Rank Rigidity of Submanifolds and Normal Holonomy of Orbits 189
Proof of Theorem 5.2.7 The group
¯
G
p
K contains the normal holonomy group Φ
p
by equation (5.11) and Lemma 5.2.10. The opposite inclusion follows from Remark
5.2.9 and Corollary 5.2.6, so Φ
p
=
¯
G
p
K altogether. Moreover, by Remark 5.2.11, the
normal holonomy algebra at p coincides with k.Thisfinishes the proof.
Remark 5.2.12 Suppose M = K · p is an orbit of an s-representation and take the
reductive decomposition k = k
p
⊕m with m = k
⊥
p
. From Lemma 3.1.5 we know
[k
p
,m]
⊥
= 0. This means that A
X
= 0ifX ∈ m. Thus, Theorem 5.2.7 gives an al-
ternative proof of Theorem 3.1.7. Namely, the normal holonomy representation of M
coincides with the slice representation, that is, the (effectivized) action of the isotropy
group K
p
on the normal space
ν
p
M.
Remark 5.2.13 Theorem 5.2.7 is used in [141] for the geometric characterization
of orthogonal representations with copolarity one (and for their classification in the
irreducible case, see Remark 2.3.13 for the definition of copolarity).
We finish the chapter by formulating the following conjecture:
Conjecture 5.2.14 (Olmos [257]) Let M be a full homogeneous submanifold of the
sphere S
n−1
with dimM ≥ 2 and which is not an orbit o f an s-representation. Then
the normal holonomy group of M acts transitively on the sphere of the normal space
(in particular the normal holonomy group acts irreducibly).
For dim M = 2 the conjecture is true, since the normal holonomy group must
always be transitive on unit normal vectors by Theorem 3.5.2. For dimM = 3the
conjecture is also true by a result of Olmos and Ria˜no-Ria˜no ( [263, Theorem B]),
but the arguments are very involved, including delicate topological considerations.
The conjecture is also true if the normal holonomy group of M acts irreducibly and
the codimension is the maximal possible one, namely
1
2
n(n+1)−1 =
1
2
(n+2)(n−1)
with n = dimM ( [263, Theorem A]). This last result is also true if one replaces the
homogeneity condition by the assumption that M is a minimal submanifold of the
sphere (see [263, Theorem C]).
5.3 Exercises
Exercise 5.3.1 Let M be an isoparametric hypersurface of the sphere S
n
and let M
i
be a focal submanifold. Prove that (
ν
M
i
)
0
= {0}.
Exercise 5.3.2 Let g : [a,b]×[c,d] →M be a p iecewise differentiable map with vari-
ables s,t and M be an immersed submanifold o f a Riemannian manifold N. Assume
that R
⊥
(
∂
g
∂
s
,
∂
g
∂
t
)=0. Let, for i ∈{1, 2}, c
i
: [0,1] → [a, b] ×[c, d] be two piecewise
differentiable curves with c
1
(0)=c
2
(0) and c
1
(1)=c
2
(1). Prove that
τ
⊥
g◦c
1
=
τ
⊥
g◦c
2
,