) c ( x y ) = ((c x) c) (c^{-2} (c y))

Since any subalgebra of the octonions generated by two elements is associative, we can rewrite the right hand side here as

(cxc)(c 1y) (c x c)(c^{-1} y)

Then the Moufang identities imply just what we want:

(cxc)(c 1y)=c(xy) (c x c)(c^{-1} y) = c (x y)

Posted by: John Baez on September 28, 2025 11:02 AM | Permalink | Reply to this

Post a New Comment