4.1

In 1914, Élie Cartan noted that the smallest of the exceptional Lie groups, , is the automorphism group of the octonions [12]. Its Lie algebra is therefore , the derivations of the octonions. Let us take these facts as definitions of and its Lie algebra, and work out some of the consequences.

What are automorphisms of the octonions like? One way to analyze this
involves subalgebras of the octonions. Any octonion whose square
is generates a subalgebra of isomorphic to . If we then
pick any octonion with square equal to that anticommutes with
, the elements generate a subalgebra isomorphic to .
Finally, if we pick any octonion with square equal to that
anticommutes with and , the elements
generate all of . We call such a triple of octonions a **basic
triple**. Given any basic triple, there exists a unique way to define
so that the whole multiplication table in Section
2 holds. In fact, this follows from the remarks on the
Cayley-Dickson construction at the end of Section 2.3.

It follows that given any two basic triples, there exists a unique automorphism of mapping the first to the second. Conversely, it is obvious that any automorphism maps basic triples to basic triples. This gives a nice description of the group , as follows.

Fix a basic triple . There is a unique automorphism
of the octonions mapping this to any other basic triple, say
. Now our description of basic triples so far has
been purely algebraic, but we can also view them more geometrically
as follows: a basic triple is any triple of unit imaginary
octonions (i.e.octonions of norm one) such that each is
orthogonal to the algebra generated by the other two. This means that
our automorphism can map to any point on the 6-sphere of
unit imaginary octonions, then map to any point on the
5-sphere of unit imaginary octonions that are orthogonal to , and
then map to any point on the 3-sphere of unit imaginary
octonions that are orthogonal to and . It follows
that

The triality description of the octonions in Section 2.4
gives another picture of . First, recall that is the
automorphism group of the triality
. To construct the octonions from this triality we need to
pick unit vectors in any two of these spaces, so we can think of
as the subgroup of fixing unit vectors in and .
The subgroup of fixing a unit vector in is just
, and when we restrict the representation to
, we get the spinor representation . Thus is the
subgroup of fixing a unit vector in . Since
acts transitively on the unit sphere in this spinor representation
[1], we have

It follows that

This picture becomes a bit more vivid if we remember that after choosing unit vectors in and , we can identify both these representations with the octonions, with both unit vectors corresponding to . Thus what we are really saying is this: the subgroup of that fixes in the vector representation on is ; the subgroup that fixes in both the vector and right-handed spinor representations is . This subgroup also fixes the element in the left-handed spinor representation of on .

Now, using the vector representation of on , we
get homomorphisms

where is the rotation group of the octonions, viewed as a real vector space with the inner product . The map from to is two-to-one, but when we restrict it to we get a one-to-one map

At the Lie algebra level, this construction gives an inclusion

where is the Lie algebra of skew-adjoint real-linear transformations of the octonions. Since is 14-dimensional and is 28-dimensional, it is nice to see exactly where the extra 14 dimensions come from. In fact, they come from two copies of , the 7-dimensional space consisting of all imaginary octonions.

More precisely, we have:

(a direct sum of vector spaces, not Lie algebras), where is the space of linear transformations of given by left multiplication by imaginary octonions and is the space of linear transformations of given by right multiplication by imaginary octonions [77]. To see this, we first check that left multiplication by an imaginary octonion is skew-adjoint. Using polarization, it suffices to note that

for all and . Note that this calculation only uses the alternative law, not the associative law, since and all lie in the algebra generated by the two elements and . A similar argument shows that right multiplication by an imaginary octonion is skew-adjoint. It follows that , and all naturally lie in . Next, with some easy calculations we can check that

and

Using the fact that the dimensions of the 3 pieces adds to 28, equation (4.1) follows.

We have seen that sits inside , but we can do better: it
actually sits inside . After all, every automorphism of the
octonions fixes the identity, and thus preserves the space of octonions
orthogonal to the identity. This space is just , so
we have an inclusion

where is the rotation group of the imaginary octonions. At the Lie algebra level this gives an inclusion

Since is 14-dimensional and
is 21-dimensional, it
is nice to see where the 7 extra dimensions come from. Examining
equation (4.1), it is clear that these extra dimensions must
come from the transformations in
that
annihilate the identity . The transformations that do this are
precisely those of the form

for . We thus have

where is the 7-dimensional space of such transformations.

We may summarize some of the above results as follows:

and we have

where the Lie brackets in and are built from natural bilinear operations on the summands.

As we have seen, has a 7-dimensional representation . In fact, this is the smallest nontrivial representation of , so it is worth understanding in as many ways as possible. The space has at least three natural structures that are preserved by the transformations in . These give more descriptions of as a symmetry group, and they also shed some new light on the octonions. The first two of the structures we describe are analogous to more familiar ones that exist on the 3-dimensional space of imaginary quaternions, . The third makes explicit use of the nonassociativity of the octonions.

First, both and are closed under the commutator. In
the case of , the commutator divided by 2 is the familiar **cross product** in 3 dimensions:

We can make the same definition for , obtaining a 7-dimensional analog of the cross product. For both and the cross product is bilinear and anticommutative. The cross product makes into a Lie algebra, but not . For both and , the cross product has two nice geometrical properties. On the one hand, its norm is determined by the formula

or equivalently,

where is the angle between and . On the other hand, is orthogonal to and . Both these properties follow from easy calculations. For , these two properties are enough to determine up to a sign. For they are not — but they become so if we also use the fact that lies inside a copy of that contains and .

It is clear that the group of all real-linear transformations of
preserving the cross product is just , which is also
the automorphism group of the quaternions. One can similarly show that
the group of real-linear transformations of preserving the
cross product is exactly . To see this, start by noting that any
element of preserves the cross product on , since the
cross product is defined using octonion multiplication. To show that
conversely any transformation preserving the cross product lies in
, it suffices to express the multiplication of imaginary octonions
in terms of their cross product. Using this identity:

it actually suffices to express the inner product on in terms of the cross product. Here the following identity does the job:

where the right-hand side refers to the trace of the map

Second, both and are equipped with a natural 3-form,
or in other words, an alternating trilinear functional. This is given
by

In the case of this is just the usual volume form, and the group of real-linear transformations preserving it is . In the case of , the real-linear transformations preserving are exactly those in the group . A proof of this by Robert Bryant can be found in Reese Harvey's book [50]. The 3-form is important in the theory of 'Joyce manifolds' [56], which are 7-dimensional Riemannian manifolds with holonomy group equal to .

Third, both and are closed under the associator. For this is boring, since the associator vanishes. On the other hand, for the associator is interesting. In fact, it follows from results of Harvey [50] that a real-linear transformation preserves the associator if and only if lies in . Thus the symmetry group of the associator is slightly bigger than : it is .

Now we must make an embarrassing admission: these three structures on
are all almost the same thing! Starting with the cross
product

we can recover the usual inner product on by equation (4.1). This inner product allows us to dualize the cross product and obtain a trilinear functional, which is, up to a constant, just the 3-form

The cross product also determines an orientation on (we leave this as an exercise for the reader). This allows us to take the Hodge dual of , obtaining a 4-form , i.e.alternating tetralinear functional

Dualizing yet again, this gives a ternary operation which, up to a constant multiple, is the associator:

We conclude this section with a handy explicit formula for all the
derivations of the octonions. In an associative algebra , any
element defines an **inner derivation**
by

where the bracket denotes the commutator . In a nonassociative algebra, this formula usually does not define a derivation. However, if is alternative, any pair of elements define a derivation by

where denotes the associator . Moreover, when is a normed division algebra, every derivation is a linear combination of derivations of this form. Unfortunately, proving these facts seems to require some brutal calculations [77].

© 2001 John Baez