I NTERNATIONAL E LECTRONIC J OURNAL OF G EOMETRY

V OLUME XX N O . X PAGE 1–9 (20XX)

Characterizations and Local Approachs On

Isoparametric Surfaces
Emre Öztürk
(Communicated by Kazım İLARSLAN)

A BSTRACT
In this study, fundamentally we give the characterization of Gauss map of isoparametric surfaces
by semi symmetric matrix in Lorentz space and we determine the isoparametric surfaces through
the chord property. Finally we consider the chord property locally and give some examples in
Euclidean space.

Keywords: Isoparametric Surface

AMS Subject Classification (2010): Primary: 14J25

1. Introduction

Isoparametric surfaces (surfaces with constant principal curvatures) are studied in ([3], [5] and [6]) by idea of
chord property and helical point in Euclidean space. In [1] the unit disk characterized by the following Lemma,
Lemma 1.1. [1] The only bounded, smooth and simply-connected plane region whose Szegö kernel coincides with the
Cauchy kernel is the disc.
Kerzman and Stein used complex analysis technics on chord of the curve for proving the Lemma 1.1 and
Boas [2], extended the idea to n−dimensional Euclidean space. By the help of Bochner-Martinelli kernel, Boas
gives the following theorem:
Theorem 1.1. [2] Ball is the only bounded C 1 domain in Rm such that given any two points of the boundary, the chord
joining them meets the normals at the two endpoints with equal angles.
In ([2], Proof of Theorem 2, pp. 277-278), the chord property idea developed to the hypersphere in
n−dimension. Boas [3], gave the global and local characterization theorem as a result of chord property and
extended his study [2], to all isoparametric surfaces. He gives the local characterization theorem for all surfaces
M whose are satisfy the following property:
D −→ E D −→ E
(x − y) , ∇r(x) = (y − x) , ∇r(y) (1.1)
−→
where x, y ∈ M and ∇r is the unit normal (gradient) vector of the surface. Wegner [4], gave the short proof for
([3], Local characterization theorem, pp.120). Similar to [2] and [3], in [5], the equation of (1.1) considered for
surfaces such that, on a hypersurface, a unit normal vector field G is naturally defined on M . Such G is also
called the Gauss map of M , and for a hypersphere, the chord joining any two points on it meets the sphere at
the same angle at the two points, that is, the sphere satisfies the condition

hy − x, G(x) + G(y)i = 0 (1.2)

In [5], Kim D.S. and Kim Y.H. considered the following question:
“What are the hypersurfaces of Euclidean space which satisfy the (1.2)?”
Kim D.S. and Kim Y.H. used some algebraic approachs and they got the hyperspheres, cylinders and spherical

Received : dd–month–yyyy, Accepted : dd–month–yyyy

* Corresponding author
 Characterizations and Local Approachs On Isoparametric Surfaces

cylinders as an isoparametric surfaces. They also got more information about these surfaces such as Gauss map
can be written by G(x) = Ax + b for symmetric matrix A and column vector b. In [10], some special curves are
defined and relations between these curves and isoparametric surfaces are given in Lorentz-Minkowski space.
In this study, we consider the following questions:
“What are the hypersurfaces of Lorentz space whose are satisfy the (1.2)?”
and
“What are the hypersurfaces of Euclidean space whose are satisfy the (1.2) locally?”
Besides, we characterize the Gauss map of isoparametric surfaces by semi-symmetric matrix.

2. Preliminaries

Any linear operator in complex space Cn is given by

Zb
ζ(f, x) = f (x)K(t, x)dt
a

such that K(t, x) is continuous according to t parameter and it is called the kernel of the ζ operator. Suppose B
is a bounded smooth domain in the complex plane. The Cauchy kernel represents holomorphic functions f in
B in terms of the boundary values on γ. Here γ is the boundary of the domain B. Cauchy integral operator on
any B domain whose bounded by the curve γ in complex plane is given by
Z
1 f (z) ·
ζ(z, w) = γ(z)dσ(z)
2πi z−w
z∈γ

·
where w ∈ C is on γ . Here dz = γ(z)dσ, z = γ(s) unit speed curve and dσ is Lebesgue measure (arc lenght).
1 1 ·
Hence, Cauchy kernel of the ζ operator is γ(z). Similarly we give the S(z, w) Szegö integral operator
2πi z − w
as Z
1
ζ(z, w) = f (z)S(z, w)dσ(z)
2πi
z∈γ

where the kernel is considered the orthogonal projection of S : L2 (γ, dσ) → H2 (γ). Here H2 (γ) closed subspace
of L2 (γ, dσ) of boundary values of holomorphic functions in B. It is easy to see that Sζ = ζ holds identically.
If we consider the proposition of ([1], Lemma 7.1), we can easily find that γ is non-null hyperbolic curves in
Lorentz plane.
In Lorentz space, vectors have different causal characters such as if hv, vi > 0 or v = 0, hv, vi < 0 and hv, vi = 0
(v 6= 0) then v is called by spacelike, timelike and lightlike (or null) vector respectively. The number of timelike
vectors of the orthonormal basis of the vector space is called the index of space and usually denoted by ν.
Definition 2.1. [7] Let X = (x1 , x2 , ..., xn ) and Y = (y1 , y2 , ..., yn ) be a two vector distinct from zero in
n−dimensional real vector space Rn . Following inner product,
n
X
hX, Y i = −x1 y1 + xi yi (2.1)
i=2

is called by Lorentzian inner product of X and Y and h, i is called metric tensor of vector space. (Rn , h, i) is
called Lorentz space and denoted by Ln or Rn1 .
p
Norm of vector v ∈ Rn1 is given by kvk = |hv, vi| [9]. Let index of n−dimensional vector space of V be
0 ≤ ν ≤ n and ε1 = ε2 = ... = εν = −1 but εν+1 = εν+2 = ... = εn = 1. Therefore the diagonal matrix (δij εj ) is
called by the sign matrix of V such that

1, i = j
δij = , 1 ≤ i, j ≤ n
0, i 6= j

is Kronecker delta [9].

www.iejgeo.com 2
 Emre Öztürk

Definition 2.2. [9] Let n ≥ 2 and 0 ≤ ν ≤ n,

(1) The pseudosphere of radius r > 0 in Rn+1
ν is the hyperquadric

Snν (r) = P ∈ Rn+1ν | hP, P i = r2

with dimension n and index ν.

(2) The pseudohyperbolic space of radius r > 0 in Rn+1
ν+1 is the hyperquadric

Hnν (r) = P ∈ Rn+1

 2

ν+1 | hP, P i = −r

with dimension n and index ν.

Definition 2.3. [9] Let M be a hypersurface and → −
n be a unit normal vector field of M . If h→ −
n,→
−n i < 0 or
→
− →
−
h n , n i > 0 then M is called spacelike or timelike surface respectively.
Lemma 2.1. [9] Let S be a shape operator (Weingarten map) of M, and v be a tangent vector on M . S(v) = −∇ → −
n and
v
for all P ∈ M, linear operator of S is self-adjoint on TP M. Here ∇ is Levi-Civita connection on M in Rn1 .

3. Isoparametric Surfaces in Lorentz Space

In this section we give some characterizationson isoparametric surfaces by the help of chord properties of it
through the [5] and [10]. A new term for this surfaces can be given by the following definition:
Definition 3.1. Let M be a non-null hypersurface and G is Gauss map of M . If

hQ − P, G(P ) + G(Q)i = 0 (G)

for all P, Q ∈ M then, M is called by G−surface.

Theorem 3.1. Let M be a G−surface. Gauss map of this surface is given by

G(x) = Ax + b

where A is semi-symmetric matrix and b ∈ E1n column vector.

Proof. Let M fully lies in space and consider the points y0 , y1 , ..., yn on M such that {yj− y0 | 1 ≤ j ≤ n} spans
Rn1 . Similar to [5], we find
AT = ε Bj A−1


j ε (3.1)
where ε = diag(−1, 1, ..., 1) is the sign matrix, Aj and Bj are n × n matrices whose accepts the yj− y0 and
n
P
G(yj ) − G (y0 ) as j−coloumn respectively. Also we have b = bk αk such that bk is given by
k=1

hα1 , α1 i hα1 , α2 i ... hα1 , αn i b1 c1

    
 hα2 , α1 i . ... hα2 , αn i   b2   c2 
 ... . . ...   ...  =  ... 
hαn , α1 i hαn , α2 i ... hαn , αn i bn cn

where αj = yj − y0 and cj = hG (y0 ) , y0 i − hG (yj ) , yj i . Therefore we write

cj = hαj , α1 i b1 + hαj , α2 i b2 + ... + hαj , αn i bn (3.2)

Here (3.1) and (3.2) implies

G(x) = Ax + b (3.3)
for all x ∈ M. Now we prove the A is semi-symmetric matrix. Let X ∈ Tx M be a tangent vector. By Lemma 2.4
we write
∇X G(X) = AX = −S(X) (3.4)
and
hS(X), Y i = hX, S(Y )i (3.5)

3 www.iejgeo.com
 Characterizations and Local Approachs On Isoparametric Surfaces

for X, Y ∈ Tx M. From (3.4) and (3.5) we get

h−AX, Y i = hX, −AY i ⇔ hAX, Y i = hX, AY i


⇔ X, (εAT ε)Y = hX, AY i
⇔ A = εAT ε

which means the A is semi symmetric matrix and this completes the proof.
Theorem 3.2. For non-null hypersurface M, the following statements are equivalent:

a) M is the G-surface.

b) M is an isopametric surface.

c) M is the open part of non-null hyperplane, pseudosphere, pseudohyperbolic space, pseudospherical cylinder or
pseudohyperbolic cylinder.

Proof. Let M be a G−surface and {E1 , E2 , ..., En−1 } orthonormal frame on surface such that Ei , 1 ≤ i ≤ n − 1
are characteristic vectors corresponding to µi , 1 ≤ i ≤ n − 1 characteristic values of shape operator. For all
j, 1 ≤ j ≤ n − 1, S(Ej ) = µj Ej and due to Theorem 3.2, G(x) = Ax + b for some semi-symmetric A matrix. (3.4)
implies
AEj (x) = −S(Ej (x)) = −µj (x)Ej (x)
or
(A + µj I) Ej (x) = 0
According to the existence of non-zero characteristic vectors,

det(A + µj I) = 0 (3.6)

It is obvious from equation (3.6), µj , 1 ≤ j ≤ n − 1 are constant. Therefore M is an isoparametric surface.

Let M be an isoparametric surface. For unit normal vector G(x) = Ax + b we define

f (x) = hAx + b, Ax + bi (3.7)

such that f : Rn1 → R and M ⊂ f −1 (±1) that depends to causal character of the surface. Explicitly equation (3.7)
written as
f (x) = x, A2 x + 2 hx, Abi + hb, bi



−→
and by straightforward calculations we get ∇f (x) = 2A(Ax + b). Because of M ⊂ f −1 (±1) gradient of f and G
is proportional. For some λ(x) functions

A(Ax + b) = λ(x)(Ax + b), x ∈ M (3.8)

Reformulating the (3.8) we get (A − λ(x)I)(Ax + b) = 0 and so det(A − λ(x)I) = 0. Obviously λ(x) is constant.
Let consider λ(x) = λ, V = {Ax + b | x ∈ M } and U as a characteristic space of corresponding to λ characteristic
value with respect to A. In particular V ⊆ U and Sp {V } is normal space at x ∈ M. Now we can determine the
surface M depends on norm of V.

Case 1. Let kV k = 1. In this case G(x) = b is constant and M is the open part of non-null hyperplane.
1 1 1
Case 2. If kV k = n then V = Im(A). For some r > 0 and λ = ± we have A |V = ± I such that G(x) = ± x + b.
r r r
Depends on causal character of Sp {V } we get Sn−1 1 (pseudosphere) or H1n−1 (pseudohyperbolic space).
Case 3. Let kV k = p, 2 ≤ p ≤ n − 1. Dimension of V ⊥ orthogonal complement is n − p and V ⊥ ⊆ Tx M. In the
neighborhood of x0 ∈ M, we choose {E1 (x), E2 (x), ..., En−1 (x)} orthonormal frame such that E1 , E2 , ..., En−p
are constant in V ⊥ . Therefore we write

V = Sp {En−p+1 (x), En−p+2 (x), ..., En−1 (x), G(x)}

www.iejgeo.com 4
 Emre Öztürk

The tangent subspace spanned by {En−p+1 (x), En−p+2 (x), ..., En−1 (x)} is integrable and integral submanifold
of M1 through the point x0 is given as M1 = M ∩ (x0 + V ) . Therefore Rn−p
1 = V ⊥ and M = M1 × Rn−p1 where
p
M1 is hypersurface in R1 and Gauss map of M1 satisfy the G1 (x) = G(x). Hence A1 = A |V satisfy
G1 (x) = A1 x + b (3.9)
1
Due to (3.8) and (3.9) we write A1 = ± I . Therefore M1 is Sp−1 1 (r) pseudosphere or Hp−1
1 (r) pseudohyperbolic
r
p−1 p−1
space and M is open part of S1 (r) × Rn−p pseudospherical cylinder or H1 (r) × Rn−p pseudohyperbolic
cylinder respectively.
The hyperplane M thinkable as an isometric immersion, f1 : M → Rn1 with respect to [x1 , x2 , ..., xn ] rectangular
coordinate system. If we think x = (x1 , x2 , ..., xn ) ∈ M and real number c then f1 is given by
f1 (x1 , x2 , ..., xn ) = a1 x1 + a2 x2 + ... + an xn = c (3.10)
where ai , 1 ≤ i ≤ n are constant coefficients. At any point P on M, Gauss map of the surface is given by
−→
∇f 1 1
G(P ) := −→ |P =
(−a1 , a2 , ..., an ) |P
∇f 1
m

p
where m = |−a21 + a22 + ... + a2n |. It can be easily seen
hQ − P, G(P ) + G(Q)i = 0
for all P, Q ∈ f1 . Therefore hyperplanes are G−surface. By Definition 2.2, pseudosphere with center x0 and
radius r is given by
Sn−1 (r) = x ∈ Rn1 : hx − x0 , x − x0 i = r2

1
Without loss of generality if we consider x0 = (0, 0, ..., 0) and x = (x1 , x2 , ..., xn ) than similar to previous we
write
f2 (x1 , x2 , ..., xn ) = −x21 + x22 + ... + x2n = r2 (3.11)
such that f2 : S1n−1 → Rn1 is an isometric immersion and then Gauss map of the surface is given by
−→
∇f 2 1
G(P ) := −→ |P = (x1 , x2 , ..., xn ) |P
∇f 2
r

By straightforward calculation we get

hQ − P, G(P ) + G(Q)i = 0
for all P, Q ∈ f2 . Therefore pseudosphere is G−surface (similarly H1n−1 (r) is G−surface). Similar to f1 and f2
we consider Sp−1
1 (r) × Rn−p by the following equation,
−x21 + x22 + ... + x2p = r2 , xp+1 = up+1 , xp+2 = up+2 , ..., xn = un

where ui , p + 1 ≤ i ≤ n are real variables. By the help of natural Lorentz projection πL : Rp1 × Rn−p
1 → Rp1 ,→ Rn1
p−1
(onto S1 ), we have
1
G(P ) = (x1 , x2 , ..., xp , 0, ..., 0) |P
r
p−1 n−p
Therefore P, Q ∈ S1 (r) × R satisfy
hQ − P, G(P ) + G(Q)i = 0
Proof is similar for Hp−1
1 (r) × Rn−p .

Corollary 3.1. Let M be an isoparametric surface. Gauss map of this surface is given by
G(x) = Ax + b
where A is semi symmetric matrix.

5 www.iejgeo.com
 Characterizations and Local Approachs On Isoparametric Surfaces

4. Locally G−Surfaces in Euclidean Space

Chord property of isoparametric surfaces are examined globally by [5]. In this section we introduce with
approach to G−surfaces locally. Here we mean by “locally” that some of points on the surface satisfy (1.2) but
not all.
Definition 4.1. Let M be a hypersurface and h, i is Euclidean metric tensor of the space. If some points P, Q ∈ M
satisfy
hQ − P, G(P ) + G(Q)i = 0
then M is called by locally G−surface.
Through the definition we can give some examples by following theorems:
Theorem 4.1. The helicoid surface of
Φ(s, t) = (0, 0, bs) + t (cos s, sin s, 0) , b 6= 0

is locally G−surface.
Proof. Let we consider the surface ϕ(U ) = Φ for U ⊆ R2 and ϕ : U → Φ differentiable map. Let P, Q ∈ Φ two
points on surface such that ϕ(p1 , p2 ) = P and ϕ (q1 , q2 ) = Q. By straightforward calculations we get
Φs (s, t) = (−t sin s, t cos s, b)

and
Φt (s, t) = (cos s, sin s, 0)
Unit orthogonal normal vector field Z of helicoid is given by
 
Φs (s, t) × Φt (s, t) b b 1
Z ◦ϕ= = − √ sin s, √ cos s, − √ t
kΦs (s, t) × Φt (s, t)k b2 + t2 b2 + t2 b2 + t2
Therefore
G(P ) + G(Q) = (α1 , α2 , α3 ) (4.1)
where
b b
α1 = − p sin p1 − p sin q1
2
b2 + p2 b2 + q22
b b
α2 = p cos p1 + p cos q1
2
b + p22 b + q22
2
b 1
α3 = − p p2 − p q2
2
b + p22 b + q22
2

We write
Q − P = (q2 cos q1 − p2 cos p1 , q2 sin q1 − p2 sin p1 , bq1 − bp1 ) (4.2)
and by the help of (4.1) and (4.2) we get
!
p2 q2
hQ − P, G(P ) + G(Q)i = b p +p (sin(q1 − p1 ) + (p1 − q1 ))
b2 + p22 b2 + q22

Obviously hQ − P, G(P ) + G(Q)i = 0 if and only if p1 = q1 or p2 = −q2 . By Definition 4.1, Φ is locally

G−surface.
Theorem 4.2. The hyperbolic paraboloid surface of
v2 u2
Φ(u, v) = (u, v, − ), a, b ∈ R\ {0}
b2 a2
is locally G−surface if and only if  
a  p1 − q1 
=
b p2 − q2 
where ϕ(p1 , p2 ) = P, ϕ (q1 , q2 ) = Q and ϕ : U ⊆ R2 → Φ.

www.iejgeo.com 6
 Emre Öztürk

Proof. Similar to Theorem 4.1 by straightforward calculations we get

 
 2u 2v 1 
Z ◦ϕ=
− r , r , −r 
2 2 2 2 2 2
4u 4v 4u 4v 4u 4v

a2 1+ 4
+ 4 b2 1+ 4
+ 4 1+ 4
+ 4
a b a b a b
and !
  r 4p21 4p22
r
4q12 4q22
2 2 2 2
−b (p1 − q1 ) + a (p2 − q2 ) 1+ 4 + 4 − 1+ 4 + 4
a b a b
hQ − P, G(P ) + G(Q)i = r r
4p2 4p2 4q 2 4q 2
a2 b2 1 + 41 + 42 1 + 41 + 42
a b a b

Therefore hQ − P, G(P ) + G(Q)i = 0 if and only if

 
a  p1 − q1 
=
b p2 − q2 

and this completes the proof.

In [11], by the help of E. Study map, some properties of Viviani ruled surface are examined and for special
case, the parametric equation of this surface is given as in the following theorem:
Theorem 4.3. The Viviani ruled surface of
5 5 5 u u
Φ(u, v) = ( + cos u, sin v, 5 sin ) + 4v(1 + cos u, sin u, 2 sin )
2 2 2 2 2
is locally G−surface if and only if
p1 = q1 + 4kπ, k ∈ Z
where ϕ(p1 , p2 ) = P, ϕ (q1 , q2 ) = Q and ϕ : U ⊆ R2 → Φ.
Proof. Similar to previous operations we get
√
u u 3u √ u

2 sin3 −5 cos + cos 2 cos2
Z ◦ ϕ = √
 2 , √ √2 2 ,√ 2 
3 + cos u 2 2 3 + cos u 3 + cos u

and
p1 − q1
hQ − P, G(P ) + G(Q)i = α sin2 ( )
4

where α is non-zero constant. Therefore hQ − P, G(P ) + G(Q)i = 0 if and only if

p1 = q1 + 4kπ, k ∈ Z

and this completes the proof.

In [11], the above surface illustrated by the author with Figure 1.

7 www.iejgeo.com
 Characterizations and Local Approachs On Isoparametric Surfaces

Figure 1. Viviani Ruled Surface

5. Conclusions
In this study we considered isoparametric surfaces locally with different perspective from the previous
studies in Euclidean space. Examples of locally isoparametric surfaces can be extend in three or high dimension
in any space. It will be more interesting to define new conditions for special surfaces like isoparametric surfaces
or more.

References
[1] Kerzman, N. and Stein, E.M., The Cauchy Kernel, the Szegö Kernel, and the Riemann Mapping Function, Math.Ann.236, (1978), 85-93.
[2] Boas, H.P., A Geometric Characterization of the Ball and the Bochner-Martinelli Kernel, Math. Ann. 248, (1980), 275-278.
[3] Boas, H.P., Spheres and Cylinders: A Local Geometric Characterization, Illinois Journal of Mathematics, Volume 28, Number 1, Spring
(1984), 120-124.
[4] Wegner, B., A Differential Geometric Proof of the Local Geometric Characterization of Spheres and Cylinders by Boas, Mathematica
Balkanica Vol. 2, (1988), pp. 294-295.
[5] Kim D.S. and Kim Y.H., New Characterizations of Spheres, Cylinders and W −Curves, Linear Algebra and Its Applications 432, (2010),
pp. 3002-3006.
[6] Kim, H. Y. and Lee, K. E., Surfaces of Euclidean 4-Space Whose Geodesics are W-Curves, Nihonkai Math. J., Vol. 4, (1993), pp. 221-232.
[7] İyigün, E. and Arslan, K., On Harmonic Curvatures Of Curves In Lorentzian n−Space, Commun. Fac. Sci. Univ. Ank. Series A1, Volume
54, Number 1, (2005), pp. 29-34.
[8] Kühnel, W., Differential Geometry Curves-Surfaces-Manifolds, American Mathematical Society, (2006).
[9] O’Neill, B., Semi-Riemann Geometry with Applications to Relativity, Academic Press. Inc., (1983).
[10] Öztürk, E. and Yaylı, Y., W-Curves In Lorentz-Minkowski Space, Mathematical Sciences and Applications E-Notes, Volume 5 (2), (2017),
76-88.
[11] Ekici, C., Özüsağlam, E., Çimdiker, M. and Öztürk, E., On The Curvatures of Viviani Ruled Surface, Ciencia e Tecnica Vitivinicola, Volume
29 (7), (2014), 449-475

www.iejgeo.com 8
 Emre Öztürk

Affiliations

E MRE Ö ZTÜRK
A DDRESS : Turkish Court of Accounts, 06520, Ankara, TURKEY.
E - MAIL : emreozturk1471@gmail.com

9 www.iejgeo.com