FOCAL SURFACES OF FRONTS 3
Definition 2.1. Let 𝑓:(Σ, 𝑝) → R3be a 𝐶∞map germ and 𝑝a singular point of 𝑓. Then
(1) 𝑓is a cuspidal edge at 𝑝if 𝑓is A-equivalent to the germ (𝑢, 𝑣) ↦→ (𝑢, 𝑣2, 𝑣3)at
the origin.
(2) 𝑓is a swallowtail at 𝑝if 𝑓is A-equivalent to the germ (𝑢, 𝑣) ↦→ (𝑢, 4𝑣3+2𝑢𝑣, 3𝑣4+
𝑢𝑣2)at the origin.
(3) 𝑓is a cuspidal butterfly at 𝑝if 𝑓is A-equivalent to the germ (𝑢, 𝑣) ↦→ (𝑢, 5𝑣4+
2𝑢𝑣, 4𝑣5+𝑢𝑣2)at the origin.
(4) 𝑓is a cuspidal lips at 𝑝if 𝑓is A-equivalent to the germ (𝑢, 𝑣) ↦→ (𝑢, 3𝑣4+
2𝑢2𝑣2, 𝑣3+𝑢2𝑣)at the origin.
(5) 𝑓is a cuspidal beaks at 𝑝if 𝑓is A-equivalent to the germ (𝑢, 𝑣) ↦→ (𝑢, 3𝑣4−
2𝑢2𝑣2, 𝑣3−𝑢2𝑣)at the origin.
Here, two map germs 𝑓1, 𝑓2:(R2,0)→(R3,0)are A-equivalent if there exist diffeomor-
phism germs 𝜑:(R2,0) → (R2,0)on the source and Φ:(R3,0) → (R3,0)on the target
such that 𝑓2= Φ ◦𝑓1◦𝜑−1.
A cuspidal edge, a swallowtail and a cuspidal butterfly are non-degenerate front singular
points. Moreover, a cuspidal edge and a swallowtail are generic singularities of fronts in
R3, and a cuspidal lips/beaks and a cuspidal butterfly are generic corank one singularities
of one-parameter bifurcation of fronts (cf. [2,37]). Further, a cuspidal edge is a singular
point of the first kind, and a swallowtail and a cuspidal butterfly are of the admissible
second kind (cf. [21]). For these singular points, the following characterizations are
known.
Fact 2.2 ([13,14,18,33]).Let 𝑓:Σ→R3be a front and 𝑝∈Σa corank one singular
point of 𝑓. Then we have the following.
(1) 𝑓is a cuspidal edge at 𝑝if and only if 𝜂𝜆(𝑝)≠0.
(2) 𝑓is a swallowtail at 𝑝if and only if 𝑑𝜆(𝑝)≠0,𝜂𝜆(𝑝)=0and 𝜂𝜂𝜆(𝑝)≠0.
(3) 𝑓is a cuspidal butterfly at 𝑝if and only if 𝑑𝜆(𝑝)≠0,𝜂𝜆(𝑝)=𝜂𝜂𝜆(𝑝)=0and
𝜂𝜂𝜂𝜆(𝑝)≠0.
(4) 𝑓is a cuspidal lips at 𝑝if and only if 𝑑𝜆(𝑝)=0and det(H𝜆)( 𝑝)>0, that is, 𝜆
has a Morse type singularity of index zero or two at 𝑝.
(5) 𝑓is a cuspidal beaks at 𝑝if and only if 𝑑𝜆(𝑝)=0,𝜂𝜂𝜆(𝑝)≠0and det(H𝜆)( 𝑝)<
0, that is, 𝜆has a Morse type singularity of index one and 𝜂𝜂𝜆 ≠0at 𝑝.
Here, 𝜆is the signed area density function of 𝑓as in (2.1),𝜂is a null vector field, 𝜂𝜆
means the directional derivative of 𝜆in the direction 𝜂and det(H𝜆)is the Hessian of 𝜆.
2.2. Geometric invariants. We recall geometric invariants of fronts.
2.2.1. Geometric invariants of cuspidal edges. First we consider the case of cuspidal
edges. Let 𝑓:Σ→R3be a front, 𝜈its Gauss map and 𝑝a cuspidal edge of 𝑓. Let 𝛾(𝑡)
be a singular curve passing through 𝑝and 𝜂a null vector field of 𝑓. Then one can define
the following geometric invariants: the singular curvature 𝜅𝑠([34]), the limiting normal
curvature 𝜅𝜈([21,34]), the cuspidal curvature 𝜅𝑐([21]) and the cuspidal torsion 𝜅𝑡([20]).
We note that 𝜅𝑠is an intrinsic invariant and its sing has a geometrical meaning (see [9,34]).
It is known that 𝜅𝑐does not vanish when 𝛾consists of cuspidal edges ([20,21]). We remark
that these invariants can be defined at singular points of the first kind for frontals but not
fronts. In such cases, 𝜅𝑐vanishes at non-front singular points (cf. [21, Proposition 3.11]).
The limiting normal curvature 𝜅𝜈relates to the boundedness of the Gaussian curvature
of a front with a cuspidal edge. In fact, the Gaussian curvature 𝐾of a front 𝑓is bounded
on a sufficiently small neighborhood 𝑈of a cuspidal edge 𝑝if and only if 𝜅𝜈vanishes
along the singular curve 𝛾through 𝑝([21, Theorem 3.9]). In this case, 𝐾can be extended
as a 𝐶∞function on 𝑈. Moreover, the following property holds.