GEOMETRY OF SURFACES IN R5 THROUGH PROJECTIONS

AND NORMAL SECTIONS

J. L. DEOLINDO SILVA, R. OSET SINHA

5
Abstract. We study the geometry of surfaces in R by relating it to the geom-
arXiv:2010.10976v1 [math.DG] 21 Oct 2020

4
etry of regular and singular surfaces in R obtained by orthogonal projections.
In particular, we obtain relations between asymptotic directions, which are not
5 4
second order geometry for surfaces in R but are in R . We also relate the
umbilic curvatures of each type of surface and their contact with spheres. We
6
then consider the surfaces as normal sections of 3-manifolds in R and again re-
late asymptotic directions and contact with spheres by defining an appropriate
umbilic curvature for 3-manifolds.

1. Introduction
The study of the differential geometry of manifolds in Euclidean spaces of di-
mension greater than 3 is a classical subject. In his seminal paper [14], Little in-
troduced certain objects to study immersed manifolds such as the curvature ellipse
for surfaces in R4 . However, it’s been in the last 20 years when the introduction
of Singularity Theory techniques has given a tremendous impulse to the subject
(see [5, 7, 11, 16, 17, 18], for example). As the dimension and codimension of
the immersed manifolds grow, deeper singularity theory concepts are needed. Be-
sides this, the attention has recently changed to singular manifolds Msing k ⊂ Rn ,
n > k ≥ 2, ([3, 4, 15, 23]) and the relation of their geometry with regular manifolds
k ⊂ Rn , n > k ≥ 2 ([1, 2, 6, 19, 20, 21]).
Mreg
Our main interest in this paper is the geometry of regular surfaces in R5 , i.e.
Mreg ⊂ R5 . This has been studied in [8], [16] and [22], however, our motivation
2

is slightly different. In [2] a commutative diagram between regular and singular

surfaces and 3-manifolds through projections and normal sections was established.
This diagram induces a commutative diagram between the corresponding curvature
loci. The curvature locus is the image in the normal space by the second fundamen-
tal form of the unit vectors in the tangent space and it contains all the second order
geometry of the manifold at a point. This means that the second order geometry of

2000 Mathematics Subject Classification. Primary 57R45; Secondary 53A05, 58K05.

Key words and phrases. surfaces in 5-space, singular surfaces in 4-space, 3-manifolds in 6-space,
projections, normal sections, umbilic curvature.
Work of J. L. Deolindo Silva partially supported by CAPES/JSPS Grant number
88887.357189/2019–00.
Work of R. Oset Sinha partially supported by MICINN Grant PGC2018-094889-B-I00.
1
2 J. L. DEOLINDO SILVA, R. OSET SINHA

all these manifolds is related. For example, in [2] a relation between the asymptotic
directions at a point p ∈ Mreg n ⊂ R2n and at π (p) ∈ M n 2n−1
u sing ⊂ R , where πu is
the orthogonal projection along the tangent direction u, is given. Namely, since the
second fundamental form and the height functions coincide, a direction is asymp-
2 ⊂ R4 projected
totic for p if and only if it is asymptotic for πu (p). In fact, for Mreg
2
to Msing ⊂ R3 it is shown in [1, 21] that a point p is hyperbolic/elliptic/parabolic
if and only if πu (p) is hyperbolic/elliptic/parabolic. However, for Mreg 2 ⊂ R5 , the

definition of asymptotic directions does not depend only on the second order geom-
2
etry, and for Msing ⊂ R4 a degenerate normal direction is binormal only if it is in
Ep , the vector space parallel to the affine space which contains the curvature locus.
It is therefore not clear how to relate asymptotic directions for this situation.
In the present study we show how a point p ∈ Mreg 2 ⊂ R5 where the second

fundamental form has rank 2 behaves like a point in a surface in R4 . More precisely,
there are distinguished asymptotic directions at p which coincide and have the
same properties as the asymptotic directions at πu (p), both when u is a tangent
direction (and the projection is a singular manifold) or a normal direction (and the
projection is regular). We also establish relations between the umbilic curvatures
of the surfaces in R5 and of their projected surfaces. We study the contact with
2
spheres of Msing ⊂ R4 and show that there exists a unique umbilic focal hypersphere
at p if and only if there exists a unique umbilic focal hypersphere at πu (p).
Surfaces in R5 can also be obtained as normal sections of 3-manifolds in R6 , but
how to relate the geometry has not been clear until now. As for the projections, we
relate asymptotic directions at a point in the 3-manifold with asymptotic directions
at the corresponding point in the normal section. We define an appropriate umbilic
curvature for 3-manifolds, study the contact with spheres using this invariant and
relate it to the contact with spheres of the surface in R5 obtained as a normal
section.
In Section 2 we establish the notation and preliminary constructions. In Section
3 we deal with Mreg 2 ⊂ R5 projected orthogonally along a tangent direction to
2 4
Msing ⊂ R . In Section 4 we study regular projections. Section 5 establishes
relations between the contact with spheres of Mreg 2 ⊂ R5 and M 2 4
sing ⊂ R . Finally,
Section 6 deals with Mreg 2 ⊂ R5 obtained as a normal section of M 3 ⊂ R6 .
reg
Aknowledgements: The authors thank their families for understanding, since
this work was developed mostly during confinement. The authors also thank Farid
Tari for useful conversations. The first author would like to express his gratitude
to the Universitat de València, where this work was partially carried out, for its
hospitality.

2. The geometry of regular surfaces in R4 and R5 , regular

3-manifolds in R6 , and singular surfaces in R4
In this section, we present some of the notation and basic aspects of regular
surfaces in Rn (n = 4, 5), regular 3-manifolds in R6 , and corank 1 surfaces in R4 .
 GEOMETRY OF SURFACES IN R5 THROUGH PROJECTIONS AND NORMAL SECTIONS 3

These subjects have been studied extensively but we mainly follow the approach
in [3, 5, 8, 10, 12, 14, 16].
2.1. Regular k-dimensional manifold in Rn . Given a smooth surface Mreg k ⊂

R , n > k ≥ 2 and f : U → R a local parametrisation of Mreg with U ⊂ R2

n n k

an open subset, let {e1 , . . . , en } be an orthonormal frame of Rn such that at any

k and {e
u ∈ U , {e1 (u), . . . , ek (u)} is a basis for Tp Mreg k+1 (u), . . . , en (u)} is a basis
k
for Np Mreg at p = f (u).
The second fundamental form of Mreg k at a point p is defined by II : T M k ×
p p reg
k → N M k given by II (u, w) = π (d2 f (u, w)), where π : T Rn → N M k
Tp Mreg p reg p 2 2 p p reg
is the canonical projection on the normal space. The curvature locus is the image
of the unit vectors in Tp Mreg k given by the map η : Sk−1 ⊂ T M k → N M k ,
p reg p reg
where η(u) = IIp (u, u). The second fundamental form of Mreg k at p along a normal

vector field ν is the bilinear map IIpν : Tp M k × Tp M k → R given by IIpν (v, w) =

hν, IIp (v, w)i.
2.1.1. Case k = 2: Taking w = w1 e1 + w2 e2 ∈ Tp Mreg 2 and (x, y) local coordinates

in R2 , li = hfxx , e2+i i, mi = hfxy , e2+i i and ni = hfyy , e2+i i, for i = 1, . . . , n − 2,

are called the coefficients of the second fundamental form. IIp is a quadratic form
which can be written as
n−2
X
IIp (w, w) = (li w12 + 2mi w1 w2 + ni w22 )e2+i .
i=1
The matrix of the second fundamental form with respect to the orthonormal frame
above is given by  
l1 m1 n1
αp =  ... .. ..  .

. . 
ln−2 mn−2 nn−2
2 is parametrized by θ ∈ [0, 2π], the curvature locus
If the unit circle in Tp Mreg
2 given by the image of the map
forms an ellipse in the normal space Np Mreg
n−2
X
(1) η(θ) = (li cos(θ)2 + 2mi cos(θ) sin(θ) + ni sin(θ)2 )e2+i .
i=1
It is denoted by ∆ne
and called the curvature ellipse. In particular, if u = cos(θ)e1 +
sin(θ)e2 belongs to unit sphere S 1 ⊂ R2 , we get IIp (u, u) = η(θ).

Case n = 4: The points on a surface Mreg 2 ⊂ R4 are classified according to the po-

sition of the point p with respect to the curvature ellipse ∆4e (Np Mreg 2 is viewed as

an affine plane through p). The point p is called elliptic/parabolic/hyperbolic if it is

inside/on/outside the curvature ellipse at p, respectively. Moreover, p is an inflec-
tion point if the curvature ellipse is a degenerate ellipse, i.e. rank αp = 1 ([14]). The
directions along θ such that the vector η(θ) is tangent to the curvature ellipse ∆4e
4 J. L. DEOLINDO SILVA, R. OSET SINHA

2 in R4 at p. There are 2/1/0/infinite asymp-

are called asymptotic directions of Mreg
totic directions at p depending on p being a hyperbolic/parabolic/elliptic/inflection
point.

Case n = 5: If Af fp denotes the affine space which contains the curvature el-
2 and the orthogonal
lipse, Ep is the vector subspace parallel to Af fp in Np Mreg
⊥
complement of Ep in Np M is denoted by Ep . The distance from p to Af fp is the
non-negative real number κreg
u called umbilic curvature ([8]).
2
Consider the subsets of Mreg

2
Mi = {p ∈ Mreg | rank αp = i}, i ≤ 3.

Points in these subspaces are called Mi -points and can be characterized in terms
of the relative positions of Ep and Af fp : If p ∈ M3 , Af fp is a plane that does
not contain the origin of Np Mreg2 . If p ∈ M , Af f is either a plane through the
2 p
2 (and thus coincides with E ), or a line that does not contain the
origin of Np Mreg p
2 . It is shown in [16] that for generically immersed surface in R5 ,
origin of Np Mreg
2
Mreg = M3 ∪ M2 , with M2 a regular curve on Mreg 2 . When p ∈ M , Af f is a
1 p
line through the origin p of Np Mreg2 , that is Af f = E .
p p
A point p ∈ Mreg 2 is called semiumbilic if the curvature ellipse ∆5 degenerates
e
to a line segment that does not contain p. In this sense, points in M2 are either
semiumbilics, or points at which Af fp passes through the origin p of Np Mreg 2 .
p
Moreover, a point p ∈ M2 can be classified as Me2 , Mh2 or M2 according to
whether the origin p of Np Mreg2 lies inside, outside, or on the curvature ellipse ∆5
e
at p. In particular, semiumbilic points can be considered as points of type Mh2 .
A geometrical characterization of points on Mreg 2 using singularity theory is car-

ried out in [16] via the family of height functions h : Mreg 2 × S 4 → R given by

h(p, ν) = hp, νi, where S 4 denotes the unit sphere in R5 . For ν fixed, the height
function hν (p) = h(p, ν) is singular if and only if ν ∈ Np Mreg 2 . It is shown in [16]

that for a generic surface, p ∈ M3 if and only if hν has only Ak -singularities for
any ν ∈ Np Mreg2 . A point p ∈ Mh ∪ Me (respectively, p ∈ Mp ) if and only if there
2 2 2
exists ν ∈ Np M such that hν has a singularity of type D4± (respectively, D5 ) at p.
This direction ν is called the flat umbilic direction and it is perpendicular to Ep .
([16], see also Remark 3.6).
The quadratic form IIpν can be identified with Hess(hν )(p) at a ν ∈ Np Mreg 2 ,

up to smooth local changes of coordinate in Mreg 2 ([5, 16, 22]). A unit vector
2
ν ∈ Np Mreg is called a degenerate direction if hν has a singularity of type A2 or
worse at p and it is a binormal direction if the singularity is of type A3 or worse at p.
A unit vector in u ∈ Tp Mreg2 is called a contact direction associated to a degenerate

direction ν if u ∈ ker(Hess(hν )(p)). For M3 -points, an asymptotic direction at p, is

a contact direction associated to the binormal direction ν. For M2 -points where hν
has an Ak≥3 -singularity, any contact direction associated to ν is called asymptotic.
 GEOMETRY OF SURFACES IN R5 THROUGH PROJECTIONS AND NORMAL SECTIONS 5

2 , but only
However, when ν is a flat umbilic direction ker(Hess(hv )(p)) = Tp Mreg
certain directions are called asymptotic (see Section 3 for details).
2.1.2. Case k = 3 and n = 6: In this case, the curvature locus is denoted by
∆v and is studied in [5]. Here, a direction u ∈ Tp Mreg 3 is called an asymptotic
3 3 such that II (u, v) =
direction of Mreg at p if there is a non zero vector ν ∈ Np Mreg ν
hII(u, v), νi = 0 for all v ∈ Tp Mreg 3 . Moreover, in such case, we say that ν is

a binormal direction. Equivalently, u is asymptotic if and only if there exists

ν ∈ Np Mreg 3 such that the height function h has a degenerate singularity and
ν
u ∈ ker Hess hν (see [10]). More details about Mreg 3 can be found in Section 6.

2.2. Singular surfaces in R4 . Let Msing 2 be a corank 1 surface in R4 , at p. The

2
singular surface Msing will be taken as the image of a smooth map g : M̃ → R4 ,
where M̃ is a smooth regular surface and q ∈ M̃ is a corank 1 point of g such
that g(q) = p. Also, consider φ : U → R2 a local coordinate system defined in an
open neighborhood U of q at M̃ . Using this construction, we may consider a local
parametrisation f = g ◦ φ−1 of Msing 2 at p.
2
The tangent line of Msing at p, Tp Msing 2 , is given by Im dg , where dg : T M̃ →
q q q
4 2
Tp R is the differential map of g at q. The normal space of Msing at p, Np Msing 2 , is
2 2
the subspace satisfying Tp Msing ⊕ Np Msing = Tp R . 4

The first fundamental form of Msing 2 at p, I : Tq M̃ × Tq M̃ → R is given by

I(u, v) = hdgq (u), dgq (v)i, ∀ u, v ∈ Tq M̃ and the second fundamental form of Msing 2

at p, II : Tq M̃ × Tq M̃ → Np M in the basis {∂x , ∂y } of Tq M̃ is given by

II(∂x , ∂x ) = π2 (fxx (φ(q))), II(∂x , ∂y ) = π2 (fxy (φ(q))), II(∂y , ∂y ) = π2 (fyy (φ(q)))
where π2 : Tp R4 → Np Msing
2 is the orthogonal projection and is extended to the
whole space uniquely as a symmetric bilinear map.
Given a normal vector ν ∈ Np Msing 2 , the second fundamental form along ν,

IIν : Tq M̃ × Tq M̃ → R is given by IIν (u, v) = hII(u, v), νi, for all u, v ∈ Tq M̃ .

Thus, if u = α∂x + β∂y ∈ Tq M̃ and fixing an orthonormal frame {ν1 , ν2 , ν3 } of
2 ,
Np Msing
3
X 3
X
II(u, u) = IIνi (u, u)νi = (α2 lνi (q) + 2αβmνi (q) + β 2 nνi (q))νi .
i=1 i=1

Let Cq ⊂ Tq M̃ be the subset of unit tangent vectors and let η : Cq → Np M be

the map given by η(u) = II(u, u). The curvature parabola of Msing 2 at p, denoted
by ∆p , is the subset η(Cq ). The curvature parabola is a plane curve, and it can
degenerate into a half-line, a line or even a point. This special curve plays a similar
role as the curvature ellipse ∆4e does for regular surfaces in R4 . In fact, in [1] there
is a relation between them.
The minimal affine space which contains the curvature parabola is denoted by
Af fp . The plane denoted by Ep is the vector space: parallel to Af fp when ∆p is
6 J. L. DEOLINDO SILVA, R. OSET SINHA

a non degenerate parabola, the plane through p that contains Af fp when ∆p is

a non radial half-line or a non radial line and any plane through p that contains
Af fp when ∆p is a radial half-line, a radial line or a point. The non negative real
number κsingu (p) = d(p, Af fp ) is called umbilic curvature.
A non-zero direction u ∈ Tq M̃ is called asymptotic if there is a non-zero vector
ν ∈ Ep such that IIν (u, v) = hII(u, v), νi = 0 for all v ∈ Tq M̃ . In such case, ν is said
be a binormal direction. A point in Msing 2 is hyperbolic/elliptic/parabolic/inflection
if it has 2/0/1/infinte asymptotic directions.
The normal directions ν ∈ Np Msing2 which are not in the plane Ep but also satisfy
the condition IIν (u, v) = hII(u, v), νi = 0, are called degenerate directions. The
subset of degenerate directions in Np Msing 2 is a cone and the binormal directions
are those in the intersection of this cone with Ep .
It is possible to take a coordinate system φ and make rotations in the target
in order to obtain f (x, y) = g ◦ φ−1 (x, y) = (x, f2 (x, y), f3 (x, y), f4 (x, y)), where
∂fi ∂fi
∂x (φ(q)) = ∂y (φ(q)) = 0 for i = 2, 3, 4. Taking an orthonormal frame {ν2 , ν3 , ν4 }
2 , the curvature parabola ∆ can be parametrized by
of Np Msing p
4
X
η(y) = (lνi + 2mνi y + nνi y 2 )νi .
i=2
where y ∈ R corresponds to a unit tangent direction u = (1, y) ∈ Cq .
A point p belongs to the subspace Mi if the rank of the second fundamental
form at p is i, i = 0, 1, 2, 3. Following [3], when ∆p is a non degenerate parabola,
p ∈ M2 or p ∈ M3 according to Af fp = Ep or not, respectively. In particular,
p ∈ M3 iff κsing
u (p) 6= 0. If ∆p is a half-line or a line, p ∈ M1 or p ∈ M2 depending
on ∆p being radial or not, respectively. Here, p ∈ M2 iff κsing u (p) 6= 0. Finally,
when ∆p is a point, p ∈ M0 or p ∈ M1 according to ∆p is p or not, respectively.
In this case, p ∈ M1 iff κsing
u (p) 6= 0.

3. Projecting surfaces in R5 along a tangent direction

2 be a point in a smooth surface in R5 and consider π : R5 → R4 the
Let p ∈ Mreg u
2 to R4 . The projection π (M 2 )
orthogonal projection along a direction u ∈ Tp Mreg u reg
2 4
is a singular surface Msing ⊂ R .
It is natural to ask whether there is a relation between the umbilic curvatures
κreg sing
u and κu .

Example 3.1. Consider a regular surface Mreg 2 whose 2-jet is given by (x, y, x2 , 2xy, y 2 )
2
and (X, Y, Z) the coordinates of Np Mreg . Its curvature ellipse is given by ηe (θ) =
2 cos2 (θ), sin(θ) cos(θ), sin2 (θ)). Here Af fp is the plane X + Z = 2 and, conse-
√
quently kureg (p) = 2 2. When projecting the surface along the tangent direction
u = (0, 1) we obtain a singular surface whose 2-jet of the parametrisation is A 2 -
equivalent to (x, xy, y 2 , 0) (i.e. equivalent by 2-jets of smooth changes of coordinates
in the source and target) and the curvature parabola is given by ηp (y) = 2 1, y, y 2 ).
 GEOMETRY OF SURFACES IN R5 THROUGH PROJECTIONS AND NORMAL SECTIONS 7

Here Af fπu(p) is the plane X = 2 and kusing (πu (p)) = 2, so Af fp 6= Af fπu(p) and
κreg sing
u 6= κu .

Definition 3.2. A point q ∈ Msing 2 ⊂ R4 is called semiumbilic if the curvature

parabola ∆p is a half-line or a line which is not radial.
Proposition 3.3. Let p ∈ Mreg 2 ⊂ R5 , u ∈ Tp Mreg
2 and πu be the orthogonal
projection such that πu (p) ∈ Msing ⊂ R . Let κu and κsing
2 4 reg
u be the corresponding
umbilic curvatures.
i) If p ∈ M3 , then κreg
u κu
sing
6= 0.
ii) If p ∈ M2 , the following are equivalent
1) κreg
u = κu
sing
= 0,
5
2) ∆e and ∆p are non-degenerate,
3) Both p and πu (p) are not semiumbilic.
The following are equivalent too
1’) κreg sing
u κu 6= 0
2’) ∆5e and ∆p are degenerate (not a point),
3’) Both p and πu (p) are semiumbilic.
Proof. First of all notice that p ∈ Mi if and only if πu (p) ∈ Mi , i = 0, . . . , 3.
For an M3 -point, Af fp and Af fπu (p) are planes that do not contain the origin,
so κreg
u and κu
sing
are non zero. Item ii) follows directly from the considerations in
Section 2 about the types of ellipses/parabolas in M2 -points and the definition of
umbilic curvature. 
Asymptotic directions on Mreg 2 ⊂ R5 can be described via the A -singularities of

the projections of Mreg 2 to 4-spaces (i.e. under the action of the group A ). If T S 4

denotes the tangent bundle of the 4-sphere S 4 , the family of projections to 4-planes
is given by
2 × S4 → T S4
Π : Mreg
(p, u) 7→ (u, πu (p))
where πu (p) = p−hp, uiu. For a given u ∈ S 4 , the map πu can be considered locally
as a germ of a smooth map R2 , 0 → R4 , 0. Then the generic A -singularities of πu
are those that have Ae -codimension less than or equal to 4 (which is the dimension
of S 4 ). These are listed in Table 1 (see [13]).

Table 1. Local singularities of projections of surfaces in R5 to 4-spaces.

Type Normal form Ae -codimension

Immersion (x, y, 0, 0) 0
Ik (x, xy, y 2 , y 2k+1 ), k = 1, 2, 3, 4 k
II2 (x, y 2 , y 3 , xk y), k = 2 3
III2,3 2 3 k l
(x, y , y ± x y, x y), k = 2, l = 3 4
VII1 (x, xy, xy 2 ± y 3k+1 , y 3 ), k = 1 4
8 J. L. DEOLINDO SILVA, R. OSET SINHA

Following [22], for a generic surface Mreg2 ⊂ R5 , a direction u ∈ T M 2 , with

p reg
p ∈ M3 , is asymptotic if and only if the projection of Mreg2 along u to a transverse

4-space has an A -singularity of type I2 or worse. At generic M3 -points, there are

at most 5 and at least 1 asymptotic directions u where πu has an A -singularity of
type I2 , which correspond to Ak≥3 -singularities of the height functions hν along the
associated binormal direction ν. If p is a generic M2 -point, then there at most 3 and
at least 1 asymptotic directions where πu has an A -singularity of type I2 . These
are dual to the flat umbilic normal direction. There are also two special asymptotic
directions (respectively, none) where πu has an A -singularity of type II2 if p ∈ Mh2
(respectively, p ∈ Me2 ), and one direction where πu has an A -singularity of type
VII1 if p ∈ Mp2 . These directions correspond to Ak≥3 -singularity of the height
functions along the binormal direction associated to u.
Proposition 3.4. For M3 -points πu (p) is a singularity of type Ik (even for non
generic surfaces). In particular, u is asymptotic if and only if πu (p) is a singularity
of type Ik , k > 1.
Proof. For generic surfaces this is proved in Theorem 3.6 in [22] by direct com-
putation. If p and πu (p) belong to M3 , then ∆5e and ∆p are a non-degenerate
ellipse and parabola, respectively. By Theorem 3.6 in [3], the curvature parabola
for πu (p) is non-degenerate if and only if j 2 πu f (0) ∼A 2 (x, y 2 , xy, 0), and the only
singularities in Table 3 in this 2-jet class are the Ik singularities. The rest follows
by Theorem 3.6 in [22]. 
Definition 3.5. For Mreg 2 ⊂ R5 a direction u ∈ T M 2 is called A -asymptotic if
p reg k
it is a contact direction associated to a binormal direction ν such that hν has an
Ak≥3 -singularity. If it is associated to a binormal direction such that hν has an
Dk -singularity (the flat umbilic) then it is called Dk -asymptotic.
With this definition, for p ∈ Mh2 the Dk -asymptotic directions u are those such
that πu (p) has an Ik -singularity and the Ak -asymptotic directions are those such
that πu (p) is an II2 -singularity.
Remark 3.6. Following [8], the cone of degenerate directions Cp at a point p ∈ Mh2
is two planes which intersect in a line which is precisely ker αp ⊆ Ep⊥ . At a point
p ∈ Mp2 it is a plane containing ker αp and at a point p ∈ Me2 it is exactly ker αp .
By Corollary 6 in [16] the flat umbilic direction is perpendicular to Ep and lies in
ker αp . All the degenerate directions in a given plane of Cp which are not the flat
umbilic have a same contact direction. This is because to be a degenerate direction
we need that hν has a singularity of type A2 or worse, i.e. det Hess(hν ) = 0,
and to be the associated contact direction we need that u ∈ ker(Hess(hν )). All
those degenerate directions have the same Hessian, but only one will be binormal.
This means that the unique contact direction associated to all the non-flat umbilic
degenerate directions in a given plane of Cp is an Ak -asymptotic direction.
Proposition 3.7. Let p ∈ M2 . The unitary tangent vector u is an Ak -asymptotic
direction if and only if ηe (u) is tangent to ∆5e .
 GEOMETRY OF SURFACES IN R5 THROUGH PROJECTIONS AND NORMAL SECTIONS 9

Proof. If p ∈ Mh2 (resp. Me2 and Mp2 ) then ∆5e is a non-degenerate ellipse with p
lying outside the ellipse (resp. inside and on the ellipse), and Af fp = Ep is a plane
passing through the origin that contains the image of the second fundamental form.
Let S 1 ⊂ Tp Mreg
2 be the unit sphere parametrized by θ and u = u(θ). We have
∂ηe (u)
∂θ = II(u, u)θ = 2II(u, uθ ). Since {u, uθ } is linearly independent, the tangency
happens if and only if {II(u, u), 2II(u, uθ )} is linearly dependent. This happens
if and only if there exists a non zero w ∈ Tp Mreg 2 such that II(u, w) = 0, i.e.

w ∈ ker II(u, ·). This means that the image of the map II(u, ·) lies in a line in Ep
and so there exists ν ∈ Ep such that hII(u, w), νi = 0 for all w ∈ Tp Mreg 2 , and this

is equivalent to u ∈ ker(Hess(hν )) with ν ∈ Ep .

We have proved that ηe (u) is tangent to ∆5e if and only if there exists a degenerate
direction ν ∈ Ep such that u is its associated contact direction. By the above
remark, u is in fact an Ak -asymptotic direction and ν is the intersection of Cp with
Ep . 
Remark 3.8. In this sense, the Ak -asymptotic directions at M2 -points play the
role of asymptotic directions at points in regular surfaces in R4 (this will be stud-
ied further in the next section). The cone Cp intersects the plane Ep in exactly
two directions which are orthogonal to the tangent rays through the origin to the
curvature ellipse (Kp in [8]). The Ak -asymptotic directions are dual to the two
directions in the intersection of Cp with Ep and, similarly to the case of surfaces in
R4 , by the Proposition 3.7, the image of the asymptotic directions by the second
fundamental form is tangent to the curvature ellipse, i.e. these images are precisely
the rays Kp .
2 is a generic surface, then
Theorem 3.9. Suppose Mreg
i) p ∈ Mh2 if and only if πu (p) is a hyperbolic point,
ii) p ∈ Me2 if and only if πu (p) is an elliptic point,
iii) p ∈ Mp2 if and only if πu (p) is a parabolic point.
Moreover, v is an Ak -asymptotic direction at p if and only if v is asymptotic for
πu (p). That is, the two (resp. one) Ak -asymptotic directions for a point p ∈ Mh2
(resp. p ∈ Mp2 ) are exactly the only two (resp. one) asymptotic directions at πu (p).
Proof. We consider Mreg 2 ⊂ R5 parametrized in Monge form, i.e. f (x, y) =

(x, y, g1 (x, y), g2 (x, y), g3 (x, y)), where ∂g ∂gi

∂x (p) = ∂y (p) = 0 for i = 1, 2, 3. By
i

rotation in the normal space generated by [ν3 , ν4 , ν5 ] we can take Af fp to be the

plane ν5 = 0. The 2-jet of the parametrization now has the form (0, 0, 21 (a20 x2 +
2a11 xy + a02 y 2 ), 21 (b20 x2 + 2b11 xy + b02 y 2 ), 0). The curvature ellipse ∆5e can be
parametrized by
ηe (θ) = a20 cos(θ)2 + 2a11 sin(θ) cos(θ) + a02 sin(θ)2 ,

(2)
b20 cos(θ)2 + 2b11 sin(θ) cos(θ) + b02 sin(θ)2 , 0 .
Now, by rotation in the tangent space we can take u to be (0, 1). These rotations
will affect the projection but the topological type and position with respect to
10 J. L. DEOLINDO SILVA, R. OSET SINHA

the origin of the curvature parabola will remain the same. By [1, Proposition
3.8], the curvature parabola of the projection along the tangent direction (0, 1) is
parametrized by
ηp (y) = (a20 + 2a11 y + a02 y 2 , b20 + 2b11 y + b02 y 2 , 0).
Suppose first that u is a non-asymptotic direction or a Dk -asymptotic direction,
i.e. an asymptotic direction such that the projection πu (p) is an Ik singularity.
Then ∆πu (p) is a non-degenerate parabola. The point p lies on the curvature ellipse
if the resultant of the two quadratic polynomials given by the first two components
of ηe (θ) is 0, i.e. they have a common root. Similarly the point lies outside/inside
the ellipse if the resultant is > 0/ < 0. This resultant is the discriminant used to
distinguish hyperbolic/elliptic/parabolic points of regular surfaces in R4 ([14, 17]).
Notice that the same resultant gives the common roots of the quadratic polynomials
in ηp (y) and so the point p lies inside/outside/on the ellipse if and only if the point
πu (p) lies inside/outside/on the parabola. The result follows.
Suppose now that u is an Ak -asymptotic direction. We distinguish into cases:
If p ∈ Mh2 then, by Proposition 3.9 in [22] there are exactly 2 Ak -asymptotic
directions ui , i = 1, 2, in which the projection has an II2 singularity. Therefore
j 2 πui f (0) ∼A 2 (x, y 2 , 0, 0) and by Theorem 3.6 in [3] the curvature parabola is a
half-line. Since the point is in M2 , this half line is not radial and therefore has 2
asymptotic direction, i.e. the projection is a hyperbolic point.
If p ∈ Me2 there are no Ak -asymptotic directions.
Finally, if p ∈ Mp2 and u is the only Ak -asymptotic direction where the projection
is an VII1 singularity, the 2-jet of the projection is equivalent to (x, xy, 0, 0) and
the curvature parabola is a non-radial line, which means that there is only one
asymptotic direction and the projection is a parabolic point.
Now, when projecting along u, as seen above, Ep = Eπu (p) . The degenerate
direction ν dual to the Ak -asymptotic direction v which lies in Ep may be binormal
or not (depending on whether the singularity of hν is of type Ak≥3 or A2 ). In any
case, since the height functions for p ∈ Mreg 2 and for π (p) ∈ M 2
u sing coincide, ν
is a binormal direction at πu (p). The associated contact directions are therefore
asymptotic at πu (p) and coincide with the Ak -asymptotic directions at p. 

4. Projecting surfaces in R5 along a normal direction to 4-space

Consider ν ∈ Np Mreg2 and let S be the surface patch obtained by projecting M 2
ν reg
orthogonally along ν. The projection is a regular surface in R4 . We are interested
in establishing some relation between the special directions in Mreg 2 with S . In
ν
2 ⊂ R5
this sense, since the definition of binormal (or asymptotic) directions in Mreg
depend on higher (than second) order singularities of the height function, we need
to consider higher order terms of Mreg 2 .

Following [22], we choose local coordinates at p and consider the surface taken
in Monge form
(3) (x, y, Q1 + f 1 (x, y), Q2 + f 2 (x, y), Q3 + f 3 (x, y)),
 GEOMETRY OF SURFACES IN R5 THROUGH PROJECTIONS AND NORMAL SECTIONS 11

where Q = (Q1 , Q2 , Q3 ) = (x2 , xy, y 2 ) at M3 points, Q = (x2 , y 2 , 0) at Mh2 points,

Q = (x2 , xy, 0) at Mp2 points, Q = (x2 − y 2 , xy, 0) at Me2 points, and
X X X
f 1 (x, y) = aij xi y j , f 2 (x, y) = bij xi y j , f 3 (x, y) = cij xi y j .
i+j≥3 i+j≥3 i+j≥3

Consider a unit normal vector ν = (ν3 , ν4 , ν5 ). Let us analyze the projection of

points p ∈ M3 ∪ M2 along ν:

M3 -points case: In this case, we take the Monge form as in (3). Without loss
of generality, we assume ν = (0, 0, 1) belongs to a unit sphere S 2 ⊂ R3 and
(Z, W, T ) the coordinates in R3 of Np Mreg 2 . We parametrize the directions near
p
to ν byp η = (ν3 , ν4 , 1 − ν32 − ν42 ). Instead of the regular projection to the plane
(ν3 , ν4 , 1 − ν32 − ν42 )⊥ , we project to the fixed plane (Z, W ). The modified family
of the orthogonal projections is given by
π: (R2 × R2 , 0) → (R4 , 0)
((x, y), (ν3 , ν4 )) 7→ πη (x, y),
where
 
πη (x, y) = x, y, x2 + f 1 (x, y) − ν3 (y 2 + f 3 (x, y)), xy + f 2 (x, y) − ν4 (y 2 + f 3 (x, y))

and πν (x, y) = (x, y, x2 +f 1 (x, y), xy+f 2 (x, y)) are parametrizations of the surfaces
Sη and Sν , respectively. Moreover, at generic points, by [22], the Ak -asymptotic
directions u = (u1 , u2 ) are given as solutions of the following equation
c30 u51 − (2b30 − c21 )u41 u2 + (c12 − 2b21 + a30 )u31 u22 + (c03 − 2b12 + a21 )u21 u32
(4)
+(a12 − 2b03 )u1 u42 + a03 u52 = 0.
For any normal direction ν (not necessarily a binormal direction), we can associate
the number of Ak -asymptotic directions with the geometry of Sη and Sν as follows.
Proposition 4.1. With the above conditions, given p ∈ M3 , we have the following:
i) πη (p) is a hyperbolic point Sη if and only if ν42 − ν3 < 0.
ii) πη (p) is an elliptic point Sη if and only if ν42 − ν3 > 0.
iii) πη (p) is a parabolic point Sη if and only if ν42 − ν3 = 0. There are at most
5 and at least 1 normal directions η where the surface Sη has a P3 (c)-point
at the origin. In addition, the number of the Ak -asymptotic directions in
M3 -points and the number of the points where the asymptotic direction of
Sη is tangent to its parabolic set are the same. These are the solutions of
the following equation
c30 ν45 − (2b30 − c21 )ν44 + (c12 − 2b21 + a30 )ν43 + (c03 − 2b12 + a21 )ν42
(5)
+(a12 − 2b03 )ν4 + a03 = 0.
Moreover, the unique asymptotic direction at P3 (c)-point in Sν is an Ak -
5 .
asymptotic direction in Mreg
12 J. L. DEOLINDO SILVA, R. OSET SINHA

Proof. Consider (x, y, x2 −ν3 y 2 , xy−ν4 y 2 ) the 2-jet of πη at the origin. The resultant
of the last two coordinates of j 2 πη (0) is ∆ = ν42 − ν3 , so a point at surface is
hyperbolic/elliptic/parabolic if and only if ∆ </>/= 0. The other part of the
proof of statement iii) follows by making changes of coordinates in order to reduce
the appropriate jet of πη . When projecting a regular surface in R4 along the
unique asymptotic direction u at a parabolic point, the projection has a P3 (c)-
singularity at isolated points on the parabolic set, i.e. it is A -equivalent to (x, y) 7→
(x, x2 y + cy 4 , xy + y 3 ) with c ∈ R \ {0, 21 , 1, 32 } (see [9]). In particular, the unique
asymptotic direction is tangent to the parabolic set of Sη at a P3 (c)-point. As the
discriminant of (4) coincides with the discriminant of (5), we have the statement.
Moreover, the projection along ν = (0, 0, 1) gives a regular surface Sν which has
a P3 (c)-point if and only if a03 = 0. In this case, the unique asymptotic direction
u = (0, 1) at πν (p) ∈ Sν is an Ak -asymptotic direction at p in Mreg 5 . 
M2 -points case: In [3, Theorem 4.14], the authors associate a regular surface in
R4 to any singular surface in R4 . To do so they consider the regular surface in R5
that gives the singular surface in R4 when projected along a tangent direction u
2 ⊕ π −1 (E
and project it to the 4-space given by Tp Mreg u πu (p) ). This regular surface
⊥
is Sν when ν ∈ Ep (see the following diagram).
2 ⊂ R5
Mreg
:: ▲▲▲
✈✈ ▲▲π▲ν
i ✈✈✈ πu
✈ ▲▲▲
✈✈  ▲&&
φ ✈✈ g
R2 oo M̃ // M 2 ⊂ R4 Sν ⊂ R 4
f
77 sing

The authors prove that πu (p) ∈ Msing 2 is an elliptic/hyperbolic/parabolic if and

only πν (p) ∈ Sν is an elliptic/hyperbolic/parabolic, respectively. Furthermore, a
2
direction v is an asymptotic direction of Msing if and only if it is also an asymptotic
direction of the associated regular surface Sν .
2 , when ν ∈ E ⊥ it is a flat umbilic direction.
In particular, at M2 -points of Mreg p
Then we have the following.
Proposition 4.2. Let p ∈ M2 ⊂ Mreg 2 ⊂ R5 and consider a direction of projection
⊥ 2
ν in Ep . A direction v ∈ Tp Mreg is Ak -asymptotic at p if and only if v is asymptotic
at πν (p) in Sν . Moreover, the second order geometry of Mreg 2 and S is the same.
ν

Proof. The proof follows from Theorem 3.9 and Theorem 4.14 in [3]. The second
2 and
order geometry is the same because the second fundamental forms at p ∈ Mreg
at πν (p) ∈ Sν coincide. 
From Proposition 4.2 we obtain as a corollary [16, Theorem 7] with some extra
information about the asymptotic directions:
Corollary 4.3. If p ∈ M2 and the normal direction ν ∈ Ep⊥ , then
i) p ∈ Mh2 if and only if πν (p) is a hyperbolic point of Sν .
 GEOMETRY OF SURFACES IN R5 THROUGH PROJECTIONS AND NORMAL SECTIONS 13

ii) p ∈ Me2 if and only if πν (p) is an elliptic point of Sν

iii) p ∈ Mp2 if and only if πν (p) is a parabolic point of Sν .
Moreover, the Ak -asymptotic directions at p coincide with the asymptotic directions
at πν (p).
Remark 4.4. The previous two results show further how M2 -points of surfaces in
R5 behave as points of regular surfaces in R4 .
Now, we project M2 -points along binormal directions ν that are associated to
Ak -asymptotic directions at p. We have the following result which characterizes
when a binormal direction lies in Cp ∩ Ep .
Proposition 4.5. Let p ∈ M2 ⊂ Mreg 2 and ν be a binormal direction which is

not a flat umbilic. Then ν ∈ Ep if and only if the regular surface Sν obtained by
2 along ν has an inflection point at π (p).
projecting Mreg ν

Proof. Consider M2 ⊂ Mreg 2 in Monge form as in (3) at p. If ν is not a flat umbilic,

then it is associated to an Ak -asymptotic direction. At Mh2 -points, the two Ak -

asymptotic directions are dual to the binormal directions ν1 = (c03 , 0, −a03 ) and
ν2 = (0, c30 , −b30 ). At Mp2 -points, the unique Ak -asymptotic direction is associ-
ated to the dual direction ν1 . There is no Ak -asymptotic direction at Me2 -points.
Consider Sν the regular projection along the binormal direction ν = (c03 , 0, −a03 )
(the case ν = (0, c30 , −b30 ) is analogous). The projection for a point in Mh2 (resp.
Mp2 ) is locally the regular surface Sν parametrized by
(x, y, −a03 (Q1 + f 1 (x, y)) − c03 (Q3 + f 3 (x, y)), Q2 + f 2 (x, y)).
Since the discriminant of Sν is ∆ = a203 > 0 (resp. = 0) (see [14]), we get that given
p ∈ Mh2 (resp. Mp2 ), a point in Sν is a hyperbolic (resp. parabolic) if and only if
a03 is non zero. On the other hand, a03 = 0 if and only if πν (p) is an inflection
point. In particular, a03 = 0 if and only if ν ∈ Ep . 

5. Contact with spheres of regular surfaces in R5 and singular

surfaces in R4
In [8], some aspects of the geometry of regular surfaces in R5 are given by the
2
family of distance squared functions on Mreg
2 × S4 → R
D : Mreg
(p, ν) 7→ ||p − ν||2 = dν (p).
A distance squared function dν has a singularity at p ∈ Mreg 2 if and only if ν ∈
2 2
Np Mreg . A focal center at p ∈ Mreg is a point at which dν has a degenerate
singularity (i.e., dν has a singularity A -equivalent to A2 or worse). The directions
lying in the kernel of the corresponding Hessian quadratic form are said to be
spherical contact directions at p. A focal center is said to be an umbilical focus at
p if dν has a singularity of corank 2 at p. The corresponding focal hypersphere
is called umbilical focal hypersphere. The set of the focal centres of the points of
14 J. L. DEOLINDO SILVA, R. OSET SINHA

2 is said to be the focal set of M 2 and denoted by F. At p ∈ M , there

Mreg reg 3
1
exists a unique umbilical focus ν = p + reg vp with vp an unit normal vector in
ku
Ep⊥ . If p ∈ M2 is not semiumbilic, the umbilical focus lies at infinity and the
umbilical focal hypersphere becomes an osculating hyperplane, and in the opposite
case, there is a line of umbilical foci at p contained in the vector plane orthogonal
to the affine line Ep ([8]).
2
For Msing ⊂ R4 , the definitions of focal set and umbilical focus at a singular
point p can be extended in a natural way from those of Mreg 2 ⊂ R5 . We also
2
call a frame [ν2 , ν3 , ν4 ] an adapted frame of Np Msing if it is a positively oriented
orthonormal frame, ν4 ∈ Ep⊥ (Ep is always a plane for Msing 2 ⊂ R4 ), [ν2 , ν3 ] is a
positively oriented frame of Ep and: ν2 is parallel to the direction of ∆p when ∆p
is a line or a half-line; ν3 is the direction of ∆p when ∆p is a point different from p.
Theorem 5.1. Let Msing 2 ⊂ R4 be a surface with a singularity of corank 1 at
2
p ∈ Msing 2 . The function d is
and let [ν2 , ν3 , ν4 ] be an adapted frame of Np Msing ν
singular at p if and only if ν ∈ Np M . Furthermore, the following possibilities hold:
i) If p ∈ M3 then the focal set of Msing 2 at p is a quadric. Moreover, it
intersects the set of degenerate directions on a line ν = p + r(ν2 + ν4 ),
r ∈ R. In addition, the umbilical focus at p is given by
1
ν = p + sing ν4 ∈ Ep⊥ .
κu
ii) If p ∈ M2 is semiumbilic and
2
1) ∆p is a half-line, then the focal set of Msing at p is the union of two
transverse planes intersecting in a line of umbilical foci at p given by
1
ℓ = p + sing ν3 + rν4 , r ∈ R.
κu
2
2) ∆p is a line, then the focal set of Msing at p is equal to the set of the
⊥
degenerate directions (Ep ). In addition, ℓ is a line of umbilical foci at
p.
2
Moreover, the focal set of Msing at p intersects the set of degenerate direc-
tions when ν ∈ Ep .⊥

iii) If p ∈ M1 and
2
1) ∆p is a half-line, then the focal set of Msing at p is the union of two
2
parallel planes in Np Msing , which can be coincident.
2
2) p is umbilic non flat, then the focal set of Msing at p is the hyperplane
2
Np Msing . In addition, there is a plane of umbilical foci at p given by
1
ν = p + tν2 + ν3 + rν4 , t, r ∈ R.
κsing
u
2
Proof. The first assertion that dν is singular if and only if ν ∈ Np Msing is obvious.
2
In all the cases we assume that p = 0 and Msing is parametrised by f (x, y) =
 GEOMETRY OF SURFACES IN R5 THROUGH PROJECTIONS AND NORMAL SECTIONS 15

(x, f2 (x, y), f3 (x, y), f4 (x, y)) with j 2 fi (0, 0) = 0, for i = 2, 3, 4. After smooth
changes of coordinates in the source and isometries in the target, the surface can
be taken with one of the following parametrisations (see [3]):
• (x, xy + p(x, y), b20 x2 + b11 xy + b02 y 2 + q(x, y), c20 x2 + r(x, y)) iff ∆p is a
non-degenerate parabola;
• (x, a20 x2 + y 2 + p(x, y), b20 x2 + q(x, y), r(x, y)) iff ∆p is a half-line;
• (x, xy + p(x, y), b20 x2 + q(x, y), r(x, y)) iff ∆p is a line;
• (x, p(x, y), b20 x2 + q(x, y), r(x, y)) iff ∆p is a point,
where aij , bij , cij ∈ R, b02 > 0 and p, q, r ∈ M32 . If we denote the coordi-
nates in R4 by (v1 , v2 , v3 , v4 ), in all the four cases we have that [ν2 , ν3 , ν4 ] =
2
[(0, 1, 0, 0), (0, 0, 1, 0), (0, 0, 0, 1)] is the adapted frame of Np Msing and take ν =
(v1 , v2 , v3 , v4 ). Thus dν is singular at p if and only if v1 = 0.
If p ∈ M3 then ∆p is a non-degenerate parabola. The degenerate directions are
given when the height function hν has singularity A2 or worse, i.e. are solutions of
the equation
−v22 − 2b11 v2 v3 − 2v2 v4 + (4b02 b20 − b211 )v32 + (4b02 c20 − 2b11 )v3 v4 − v42 = 0.
Furthermore, dv is given by
dν (x, y) = (v2 − xy)2 + (v3 − (b20 x2 + b11 xy + b02 y 2 ))2 + (v4 − c20 x2 )2 + s(x, y),
2
where s ∈ M25 . The focal set of Msing at p is given by solutions of the equation
−v22 − 2b11 v2 v3 − 2v2 v4 + (4b02 b20 − b211 )v32 + (4b02 c20 − 2b11 )v3 v4 − v42 − 2b02 v3 = 0
which is a quadric and it intersects the set of degenerate directions when v3 = 0.
Moreover
 dv has a singularity of corank 2 if and only if κsing
u = 2c20 6= 0 and
1
ν = 0, 0, 0, sing .
κu
Suppose that p ∈ M2 is semiumbilic. If ∆p is a half-line then the degenerate
directions are given by v2 (a20 v2 + b20 v3 ) = 0 and
dν (x, y) = (v2 − (a20 x2 + y 2 )2 + (v3 − b20 x2 )2 + v42 + s(x, y),
where s ∈ M25 . The focal set of Msing 2 at p is given by solutions of the equation
v2 (2a20 v2 + 2b20 v3 − 1) = 0 that intersect the set of degenerate directions in the
plane Ep⊥ (v2 = 0). In this case, if κsing
u = 2b20 6= 0 we have two transverse planes,
intersecting on the line (0, 0, sing , v4 ). Otherwise, if κsing
1
u = 0, then p ∈ M1 we
κu
have two parallel planes. In both cases, v2 = 0 is one of the planes. Moreover,
 there

sing 1
is a corank 2 singularity only in the case that κu 6= 0 and ν = 0, 0, sing , v4 .
κu
Now suppose that ∆p is a line. The distance-squared function is given by
dν (x, y) = (v2 − xy)2 + (v3 − b20 x2 )2 + v42 + s(x, y)
where s ∈ M25 . The focal set and the set of degenerate directions are equal to Ep⊥
(v2 = 0). Furthermore, the singularity is of corank 2 if and only if κsing
u = 2b20 6= 0
1
and ν = 0, 0, sing , v4 .
κu
16 J. L. DEOLINDO SILVA, R. OSET SINHA

Finally, suppose p ∈ M1 is umbilic non-flat. Here ∆p is a point different of p,

then
dν (x, y) = v22 + (v3 − b20 x2 )2 + v42 + s(x, y),
where s ∈ M25 . The focal set and the set of degenerate directions are equal to
sing
Np M . Moreover,
 dν has a singularity of corank 2 if and only if κu = 2b20 6= 0
1
and ν = 0, v2 , , v4 . 
κsing
u

Corollary 5.2. Let p ∈ Mreg2 ⊂ R5 , u ∈ T M 2 and π be the orthogonal projection

p reg u
such that πu (p) ∈ Msing ⊂ R4 .
2

i) There exists a unique umbilical focus at p ∈ Mreg 2 if and only if there exists
2
a unique umbilical focus at πu (p) ∈ Msing .
ii) The point p ∈ M2 is not semiumbilic and the umbilical focus of Mreg 2 lies

at infinity if and only if the point πu (p) ∈ M2 is not semiumbilic and

2
the umbilical focus of Msing lies at infinity. Moreover, in both cases the
umbilical focal hypersphere becomes an osculating hyperplane.
iii) The point p ∈ M2 is semiumbilic and there is a line of umbilical foci at p
in Ep⊥ if and only if the point πu (p) ∈ M2 is semiumbilic and there is a
line of umbilical foci at πu (p) in Ep⊥ . Furthermore, the line is a degenerate
direction.
Proof. The proof follows from Theorem 5.1 and the above comments. In case ii), in
both cases there is no proper umbilical focal hypersphere, but the height function in
the direction ν has a corank 2 singularity at p and πu (p), and hence, the hyperplane
with orthogonal direction ν is an osculating hyperplane (with corank 2 contact)
that can be considered as a degenerate umbilical focal hypersphere. 

6. Surfaces in R5 as normal sections of 3-manifolds in R6

3 ⊂ R6 is given in Monge form by
Following [2], if Mreg
(x, y, z, f1 (x, y, z), f2 (x, y, z), f3 (x, y, z)),
3 and let π be the projection along the direction u.
let u = (0, 0, 1) ∈ Tp Mreg u
Consider the normal section given by {Y = 0}. Let i1 , i2 be the immersions of the
normal sections in R6 and R5 respectively. Let u′ = i−1
1∗ (u) = (0, 1) ∈ Ti−1
2 .
Mreg
1 (p)
Then the following diagram is commutative
3 ⊂ R6 − π3
Mreg −−u−→ Msing ⊂ R5
x x

i1 
i
2
2 ⊂ R5 −
Mreg 2
−−−→ Msing ⊂ R4
π u′
2 = M 3 ∩ {Y = 0} and M 3 , M 2
where Mreg reg sing sing are the corresponding singular
projections, and induces a commutative diagram amongst the curvature loci of the
four manifolds.
 GEOMETRY OF SURFACES IN R5 THROUGH PROJECTIONS AND NORMAL SECTIONS 17

Theorem 6.1. Suppose that q ∈ M2 ⊂ Mreg 2 ∈ R5 is not semiumbilic and suppose

that u′ ∈ Tq Mreg 2 is not an A -asymptotic direction.The following are equivalent:

k
i) v is Ak -asymptotic at q
ii) v is asymptotic at πu′ (q)
iii) i∗2 (v) is asymptotic at i2 (πu′ (q)) = πu (i1 (q))
iv) i∗1 (v) is asymptotic at i1 (q)
Proof. Equivalence between i) and ii) is proven in Theorem 3.9. Since q is not
semiumbilic the curvature ellipse at q is non degenerate, and since u′ is not an Ak -
asymptotic direction, πu′ (q) is an Ik singularity. Therefore the curvature parabola
at πu′ (q) is non degenerate and by Proposition 3.3 πu′ (q) is not semiumbilic. Thus
we have Af fπu′ (q) = Eπu′ (q) and by Theorem 5.9 in [2] we get equivalence between
ii) and iii). Equivalence between i) and iv) follows similarly to Theorem 5.9 in [2]
since by the Proposition 3.7 the image of the asymptotic directions is tangent to
the curvature locus. Equivalence between iii) and iv) follows from the fact that the
second fundamental forms are the same and the condition for being an asymptotic
direction depends on A2 -singularities of the height function in the dual direction
in both cases. 
Example 6.2. Consider a 3-manifold whose 2-jet is given by (x, y, z, x2 +1/2z 2 , xz, yz)
and consider the projection in √ the tangent direction (0, 0, 1) and the normal
√ sec-
tion given by {Y = 0}. (1, ± 2) is asymptotic at q and πu′ (q) and (1, 0, ± 2) is
asymptotic at i2 (πu′ (q)) = πu (i1 (q)) and at i1 (q).
3 ⊂ R6 the matrix of the second
For p ∈ Mreg fundamental form with respect to
the frame given in Section 2 is
 
a1 b1 c1 d1 r1 s1
αp =  a2 b2 c2 d2 r2 s2  .
a3 b3 c3 d3 r3 s3
3 at p can be parametrised by
In [5] it is shown that the curvature locus of Mreg
2 3 3
η : S ⊂ Tp Mreg → Np Mreg , (θ, φ) 7→ η(θ, φ), where
η(θ, φ) = H + (1 + 3 cos(2φ))B1 + cos(2θ) sin2 θB2
+ sin(2θ) sin2 φB3 + cosθ sin(2φ)B4 + sin θ sin(2φ)B5
with
1 P3 1 P3
H= i=1 (ai
+ di + si )ei+3 , B1 = 12 (−a − di + 2si )ei+3 ,
3
1 P3 P3 i
i=1
B2 = 2 i=1 (ai − di )ei+3 , B3 = i=1 (bi )ei+3 ,
B4 = 3i=1 (ci )ei+3 , B5 = 3i=1 (ri )ei+3 .
P P

3 = hH, B , B , B , B , B i . The spaces Af f

The first normal space is Np1 Mreg 1 2 3 4 5 (p) p
and Ep are defined as in all other regular cases. In fact, Ep = hB1 , B2 , B3 , B4 , B5 i(p) .
3 ) = 3, and that if dim E = 2 (i.e.
In [5], it is shown that, generically, dim(Np1 Mreg p
H∈ / Ep ) the curvature locus is a planar region contained in the plane Af fp. This
motivates the following.
18 J. L. DEOLINDO SILVA, R. OSET SINHA

Definition 6.3. If dim Ep = 2 we define the umbilic curvature at a point p ∈

3 ⊂ R6 as
Mreg
κ6u (p) = d(p, Af fp ).
Remark 6.4. The vector H is called the mean curvature vector and lies in Af fp .
If we denote by ν the unit vector orthogonal to Ep , then κ6u (p) = hH, νi.
2 ⊂ R5 .
From now on we shall denote by κ5u the umbilic curvature of Mreg
Proposition 6.5. Consider p ∈ Mreg 3 ⊂ R6 and q ∈ M 2 ⊂ R5 obtained as a
reg
normal section. Suppose dim Ep = 2 and that q is not semiumbilic, then κ6u = κ5u .
Proof. In [2] the authors show that the curvature locus of the 3-manifold can be
generated by the curvature loci of all the possible normal sections. If dim Ep = 2,
the curvature locus of the 3-manifold is a planar region contained in the plane Af fp ,
therefore the curvature locus of any normal section will be a curve contained in the
same plane. Since q is not semiumbilic, ∆5e is non-degenerate and so Af fq = Af fp .
The result follows. 
6.1. Contact with spheres. As we did for singular surfaces in R4 , we consider
3
the family of distance squared functions on Mreg
3 × R6 → R
D : Mreg
(p, a) 7→ ||p − a||2 = da (p).
Following [5], a point in the focal set (points where the distance squared function
has a degenerate singularity) where Hess da has corank 3 is called an umbilical
focus. For a generic 3-manifold the rank of the second fundamental form is 3 at
every point and the umbilical foci lie at isolated points where the curvature locus
is of planar type.
Proposition 6.6. Let p ∈ Mreg 3 be a point in a generic 3-manifold and suppose

dim Ep = 2 (i.e. the curvature locus is a planar region), then there exists a unique
umbilical focus at p given by
1
a =p+ 6v
κu
where v is a unit vector in Ep⊥ .
3 and a = p + λv we have
Proof. For v = v4 e4 + v5 e5 + v6 e6 ∈ Np Mreg
1 − λ 3i=1 ai vi λ 3i=1 bi vi λ 3i=1 ci vi
 P P P 

Hess da =  λ 3i=1 bi vi 1 − λ 3i=1 di vi λ 3i=1 ri vi  .

P P P
P3 P3
1 − λ 3i=1 si vi
P
λ i=1 ci vi λ i=1 ri vi
So corank Hess da is 3 if and only if
3
X 3
X 3
X
1−λ ai vi = 1 − λ di vi = 1 − λ si vi = 0,
i=1 i=1 i=1
 GEOMETRY OF SURFACES IN R5 THROUGH PROJECTIONS AND NORMAL SECTIONS 19

3
X 3
X 3
X
bi vi = ci vi = ri vi = 0.
i=1 i=1 i=1
This is equivalent to
3
X 3
X 3
X 3
X 3
X
(ai − di )vi = (−ai − di + 2si )vi = bi vi = ci vi = ri vi = 0,
i=1 i=1 i=1 i=1 i=1
which is equivalent to
hv, Bj i = 0
for j = 1, . . . , 5. This means that v is in Ep⊥ .
Furthermore, we get
3
1 1X
= (ai + di + si )vi = hH, vi,
λ 3
i=1
1 1
so λ = hH,vi = κ6u
. 

Corollary 6.7. Let p ∈ Mreg 3 be a point in a generic 3-manifold and consider q ∈

2 ⊂ R5 given by a generic normal section. There exists a unique umbilical focus
Mreg
at p if and only if there exists a unique umbilical focus at q = i−1
1 (p). Moreover,
the radii of the umbilical focal hyperspheres are the same.
Proof. By Proposition 6.6, for 3-manifolds we have a unique umbilical focus if and
only if the curvature locus is a planar region and Af fp does not contain p. Taking
a generic normal section we obtain a surface in R5 whose curvature ellipse lies in
the plane Af fp . Since the section is generic, the ellipse will be non-degenerate
(taking a non generic normal section could make q a semiumbilic point). Therefore
Af fp = Af fq and q is an M3 -point. From [8] we know that at M3 -points (and
only at these points) there is a unique umbilical focus. The statement about the
radii follows from Proposition 6.5. 

References
[1] P. Benedini Riul and R. Oset Sinha A relation between the curvature ellipse and the
curvature parabola. Adv. Geom. 19 (2019), no. 3, 389–399.
[2] P. Benedini Riul and R. Oset Sinha Relating second order geometry of manifolds through
projections and normal sections. To appear in Publicacions Matemàtiques. arXiv:1909.07307
[3] P. Benedini Riul, R. Oset Sinha and M. A. S. Ruas The geometry of corank 1 surfaces
in R4 . Q. J. Math. 70 (2019), no. 3, 767–795.
[4] P. Benedini Riul, M. A. S. Ruas and A. de Jesus Sacramento Singular 3-manifolds in
R5 . Preprint. arXiv:1911.00360
[5] R. R. Binotto, S. I. Costa and M. C. Romero Fuster The curvature Veronese of a 3-
manifold in Euclidean space. Real and complex singularities: Amer. Math. Soc., Providence,
RI (2016), (Contemp. Math., v. 675), p. 25–44.
[6] J. W. Bruce and A. C. Nogueira Surfaces in R4 and duality. Quart. J. Math. Oxford Ser.
49 (1998), 433–443.
[7] J. W. Bruce and F. Tari Families of surfaces in R4 . Proc. Edinb. Math. Soc. (2) 45 (2002),
no. 1, 181–203.
20 J. L. DEOLINDO SILVA, R. OSET SINHA

[8] S. I. R. Costa, M. S. Moraes and M. C. Romero Fuster Geometric contact of surfaces

immersed in Rn , n > 5. Differential Geom, Appl. 27 (2009), 442–454.
[9] J. L. Deolindo-Silva Cross-ratio invariants for surfaces in 4-space. Bull Braz Math Soc,
New Series (2020). https://doi.org/10.1007/s00574-020-00221-w.
[10] D. Dreibelbis Self-conjugate vectors of immersed 3-manifolds in R6 . Topology Appl. 159
(2012), no. 2, 450–456.
[11] R. Garcia, D. K. H. Mochida, M. C. Romero Fuster and M. A. S. Ruas Inflection
points and topology of surfaces in 4-space. Trans. Amer. Math. Soc. 352 (2000), 3029–3043.
[12] S. Izumiya, M. C. Romero Fuster, M. A. S. Ruas and F. Tari Differential Geometry
from Singularity Theory Viewpoint. World Scientific Publishing Co. Pte. Ltd., Hackensack,
NJ, 2016. xiii+368 pp. ISBN: 978-981-4590-44-0
[13] C. Klotz, O. Pop and J. H. Rieger Real double-points of deformations of A-simple map-
germs from Rn to R2n . Math. Proc. Camb. Phil. Soc. (2007), 142-341.
[14] J. A. Little On singularities of submanifolds of higher dimensional Euclidean spaces. Ann.
Mat. Pura Appl. 83 (4) (1969), 261–335.
[15] L. F. Martins and J. J. Nuño-Ballesteros, Contact properties of surfaces in R3 with
corank 1 singularities. Tohoku Math. J. 67 (2015), 105–124.
[16] D. K. H. Mochida, M. C. Romero Fuster and M. A. S. Ruas, Inflection points and
nonsingular embeddings of surfaces in R5 , Rocky Mountain J. Math. 33 (2003) 995–1010.
[17] D. K. H. Mochida, M. C. Romero Fuster and M. A. S. Ruas, The geometry of surfaces
in 4-space from a contact viewpoint. Geom. Dedicata 54 (1995), 323–332.
[18] D. K. H. Mochida, M. C. Romero Fuster and M. A. S. Ruas, Osculating hyperplanes and
asymptotic directions of codimension two submanifolds of Euclidean spaces. Geom. Dedicata
77 (1999), 305–315.
[19] J. J. Nuño-Ballesteros and F. Tari, Surfaces in R4 and their projections to 3-spaces.
Proc. Roy. Soc. Edinburgh Sect. A, 137 (2007), 1313–1328.
[20] R. Oset Sinha and K. Saji, On the geometry of folded cuspidal edges. Rev. Mat. Complut.
31 (2018), no. 3, 627–650.
[21] R. Oset Sinha and F. Tari, Projections of surfaces in R4 to R3 and the geometry of their
singular images. Rev. Mat. Iberoam. 32 (2015), no. 1, 33–50.
[22] M. C. Romero Fuster, M. A. S. Ruas and F. Tari, Asymptotic curves on surfaces in
R5 . Communications in Contemporary Maths. 10 (2008), 1–27.
[23] K. Saji, M. Umehara, and K. Yamada, The geometry of fronts. Ann. of Math (2) 169
(2009), 491–529.

Departamento de Matemática. Universidade Federal de Santa Catarina - UFSC,

89036-004 - Blumenau-SC, Brazil
Email address: jorge.deolindo@ufsc.br

Departament de Matemàtiques, Universitat de València, Campus de Burjassot,

46100 Burjassot, Spain
Email address: raul.oset@uv.es