Chapter 1

Surfaces: basic definitions

A regular parametrized surface is a smooth mapping ϕ : U → R3 , where U is

an open subset of R2 , of maximal rank. This is equivalent to saying that the
rank of ϕ is 2 or that ϕ is an immersion. Such a ϕ is called a parametrization.
Let (u, v) be coordinates in R2 , (x, y, z) be coordinates in R3 . Then

ϕ(u, v) = (x(u, v), y(u, v), z(u, v)),

x(u, v), y(u, v), z(u, v) admit partial derivatives of all orders and the Jacobian
matrix
 ∂x ∂x 
∂u ∂v
 ∂y ∂y 
∂u ∂v
∂z ∂z
∂u ∂v

has rank two. This is equivalent to

∂(x, y) ∂(y, z) ∂(z, x)

6= 0 or 6= 0 or 6= 0.
∂(u, v) ∂(u, v) ∂(u, v)

A surface is a subset S of R3 satisfying:

(1) S = ∪i∈I , where Vi is an open subset of S and ϕI : Ui ⊂ R2 → ϕi (Ui ) = Vi

is a parametrization. In other words, every point p ∈ S lies in an open
subset W ⊂ R3 such that W ∩ S is the image of a smooth immersion of
an open subset of R2 into R3 .

(2) Each ϕi : Ui → Vi is a homeomorphism. The continuity of ϕ−1 i means

that for any Vi , given ǫ > 0, there exists δ > 0 such that

ϕ−1
i ( B(q, δ) ∩Vi ) ⊂ B(ϕ−1 (q), ǫ) .
| {z } | i {z }
bola em R3 bola em R2

1
2 CHAPTER 1. SURFACES: BASIC DEFINITIONS

1.1 Examples
1. The graph of a smooth function f : U → R, where U ⊂ R2 is open, is a
regular parametrized surface, where the parametrization is given by ϕ(u, v) =
(u, v, f (u, v)). Note that
 
1 0
(dϕ) =  0 1 
∂f ∂f
∂u ∂v

has rank two.

2. If S ⊂ R3 is a union of graphs as in (1) (with respect to any one of the
three coordinate planes), then S is a surface. It only remains to check that the
parametrizations constricted in (1) are homeomorphisms. But this follows from
the fact that ϕ−1 = π|ϕ(U ) is continuous, where π(x, y, z) = (x, y) is continuous.
3. The unit sphere is defined as

S 2 = {(x, y, z) ∈ R3 |x2 + y 2 + z 2 = 1}.

p √
It is clearly
p a union of (six) graphs: z = ± 1 − x2 − y 2 , y = ± 1 − x2 − z 2 ,
x = ± 1 − y2 − z2 .

1.2 Inverse images of regular values

Let F : W → R3 be a smooth map, where W ⊂ R3 is open. A point p ∈ W is
called a critical point of F is dFp = 0; otherwise, it is called a regular point. A
point q ∈ R is called a critical value of F if there exists a critical point of F in
F −1 (q); otherwise, it is called a regular value.

Theorem 1.1 If q is a regular value of F and F −1 (q) 6= ∅, then S = F −1 (q)

is a surface.

Proof. It suffices to show that S is a union of graphs. Let p = (x0 , y0 , z0 ) ∈

 ∂F 
∂x (p)
∂F
S. Then dFp =  ∂y (p)  6= 0. Without loss of generality, asssume that
∂F
∂z (p)
∂F
∂z (p) =
6 0. By the implict function theorem, tehre exist open neighborhoods Ṽ
of (x0 , y0 , z0 ) in R and U of (x0 , y0 ) in R2 and a smooth function f : U → R
3

such that F (x, y, z) = q, (x, y, z) ∈ Ṽ if and only if z = f (x, y), (x, y) ∈ U .

Hence p ∈ V = Ṽ ∩ S is the graph of f . 

1.3 More examples

5. Spheres cam also be seen as inverse images of regular values. Let F (x, y, z) =
x2 + y 2 + z 2 . Then (dF(x,y,z) )t = (2x 2y 2z) = (0 0 0) if and only if (x, y, z) =
1.4. CHANGE OF PARAMETERS 3

(0, 0, 0). Since (0, 0, 0) 6∈ F −1 (r2 ) for r > 0, we have that the spehre F −1 (r2 )
2 2 2
of radius r > 0 is a surface. Similarly, the ellipsoides xa2 + yb2 + zc2 = 1 (a, b,
c > 0) are surfaces.
6. The hyperboloids x2 + y 2 − z 2 = r2 (one sheet) and x2 + y 2 − z 2 = −r2
(two sheets) are surfaces, r > 0. The cone x2 + y 2 − z 2 = 0 is not a surface in
a neighborhood of its vertex (0, 0, 0). p
7. The tori of revolution are surfaces given by the equation z 2 + ( x2 + y 2 −
a)2 = r2 , where a, r > 0.
8. More generally, one can consider surfaces of revolution. Let γ(t) =
(f (t), 0, g(t)) be a regular parametrized curve, t ∈ (a, b). Define

ϕ(u, v) = (f (u) cos v, f (u) sin v, g(u))

where (u, v) ∈ (a, b) × (v0 , v0 + 2π). One can cover the surface by varying v0 in
R. But there are conditions on γ for ϕ to be an immersion. One has
∂(x, y) ∂(y, z) ∂(z, x)
= f f ′, = −f g ′ cos v, = −f g ′ sin v,
∂(u, v) ∂(u, v) ∂(u, v)
so  2  2  2
∂(x, y) ∂(y, z) ∂(z, x)
+ + = f 2 ||γ̇||2 ,
∂(u, v) ∂(u, v) ∂(u, v)
and ϕ is an immersion if and only if f > 0. Note also that ϕ is injective if
and only if γ is injective. One also checks that ϕ−1 is continuous by writing its
explicit expression.

1.4 Change of parameters

Theorem 1.2 Let S ⊂ R3 be a surface and let ϕ : U → ϕ(U ), ψ : V → ψ(V )
be two parametrizations of S, where U , V ⊂ R2 are open. Then the change of
parameters h = ϕ−1 ◦ ψ : ψ −1 (ϕ(U )) → ϕ−1 (ψ(V )) is a diffeomorphism between
open sets of R2 .

Proof. h is a homeomorphism because it is the composite map of two home-

omorphisms. Note that a similar argument cannot be used to say that h is
smooth, because it does not make sense (yet) to say that ϕ−1 is smooth.
Let p = (u0 , v0 ) ∈ U , q ∈ V , ϕ(p) = ψ(q). Since ϕ is an immersion, dϕ
∂(x,y)
has rank two and we may assume WLOG that ∂(u,v) (p) 6= 0. Write ϕ(u, v) =
(x(u, v), y(u, v), z(u, v)), (u, v) ∈ U and define

Φ(u, v, w) = (x(u, v), y(u, v), z(u, v) + w),

where (u, v, w) ∈ U × R. Then Φ : U × R → R3 is smooth and

∂x ∂x
0
∂u
∂y
∂v
∂y ∂(x, y)
det(dΦ(u0 ,v0 ,0) ) = ∂u ∂v 0 = (p) 6= 0.
∂(u, v)
0 0 1 (u0 ,v0 ,0)
4 CHAPTER 1. SURFACES: BASIC DEFINITIONS

Since dΦ(u0 ,v0 ,0) is non-singular, by the inverse function theorem, Φ−1 is defined
and is smooth on some open neighborhood W of ϕ(p) in R3 . Since Φ|U ×{0} = ϕ,
we have that Φ−1 ϕ(U )∩W = ϕ
−1
|ϕ(U )∩W . Since W ∩ ϕ(U ) is open in S and ψ is a
−1
homeomorphism, ψ (W ∩ ϕ(U )) ⊂ V is open. Now

h|ψ−1 (W ) = ϕ−1 ◦ ψ|ψ−1 (W ∩ϕ(U )) = Φ−1 ◦ ψ|ψ−1 (W ∩ϕ(U ))

is smooth, because it is the composite map of smooth maps.

Similarly, one sees that h−1 is smooth by reversing the rôles of ϕ and ψ in
the argument above. Hence h is a diffeomorphism. 

Corollary 1.3 Let S ⊂ R3 be a surface and suppose f : W → R3 is a smooth

map defined on the open subset W ⊂ Rm such that f (W ) ⊂ S. Then ϕ−1 ◦ f :
W → R2 is smooth for every parametrization ϕ : U → ϕ(U ) of S.

Proof. If Φ is as in the proof of the theorem, we have that ϕ−1 ◦ f = Φ−1 ◦ f

is the composite of smooth maps between Euclidean spaces. 
As an application of the smoothness of change of parameters, we can make
the following definition. Let S be a surface. An application f : S → Rn is
smooth at a point p ∈ S if f ◦ ϕ : U → Rn is smooth at ϕ−1 (p) ∈ U , for some
parametrization ϕ : U → ϕ(U ) of S with p ∈ ϕ(U ). Note that if ψ : V → ψ(V )
is another parametrization of S with p ∈ ψ(V ), then f ◦ ψ is smooth at ψ −1 (p)
if and only if f ◦ ϕ is smooth at ϕ−1 (p), because

f ◦ ψ −1 = (f ◦ ϕ−1 ) ◦ (ϕ−1 ◦ ψ)

and the change of parameters ϕ−1 ◦ ψ is smooth.

Example 1.4 If S is a surface and F : R3 → R is smooth, then the restriction

F |S : S → R is smooth. In fact, f ◦ ϕ = F ◦ ϕ is smooth for any parametrization
ϕ of S. As special cases, we can take the height function relative to a, F (x) =
hx, ai, where a ∈ R3 is a fixed vector; or the distance function from q, F (x) =
||x − q||2 , where q ∈ R3 is a fixed point.

In particular, if f : S → R3 is smooth at p ∈ S and S̃ ⊂ R3 is a surface

such that f (S) ⊂ S̃, then we say that f : S → S̃ is smooth at p.

1.5 Tangent plane

Let S ⊂ R3 be a surface. Recall that a smooth curve γ : I ⊂ R → S is simply
a smooth curve γ : I → R3 such that γ(I) ⊂ S. Fix a point p ∈ S. A tangent
vector to S at p is the tangent vector γ̇(0) to a smooth curve γ : (−ǫ, ǫ) → S
such that γ(0) = p. The tangent plane to S at p is the collection of all tangent
vectors to S at p.
1.5. TANGENT PLANE 5

Proposition 1.5 The tangent space Tp S is the image of the differential

dϕa : R2 → R3 , (1.6)

where ϕ : U → ϕ(U ) is any parametrization of S with p = ϕ(a) and a ∈ U .

Proof. Any vector in the image of (1.6) is of the form dϕa (w0 ) for some w0 ∈
R2 and therefore is the tangent vector at 0 of the smooth curve t 7→ ϕ(a + tw0 ).
Conversely, suppose w = γ̇(0) is tangent to a smooth curve γ : (−ǫ, ǫ) → S
such that γ(0) = p. By Corollary 1.3, η := ϕ−1 ◦ γ : (−ǫ, ǫ) → U ⊂ R2 is a
smooth curve in R2 with η(0) = a. Note that γ = ϕ ◦ η. By the chain rule

v = dϕa (η̇(0)) (1.7)

lies in the image of (1.6). 

Corollary 1.8 The tangent plane Tp S is a 2-dimensional vector subspace of

R3 . For any parametrization ϕ : U → ϕ(U ) of S with p = ϕ(a), a ∈ U ,

{ ∂ϕ ∂ϕ
∂u (a), ∂v (a)} (1.9)

is a basis of Tp S.

It is also convenient to write ϕu := ∂ϕ ∂ϕ

∂u and ϕv := ∂v .
Consider a tangent vector w ∈ Tp S, say w = γ̇(0) where γ : (−ǫ, ǫ) → S
is a smooth curve with γ(0) = p, as in the proof of Proposition 1.5. Then
η = ϕ−1 ◦ γ is a smooth curve in R2 which we may write as η(t) = (u(t), v(t)).
Since η̇(0) = (u′ (0), v ′ (0)), eqn. (1.7) yields that

w = u′ (0)ϕu (a) + v ′ (0)ϕv (a),

namely, u′ (0), v ′ (0) are the coordinates of w in the basis (1.9). This remark also
shows that another smooth curve γ̄ : (−ǫ, ǫ) → S represents the same w if and
only if η̄(t) = ϕ−1 ◦ γ̄(t) = (ū(t), v̄(t)) satisfies (ū′ (0), v̄ ′ (0)) = (u′ (0), v ′ (0)).
With the same notation as above, suppose now that f : S → S̃ is a smooth
map at p ∈ S. Note that f ◦ γ is a smooth curve in S̃. The differential of f at
p is the map
dfp : Tp S → Tf (p) S̃
that maps w = γ̇(0) ∈ Tp S to the tangent vector γ̃(0),˙ where γ̃ = f ◦γ. We check
that dfp (w) does not depend on the choice of curve γ. Let ϕ : U → ϕ(U ) = V ,
ϕ̃ : Ũ → ϕ̃(Ũ ) = Ṽ be parametrizations of S, S̃, resp., with p = ϕ(a), a ∈ U ,
f (p) = ϕ̃(ã), ã ∈ Ũ , and such that f (V ) ⊂ Ṽ . Consider the local representation
of f ,
g = ϕ̃−1 ◦ f ◦ ϕ : U → Ũ ,
and write
g(u, v) = (g1 (u, v), g2 (u, v))
6 CHAPTER 1. SURFACES: BASIC DEFINITIONS

for (u, v) ∈ U ⊂ R2 . Then

γ̃(t) = ϕ̃ ( g1 (u(t), v(t)), g2 (u(t), v(t)) ) ,

so    
˙ = ∂g1
u′ (0) + ∂g1
v ′ (0) ϕ̃ũ + ∂g ′ ∂g2
v ′ (0) ϕ̃ṽ .
γ̃(0) ∂u u (0) +
2
∂u ∂v ∂v

˙
This relation shows that γ̃(0) depends only on u′ (0), v ′ (0) and hence has the
same value for any smooth curve representing w. This relation can also be
rewritten as  ∂g1 ∂g1   ′ 
∂u ∂v u (0)
dfp (w) = ,
∂g2
∂u
∂g2
∂v
v ′ (0)
which shows that dfp is a linear map whose matrix with respect to the bases
{ϕu , ϕv }, {ϕ̃ũ , ϕṽ } is the 2 by 2 matrix above.

Example 1.10 If S is a surface given as the inverse image under F : R3 → R of

a regular value, then Tp S = ker(dFp ) for every p ∈ S. In fact, if γ : (−ǫ, ǫ) → S
is a smooth curve with γ(0) = p, then F (γ(t)) is constant for t ∈ (−ǫ, ǫ). By
the chain rule, dFp (γ̇(0)) = 0. This proves the inclusion Tp S ⊂ ker(dFp ) and
hence the equality by dimensional reasons.