4 MARTIN VAN BEEK
[BMO12]. Note also that we manage to capture a large number of fusion systems
at odd primes associated to sporadic simple groups. Indeed, as can be witnessed in
the tables provided in [AH12], almost all of the p-fusion categories of the sporadic
simple groups at odd primes are either constrained, supported on an extraspecial
group of exponent pand so are classified in [RV04], or satisfy the hypothesis of
the Main Theorem.
It is surprising that in the conclusion of the Main Theorem there are so few exotic
fusion systems. It has seemed that, at least for odd primes, exotic fusion systems
were reasonably abundant. Perhaps an explanation for the apparent lack of exotic
fusion system is that the setup from the Main Theorem somehow reflects some
of the geometry present in rank 2 groups of Lie type. Additionally, we remark
that in the two exotic examples in the classification, the fusion systems are ob-
tained by “pruning” a particular class of essential subgroups, as defined in [PS21].
Indeed, these essential subgroups, along with their automizers, seem to resemble
Aschbacher blocks, the minimal counterexamples to the Local C(G, T )-theorem
[BHS06]. Most of the exotic fusion systems the author is aware of either have a
set of essentials resembling blocks, or are obtained by pruning a class of essentials
resembling blocks out of the fusion category of some finite group. For instance,
pearls in fusion systems, investigated in [Gra18] and [GP22], are the smallest ex-
amples of blocks in fusion systems.
The work we undertake in the proof of the Main Theorem may be regarded as a
generalization of some of the results in [AOV13], where only certain configurations
at the prime 2 are considered. There, the authors exhibit a situation in which a
pair of subgroups of the automizers of pairs of essential subgroup generate a sub-
system, and then describe the possible actions present in the subsystem, utilizing
Goldschmidt’s pioneering results in the amalgam method. With this in mind, we
provide the following corollary along the same lines which, at least with regards
to essential subgroups, may also be considered as the minimal situation in which
a saturated fusion system satisfies Op(F) = {1}.
Corollary A. Suppose that Fis a saturated fusion system on a p-group Ssuch
that Op(F) = {1}. Assume that Fhas exactly two essential subgroups E1and
E2. Then NS(E1) = NS(E2)and writing F0:= hNF(E1), NF(E2)iNS(E1),F0is a
saturated normal subsystem of Fand either
(i) F=F0is determined by the Main Theorem;
(ii) pis arbitrary, F0is isomorphic to the p-fusion category of H, where
F∗(H)∼
=PSL3(pn), and Fis isomorphic to the p-fusion category of G
where Gis the extension of Hby a graph or graph-field automorphism;
(iii) p= 2,F0is isomorphic to the 2-fusion category of H, where F∗(H)∼
=
PSp4(2n), and Fis isomorphic to the 2-fusion category of Gwhere Gis
the extension of Hby a graph or graph-field automorphism; or