3 2 P 3 P P P C
3 2 P 3 P P P C
3 2 P 3 P P P C
T. A. MEDINA-TEJEDA
1. Introduction
In recent years, there is a great interest in the geometry of a special type of
singular surface, namely, frontal. Many papers are dedicated to the study of frontals
from singularity theory and geometry viewpoints [8, 9, 11], in particular wave fronts
a subclass of these [1, 2, 3, 6]. The word ”front” comes from physical fronts,
bounding a domain in which a physical process propagates at a fixed moment
in time. For instance, a wave propagating in the 3-Euclidean space with constant
speed starting from each point of an ellipsoid in direction of the interior of this (the
initial domain to be perturbed) creates a equidistant surface at time t bounding
an interior part of the ellipsoid that it has not been perturbed at time t. In this
case, the complete equidistant surface is called the wave front, this changes as time
passes leading to the formation of singularities along the whole equidistant surface
in any time [1]. The notion of ”frontal” surged as a natural generalization of wave
front in the case of hypersurfaces and a generalized definition with equivalences
can be found in [9]. A smooth map x : U → R3 defined in an open set U ⊂ R2 is
called a frontal if, for all p ∈ U there exists a unit normal vector field ν : Vp → R3
along x, where Vp is an open set of U , p ∈ Vp . This means, |ν| = 1 and it is
orthogonal to the partial derivatives of x for each point (u, v) ∈ Vp . If also the
singular set Σ(x) = {p ∈ U : x is not immersive at p} has empty interior we call x
a proper frontal. Since Σ(x) is closed, this is equivalent to have Σ(x)c being dense
and open in U . A frontal x is a wave front or simply front if the pair (x, ν) is an
2010 Mathematics Subject Classification. Primary 53A40; Secondary 53A05, 57R45.
Key words and phrases. singular surface, frontal, front, relative curvature, first fundamental
form, second fundamental form, singular compatibility equations.
The author is supported by CAPES Grant no. PROEX-10359340/D.
1
2 T. A. MEDINA-TEJEDA
immersion for all p ∈ U . There are many examples of frontals which are not wave
fronts, cuspidal Sk singularities for instance [8]. The existence of a smooth normal
vector field on these singular surfaces determines a plane (the orthogonal space) at
singular points that can be understood as a limiting plane of the tangent planes on
regular points around them (see Figure 1).
The cuspidal edge and the swallowtail (see Figure 1 and 2) are two types of
singular points that represent the generic singularities in the space of wave fronts
with the Whitney C ∞ -topology. For this reason, all the re-parametrizations and
diffeomorphic singular surfaces to these are the most studied and there exist crite-
rias to recognize them [10, 11]. However, these singularities are not generic in the
space of all frontals (in fact proper frontals are not generic either)[9]. There are
some non-proper frontals which are not ”surfaces”, x(u, v) = (uv, 0, 0) for instance
and others whose entire image is a surface but locally at some singular points the
image of a neighborhood at these is a constant (see example 2.5 [9]). Here we treat
frontals in general, but our main result aim to proper frontals.
satisfy formally the Gauss and Mainardi-Codazzi equations, then for each p ∈ U
there exist a neighborhood V ⊂ U of p and a diffeomorphism x : V → x(V ) ⊂ R3
such that the regular surface x(U ) has E, F, G and e, f, g as coefficients of the first
and second fundamental forms, respectively. Furthermore, if U is connected and if
x̄ : U → x̄(U ) ⊂ R3 is another diffeomorphism satisfying the same conditions, then
there exist a translation T and a proper linear orthogonal transformation ρ in R3
such that x̄ = T ◦ ρ ◦ x.
Gauss equation:
(1) Γ212u − Γ211v + Γ112 Γ211 + Γ212 Γ212 − Γ211 Γ222 − Γ111 Γ212 = −EK
Mainardi-Codazzi equations:
(2a) ev − fu = eΓ112 + f (Γ212 − Γ111 ) − gΓ211
(2b) fv − gu = eΓ122 + f (Γ222 − Γ112 ) − gΓ212
Definition 2.1. We call moving base a smooth map Ω :U → M3×2 (R) in which
the columns w1 , w2 : U → R3 of the matrix Ω = w1 w2 are linearly independent
smooth vector fields.
Definition 2.2. We call a tangent moving base of x a moving base Ω = w1 w2
such that xu , xv ∈ hw1 , w2 i, where h, i denotes the linear span vector space.
Let x : U → R3 be a frontal with a global normal vector field ν : U → R3 .
Denoting the inner product by (·) and ()T the operation of transposing a matrix,
we set the matrices:
E F xu · xu xu · xv
(3a) I= :=
F G xu · xv xv · xv
e f ν · xuu ν · xuv
(3b) II = :=
f g ν · xuv ν · xvv
1
Eu (Fu − 12 Ev ) −1
1
Γ11 Γ211
(3c) Γ1 = := 1 2 I
Γ121 Γ221 2 Ev
1
2 Gu
1 1
1
Γ12 Γ212
2 Ev 2 Gu −1
(3d) Γ2 = := I
Γ122 Γ222 (Fv − 12 Gu ) 12 Gv
(3e) α := −IIT I−1
The matrices I and II in a non-singular point p ∈ U coincide with the matrix
representation of the first fundamental form and of the second fundamental form
respectively. Γ1 , Γ2 and α are defined in Σ(x)c , they are the Christoffel symbols
and the Weingarten matrix. Also observe that, we can compute these matrices in
this way:
(4a) I = DxT Dx
(4b) II = −DxT Dν
(4c) Γ1 = (DxTu Dx)I−1
(4d) Γ2 = (DxTv Dx)I−1
w1 ×w2
Let Ω = w1 w2 be a moving base, we denote by n := kw1 ×w2 k and we set the
matrices:
EΩ FΩ
(5a) IΩ = := ΩT Ω
FΩ GΩ
eΩ f1Ω
(5b) IIΩ = := −ΩT Dn
f2Ω gΩ
1 2
T11 T11
(5c) T1= 1 T2 := (ΩTu Ω)I−1
Ω
T21 21
1 2
T12 T12
(5d) T2= 1 2 := (ΩTv Ω)I−1
Ω
T22 T22
(5e) µ := −IITΩ I−1
Ω
THE FUNDAMENTAL THEOREM FOR SINGULAR SURFACES 5
Notice that, these last matrices coincide with I, II, Γ1 , Γ2 and α when Ω = Dx is
a moving base. Since n · w1 = 0 and n · w2 = 0, then we have −nu · w1 = n · w1u ,
−nv · w1 = n · w1v , −nu · w2 = n · w2u and −nv · w2 = n · w2v . Therefore,
n · w1u n · w1v
(6) IIΩ =
n · w2u n · w2v
Also, as nu , nv ∈ hw1 , w2 i, there exist real functions (µ̄ij ) i, j ∈ {1, 2} defined on
U , such that:
(7a) nu = µ̄11 w1 + µ̄12 w2
(7b) nv = µ̄21 w1 + µ̄22 w2
Then, Dn = Ωµ̄T , where µ̄ = (µ̄ij ). Thus, using (5b) IIΩ = −ΩT Dn =
−ΩT Ωµ̄T = −IΩ µ̄T , therefore µ̄ = −IITΩ I−1
Ω = µ and we have:
(8) Dn = ΩµT
By last, w1 and w2 are linearly independent, the positive-definite quadratic form
(·) restricted to hw1 , w2 i has IΩ = ΩT Ω as its matrix representation in the base
{w1 , w2 } and therefore det(IΩ ) > 0.
The following is a particular version of Frobenius theorem that can be found in
(appendix B[13]) or [14].
Theorem 2.3 (Frobenius). Let Θ, Ξ : U × V → Rn be smooth vector fields, where
U ⊂ R2 and V ⊂ Rn are open sets. Let (u0 , v0 ) ∈ U be a fixed point. Then for
each point p ∈ V the system of PDE:
∂x
(9a) = Θ(u, v, x(u, v)),
∂u
∂x
(9b) = Ξ(u, v, x(u, v)),
∂v
(9c) x(u0 , v0 ) = p,
has a unique smooth solution x : U0 → Rn defined on a neighborhood U0 of
(u0 , v0 ) ∈ U0 if and only if, it satisfies the compatibility condition:
∂Θ ∂Θ ∂Ξ ∂Ξ
(10) + Ξ= + Θ
∂v ∂x ∂u ∂x
Corollary 2.4. Let S, T : U → Mn×n (R) be smooth vector fields, where U is an
open set in R2 . Let (u0 , v0 ) ∈ U be a fixed point. Then for each point A ∈ GL(n)
the system of PDE:
∂G
(11a) = SG,
∂u
∂G
(11b) = TG,
∂v
(11c) G(u0 , v0 ) = A,
6 T. A. MEDINA-TEJEDA
∂S ∂T
(12) − + [S, T] = 0,
∂v ∂u
where [S, T] = ST − TS is the Lie bracket.
2
Proof. Identifying Mn×n (R) ≡ Rn and defining Θ(u, v, X) := SX and Ξ(u, v, X) :=
TX for X ∈ Mn×n (R), the compatibility condition (10) is equivalent to (12) and
by theorem 2.3 follows the result.
From now on, as we want to describe local properties and tangent moving bases
exist locally, we can suppose that we have a global tangent moving base for a
frontal restringing the domain if necessary. If x is a frontal and Ω a tangent
moving base of x, we denote Λ := DxT Ω(IΩ )−1 , λΩ := det(Λ) and TΩ as the
principal ideal generated by λΩ in the ring C ∞ (U, R). Thus, we have globally
Dx = ΩΛT , Σ(x) = λ−1 Ω (0)
and rank(Dx) = rank(Λ). Also, with a tangent
moving base Ω = w1 w2 given, we always choose as unit normal vector field
w1 ×w2
ν : U → R3 along x, the induced by Ω (i.e n = kw 1 ×w2 k
).
T
E F λ11 λ12 EΩ FΩ λ11 λ12
(13a) =
F G λ21 λ22 FΩ GΩ λ21 λ22
e f λ11 λ12 eΩ f1Ω
(13b) = ,
f g λ21 λ22 f2Ω gΩ
in which all the components are smooth real functions defined on U , EΩ > 0,
GΩ > 0, EΩ GΩ − FΩ2 > 0, rank(Dx) = rank(Λ), Σ(x) = λ−1Ω (0) and
From (3c) and (4c) we have DxTu Dx − 12 Iu = 21 A1 , then using that I = ΛIΩ ΛT ,
Dx = ΩΛT and developing derivatives,
1 1
(ΛΩTu + Λu ΩT )ΩΛT = (Λu IΩ ΛT + ΛIΩu ΛT + ΛIΩ ΛTu ) + A1 .
2 2
Substituting IΩ = ΩT Ω and IΩu = ΩTu Ω + ΩT Ωu , we can group and cancel similar
terms, getting
ΛB1 ΛT = ΛIΩ ΛTu − Λu IΩ ΛT + A1 .
8 T. A. MEDINA-TEJEDA
T
multiplying the equality by left side with 1 0 and by the right side with 0 1 ,
we obtain,
0 −τ1
−τ1 λΩ = Λ(1) ΛT(2) = Λ(1) IΩ ΛT(2)u − Λ(1)u IΩ ΛT(2) − (Ev − Fu )
τ1 0
and from it follows (14a). Setting the matrices:
0 −(Fv − Gu ) 0 −τ2
A2 := , B2 = := ΩTv Ω − ΩT Ωv
Fv − Gu 0 τ2 0
Observing that, DxTv Dx − 12 Iv = 21 A2 and proceeding similarly as before, it follows
(14b).
The conditions (14a) and (14b) in theorem 3.5 may seem kind of strange, but
we will see in proposition 3.14 why these are so important. Also these expressions
can be reduced depending on the type of Ω chosen. If we have a tangent moving
base of a frontal, we always can construct an orthonormal one applying Gram-
Schmidt orthonormalization, then the decompositions in theorem 3.5 are reduced
and follows easily the corollary:
Corollary 3.6. Let x : U → R3 be a frontal and Ω a orthonormal tangent moving
base of x, then the matrices defined by equations 3a and 3b have the following
decomposition:
T
E F λ11 λ12 λ11 λ12
(15a) =
F G λ21 λ22 λ21 λ22
e f λ11 λ12 eΩ f1Ω
(15b) = ,
f g λ21 λ22 f2Ω gΩ
in which all the components are smooth real functions defined on U , rank(Dx) =
rank(Λ), Σ(x) = λ−1Ω (0) and
Denoting Λ̂T := E−1 · · · E−1 E−1 ΛT and x = (a, b, c), we can multiply the last
m 2 1
1 0 0
equality by to get:
0 1 0
1 0
au av 1 0 0 1 0 0
D(a, b) = = Dx = 0 1 Λ̂T = I2 Λ̂T = Λ̂T .
bu bv 0 1 0 0 1 0
g1 g2
1 + g12 g1 g2
g1u g1v 1
IΩ̂ = 2 , IIΩ̂ = (1 + g12 + g22 )− 2 ,
g1 g2 1 + g2 g2u g2v
1
and since Dn = Ω̂µT with n = (−g1 , −g2 , 1)det(IΩ )− 2 , reasoning as before we get
1 1
that D(−g1 det(IΩ )− 2 , −g2 det(IΩ )− 2 ) = µT .
Example 3.9. The cuspidal cross-cap (see Figure 3) can be decomposed in this
way:
10 T. A. MEDINA-TEJEDA
on the other hand, since ΛT is an exact differential, ΛTu e2 = ΛTv e1 . Thus, ΩΛTu e2 =
ΩΛTv e1 and adding this equality to the above one, we get:
(ΩΛT )u e2 = Ωu ΛT e2 + ΩΛTu e2 = Ωv ΛT e1 + ΩΛTv e1 = (ΩΛT )v e1
Denoting by z1 and z2 the first and second columns of ΩΛT respectively, fixing
(u0 , v0 ) ∈ U and p ∈ R3 the last equality is equivalent to z2u = z1v , which is the
compatibility condition of the system:
(19a) xu = z 1
(19b) xv = z 2
(19c) x(u0 , v0 ) = p,
By theorem 2.3, this system of PDE has a solution x : V → R3 , V ⊂ U , V a
neighborhood of (u0 , v0 ). Therefore Dx = ΩΛT and by proposition 3.2, x : V → R3
is a frontal. Now, the first fundamental form is DxT Dx = ΛΩT ΩΛT = I as
1
we wished. Using that n = (−g1 , −g2 , 1)(1 + g12 + g22 )− 2 and (6), the second
1
fundamental form is ΛIIΩ = D(a, b)T D(g1 , g2 )(1 + g12 + g22 )− 2 .
Proposition 3.11. Let x : U → R3 be a frontal and Ω a tangent moving base of x,
then the matrices T 1 , T 2 satisfies IΩ T T1 + T 1 IΩ = IΩu and IΩ T T2 + T 2 IΩ = IΩv .
Proof. IΩu = ΩTu Ω + ΩT Ωu = ΩTu ΩI−1 −1 T T
Ω IΩ + IΩ IΩ Ω Ωu = T 1 IΩ + IΩ T 1 . For IΩv
is analogous.
Proposition 3.12. Let x : U → R3 be a proper frontal and Ω a tangent moving
base of x, then the Christoffel symbols defined on U − λ−1
Ω (0) have the following
decomposition:
Γ1 = (ΛT 1 + Λu )Λ−1 and Γ2 = (ΛT 2 + Λv )Λ−1
Where the right sides are restricted to the open set Σ(x)c . As Γ1 and Γ2 are
expressed in terms of E, F , G and these by (13a) are expressed in terms of EΩ ,
FΩ , GΩ and λij , then T 1 and T 2 can be expressed just using EΩ , FΩ , GΩ and λij
12 T. A. MEDINA-TEJEDA
2 0 2 0
u 0 T u 0
Dx = 0 2 = ΩΛ , where Ω = 0 2 , Λ = ,
0 v 0 v
3u 3v 3u 3v
1
being Ω a tangent moving base of x, then we have n = (−6u, −6v, 4)− 2 , w1u =
(0, 0, 3), w1v = (0, 0, 0), w2u = (0, 0, 0) and w2v = (0, 0, 3). Thus
1
!
4 + 9u2 12− 2
9uv 0
IΩ = , IIΩ =
9uv 4 + 9v 2 0
1
12− 2
where = 36u2 + 36v 2 + 16. Also, KΩ (u, v) = 144(36u2 + 36v 2 + 16)−2 6= 0 and
HΩ (0, 0) = 0, then by corollary 3.23, x is a front.
2
Example 3.25. Let x : R2 → R3 defined by x(u, v) := (ueu , v 2 , ( u2 + u)v 3 ),
this is a frontal with rank(Dx(p)) = 0 on p = (−1, 0) (Figure 5). We have the
decomposition:
eu
0
1+u 0
Dx = 0 2 = ΩΛT ,
3 u2 0 v
v ( 2 + u)3v
u
e 0
1+u 0
where Ω = 0
2 ,Λ = ,
2 0 v
v 3 ( u2 + u)3v
2 1
being Ω a tangent moving base of x, then we have n = (−2v 3 , −eu ( u2 +u)3v, 2eu )δ − 2 ,
2
w1u = (eu , 0, 0), w1v = (0, 0, 3v 2 ), w2u = (0, 0, (u + 1)3v) and w2v = (0, 0, 3( u2 +
18 T. A. MEDINA-TEJEDA
u)). Thus
2
!
e2u + v 6 3( u2 + u)v 4 −2v 3 eu 6eu v 2
1
IΩ = 2 2 , IIΩ = u u u2 δ− 2
3( u2 + u)v 4 4 + 9( u2 + u)2 v 2 6(1 + u)e v 6e ( 2 + u)
2
where δ = 4v 6 + e2u (9( u2 + u)2 v 2 + 4). Also, KΩ (−1, 0) = 0 and HΩ (−1, 0) = 0,
then by corollary 3.23, x is not a front.
(31a) Wu = WPT
(31b) Wv = WQT
which is equivalent to:
(32a) WuT = PWT
(32b) WvT = QWT
then, we have that P = WuT (WT )−1 = WuT WW−1 (WT )−1 = WuT W(WT W)−1
and Q = W T T −1 = WT WW−1 (WT )−1 = WT W(WT W)−1 . Considering
v (W ) v v
W = Ω n as a block matrix, we have:
T
ΩT
−1 Ωu
T T
Ω n )−1
(33) P = Wu W(W W) = Ω n ( T
nTu n
T −1 T −1
Ωu Ω ΩTu n
T
Ω Ω ΩT n Ωu Ω ΩTu n
IΩ 0
= =
nTu Ω nTu n nT Ω nT n nTu Ω 0 0 1
−1 −1
T T
T T
Ωu Ω Ωu n IΩ 0 Ωu ΩIΩ Ωu n
= =
nTu Ω 0 0 1 nTu ΩI−1
Ω 0
THE FUNDAMENTAL THEOREM FOR SINGULAR SURFACES 19
ΩTu n
T −1
Ωu ΩIΩ ΩTu n T1 T 1 ΩTu n
(34) P= = =
nTu ΩI−1
Ω 0 µT(1) ΩT ΩI−1
Ω 0 µT(1) 0
Finally, using (6) and by analogy with the same procedure for Q, we get:
1 2
T11 T11 eΩ
1 2
(35) P = T21 T21 f2Ω
µ11 µ12 0
1 2
T12 T12 f1Ω
1 2
(36) Q = T22 T22 gΩ
µ21 µ22 0
Substituting in (57), IΩu and IΩu by (51) and (52), we obtain canceling similar
terms that, the right side of (54) is equal to the right side of (56). Then, A is
skew-symmetric having the form
0 −a
A= ,
a 0
hence (53) is satisfied if and only if, a + b=0 as also, (47) can be expressed in this
form,
0 −a − b −1
(58) IΩ = 0.
a+b 0
Computing the component (1, 2) of (58) we have EΩ (−a − b) det(IΩ )−1 = 0, then
a + b = 0 if and only if, the component (1, 2) of (47) is satisfied, which is the
equation (38b) that is simplified to (39).
Proposition 4.2. Let IΩ , IIΩ , T 1 , T 2 : U → M2×2 (R) be arbitrary smooth maps
and IΩ symmetric non-singular. If we set µ := −IITΩ I−1
Ω and we have that
(59) IΩ T T1 + T 1 IΩ = IΩu ,
(60) IΩ T T2 + T 2 IΩ = IΩv ,
then, the equation (49) is satisfied if and only if, the equation (48) is satisfied.
that is satisfied if and only if, the resulting equation of multiply this by the right
side with IΩ is satisfied, in which we can later substitute (I−1 −1
Ω )u IΩ = −IΩ IΩu ,
−1
(IΩ )v IΩ = −I−1
Ω IΩv , factorize similar terms in both sides and get
T
E F λ11 λ12 EΩ FΩ λ11 λ12
(61a) =
F G λ21 λ22 FΩ GΩ λ21 λ22
e f λ11 λ12 eΩ f1Ω
(61b) =
f g λ21 λ22 f2Ω gΩ
in which all the components are smooth real functions defined in U , EΩ > 0, GΩ >
0, EΩ GΩ − FΩ2 > 0, λ−1
Ω (0) has empty interior and
E Ω FΩ EΩ FΩ
(62a) Λ(1)u ΛT(2) − Λ(1) ΛT(2)u + Ev − Fu ∈ TΩ
FΩ G Ω FΩ GΩ
EΩ FΩ T EΩ FΩ
(62b) Λ(1)v Λ(2) − Λ(1) ΛT(2)v + Fv − Gu ∈ TΩ ,
FΩ GΩ FΩ GΩ
where Λ = (λij ), λΩ = det(Λ) and TΩ is the principal ideal generated by λΩ in the
ring C ∞ (U, R). E, F, G, e, f, g formally satisfy the Gauss and Mainardi-Codazzi
equations for all (u, v) ∈ U − λ−1Ω (0). Then, for each (u0 , v0 ) ∈ U there exists a
neighborhood V ⊂ U of (u0 , v0 ) and a frontal x : V → x(V ) ⊂ R3 with a tangent
moving base Ω such that Dx = ΩΛT ,
EΩ FΩ eΩ f1Ω
IΩ = , IIΩ =
FΩ GΩ f2Ω gΩ
and the frontal x has E, F, G and e, f, g as coefficients of the first and second
fundamental forms, respectively. Furthermore, if U is connected and if
x̄ : U → R3 and Ω̄ : U → R3
are another frontal and a tangent moving base satisfying the same conditions, then
there exist a translation T and a proper linear orthogonal transformation ρ in R3
such that Ω̄ = ρΩ and x̄ = T ◦ ρ ◦ x.
Lemma 5.2. If we have:
(63a) Γ̄1 Λ̄ − Λ̄u = Λ̄T̄ 1
(63b) Γ̄2 Λ̄ − Λ̄v = Λ̄T̄ 2
in which Λ̄, T̄ 1 , T̄ 2 : U → Mn×n (R) and Γ̄1 , Γ̄2 : U − det−1 (Λ̄)(0) → Mn×n (R)
are smooth maps with int(det−1 (Λ̄)(0)) = ∅. Then,
T
By density of U − det−1 (Λ̄)(0), ĪΩ T̄ 1 + T̄ 1 ĪΩ = ĪΩu holds on U . The converse is
obtained in the same way.
Proof. Teorema 5.1(Existence). By proposition 3.14 there exist T 1 , T 2 : U →
M2×2 (R) smooth maps such that on (λ−1 c
Ω (0)) ,
(72a) T 1 = Λ−1 (Γ1 Λ − Λu ),
(72b) T 2 = Λ−1 (Γ2 Λ − Λv ).
Let us construct T̄ 1 and T̄ 2 as the matrices P and Q in (35) and (36) respectively,
using (5e), (5a) and (5b). By (72a), (72b) and since αΛ = µ on (λ−1 c
Ω (0)) (caused
by (61a) and (61b)) we have for all (u, v) ∈ (λ−1 c
Ω (0)) ,
Γ̄1 Λ̄ − Λ̄u = Λ̄T̄ 1 and Γ̄2 Λ̄ − Λ̄v = Λ̄T̄ 2
where,
1 1
−1
Γ111 Γ211 e
2 Eu (Fu − 2 Ev ) e E F 0
(74) Γ̄1 = Γ121
Γ221 f = 12 Ev 1
2 G u f F G 0
α11 α12 0 −e −f 0 0 0 1
1 1 1
−1
Γ122 α21
Γ12 2 Ev 2 Gu −f E F 0
(75) Γ̄2 = Γ212 Γ222 α22 = (Fv − 12 Gu ) 21 Gv −g F G 0
f g 0 f g 0 0 0 1
λ11 λ12 0
(76) Λ̄ = λ21 λ22 0
0 0 1
Let (u0 , v0 ) ∈ U , q ∈ R3 be fixed points and since EΩ GΩ − FΩ2 > 0 we can find z1 ,
z2 , z3 fixed vectors of R3 linearly independent and positively oriented such that
z1 · z1 = EΩ (u0 , v0 ), z1 · z2 = FΩ (u0 , v0 ), z2 · z2 = GΩ (u0 , v0 ), z3 · z3 = 1 and
z3 · zi = 0 for i = 1, 2. Consider the system of PDE,
(77a) WuT = T̄ 1 WT
(77b) WvT = T̄ 2 WT
(77c) W(u0 , v0 ) = z1 z2 z3
It is known in classical differential geometry that, the Gauss and Mainardi-Codazzi
equations are equivalent to Γ̄1v − Γ̄2u + [Γ̄1 , Γ̄2 ] = 0, then as this is satisfied, by
lemmma 5.2 T̄ 1v − T̄ 2u + [T̄ 1 , T̄ 2 ] = 0 in U which is the compatibility condition of
the above system of equations. By corollary 2.4, this system has a unique solution
W : V̄ → GL(3), where V̄ is a neighborhood of (u0 , v0 ). Since det(W(u0 , v0 )) > 0,
restricting V̄ if it is necessary, we can suppose that det(W) > 0 on V̄ . Setting the
matrices,
E F 0 EΩ FΩ 0
Ī := F G 0 , ĪΩ := FΩ GΩ 0
0 0 1 0 0 1
26 T. A. MEDINA-TEJEDA
(78) Y := WT W
then,
1 2
1 2
T11 T11 −1 T12 T12
T1= 1 T2
T
= (Ωu Ω)IΩ and T 2 = 1 T2 = (ΩTv Ω)I−1
Ω
T21 21 T22 22
n · w1u n · w1v eΩ f1Ω
IIΩ = =
n · w2u n · w2v f2Ω gΩ
Let us consider the system of PDE restricted to V̂ ,
(85a) xu = λ11 w1 + λ12 w2
(85b) xv = λ21 w1 + λ22 w2
(85c) x(u0 , v0 ) = q
As,
0 1 (ΛT 1 + Λu ) = 0 1 Γ1 Λ = 1 0 Γ2 Λ = 1 0 (ΛT 2 + Λv )
for (u, v) ∈ (λ−1 (0))c , then by density
0 1 (ΛT 1 + Λu ) = 1 0 (ΛT 2 + Λv )
on the entire U , as also, by (61b) ΛIIΩ is symmetric, then the singular compatibility
equations (44a), (44b) and (44c) are satisfied, which are the compatibility condition
of the system (85). Therefore by theorem 2.3, this system has a solution x : V →
x(V ) ⊂ R3 , where V ⊂ V̂ is a neighborhood of (u0 , v0 ). As Dx = ΩΛT , by
proposition 3.2, x is a frontal with Ω being a tangent moving base of it, satisfying
what we wished.
Proof. Teorema 5.1(Rigidity). Let x̄ : U → x̄(U ) ⊂ R3 be a frontal, U connected,
with Ω̄ a tangent moving base of x̄ satisfying the same conditions of x and Ω.
As IΩ = IΩ̄ , exists a rotation ρ ∈ SO(3) such that ρΩ(u0 , v0 ) = Ω̄(u0 , v0 ). Set
a := x̄(u0 , v0 ) − ρx(u0 , v0 ), x̂ := ρx + a and Ω̂ := ρΩ. Observe that, x̄(u0 , v0 ) =
x̂(u0 , v0 ), Ω̂(u0 , v0 ) = Ω̄(u0 , v0 ), Dx̂ = Ω̂ΛT , IΩ = IΩ̂ and IIΩ = IIΩ̂ (caused by
ρw1 × ρw2 = ρ(w1 × w2 )). Also by remark 3.13 T i = T̄ i = T̂ i . We want to prove
that x̄ = x̂ on U , so, let us define the set,
B := {(u, v) ∈ U : Ω̄(u, v) = Ω̂(u, v)}
B is not empty
and closed by continuity. For each (ū, v̄) ∈ B, as we saw in section
4, Ω̄ n̄ is a solution of the system:
(86a) WuT = PWT
(86b) WvT = QWT
(86c) W(ū, v̄) = Ω̄(ū, v̄) n̄(ū, v̄)
As the matrices P (35)
and Q (36) are constructed with the coefficients of IΩ , IIΩ
and T i , then Ω̂ n̂ is solution of the system as well and by uniqueness, Ω̂ = Ω̄
on a neighborhood of (ū, v̄). We have that B is open and since U is connected,
28 T. A. MEDINA-TEJEDA
References
0
[1] V. I. Arnol d, Singularities of caustics and wave fronts, volume 62 of Mathematics and its
Applications (Soviet Series). Kluwer Academic Publishers Group, Dordrecht, 1990.
[2] M. Kossowski, Realizing a singular first fundamental form as a nonimmersed surface in Eu-
clidean 3-space, J. Geom. 81 (2004), no. 1-2, 101–113.
[3] K. Saji, M. Umehara, and K. Yamada, The geometry of fronts, Ann. of Math. 169 (2009),
491–529.
[4] M. Hasegawa, A. Honda, K. Naokawa, K. Saji, M. Umehara and K. Yamada, Intrinsic prop-
erties of surfaces with singularities, Internat. J. Math. 26 (2015), no. 4, 1540008, 34 pp.
[5] L.F. Martins, K. Saji, K. Teramoto, Singularities of a surface given by Kenmotsu-type formula
in Euclidean three-space, arXiv:1804.01671.
[6] L. F. Martins, K. Saji, M. Umehara and K. Yamada, Behavior of Gaussian curvature and
mean curvature near non-degenerate singular points on wave fronts, Geometry and Topology
of Manifold, Springer Proc. Math. & Statistics, 2016, 247–282.
[7] S. Fujimori, K. Saji, M. Umehara and K. Yamada, Singularities of maximal surfaces, Math.
Z., 259(2008), no. 4, 827–848.
[8] K. Saji, Criteria for cuspidal Sk singularities and their applications. J. Gokova Geom. Topol.
GGT 4, 67-81 (2010).
[9] G. Ishikawa, Singularities of Frontals, arXiv:1609.00488v1.
[10] M. Kokubu, W. Rossman, K. Saji, M. Umehara and K. Yamada, Singularities of flat fronts
in hyperbolic 3-space, Pacific J. of Math. 221 (2005), 303–351.
[11] G. Ishikawa, Recognition Problem of Frontal Singularities, arXiv:1808.09594.
[12] M. P. Do Carmo, Differential Geometry of Curves and Surfaces, Prentice Hall, Inc, 1976.
[13] J. J. Stoker, Differential Geometry, Wiley-Interscience, New York, 1969.
[14] C. L. Terng, Lecture notes on Curves and Surfaces Part I, 2005, www.math.uci.edu/ cterng/.