The Dynkin diagram/Dynkin quiver D4 has an S3-symmetry group
(graphics grabbed from Wikipedia here)
This manifests itself in “triality” symmetries enjoyed by the objects that are labeled by D4 in an ADE-classification.
Specifically for the simple Lie group SO(8) corresponding to D4, triality is the statement that the two spin representations of the corresponding spin group $Spin(8)$ and the defining vector representations are all isomorphic, and in a nice way.
(Spin(7)-subgroups in Spin(8))
There are precisely 3 conjugacy classes of Spin(7)-subgroups inside Spin(8), and the triality group $Out(Spin(8))$ acts transitively on these three classes.
(Varadarajan 01, Theorem 5 on p. 6, see also Kollross 02, Prop. 3.3 (1))
(G2 is intersection of Spin(7)-subgroups of Spin(8))
The intersection inside Spin(8) of any two Spin(7)-subgroups from distinct conjugacy classes of subgroups (according to Prop. ) is the exceptional Lie group G2, hence we have pullback squares of the form
(Varadarajan 01, Theorem 5 on p. 13)
We have the following commuting diagram of subgroup inclusions, where each square exhibits a pullback/fiber product, hence an intersection of subgroups:
Here in the bottom row we have the Lie groups
Spin(5)$\hookrightarrow$ Spin(6) $\hookrightarrow$ Spin(7) $\hookrightarrow$ Spin(8)
with their canonical subgroup-inclusions, while in the top row we have
SU(2)$\hookrightarrow$ SU(3) $\hookrightarrow$ G2 $\hookrightarrow$ Spin(7)
and the right vertical inclusion $B \iota'$ is the delooping of one of the two non-standard inclusions, according to Prop. .
The square on the right is that from Prop. .
The square in the middle is Varadarajan 01, Lemma 9 on p. 10.
The statement also follows with Onishchik 93, Table 2, p. 144:
(Spin(5).Spin(3)-subgroups in SO(8))
The direct product group SO(3) $\times$ SO(5) together with the groups Sp(2).Sp(1) and $Sp(1) \cdot Sp(2)$, with their canonical inclusions into SO(8), form 3 conjugacy classes of subgroups inside SO(8), and the triality group $Out(Spin(8))$ acts transitively on these three classes.
Similarly:
(Spin(5).Spin(3)-subgroups in Spin(8))
The groups Spin(5).Spin(3), Sp(2).Sp(1) and $Sp(1) \cdot Sp(2)$, with their canonical inclusions into Spin(8), form 3 conjugacy classes of subgroups inside Spin(8), and the triality group $Out(Spin(8))$ acts transitively on these three classes.
In summary we have these subgroup inclusions
permuted by triality:
graphics grabbed from FSS 19, Sec. 3.3
Discussion of triality of subgroups of Spin(8) and SO(8):
Veeravalli Varadarajan, Spin(7)-subgroups of SO(8) and Spin(8), Expositiones Mathematicae Volume 19, Issue 2, 2001, Pages 163-177 (doi:10.1016/S0723-0869(01)80027-X, pdf)
Martin Čadek, Jiří Vanžura, On $Sp(2)$ and $Sp(2) \cdot Sp(1)$-structures in 8-dimensional vector bundles, Publicacions Matemàtiques Vol. 41, No. 2 (1997), pp. 383-401 (jstor:43737249)
Megan M. Kerr, New examples of homogeneous Einstein metrics, Michigan Math. J. Volume 45, Issue 1 (1998), 115-134 (euclid:1030132086)
Andreas Kollross, Prop. 3.3 of A Classification of Hyperpolar and Cohomogeneity One Actions, Transactions of the American Mathematical Society Vol. 354, No. 2 (Feb., 2002), pp. 571-612 (jstor:2693761)
See also
Wikipedia, Triality
Exposition with an eye to application in string theory:
Discussion in the foundations of M-theory:
Last revised on June 19, 2019 at 04:36:26. See the history of this page for a list of all contributions to it.