2024

2024 10 30

1 1
1.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2 6

3 20

4 27

i
 1

1
R2 R3

1.1
γ γ : [a, b] → Rn
1
n = 2, 3 C γ
y = y(x) F :R →R2
F (x, y) = 0
s 7→ t γ(s) := γ(t(s))

1 ′ ′
γ ′ (t) = lim (γ(t + ∆t) − γ(t)) = (γ 1 (t), . . . , γ n (t)).
∆t→0 ∆t
γ(t + ∆t) = γ(t) + ∆tγ ′ (t) + O(∆t2 )

Definition 1

Z t
L(γ)(t) := |γ ′ (t)|dt,
a

L(γ)(b)

u 7→ t
Z ′u Z ′u Z t
dγ dγ dt dγ
L(γ)(u) = | |du = | |du = | |dt.
a du a dt du a dt

arc length parameterization γ(t)

γ(s) : [0, L(γ)] → Rn , s(t) = L(γ)(t).

t s
dt ds 1
= 1/ = ′
ds dt |γ (t)|

dγ dt γ ′ (t)
γ ′ (s) = = ′
dt ds |γ (t)|

1
1.2 1

Definition 2

|γ ′′ (t)⊥ |
k(t) :=
|γ ′ (t)|2

v2
a=
R
1
R

k(t) = |γ ′′ (s)|

κ(s)
κ(s) = γ ′′ (s), n =
k(s)

n normal vector

Proposition 1

κ(s)

• κ(s) γ ′ (s)

• |κ(s)| = k(s) n(s)

κ(s) 2
γ(s + ∆s) − γ(s) = γ ′ (s)∆s + ∆s + O(∆s3 )
2

1.2

d dt d 1 d
= = ′ .
ds ds dt |γ (t)| dt

Definition 3

κ(s)
γ(s) e(s) := γ ′ (s) n(s) =
k(s)

2
1.2 1

1.2.1

n e
π
2
n = R90 (e)

Definition 4

κs (t)

< γ ′′ (t), n(γ(t)) >

κs (t) = ,
|γ ′ (t)|2

κs (s) =< γ ′′ (s), n(s) > .

Remark 1
!
0 1
R90 =
−1 0

e′ (s) = k(s)n(s) < e, n >= 0 0 =< e′ (s), n(s) > + < e(s), n′ (s) >

Proposition 2

! ! !
d e(s) 0 k(s) e(s)
=
ds n(s) −k(s) 0 n(s)

e(s), n(s) k(x) γ(s)

d ′′ d
γ (3) = γ (s) = k(s)n(s)
ds ds
= k ′ (s)n(s) + k(s)n′ (s) = k ′ (s)n(s) − k(s)2 e(s).

1.2.2

γ ′′ (s)
n(s) = .
k(x)
k=0
R3 k 6= 0 binormal vector

b = t × n.

{γ(s), t(s), n(s), b(s)} γ

3
1.2 1

Proposition 3

    
e(s) 0 k(s) 0 e(s)
d   
n(s) = −k(s)
 
0 τ (s)  
 n(s) ,
ds   
b(s) 0 −τ (s) 0 b(s)

τ (s) torsion

. [e, n, b]′ = A[e, n, b] A(s)

d
0= < e, e >= 2 < e′ , e >,
ds

0 =< e′ , n > + < e, n′ >= k(s)+ < e, n′ >,

0 =< n′ , b > + < n, b′ >,

0 =< e′ , b > + < e, b′ > .

τ :=< n′ , b >= − < b′ , n > .

Remark 2

τ γ

γ ′ (s) = e(s),
γ ′′ (s) = k(s)n(s),
γ ′′′ (s) = (kn)′ = k ′ n + k(ke + τ b),

k2 3 k k′ kτ
γ(s) − γ(0) = e(s)(s − s ) + n(s)( s2 + s3 ) + b(s) s3 + O(s4 ).
6 s 6 6
τ
k>0 τ >0 b

4
1.3 1

1.2.3

|a⊥ | e′
k= =
|v |
2 |v|
= γ ′′ (s) n
< a, R90 v >
= 2
|v 2 |
|γ ∧ γ ′′ |
′
= 3 .
|γ ′ |2

γ ′ ∧ γ ′′
ks := .
|γ ′ |2
x ∧ y := x1 y 2 − x2 y 1

< n′ , b > < γ ′ ∧ γ ′′ , γ ′′′ >

τ := = .
|v| |γ ′ ∧ γ ′′ |2

1.3
3

Remark 3

5
 2

Exercise: 1.3

A logarthmic spiral means a plane curve of the form

γ(t) = c eλt cos(t), eλt sin(t) , t ∈ R,

where c, λ ∈ R with c 6= 0. Figure 1.4 shows the restriction to [0, ∞) of a logarithmic spiral with λ < 0. Use
an improper integral to prove that such a restriction has finite arc length even though it makes infinitely many
loops around the origin.

Proof.

γ ′ (t) = ceλt (λ cos(t) − sin(t), λ sin(t) + cos(t)) ,

 1
1
kγ ′ (t)k
2 2 2
= |c|eλt (λ cos(t) − sin(t)) + (λ sin(t) + cos(t)) = |c|eλt 1 + λ2 2 ,
Z ∞ Z √

1 ∞ λt |c| 1 + λ2
kγ ′ (t)kdt = |c| 1 + λ2 2 e dt =
0 0 −λ

Exercise: 1.4

t π
The curve γ(t) = (sin(t), cos(t) + ln(tan( ))), t ∈ ( , π), is shown in Fig. 1.5. Demonstrate that for every
2 2
point p of its image, the segment of the tangent line at p between p and the y-axis has length 1. This curve is
called a tractrix (from the Latin ”to drag”), because it represents the path of an object tethered by a length-1
rope to a tractor moving upward along the positive y-axis. COMMENT: See the Wikipedia ”tractrix” page for
a nice animation of this. https://math.fandom.com/zh/wiki/

Proof. The velocity vector of γ at t is

 
cos2 (t)
γ ′ (t) = cos(t), .
sin(t)

The tangent line of γ at t is

l(s) = (x(s), y(s)) := γ(t) + sγ ′ (t).

Notice that l(0) = γ(t), x(− tan(t)) = 0, |l′ (s)| = |γ ′ (t)| = − cot(t), thus the segment of the tangent line at p
between p and the y-axis is
L = ∆(x)|l′ (s)| = 1.

6
 2

Exercise: 1.9

Let a, b > 0. Find the maximum and minimum speed of the ellipse γ(t) = (a cos t, b sin t).

Proof.

γ ′ (t) = (−a sin t, b cos t),

q q p
|γ ′ (t)| = a2 sin2 (t) + b2 cos2 (t) ≤ max{a2 , b2 }(sin2 (t) + cos2 (t)) ≤ max{a2 , b2 };
q q p
|γ ′ (t)| = a2 sin2 (t) + b2 cos2 (t) ≥ min{a2 , b2 }(sin2 (t) + cos2 (t)) ≤ min{a2 , b2 }.

Exercise: 1.11

Figure 1.6 shows a polygonal approximation of the regular curve γ : [a, b] → Rn . This polygonal approximation
is determined by a partition, a = t0 < t1 < t2 < · · · < tk = b. The sum of the lengths of the line segments
X
k−1
equals L = |γ (ti+1 ) − γ (ti )|. The mesh of the partition is defined as δ = max {ti+1 − ti }. Prove that L
i=0
converges to the arc length of γ for every sequence of partitions for which δ → 0.

HINT: For notational simplicity, suppose first that n = 2, and denote the component functions by γ(t) =
(x(t), y(t)). For each i ∈ {0, 1, . . . , k − 1}, the mean value theorem guarantees a sample point t∗i ∈ [ti , ti+1 ]
x (ti+1 ) − x (ti )
such that x′ (t∗i ) = , where ∆ti = ti+1 − ti . Similarly, there exists a sample point t∗∗ i ∈ [ti , ti+1 ]
∆ti
y (ti+1 ) − y (ti ) Xq
k−1
such that y ′ (t∗∗ x′ (t∗i ) + y ′ (t∗∗ ∗ ∗∗
2 2
i ) = . Notice that L = i ) ∆ti . If we knew that ti = ti for all
∆ti i=0
Z b
i, then this expression would be a Riemann sum approximation of |γ ′ (t)| dt. But since we can’t guarantee that
a
ti∗ = t∗∗ ′ ′
i , we must instead use the uniform continuity of x and y to show that this expression is close to such a
Riemann sum.

Proof. (1). The mean value theorem guarantees: ∀i, j, ∃tji ∈ (ti , ti+1 ), s.t.

xj (ti+1 ) − xj (tI )
= x′j (tji ), ∆ti = ti+1 − ti .
∆ti

(2). By the uniform continuity of x′i , ∃σ > 0, s.t. ∀t, s ∈ [a, b], with |a − b| < σ, holds |x′i (t) − x′i (s)| < ϵ, i =
1, · · · , n.
Z b
(3). ∀ϵ > 0, ∃0 < λ < σ, s.t. if ∆ < λ, then |γ ′ (t)|dt − Σk−1 ′ 1
i=0 |γ (ti )|δ(ti ) < ϵ.
a
Thus by (1)-(3), we have
Z b Z b
′
L− |γ (t)|dt ≤ |γ ′ (t)|dt − Σk−1 ′ 1 k−1 ′ 1
i=0 |γ (ti )|δ(ti ) + L − Σi=0 |γ (ti )|δ(t) i
a a
q q 
≤ Σk−1
i=0 Σnj=1 |xj (ti+1 ) − xj (ti )|2 − Σnj=1 |x′j (t1i )|2 δ(ti ) + ϵ

7
 2

q q 
= Σk−1
i=0 Σnj=1 |x′j (tji )|2 n ′
− Σj=1 |xj (ti )| δ(ti ) + ϵ
1 2

r !
 2
≤ Σk−1
i=0 Σnj=1 |x′j (ti )| − |x′j (t1i )| δti + ϵ
j

< [n(b − a) + 1]ϵ.

Exercise: 1.16

Let γ be a logarithmic spiral, as defined in Exercise 1.3 on page 6. Prove that the angle between γ(t) and γ ′ (t)
is a constant function of t.

Proof.


γ(t) = c eλt cos(t), eλt sin(t) , t ∈ R,
γ ′ (t) = ceλt (λ cos(t) − sin(t), λ sin(t) + cos(t)) ,
γ(t), γ ′ (t) λ
cos θ = ′
=√ .
kγ(t)kkγ (t)k 1 + λ2

Exercise: 1.23

Let γ : I → R3 be a regular space curve, and let P ⊂ R3 be a plane that does not intersect the image of γ. If
γ comes closest to P at time t0 , prove that γ ′ (t0 ) is parallel to P .


Proof. Let P = (x, y, z) ∈ R3 | h(x, y, z), ⃗ni = 0 , where ⃗n is a constant unit vector, P0 ∈ P , and

S(t) := distanceof γ(t)andP = hγ(t) − p0 , ⃗ni .

ds
∵ S (t0 ) is a locally minimal point of S(t), ∴ (t0 ) = hγ ′ (t0 ), ⃗ni = 0, i.e., γ ′ (t0 ) parallel to P .
dt

Exercise: 1.24

If t0 ∈ I is the time at which the curve γ : I → Rn is farthest from the origin, prove that γ (t0 ) is orthogonal
to γ ′ (t0 ).

Proof. Let l(t) := |γ(t)|2 , then l′ (t) = z hγ(t), γ ′ (t)i.

∵ l (t0 ) is maxional, ∴ 0 = l′ (t0 ) = 2 hγ(t0 ), γ ′ (t0 )i.

8
 2

Exercise: 1.27

 in R with |γ(t)| = C (a constant), prove that ha(t), −(t)i = |v(t)|2 . Rewrite this as
n
If
 γ is a curve
γ(t) |v(t)| 2
a(t), − = , and notice that the left side is the component of a(t) in the direction of the
|γ(t)| C
center-pointing vector. Interpret this physically in terms of centripetal force.

Proof. Let γ be a curve of unit-speed, thus a(t) = a⊥ (t) ⊥ v(t).

∵ c2 = |γ(t)|2 = hγ(t), γ(t)i,

d
∴0 = hγ(t), γ(t)i = 2hγ(t), γ ′ (t)i,
dt
d
0 = hγ(t), γ ′ (t)i = hγ ′ (t), γ ′ (t)i + hγ(t), γ ′′ (t)i ,
dt
i.e.|v(ti|2 = −hγ(t), a(t)i = −hγ(t), a⊥ (t)i = |γ(t)| · |a⊥ (t)|,

Physically, Centripetal force is the force that keeps an object moving in a circular path. It acts towards the
center of the circle or curve around which the object is moving. This inward force prevents the object from moving
in a straight line (due to inertia) and instead causes it to continuously change direction, thus maintaining a circular
motion.
If an object with mass m powered by a centripetal force F , then it has acceleration a in the same direction of
|v|2
F , i.e., F = ma. The norm of a is just , Where v is the velocity of the object, and r is the radius of the circular
r
path.

Exercise: 1.28


Find a space curve γ : R → R3 with acceleration function a(t) = t2 − 1, t3 , t2 + 1 . How unique is the
solution?

Proof.
Z  


1 3 1 4 1 3
γ(t) = t − 1, t , t + 1 dt =
2 3 2
t − t, t , t + t + ⃗v (0),
3 4 3
Z  
′ 1 4 1 2 1 5 1 4 1 2
γ(t) = γ (t)dt = t − t , t , t + t + ⃗v (0)t + ⃗γ (0)
12 2 20 12 2
∴ γ is unique up to the choice of ⃗v (0) and ⃗γ (0).

Exercise: 1.29

Consider the following pair of plane curves:

γ(s) = (cos s, sin s), s ∈ (−π, π)

 
1 − t2 2t
γ̃(t) = , , t∈R
1 + t 2 1 + t2

9
 2

Verify that γ̃ is a reparametrization of γ. HINT: t = tan(s/2).

s
Proof. ϕ : (−π, π) → R, s → tan( ), gives a smooth bijection from (π, π) to R. And
2
1 − tan2 2s 2 tan 2s
γ̃(ϕ(s)) = ( , ) = (cos s, sin s) = γ(s), s ∈ (−π, π).
1 + tan2 2s 1 + tan2 2s

Thus γ̃ is a reparametrization of γ.

Exercise: 1.30

Let γ : I → Rn be a regular curve, and let γ̃ = γ ◦ ϕ : I˜ → Rn be a reparametrization of γ, as in Definition

1.22. (1) Is it possible that I is unbounded while I˜ is bounded? (2) Is it possible that I is noncompact while
I˜ is compact? (3) Is it possible that I = (a, ∞) while I˜ = [b, ∞) ?

Proof. (1). Yes. Exercise 1.29 gives an example.

(2). No. Since a continuous function maps a compact set to a compact set.
(3). No. Suppose there is such a reparametrization, there is a smooth bijection function f with f ′ not changes
sign. Since f is monotonous, thus f maps [b, ∞) to [f (b), ∞) or (a, f (b)], contradiction!

Exercise: 1.31

Precisely state and prove the assertion that the definition of arc length is independent of parametrization for
regular curves.

Assertion : Let γ : I → Rn and its reparametrization γ̃ = γ ◦ ϕ : I˜ → Rn be two regular

Z curves, ϕZ: I → I˜ be a
smooth bijection with nonzero derivative. Then they have same arc length: L(γ) := |γ ′ (t)| dt = |γ̃ ′ (s)| ds =
I I˜
L(γ̃).

Proof.
 Z
Z 
 | γ ′ (ϕ(s)) | ϕ′ (s)ds, if ϕ′ > 0, Z Z
I˜Z
t=ϕ(s)
|γ ′ (t)| dt ====== ′ ′
= |γ (ϕ(s)) · ϕ (s)| ds = |γ̃ ′ (s)| ds.
I 
 − | γ ′ (ϕ(s)) | ϕ′ (s)ds, if ϕ′ < 0 I˜ I˜
I˜

Where the minus appear above is because of the inverse direction of the integral.

Exercise: 1.39

For constants a, b, c > 0, consider the ”generalized helix” defined as γ(t) = (a cos t, b sin t, ct), t ∈ R. Where is
the curvature maximal and minimal?

10
 2

Proof.

v = γ ′ = (−a sin t, b cos t, c)

a = γ ′′ = −(a cos t, b sin t, 0)
"
# 12
a2 b2 + c2 a2 cos2 t + b sin2 t
a|| =
a2 sin2 t + b2 cos2 t + c2
 2 2
 1
a⊥ a b + c2 a2 cos2 t + b sin2 t 2
k(t) = =
3
|ν|2 a2 sin2 t + b2 cos2 t + c2 2

1

 a 2 b2 + c 2 · a 2 2 a

 = 2 ( when cos t = 1), if a ≤ b,
 2 2
1
b + c2
(b + c ) 2
∴ k(t) ⩾
1

 a 2 b2 + c 2 · b 2 2


b
( when cost = 0), if a > b.
 1 = 2
2
(a + c ) 2 2 a + c2

1

 a 2 2
b + c 2
· b 2 2
b

 = 2 ( when cos t = 0), if a ≤ b,
 2 2
1
a + c2
⩽ (a + c ) 2
k(t)
1

 a 2 b2 + c 2 · a 2 2


a
( when cost = 1), if a > b.
 1 = 2
2
(b + c ) 2 2 b + c2

Exercise: 1.41

Let γ : I → R3 be a unit-speed space curve with component functions denoted by γ(t) = (x(t), y(t), z(t)). The
plane curve γ̄(t) = (x(t), y(t)) represents the projection of γ onto the xy-plane. Assume that γ ′ is nowhere
parallel to (0, 0, 1), so that γ̄ is regular. Let κ and κ̄ denote the curvature functions of γ and γ̄ respectively.
Let v, v denote the velocity functions of γ and γ̄ respectively.
(1) Prove that κ ≥ κ̄|v|2 . In particular, at a time t ∈ I for which v(t) lies in the xy-plane, we have
κ(t) ≥ κ̄(t).

(2) Suppose the trace of γ lies on the cylinder (x, y, z) ∈ R3 | x2 + y 2 = 1}. At a time t ∈ I for which
v(t) lies in the xy-plane (so that γ is tangent to the ”waist” of the cylinder), conclude that κ(t) ≥ 1. Is there
any upper bound for κ(t) under these conditions?
(3) Find an optimal lower bound for κ(t) at a time t ∈ I when v(t) makes the angle θ with the xy-plane.

q
′′ ā⊥
. ∴ κ̄ ≥ |γ̄”| ≥ ā⊥ = κ̄|v̄|2 .
2 2 2
Proof. (1) ∵ κ(t) = |a(t)| = |γ (t)| = (x′′ ) + (y ′′ ) + (Z ′′ ) ⩾ |γ̄”|. And κ̄ =
|v̄|2
In particular, if v(t) is in xy-plane, then |v| = |v| = 1, thus k ≥ κ̄.
(2) γ̄(t) = (cos t, sin t) with γ̄ ′ (t) = (− sin t, cos t) is a unit-speed curve, thus κ̄ = |ā(t)| = γ̄”(t) = | −
(cos t, sin t)| = 1.
If at a time t ∈ I for which v(t) lies in the xy-plane, then by (1), κ(t) ≥ κ̄(t) = 1.
There is no upper bound for κ(t) under these conditions. For example, ∀M > 0, consider the curve


γ̂ = cos t, sin t, M t2 , v̂(t) = γ̂ ′ (t) = (− sin t, cos t, 2M t).
â(0) = −(cos t, sin t, 2M ), v̂(0) = (0, 1, 0), â(0) = (−1, 0, 2M ) ⊥ v̂(0)

11
 2

p
ˆ (0) = |â(0)| /|v̂|2 =
Thus κ̄ 1 + 4M 2 > M .

(3) If γ is still lies on the cylinder as in (2), then by (1), κ(t) ≥ κ̄(t)|v̄(t)|2 = 1 · |v(t) cos θ|2 = cos2 θ.
If γ is not restricted on the cylinder, then the lower bound for κ at t may takes 0, for example γ(t) = (0, t, t tan θ).

Exercise: 1.43

Let γ : I → Rn be a regular curve. Assume that the function t → |γ(t)| has a local maximum value of r
1
occurring at time t0 . Prove that κ(t0 ) ≥ . Is there any upper bound for κ(t0 ) under these conditions?
r

Proof. Let l(t) := |γ(t)|2 , γ(t) be of arc-length parametrized. Then

l′ (t) = 2hγ, vi, l′′ = 2(hv, vi + hγ, ai) = 2(1 + hγ, ai) = 2 (1 + |γ|κ cos(arg(γ, a)).
Since |γ(t)| has a local maximum value of r occurring at time t0 , thus v(t0 ) = 0, and a(t0 ) ≤ 0.
1 1
Thus 1 + rκ(t0 ) cos(arg(γ(t0 ), a(0)) ≤ 0, κ(t0 ) ≥ ≥ .
−r cos(arg(γ(t0 ), a(0)) r
(2) No upper bound for k (t0 ) under these conditions. For example, γ(t) = β(t)|γ(t)|, |β(t)| = 1. than β ⊥ β ′ ,
and

γ′ = β ′ |γ| + β|γ|′ , γ ′ (t0 ) = β ′ (t0 ) |γ (t0 )| = rβ ′ (t0 ) .

′ ′′
γ ′′ = β ′′ |γ| + 2β ′ |γ| + β |γ| , γ ′′ (t0 ) = (rβ” + β|γ|”) (t0 ),
a⊥ r(β”)⊥ (t0 ) + |γ|”(t0 )β (t0 )
∴ κ(t0 ) = (t0 ) = .
|v| 2 r2 |β ′ (t0 ) |2

Since |γ| and β are independent with each other, and we can take |γ|”(t0 ) < 0 as any small as possible, hence κ(t0 )
has no bound.

Exercise: 1.45


For the parabola γ(t) = t, t2 , at an arbitrary time t ∈ R : (1) Compute κ(t) from Definition 1.30 by decom-
posing a(t) = a(t)∥ + a(t)⊥ . (2) Compute κ(t) using Proposition 1.34. (3) Compute κs (t) using Definition
1.38 .

hγ ′ , ai (−4t, 2)
Proof. (1) v(t) = (1, 2t), a(t) = (0, 2), a⊥ = a − 2 γ′ = .
|γ ′ | 1 + 4t2
a⊥
3
Thus κ(t) = = 2/ 1 + 4t2 2 .
|v| 2
(2)


(0, 2t) ⃗′ −4t2 ⃗t′ 2
⃗t = √ ,t = 3 , κ = = 3 .
1 + 4t 2
(1 + 4t2 ) 2 |v| (1 + 4t2 ) 2

2
(3) R90 (v) = (2t, 0), κs = ha, R90 (v)i = 3 .
(1 + 4t2 ) 2

12
 2

Exercise: 1.48

Prove that the signed curvature function of a regular plane curve described as γ(t) = (x(t), y(t)) is

x′ (t)y ′′ (t) − x′′ (t)y ′ (t)

κs (t) = 3
(x′ (t)2 + y ′ (t)2 ) 2

Proof.

γ′ = (x′ , y ′ ) , γ ′′ = (x′′ , y ′′ ) , Rqp (γ ′ ) = (−y ′ x′ ) .

hγ ′′ , R90 (γ ′ )i x′ y ′′ − x′′ y ′
ks (t) = 3 = 3
|γ ′ | 2 2 2
(x′ ) + (y ′ )

Proof. Let γ̃ := γ ◦ ϕ be a reparametrization of γ, ϕ′ > 0. Then

γ̃ ′ (s) = γ ′ (ϕ(s))ϕ′ (s), γ̃ ′′ (s) = γ ′′ (ϕ(s))(ϕ′ (s))2 + γ ′ (ϕ(s))ϕ′′ (s).

hγ̃ ′′ (s), R90 (γ̃ ′ )i hγ ′′ (ϕ(s))(ϕ′ (s))2 + γ ′ (ϕ(s))ϕ′′ (s), R90 (γ ′ (ϕ(s))ϕ′ (s))i
κs (γ̃(s)) = =
|ṽ|3 |γ ′ (ϕ(s))ϕ′ (s)|3
hγ ′′ (ϕ(s))(ϕ′ (s))2 , ϕ′ (s)R90 (γ ′ (ϕ(s)))i
= = κs (γ)(ϕ(s))
|γ ′ (ϕ(s))|3 ϕ′ (s)3

Exercise: 1.51

Find a plane curve γ : R → R2 whose global angle function is θ(t) = 2t. How unique is the answer?
Z
1
Proof. v(t) = (cos(2t), sin(2t)), thus γ(t) = v(t)dt = (sin(2t), − cos(2t)) + γ(0). It is unique up to the choice of
2
γ(0).

Exercise: 1.54

Prove that the signed curvature of a polar coordinate function r(θ) is

2 (r′ ) − rr′′ + r2
2
κs =  3/2
2
(r′ ) + r2

Proof. Let γ(θ) = (r cos θ, r sin θ), then

γ ′ = (r′ cos θ − r sin θ, r′ sin θ + r cos θ),
γ ′′ = (r′′ cos θ − 2r′ sin θ − r cos θ, r′′ sin θ + 2r′ cos θ − r sin θ).
hγ ′′ , R90 (γ ′ )i 2 (r′ ) − rr′′ + r2
2
Thus κs (θ) = 3 =  3/2 .
|r′ | 2
(r′ ) + r2

13
 2

Exercise: 1.55

(Rotation Index of Variable-Speed Curves). Let γ : [a, b] → R2 be a regular (not necessarily unit-speed) closed
plane curve. (1) Prove there exists a smooth function θ : [a, b] → R such that for all t ∈ [a, b], the unit tangent
vector satisfies

t(t) = (cos θ(t), sin θ(t))

1
and that the rotation index of γ equals (θ(b) − θ(a)). (2) Prove that the rotation index of γ equals
2π
Z b
1
rotation index = κs (t) |γ ′ (t)| dt
2π a
HINT: For both parts, verify that the formula is valid for unit-speed curves and is unchanged by
orientation-preserving reparametrizations.

Proof. (1) For γ : [a, b] → R2 , ∃ smooth bijection ϕ : Ie = [ã, b̃] → [a, b], s.t. γ̃ = γ ◦ ϕ, and ϕ > 0 and |γ̃ ′ (s)| = 1.
For umit-speed curve γ̃, by proposition 1.39,

t̃(s) = γ̃ ′ (s) = (cos θ̃(s), sin θ̃(s)),

1
and the rotation index of γ̃ equals (θ̃(b̃) − θ̃(ã)).
2π
γ ′ (t)
∵ t̃(s) = γ̃ ′ (s) = γ ′ (ϕ(s)) · ϕ′ (s), t(t) = ,
|γ ′ (t)|
γ ′ (ϕ(s))ϕ′ (s)
∴ t(ϕ(s)) = = t̃(s),
|γ ′ (ϕ(s))ϕ′ (s)|
t(t) = (cos θ̃(ϕ−1 (t)), sin θ̃(ϕ−1 (t))),
1 1
and the rotation index of γ equals (θ̃(b̃) − θ̃(ã)) = (θ(b) − θ(a)).
2π 2π

(2) Rotation index of γ = Rotation index of γ̃

Z b̃ Z b̃
1   1 1
= θ̃(b̃) − θ̃(ã) = θ̃′ (s)ds = κ̃s (s)ds
2π 2π ã 2π ã
Z b̃ Z b Z b
1 1 1
= κs (ϕ(s))ds = ks (t)(ϕ−1 )′ (t)dt = κs (t) |γ ′ (t)| dt.
2π ã 2π a 2π a

14
 2

Exercise: 1.56

Let m, n be positive integers. Find the rotation index of the Lissajous curve γ : [0, 2π] → R2 (see Fig. 1.26)
defined as
γ(t) = (cos(mt), sin(nt))

˜ =
Proof. If the greatest common divisor d = gcd(m, n) 6= 1, the curve is a d-times covering of the curve γ(s)
(cos(m1 t), sin(n1 t)), t ∈ [0, 2π], thus the result is d-times of that of γ̃. We only need to consider the case when
d = 1 below.


γ(t) = (cos mt, sin nt), v(t) = (−m sin mt, n cos nt), a(t) = − m2 cos mt, n2 sin nt ,
ha(t), Rq0 (v(t)ii m2 n cos nt cos nt + mn2 sin mt sin nt
κs (t) · |v(t)| = =
|v(t)| 2
m2 sin2 mt + n2 cos2 nt
 ′
sin mt cos2 nt
= mn 2
cos nt m2 sin mt + n2 cos2 nt
′ m ′
h:= sin mt h (t) n h (t) y:= m
n h y′
====cos=nt
== mn 2 =
2 = === == .
m h(t) + n2 m
h(t) + 1 1 + y2
n

Z Z ∞
1 X 1 X 1X
2π 2n 2n 2n
1 dy
∴ rolation index = ks (t)|v(t))|dt = δ(l) 2
= δ(l) arctan y = δ(l),
2π 0 2π Il 1 + y 2π 2
l=1 l=1 −∞ l=1

where δ(l) ∈ {−1, 0, 1} depends on the signs of sin(nt) when cos(mt) → ±0.

Exercise: 1.58

Let γ : R → R2 be a unit-speed plane curve. Let κs be its signed curvature function, and let θ be a global
angle function for γ. Prove or disprove: (1) If θ is periodic, then γ is periodic. (2) If κs is periodic, then θ is
periodic.

Proof. (1) The θ of the curve y = sin(x) is periodic, but the trace of it is not periodic.
(2) The Archimedean spiral γ defined in polar coordinates as: r(θ) = a + bθ, where a and b are constants, and
θ is the polar angle.
The κs of γ is periodic, but θ is not periodic.

Exercise: 1.60

In Example 1.41, express the signed curvature functions of γL and γR in terms of c and the signed curvature
function of γ.

Proof. Let γ be of unit-speed curve. By definition, γL = γ + cR90 (γ ′ ) , γR = γ − cR90 (γ ′ ) .

15
 2

γL′ = γ ′ + cR90 (γ ′′ ) = γ ′ + cR90 (κs R90 (γ ′ )) = (1 − cκs ) γ ′ , γL′′ = −cκ′s γ ′ + (1 − cκs ) γ ′′ ,

hγL′′ , Rq0 (γL′ )i κs
κs (γL ) = = 2.
|γL′ |3 (1 − cκs )
κs
Similarly, κs (γR ) = 2.
(1 + cκs )

Exercise: 1.62


Calculate the curvature function and torsion function for the curve γ(t) = t, t2 , t3 , t ∈ R.

Proof.


γ′ = 1, 2t, 3t2 , γ ′′ = (0, 2, 6t), γ ′′′ = (0, 0, 6)


γ′ × γ ′′ = 6t2 , −6t, 2 ,

1
|γ ′ × γ ′′ | 2 1 + 9t2 + 9t4 2
k(t) = 3 = 3
|γ ′ | (1 + 4ψ 2 + 9t4 ) 2
hγ ′ × γ ′′ , γ ′′′ i 12 3
τ (t) = 2 = = .
′
|γ × γ | ′′ 4 (1 + 9t 2 + 944 ) 1 + 9t 2 + 944

Exercise: 1.64

For the helix in Example 1.3, defined as γ(t) = (cos t sin t, t), t ∈ R, compute the torsion function. Do the
same for the helix defined as β(t) = (cos t, sin t, −t), t ∈ R. Describe the visual difference between a helix with
positive torsion and a helix with negative torsion.

Proof.

γ′ = (− sin t, cos t, 0), γ ′′ = −(cos t, sin t, 0), γ ′′′ = (sin t, − cos t, 0).
hγ ′ × γ ′′ , γ ′′′ i 1
τγ = =
|γ ′ × γ ′′ |2 2
1
Similarly, τβ = − .
2

16
 2

Exercise: 1.65

Let γ : I → R3 be a regular space curve, not necessarily parametrized by arc length. Prove that at every t ∈ I
with κ(t) 6= 0, the torsion is given by the formula

hγ ′ × γ ′′ , γ ′′′ i
τ= 2
|γ ′ × γ ′′ |
HINT: First prove the result for a unit-speed curve, and then prove that the formula is independent of
parametrization.

Proof. For unit-speed parameterized curve γ̃,

1˙ ¨
ṽ = γ̃˙ = t̃, γ̃¨ = t̃˙ = κ̃ñ, b̃ = t̃ ∧ ñ = γ̃ × γ̃ ⇒ k̃ = |γ̃˙ × γ̃|
¨
κ̃
γ̃¨ = ˜˙ + κ̃ñ˙ = κ̇ñ − κ̃2 t̃ + κ̃γ̃ b̃, ⇒ hγ ′ × γ ′′ , γ ′′′ i = k̃ 2 γ̃,
kκñ
hγ̃ ′ × γ̃ ′′ , γ̃ ′′′ i
∴τ = 2
|γ̃ ′ × γ̃ ′′ |

If γ = γ̃ ◦ ϕ is a usual reparameterizated curve, then

...
γ ′ = γ̃˙ · ϕ′ , γ ′′ = γ̃(ϕ
¨ ′ )2 + γ̃ϕ
˙ ′′ , γ ′′′ = γ̃ (ϕ′ )3 + 3γ̃ϕ
¨ ′ ϕ′′ + γ̃ϕ
˙ ′′′ .

T hus γ ′ × γ ′′ = γ̃˙ × γ̃(ϕ

¨ ′ )3 , hγ ′ × γ ′′ , γ ′′′ i = hγ̃ ′ × γ̃ ′′ , γ̃ ′′′ i (ϕ′ )6
hγ ′ × γ ′′ , γ ′′′ i hγ̃ ′ × γ̃ ′′ , γ̃ ′′′ i
and 2 = 2 = τ.
|γ ′ × γ ′′ | |γ̃ ′ × γ̃ ′′ |

Exercise: 1.66

For a space curve γ : I → R3 at a time t0 ∈ I with κ (t0 ) 6= 0, the rectifying plane means the plane P passing
through γ (t0 ) spanned by t (t0 ) and b (t0 ). Prove that there exists ϵ > 0 such that the trace of the restriction
of γ to (t0 − ϵ, t0 + ϵ) lies on a single side of P .

Proof. Let γ e of unit-speed.

′
hγ(t) − γ(t0 ), ⃗n0 i = ⃗t, ⃗nb
∆t2 ′′

γ = γ (t0 ) + ∆tγ ′ (t0 ) + γ (t0 ) + o ∆t2
2
∆t 2

= γ (t0 ) + ∆t⃗t0 + κ⃗n0 + o ∆t2
2
Since γ (t0 ) + ∆t⃗t0 lie in the rectifying plane, and κ 6= 0 in a neighborhood (t0 − ϵ, t0 + ϵ) , thus γ lie in one side of
the rectifying plane near t0 .

17
 2

Exercise: 1.68

Let γ : I → R3 be a space curve, and let t0 ∈ I with κ (t0 ) 6= 0. Let P denote the osculating plane at t0 (
translated to γ (t0 )). For t ∈ I near t0 , let β(t) denote the point of P closest to γ(t). Prove that γ and β have
the same curvature at time t0 .

Proof. Let γ be a unt-apeed curve, and {⃗t, ⃗n, ⃗b} be its Fronet frame. Denote the evaluation of {⃗t, ⃗n, ⃗b} at t0 as
{⃗t0 , ⃗n0 , ⃗b0 }, then γ := x⃗t0 + y⃗n0 + z⃗b0 , and γ̃ := x⃗t0 + y⃗n0 , is the projection of γ onto its escalating plane at t0 .

⃗t = γ ′ = x′⃗t0 + y ′⃗n0 + z ′⃗b0 ,

⃗t′ = x′′⃗t0 + y ′′⃗n0 + z ′′⃗t0 = κ⃗n.

Thus


 x′ (t0 ) = 1,


 y ′ (t ) = z ′ (t ) = 0,
0 0


n ′′
x (t0 ) = z (t0 ) = 0,


 ′′
y (t0 ) = κ (t0 ) ⃗n0 ,

and

γ̃ ′ = x′⃗t0 + y ′⃗n0 , γ̃ ′′ = x′′⃗t0 + y ′′⃗n0 ,

γ̃ ′ × γ̃ ′′
κ̃(t0 ) = (t0 ) = κ(t0 ).
|γ̃ ′ |3

Exercise: 1.69

Let γ : I → R3 be a regular space curve (possibly with points where κ = 0 and hence where τ is undefined).
Prove or disprove: (1) If the trace of γ lies in a plane, then τ equals zero everywhere it is defined. (2) If τ
equals zero everywhere it is defined, then the trace of γ lies in a plane.

Proof. Without loss of generalinity, we can assume γ to be of unit-speed.

(1) Since γ lies in a plane, thus there is a constant vector w such that

hγ − γ0 , wi = 0, ⇒ h⃗t, wi = 0, ⇒ hκ⃗n, wi = 0,

When τ has definition, κ 6= 0, thus h⃗n, wi = 0, ⇒ 0 = h−κ⃗t + τ⃗b, wi = hτ⃗b, wi.

∵ b||w, ∴ hb, wi 6= 0, ∴ τ = 0.
Another proof: Let γ = x⃗t0 + y⃗n0 , then

⃗t = x′⃗t0 + y ′⃗n0 , κ⃗n = ⃗t′ = x′′⃗t0 + y ′′⃗n0 ,

= ⃗t × κ⃗n = (x′ y ′′ − y ′ x′′ ) t⃗0 × ⃗n0 = (x′ y ′′ − x′′ y ′ ) ⃗b0 .
κ⃗b
 ′ ′′ ′
x y − x′′ y ′ ⃗ ⃗
When τ has definition, κ 6= 0, and −τ⃗n = ⃗b′ = b0 //b. Since ⃗n ⊥ ⃗b, thus τ = 0.
κ

18
 2

(2) IfD τ ≡E0 when ⃗′ ⃗ ⃗

D it E has definition,
D D b E= −τ⃗
E′ then D n = 0, Ethus Db ≡ b0 . E
0 = ⃗t, ⃗b0 = γ ′ , ⃗b0 = γ, ⃗b0 ⇒ γ, b⃗0 = γ(ti ), ⃗b ⇒ γ − γ(ti ), ⃗b = 0 i.e., the trice of every segment
of γ with τ has definition lies in a plane, but there maybe more then one such segments and they maybe lie in
different planes.

19
 3

Exercise: 2.2

Prove Proposition 2.3. HINT: Use a compactness argument to divide [a, b] into finitely many subintervals, on
each of which the image of f is completely contained in one of the following four half-circles: top, bottom,
right, left. Define a local angle function on each subinterval. Working from left to right, add the correct integer
multiple of 2π to each local angle function so they match to form a global angle function.

1
proof. Step 1 ∵ f is contionurus on [a, b], ∴ ∃δ > 0, s.t. ∀t, s ∈ [a, b], with |t − s| < δ, ⇒ |f (t) .f (s)| <
2
Thus we can take partition a = t0 < t1 < · · · < tn = b, s.t. max{ti+1 − ti } < δ, thus the length of the image of
[ti , tt→1 ] is no more than 1, thus the image of [ti , tt→1 ] is completely contained in one of the four half-circles.
After union of some sub-intervals, we may assume the image of every next sub-interval lies in different half-plane
with the former.
Step 2 Since f ([ti , ti+1 ]) lies in half-plane, thus it can be parameterized by its x-component or y-component
coordinate as:
On [si , si+1 ], defone


 arccos(x(s)) ∈ [0, π], if f ([si , si+1 ]) lie in the top half-plane;



 π + arccos(x(x)) ∈ [π, 2π], if f ([si , si+1 ]) lie in the bottom half-plane;
 h π πi
θ̃i (s) := arcsin(y(x)) ∈ − , , if f ([si , si+1 ]) lie in the right half-plane; (1)

 2 2 



 π 3π
 π + cos sin(y(x)) ∈ , , if f ([si , si+1 ]) lie in the left half-plane.
2 2

And we have f (s) = (cos(θ̃), sin(θ̃)).

Step 3 Since f ([ti , ti+1 ]), i = 0, · · · , n − 1, and f ([ti+1 , ti+2 ]), i = 0, · · · , n − 2, intersect in some one quadrant,
thus we can define a uniform continuous function.
X
On [s0 , s1 ], define θ = θ̃, and on [si+1 , si+2 ], define θ = θ̃i+1 + δk · 2π, where
k≤i+1



 1, if f ([sk , sk+1 ]) and f ([sk+1 , sk+2 ]) lie in bottom and right half-plane respectively;
δk+1 = −1, if f ([sk , sk+1 ]) and f ([sk+1 , sk+2 ]) lie in right and bottom half-plane respectively; (2)


 0, otherwise.

Then θ(t− +
i+1 ) = θ(ti+1 ), i.e., θ is continuous on [a, b], and satisfying f (t) = (cos(θ(t)), sin(θ(t))).

20
 3

Exercise: 2.4

Suppose p > q > 0 and consider the ellipse γ(t) = (p cos(t), q sin(t)). The foci of this ellipse are the two
p
points on the x-axis with x-coordinates ± p2 − q 2 , colored purple in Fig. 2.17. (1) Prove that the sum of the
distances from γ(t) to these two foci is independent of t. (2) Prove that the signed curvature function of the
ellipse is

pq
κs (t) =
23
p2 sin2 (t) + q 2 cos2 (t)
 
π 3π
(3) Prove that the critical points of the signed curvature function occur at t ∈ 0, , π, . The
2 2
corresponding points on the ellipse are its intersections with the x - and y-axes.

 p  p  p p
. (1) γ(t) − − p2 − q 2 , 0 + γ(t) − γ 2 − q 2 , 0 = p2 − q 2 cos t + p + p − p2 − q 2 cos t = 2p. (2),(3)

Exercise: 2.5

Prove that the definition of vertex is independent of parametrization.

. Let γ̃ = γ ◦ ϕ be a reparametrization of γ, then ϕ′ (s) 6= 0, and κ̃s (s) = ±κs (ϕ(s)).

Then t0 is a vertex of γ iff
dγ ′ d2 γ ′
(t0 ) = 0, and (t0 ) > 0 (or < 0),
dt dt2
iff
dγ̃ −1 dγ d2 γ̃ d2 γ
(ϕ (t0 )) = (t0 )ϕ′ (ϕ−1 (t0 )) = 0, and 2 (ϕ−1 (t0 )) = 2 (t0 )(ϕ′ (ϕ−1 (t0 )))2 > 0 (or < 0).
ds dt ds dt
i.e., iff s0 = ϕ−1 (t0 ) is a vertex of γ̃.

Exercise: 2.10

Let γ be a closed plane curve whose signed curvature does not change sign. If the rotation index of γ equals
±1, prove that γ is simple.

. ks ≥ 0 RI = 1 γ k I1 , I1′ , . . . Ik , Ik′
Ik+1 = I − ∪kj=1 (Ij ∪ Ij′ ) γj := γ|Ij ∪Ij′ γk+1 = γ|Ik+1
0 k=1 t1 < t 2 γ(t1 ) = γ(t2 ) γ1 := γ|[t1 ,t2 ]
General Hopf’s Umlaufsatz j
Z X
ks ds + αjl = 2π,
γj l

k+1 Z
ks ds = 2(k + 1)π,
γ

21
 3

RI = k RI = 1 simple

. from γ simple t1 < t 2 γ(t1 ) = γ(t2 )

v f (t) =< γ(t) − γ(t1 ), R90 v > (t1 , t2 ) (t2 , b + a − t1 )
0 t1 , t 2 t3 , t 4 v
t3 > t 1 ks > 0 θ θ(t3 ) − θ(t1 ) = 2π RI=1

Exercise: 2.13

Let γ : [a, b] → R2 be a (not necessarily closed) regular plane curve with γ(a) = γ(b). What is the minimal
possible total curvature of γ?

Solution. γ
cusp total curvature
γ(a), γ(b) ε ε γ
γε ε→0 γε total curvature γ total curvature
γε Fenchel total curvature≥ 2π
ε→0 γ γε γ ′ (b)
γ ′ (a) θ ∈ [0, π) ε→0 total curvature(γ) ≥ 2π − θ ≥ π

γε

γ(a) = γ(b)

θ
γ ′ (a)
γ ′ (b)
Z b
total curvature > π II
γ(b) = γ(a) v(t)dt = γ(b) − γ(a) = 0.
* Z + Z a
b b
γ ∀n ∈ S 1 , 0 = n, v(t)dt = hn, v(t)i dt.
a a
Z b
hn, v(t)i 6≡ 0 v v ≡ va ⊥ n v(t)dt = (b − a)va 6= 0,
a
∃t1 , t2 ∈ [a, b], hn, v(t1 )i > 0, hn, v(t2 )i < 0.
v(t1 ), v(t2 ) S1 n n
v(t) total curvature > π.
Z b
total curvature ≥ π III Hopf κs (t)dt + α = 2π, α ∈ [−π, π] γ(b) γ(a)
a

Z b Z b Z b
κ(t)dt = |κs (t)|dt ≥ κs (t)dt = |2π − α| ≥ π.
a a a

22
 3

total curvature > π IV

γ(t) = x(t)t⃗0 + y(t)n⃗0 , γ(a) = (0, 0), x′ (a) = 1, y ′ (a) = 0. γ(b) = (0, 0),
Z b
′ ′ ′
t0 ∈ (a, b) ( ) x (t0 ) = 0 y (t0 ) = 1 (or − 1) x(t0 ) = x (t)dt > 0.
a
(1) y(t0 ) > 0, t1 ∈ (t0 , b) y(t1 ) y(t1 ) > 0, y ′ (t1 ) = 0, x′ (t1 ) = ±1.
Z t1 Z t0 Z t1
κ(t)dt ≥ | κs (t)dt| + | κs (t)dt| ≥ |θ(t0 ) − θ(a)| + |θ(t1 ) − θ(t0 )| ≥ π.
a a t0

Z b
y y(t1 ) > 0 0 y ′ (t1 ) = 0, θ(b) 6= θ(t1 ), κ(t)dt > π.
a
2 y(t0 ) < 0, t0 t1 ∈ (a, t0 ), y(t1 )
Z t0
π
y(t1 ) < y(t0 ), y ′ (t1 = 0). κ(t)dt ≥ .
2
t1
Z t1
−y(t0 )
t2 ∈ (a, t1 ), γ(t2 ) γ(a)γ(t1 ) . κ(t)dt > arctan .
a x(t0 )
−y(t0 )
t0 (0, 0), γ ′ (t0 ) cot−1 .
x(t0 )
Z b
κ(t)dt > π.
a
 totalcurvature π
1
γn 1+ π
n


 (nt, 0), t ∈ [0, 1];

        
 1 1
γn (t) := (n, 1) + sin 1 + π(t − 1) , cos (t − 1) 1 + π , t ∈ (1, 2);
   n     n




 (3 − t) n + sin 1 +
1
π , 1 + cos 1 +
1
π , t ∈ (2, 3).
n n

Exercise: 2.14

Prove the n = 2 (plane curve) case of Fenchel s theorem.

.
Lemma γ : I → S1 L(γ) ≥ 2π γ S1
γ S1 ≥ 2π γ 6= S 1 I˜ ⊂ I γ̃ = γ|I−I˜
1 1
S L(γ̃) = L(S ) L(γ|I˜) > 0 L(γ) > 2π
follow γ Claim v → S 1 n ∈ S1 f (t) =< γ(t), n >
′
f (b) = f (a) t0 ∈ I Z f (t0 ) =
Z0 < v(t0 ), n >= 0 v(I) = S 1
γ k ds = |v ′ |ds = L(v) ≥ 2π v S1 θ
γ

23
 3

Exercise: 2.15

Prove that every great circle of S 2 is the image of the equator under a rigid motion.

C n O On = ez Claim
O(C) = equator p∈C O(p) ∈ O(C) ∩ equator T
< O(p), ez >=< p, O ez >=<
p, n >= 0 O(C) ⊂ equator O (equator) ⊂ C
T
equator ⊂ O(C) O(c) =
equator

Exercise: 2.16

Prove that the length of a regular closed curve with nowhere vanishing curvature satisfies

kmax
L≥
2π

. Fenchel
Z Z
kmax L = kmax ds ≥ k ds ≥ 2π.

Exercise: 2.28

If l, A ∈ Rare positive numbers for which l2 ≥ 4πA, prove that there exists a simple closed plane curve with
length l and area A.

. l
l
≥A
4π
l
0
2
l
(0, ] A
4π
r a , √ r √ r
l − l 2 − 4πA l 2 − 4πA
A = πr2 +2ar, l = 2πr+2a. r = ,a= . A, l l2 −4πA =
2π 2
4a ≥ 0.
2

24
 3

Exercise: 2.29

For p > q > 0, consider the ellipse γ(t) = (pcost, qsint). Let A denote its area and l its length. Use Green s
theorem to calculate A in terms of p and q.

Solution. P = 1, Q = 0
ZZ Z Z 2π
A= dx ∧ dy = xdy = p cos t(q cos t)dt = pqπ
D C 0

Exercise: 2.30

Use Green s theorem to find area of the region bounded by the x-axis and the trace of the curve γ(t) =
(t−sin(t), 1−cos(t)), t ∈ [0, 2π].

. γ
ZZ Z Z 2π
A= dx ∧ dy = xdy = −(t − sin t) sin t dt = 3π
D C 0

25
 3

Exercise: 2.31

1. Show that the line integral of F (x, y) = (−y, x) along the line segment from (x1 , y1 ) to (x2 , y2 ) equals
x1 y2 −x2 y1 .

2. If C is a polygon with vertices denoted by (x1 , y1 ), . . . , (xn , yn ) (ordered counterclockwise), prove that
the area A of the polygon is given by

2A = (x1 y2 − x2 y1 ) + · · · + (xn y1 − x1 yn ).

. 1. (x1 , y1 ), (x2 , y1 ) γ1 (x2 , y1 ), (x2 , y2 ) γ2

Z Z Z ZZ
( − − )ωF = dF = 0
γ γ1 γ2 D

Z
ωF = (−y2 )(x2 − x1 ) + x2 (y2 − y1 ) = x1 y2 − x2 y1
γ

(x2 , y2 )

(x1 , y1 ) (x1 , y2 )

Green
ZZ ZZ Z X
n
dF = 2dA = 2A = F = (xj yj+1 − xj+1 yj ) (xn+1 = x1 , yn+1 = y1 )
D D line segments of boundary j=1

26
 4

Exercise: 3.1(The Derivative of a Linear Transformation Is Itself )

Let f : Rm →Rn be a linear transformation. For every p ∈ Rm , prove that dfp = f .

. f
X
m X
m
f (x1 , . . . , xm ) = ( fj1 xj , . . . , fjn xj ),
j=1 j=1

fji
∂f i
(df )ij := = fji .
∂xj

Exercise: 3.3

Define f : R2 →R3 as f (x, y) = (5x2 y 3 , 2x + y 2 , x2 −y 2 ). Compute the Jacobian matrix for f at p0 = (1, −1).
What is the rank of dfp0 ?

.  
10xy 3 15x2 y 2
 

df =  2 2y 

2x −2y

 
−10 15
 
dfp0 = 
 2 −2

2 2

2×2 rank(dfp0 ) = 2

Exercise: 3.4

Is the function f : R2 → R2 defined as f (x, y) = (x3 , y) a diffeomorphism?

. f " #
3x2 0
df (x, y) = .
0 1

(x, y) = f −1 (z, w) = (z 3 , w)
1
x=0 x=0 x=0
z=0 f

27
 4

Exercise: 3.5

Let U ⊂ R2 be open and f : U → R smooth. Denote the coordinates of R2 by {u, v}. Assume that the point
q ∈ U is a critical point of f , which means that dfq (w) = 0 for all w ∈ R2 , or equivalently, fu (q) = fv (q) = 0.
Let γ : I → U be a regular curve with γ(0) = q and γ ′ (0) = w = (a, b). Prove that

(f ◦ γ)′′ (0) = a2 fuu (q) + 2abfuv (q) + b2 fvv (q)

In particular, the value (f ◦ γ)′′ (0) does not depend on the second or higher derivative of γ at 0 . In other
words, the expression Hess(f )q (w) = (f ◦ γ)′′ (0) is well defined (independent of the choice of γ with γ(0) = q
and γ ′ (0) = w). This expression is called the Hessian of f at q in the direction w.

. γ(t) = (γ u (t), γ v (t)) f ◦ γ(t) = f (γ u (t), γ v (t)),

(f ◦ γ)′ (t) = fu (γ(t))(γ u )′ (t) + fv (γ(t))(γ v )′ (t),

(f ◦ γ)′′ (t) = fuu (γ(t))((γ u )′ (t))2 + 2fuv (γ(t))(γ u )′ (t)(γ v )′ (t) + fvv (γ(t))((γ v )′ (t))2
+fu (γ(t))(γ u )′′ (t) + fv (γ(t))(γ v )′′ (t),
(f ◦ γ)′′ (0) = a2 fuu (q) + 2abfuv (q) + b2 fvv (q).

Remark 4
f (q) q Hessian γ

Exercise: 3.6

In Example 3.6, polar coordinate,

f (r, θ) = (r cos θ, r sin θ),

describe a largest-possible neighborhood of p = (2, π/4) such that f restricts to a diffeomorphism between this
neighborhood and its image.

. f r 6= 0 r ∈ (0, ∞) θ

(0, ∞) × (π/4 − 2π, π/4 + 2π).

28
 4

Exercise: 3.9

Prove that a composition of two diffeomorphisms is a diffeomorphism.

. f : A → B, g : B → C f, g f −1 , g −1 :
(1) ˜
g ◦ f (x) = g ◦ f ((x)) g, f f (x) = f (x̃), x = x̃ g◦f
2 g, f ∀c ∈ C, b∈B g(b) = c a ∈ A, f (a) = b, g ◦ f (a) = c, g ◦ f
−1
3 g◦f (g ◦ f ) = f −1 ◦ f −1

1 - 3 g◦f

Exercise: 3.11

(Tubular Neighborhood of a Plane Curve). In this exercise, you will prove Proposition 2.5, illustrated in
Fig. 2.3 on page 65. Let γ : [a, b] → R2 be a unit-speed simple closed plane curve. For t ∈ [a, b], set
N (t) = R90 (γ ′ (t)). Define φ : (−ϵ, ϵ) × [a, b] → R2 as φ(s, t) = γ(t) + s · N (t). Prove that there exists ϵ > 0
such that:
(1) φ is a diffeomorphism from (−ϵ, ϵ) × (a, b) onto its image.
(2) φ is injective on (−ϵ, ϵ) × [a, b).
HINT: Extend γ to a periodic function on R. For t0 ∈ [a, b], use the inverse function theorem to find
ϵ0 > 0 such that φ is a diffeomorphism on (−ϵ0 , ϵ0 ) × U0 , where U0 is a neighborhood of t0 in R. Since [a, b]
is compact, there are finitely many such times t1 , . . . , tn ∈ [a, b] (with corresponding values ϵ1 , . . . , ϵn and
corresponding neighborhoods U1 , . . . , Un ) such that {U1 , . . . , Un } is an open cover of [a, b]. Let δ denote the
Lebesgue number of this  open cover, as defined in Proposition A. 23 on page 355 of the appendix. Define
δ
ϵ = min ϵ1 , . . . , ϵn , .
2

. γ R γ̃
∂φ ∂φ
= N (t), = γ ′ (t) + s · N ′ (t) φ Jacobian Jφ (s, t) :
∂s ∂t
h i
Jφ (s, t) = N (t) γ ′ (t) + sN ′ (t) .
h i
Jφ (0, t) = N (t) γ ′ (t) γ |γ ′ (t)| = 1 N (t) = R90 (γ ′ (t)) ⊥ γ ′ (t), Jφ (0, t)
Wt := (−ϵt , ϵt ) × (t − ϵt , t + ϵt ) φ : Wt → φ(Wt )
[a, b] {a = t0 < t1 < · · · < tn < b} {φ(Wti ) =: φ(Wi ), i = 0, · · · , n}
γ Proposition A.23 δ>0 Lebesgue
i (c, d) ⊂ [a, b] (γ(ti − ϵi , ti + ϵi )) ∪ γ(c, d) ⊂ (φ(Wi ) ∩ γ)
φ Wi Wi

 
δ
ϵ̃ = min ϵt1 , . . . , ϵtn , ,
2

29
 4

δ̃ := inf{|γ(t1 ) − γ(t2 )| : |t1 − t2 | ≥ ϵ̃/2, ti ∈ [a, b)}.

δ̃ > 0 ( )
δ̃
ϵ = min ϵ̃, φ (−ϵ, ϵ) × (a, b) φ (−ϵ, ϵ) × (a, b)
2
φ (−ϵ, ϵ) × [a, b)
φ (−ϵ, ϵ) × [a, b) (s, t) 6= (s̃, t̃) ∈ (−ϵ, ϵ) × [a, b), φ(s, t) = φ(s̃, t̃).
|γ(t) − γ(t̃)| < min{δ, δ̃} |t − t̃| < ϵ̃/2 |φ(s, t) − φ(s̃, t̃)| < 2ϵ < δ φ(s, t), φ(s̃, t̃) Wi
φ (−ϵi , ϵi ) × (t − ϵi , t + ϵi ) |γ(t)) − γ(t̃)| ≥ min{δ, δ̃}

0 = |φ(s, t) − φ(s̃, t̃)| ≥ |γ(t) − γ(t̃)| − |sN (t)| − |s̃N (t̃)| > δ̃ − 2ϵ ≥ 0,

φ (−ϵ, ϵ) × [a, b)

Exercise: 3.12

(Tubular Neighborhood of a Space Curve). Let γ : [a, b] → R3 be a unit-speed simple closed space curve with
nowherevanishing curvature. For t ∈ [a, b], let {t(t), n(t), b(t)} denote the Frenet frame vectors at γ(t). Set

Bϵ = (x, y) ∈ R2 | x2 + y 2 < ϵ , and define

φ : [a, b] × Bϵ → R3 as φ(t, x, y) = γ(t) + x · n(t) + y · b(t)

ϕ : [a, b] × [0, 2π] → R3 as ϕ(t, θ) = γ(t) + (ϵ · cos θ) · n(t) + (ϵ · sin θ) · b(t)
Let Sϵ denote the image of ϕ (equivalently, the boundary of the image of φ ). Prove there exists ϵ > 0 such
that:
(1) φ is a diffeomorphism from (a, b) × Bϵ onto its image, and is injective on [a, b) × Bϵ .
(2) ϕ is a diffeomorphism from (a, b) × (0, 2π) onto its image, and is injective on [a, b) × [0, 2π).

(3) Sϵ = p ∈ R3 | dist(p, C) = ϵ , where C is the trace of γ and ”dist (p, C) ” denotes the distance from
p to the point of C closest to p.
Use a computer graphics application to plot Sϵ for the space curves that you plotted in Exercise 1.13 on
page 8 (for suitable choices of ϵ ).

. (1),(2) Exercise 3.11

|φ(t, x, y) − φ(t̃, x̃, ỹ)| ≥ |γ(t) − γ(t̃)| − |xn(t) + yb(t)| − |x̃n(t̃) + yb(t̃)| > δ̃ − 2ϵ.

/ n∈N φ B1/n
{(tn , zn )} {t′n , zn′ } ′ ′ ′ ′
(tn , zn ) 6= (tn , zn ), φ(tn , zn ) = φ(tn , zn ) ′
|zn |, |zn | ≤ 1/k
′ ′
{(tn , zn )} {tn , zn } tn → t, tn → t , |zn |, |zn′ | → 0
′ ′
t = t′
t 6= t′ |γ(t)−γ(t′ )| > δ > 0 |zn |, |zn′ | → 0 n |zn |, |zn′ | < δ/4 |γ(tn )−γ(t)| < δ/4
|γ(tn ‘’) − γ(t)| < δ/4

|φ(tn , zn ) − φ(t′n , zn′ )| ≥ |γ(t) − γ(t′ )| − |γ(tn ) − γ(t)| − |γ(tn ‘’) − γ(t)| − |zn | − |zn′ | > 0

t = t′ φ γ(t) U n φ(tn , zn ), φ(t′n , zn′ )

U (tn , zn ) 6= (t′n , zn′ ), φ(tn , zn ) = φ(t′n , zn′ )

30
 4

n ϵ < 1/n φ
3
|ϕ(t, θ) − γ(t)| = ϵ,

Sϵ ⊂ p ∈ R3 | dist(p, C) ≤ ϵ . |ϕ(t, θ) − γ(s)| ≥ ϵ :
t, s (ti − ϵti , ti + ϵti ), γ(s) = γ(t) + (s − t)γ ′ (t) + ⃗o(s − t),

|γ(s) − ϕ(t, θ)| = |(s − t)γ ′ (t) + ⃗o(s − t) − ϵ · cos θ · n(t) − ϵ · sin θ · b(t)|
q
≥ | (s − t)2 + ϵ2 · cos2 θ + ϵ2 · sin2 θ − o(s − t)| ≥ ϵ;

t, s (ti − ϵti , ti + ϵti ) |s − t| > δ, |s − t| > δ̃

|γ(s) − ϕ(t, θ)| = |γ(s) − γ(t) − ϵ · cos θ · n(t) − ϵ · sin θ · b(t)|

≥ ||γ(s) − γ(t)| − |ϵ · cos θ · n(t) − ϵ · sin θ · b(t)|| > δ̃ − ϵ > ϵ.

|ϕ(t, θ) − γ(s)| ≥ ϵ Sϵ ⊂ p ∈ R3 | dist(p, C) = ϵ .

p ∈ R3 | dist(p, C) = ϵ ⊂ Sϵ :
dist(p, C) = dist(p, γ(t)) = ϵ, p = γ(t) + (ϵ · cos α) · t(t) + (ϵ · sin α · cos θ) · n(t) + (ϵ · sin α · sin θ) · b(t).
2
f (s) = d (p, γ(s)) =< p−γ(s), p−γ(s) > s=t 0 = f ′ (t) = 2 < p−γ(t), γ ′ (t) >= ϵ·cos α
p = γ(t) ± [(ϵ · cos θ) · n(t) + (ϵ · sin θ) · b(t)] ∈ Sϵ .

C p, dist(p, C) = ϵ q∈C d(p, q) = ϵ

ϵ < δ/2 δ Lebesgue number q ϵ U U φ
φ(t0 , z0 ) = p, γ(t1 ) = q d(p, γ(t)) t = t1 (p − q) ⊥ t(t0 ) q−p
n(t0 ) b(t0 ) φ t0 = t1 d(p, q) = |z0 | = ε q = ϕ(t0 , z0 )
′ ′ ′ ′
{d(p, C) = ϵ} ⊂ Sϵ p ∈ Sϵ d(p , C) = ϵ q = γ(t′0 ) ∈C
q= ϕ(t′0 , ϵ′ ) ϕ ′
ϵ =ϵ Sϵ ⊂ {dist(p, C) = ϵ}

Exercise: 3.13

Prove that a set S ⊂ R3 is a regular surface if and only if each of its points has a neighborhood in S that is

diffeomorphic to the disk (x, y) ∈ R2 | x2 + y 2 < 1

. S p ∈ S δ(p), R2 O(p) ⊂ R2
fp : δ(p) → O(p)
O(p) Bϵ(p) (p) fp−1 Bϵ(p) (p) Bϵ(p) (p)

(x, y) ∈ R | x + y < 1
2 2 2
S

(x, y) ∈ R | x + y < 1
2 2 2

31
 4

Exercise: 3.14

Show that the cylinder (x, y, z) ∈ R3 | x2 + y 2 = 1 can be entirely covered by a single surface patch with
domain U = R2 − {(0, 0)}.

. coordinate chart (surface patch)

σ(x, y) := (x/r, y/r, log r),

p
r= x2 + y 2 rank=2 U = R2 − {r = 0} = R2 − {(0, 0)} cylinder

Exercise: 3.16

Let S be a regular surface. If X ⊂ S is open in S, prove that X is itself a regular surface.

. X ∀p ∈ X R3 U U ∩S R2
U ∩X U ∩S X R2
U ∩S U ∩X

Exercise: 3.21
 
∂f ∂f
Let U ⊂ R2 be open, f : U → R a smooth function, p ∈ U , and λ = f (p). If dfp = (p), (p) 6= (0, 0),
∂x ∂y
prove that there exists a neighborhood of p in the preimage C = f −1 (λ) that is the trace of a regular plane
curve.

. 3

Exercise: 3.22

Let f : R3 → R2 be smooth. Let λ ∈ R2 be such that the preimage S = f −1 (λ) is nonempty. Let p ∈ S.
Assume that dfp has rank 2. Prove that there exists a neighborhood of p in S that is the trace of a parametrized
space curve.

. constant rank theorem maximal rank

S = f −1 (λ) local chart level set f

p rank = 2 dfp R 3
R 2

0 level set f dfp kernel level set

d(f ◦ γ) ′
dfp (v) = (0) = 0, γ(0) = p, γ (0) = p dim S = dim R − rank(df ) = 3 − 2 = 1
3
dt

32
 4

p U f (p) V ϕ : U → ϕ(U ) ⊂ R3 , ψ : V → ψ(V ) ⊂ R2

f˜ = ψ ◦ f ◦ ϕ−1 (x, y, z) = (x, y) U f ◦ ϕ−1 (x, y, z) = λ ψ(x, y) = λ
−1
S∩U =ϕ ({(ψ(λ), z) | (ψ(λ), z) ∈ ϕ(U )}) z
∂(f 1 , f 2 )
dfp =Q
∂(x, y)
dfp  
∂(f 1 , f 2 ) ∂f h i
dfp = =: Q2×2 R2×1
∂(x, y) ∂z
" #−1 " #
h i Q R Q−1 −Q−1 R
I2 0 =
0 1 0 1
" #−1
h i Q R h i
Q R · = I2 0
0 1

level set

ϕ(x, y, z) = (f (x, y, z), z),

" #
Q R
dϕp =
0 1
h i
p U ϕ : U → ϕ(U ) ⊂ R3 d(f ◦ ϕ−1 ) = I2 0 f ◦ ϕ−1
(x, y, z) = ϕ ◦ ϕ−1 (x, y, z) = (f ◦ ϕ−1 (x, y, z), πz ◦ ϕ−1 (x, y, z)) f ◦ ϕ−1 (x, y, z) = (x, y)

Exercise: 3.27

The function σ : R2 → R3 defined as

 
u3 v3
σ(u, v) = u− + uv 2 , v − + vu2 , u2 − v 2
3 3
is called Enneper’s surface. It is illustrated in Fig. 3.17. Verify that σ is a parametrized surface.

.
rank(dσ)(u, v) = 2.

 
1 − u2 + v 2 2uv
 
dσ(u, v) = 
 2uv 1 − v 2 + u2 
,
2u −2v

2×2

(1 − u2 + v 2 )(1 − v 2 + u2 ) − 4u2 v 2 = 1 − (u2 + v 2 )2 ,

33
 4

u2 + v 2 6= 1 rank=2 u2 + v 2 = 1
 
2v 2 2uv
 
2uv 2u2 
 
2u −2v

u v
v u

Remark 5

     
1 − u2 + v 2 2uv 1 + u2 + v 2 0 1 + u2 + v 2 0
     
dσ(u, v) = 
 2uv 1 − v 2 + u2  
→ 2uv 1 − v 2 + u2  
→ 0 1 + v 2 + u2 
,
2u −2v 2u −2v 2u −2v

2.

Exercise: 3.28

1
Let γ : (a, b) → R3 be a unit-speed space curve whose curvature function satisfies 0 < κ(t) < for all t ∈ (a, b).
ϵ
Define ϕ : (a, b)× (0, 2π) → R3 exactly as in Exercise 3.12 on page 124 :

ϕ(t, θ) = γ(t) + (ϵ · cos θ) · n(t) + (ϵ · sin θ) · b(t)

Prove that ϕ is a parametrized surface.

. k Ex 3.12 ϕ
ε ϕ k k0
1/k0 1/k0 ϕ

k>0 τ Frenet
∂ϕ ∂ϕ
dϕ = ( , ) = (t(1 − εk cos θ) − nε sin θτ + bε cos θτ, −nε sin θ + bε cos θ) ,
∂t ∂θ
∂ϕ ∂ϕ ∂ϕ
k < 1/ε −τ = t(1 − kε cos θ)
∂t ∂θ ∂θ

34
 4

Exercise: 3.34

Definition 3.30 can be used to define Tp S when S ⊂ R3 is an arbitrary set and p ∈ S. Give a variety of
examples in which Tp S is not a subspace.

S := {(r cos θ, r sin θ, 0), r ∈ [0, +∞), θ ∈ (θ1 , θ2 ) ⊂ (0, π)}, p = (0, 0, 0).

Remark 6

p v R3 p
γ v t γ(t) (t ≥ 0) γ(t) (t ≤ 0) S

γ(−h) − γ(0)
lim .
h→0+ −h

Exercise: 3.35

Let S be a regular surface and p ∈ S. Suppose P ⊂ R3 is a plane such that p ∈ P and S lies on one (closed)
side of P . Prove that Tp S = P .

Remark 7

S {(x, y, z) | y = x3 , z ∈ R}

. P = {p + xe1 + ye2 |(x, y) ∈ R2 }, S ⊂ {p + xe1 + ye2 + ze3 |(x, y) ∈ R2 , z ≥ 0}. {e1 , e2 , e3 }

R 3

v ∈ Tp S, < v, e3 >= v3 > 0, (x, y, z) : (a, b) → S, γ(0) = p, γ ′ (0) = v = v1 e1 + v2 e2 + v3 e3 ,

′
(−ϵ, ϵ) ⊂ (a, b) z (−ϵ, ϵ) z(0) = 0 z(−ϵ) < 0,

Exercise: 3.37

Let S be a regular surface and p ∈ S. Let N ∈ R3 denote a normal vector to Tp S. If p1 , p2 , . . . is an infinite

sequence of points of S converging to p, prove that
 
pn − p
lim N, =0
n→∞ |pn − p|

Remark 8

35
 4

S R3

. P = {p + xe1 + ye2 |(x, y) ∈R2 }, S ⊂ {p + xe1 + ye2 + ze3 |(x, y, z) ∈ R3 }. {e1 , e2 , e3 } R3

pn − p
lim N, 6= 0, pn {pni }
n→∞ |pn − p|
 
pn i − p z(pni ) z(pni )
lim N, = lim = lim p = k 6= 0.
i→∞ |pni − p| i→∞ |pni − p| i→∞ x2 (pni ) + y 2 (pni ) + z 2 (pni )
pn i − p
S pni p γi γ(0) = p, γi (|pni − p|) = pni . γi′ (0)
|pni − p|
pn i − p
lim γi′ (0) = lim = (x′ (0), y ′ (0), z ′ (0)),
i→∞ i→∞ |pni − p|
z ′ (0)
p = k, γ ′ (0) 6∈ P (x′ (0), y ′ (0), z ′ (0)) 6∈ Tp S
(x′ (0))2 + (y ′ (0))2 + (z ′ (0))2

Exercise: 3.38

Let U ⊂ R2 be open, f : U → R smooth, G its graph, and q = (x, y) ∈ U . Prove that the tangent plane to G
at (x, y, f (x, y)) is the graph of dfq .

 
1 0
 
. σ(u, v) := (u, v, f (u, v)), dσq = 
 0 1  . dσq
 σ parametrized surface. G
fu (u, v) fv (u, v)
γ(u) := (u, y, f (u, y)) γ̃(v) := (x, v, f (x, v)), q (1, 0, fu (x, y)) (0, 1, fv (x, y))
Tq G
e1 = (1, 0), e2 = (0, 1) Tq U
d(f (x + t, y)) d(f (x, y + t))
dfq (e1 ) = (0) = fu (x, y), dfq (e2 ) = (0) = fv (x, y).
dt dt
dfq (e1 , dfq (e1 )) = (1, 0, fu (x, y)) (e2 , dfq (e2 )) = (0, 1, fv (x, y))

Exercise: 3.40

Let S be a path-connected regular surface, and f : S → R a smooth function. Prove that f is a constant
function if and only if dfp (v) = 0 for all p ∈ S and all v ∈ Tp S.

. f dfp dfp (v) = 0.

p 6= q ∈ S, f (p) > f (q). S p, q γ, γ(t1 ) = p, γ(t2 ) = q,
f (γ(t1 )) − f (γ(t2 )) ′ ′
t3 ∈ (t1 , t2 ) 0< = (f ◦ γ) (t3 ) = dfγ(t3 ) (γ (t3 )).
t 1 − t2

36
 4

Exercise: 3.44

Let S be a regular surface and f : S → R a smooth function. Assume p ∈ S is a critical point of f , which
means that dfp (v) = 0 for all v ∈ Tp S. Define the Hessian of f at p in the direction v as

Hess(f )p (v) = (f ◦ γ)′′ (0)

where γ is a regular curve in S with γ(0) = p and γ ′ (0) = v. Prove the Hessian is well defined in the sense
that it is independent of the choice of γ. HINT: Work in local coordinates and use Exercise 3.5 on page 123.

. U ⊂ R2 , σ : U → σ(U ) = V ⊂ S, σ(u, v) = (x, y, z), σ(q) = p. f˜ := (f ◦ σ) : U → R

γ̃ := (σ −1 ◦ γ) : I → U df˜q (w) = dfp (dσq (w)) = 0, ∀w ∈ R2 , γ̃(0) = q γ̃ ′ (0) =
dσp−1 (v) =w γ Exercise 3.5

(f ◦ γ)′′ (0) = (f˜ ◦ σ̃)′′ (0) = w12 f˜uu (q) + 2w1 w2 f˜uv (q) + w22 f˜vv (q).

(f ◦ γ)′′ (0) γ

Exercise: 3.46

g f
(The Chain Rule for Surfaces). Let S1 −
→ S2 −
→ S3 denote smooth functions between regular surfaces. Prove
that for all q ∈ S1 ,

d(f ◦ g)q = dfp ◦ dgq

where p = g(q).

. , v ∈ T q S1 , S γ γ(0) = p, γ ′ (0) = v,

d(f ◦ g)q (v) = (f ◦ g ◦ γ)′ (0) = (f ◦ γ̃)′ (0) = dfp (γ̃ ′ (0)) = dfp ◦ dgq (v),

γ̃ := g ◦ γ.

Exercise: 3.57

Let λ ∈ R and define V = span{(1, 0, 0), (0, 1, λ)} and Ṽ = span{(1, 0, 0), (0, 1, 0)}. Compute the area distortion
of g : V → Ṽ defined such that g(x, y, z) = (x, y, 0) for all (x, y, z) ∈ V.

|g(1, 0, 0) ∧ g(0, 1, λ)| |(1, 0, 0) ∧ (0, 1, 0)| |(0, 0, 1)| 1

. kgk = = = =√ .
|(1, 0, 0) ∧ (0, 1, λ)| |(1, 0, 0) ∧ (0, 1, λ)| |(0, −λ, 1)| 1 + λ2

37
 4

Exercise: 3.58

Let V, Ṽ ⊂ R3 be a pair of two-dimensional subspaces with orientations. Let g : V → Ṽ be an orientation-

g (v1 )
preserving linear isomorphism. Let {v1 , v2 } be a positively oriented orthonormal basis of V. Define ṽ1 =
|g (v1 )|
and ṽ2 = R90 (ṽ1 ), so that {ṽ1 , ṽ2 } is a positively oriented orthonormal basis of Ṽ. If v = av1 + bv2 and
g(v) = ãṽ1 + b̃ṽ2 , prove that (ã, b̃) is not a negative scalar multiple of (a, b).

. g(v) = ag(v1 ) + bg(v2 ) = a|g(v1 )|ṽ1 + bg(v2 ) = ãṽ1 + b̃ṽ2 . (ã, b̃) = −k(a, b), k > 0
−a
a|g(v1 )|ṽ1 + bg(v2 ) = −kaṽ1 − kbṽ2 , b 6= 0, g(v2 ) = (k + |g(v1 )|)ṽ1 − kṽ2 ,
b
g(v1 ) ∧ g(v2 ) = |g(v1 )|ṽ1 ∧ (−kṽ2 ) = −k|g(v1 )|ṽ1 ∧ ṽ2 .

Exercise: 3.59

Let G be the graph of f (x, y) = xy, p = (2, 3, 6) ∈ G.

1. Find basis for Tp G;

2. Find unit normal vector at p;

3. Find a function h such that G = h−1 (0)

. 1. G(f ) = {(x, y, f (x, y)) | (x, y) ∈ R2 } p γ1 = (px + t, py , f (px + t, py )) γ2 =

(px , py + t, f (px , py + t))

e1 = (1, 0, ∂x fp ) = (1, 0, 3), e2 = (0, 1, ∂y fp ) = (0, 1, 2),

Tp G
e1 e2

2.

e1 ∧ e2 (−∂x fp , −∂y fp , 1) (−3, −2, 1)

n= = p = √ .
|e1 ∧ e2 | 1 + |∇f |2 15

3. f

h(x, y, z) = z − f (x, y) = 0

h
∇h
n= .
|∇h|

38
 4

Remark 9

aux + buy = c, (x, y) ∈ D

{(x, y, u(x, y))},

(−ux , −uy , 1) (a, b, c)

< (a, b, c), (−ux , −uy , 1) >= −aux − buy + c = 0,

p (a, b, c)
Monge Monge

Exercise: 3.60

A diffeomorphism f : S → S̃ between connected oriented surfaces is orientation-preserving if and only if there

exists a point p ∈ S such that dfp is orientation-preserving.

. f det(dfp ) > 0, ∀p ∈ S
e1 , e 2 Tp S e˜1 , e˜2 Tf (p) S̃

dfp (e1 ) ∧ dfp (e2 ) = (df11 e˜1 + df12 e˜2 ) ∧ (df21 e˜1 + df22 e˜2 ) = (df11 df22 − df12 df21 )e˜1 ∧ e˜2 = det(df )e˜1 ∧ e˜2 ,

det(df ) f
det(df ) > 0
S f det(df ) 6= 0
dfp

Exercise: 3.62

E = {(x/a)2 + (y/b)2 + (z/c)2 = 1}

f (x, y, z) = (ay/b, bz/c, cx/a) : E → E kdf k

. 1. f = (f1 , f2 , f3 ), f −1 (f1 , f2 , f3 ) = (af3 /c, bf2 /a, cf1 /b), f

 
0 a/b 0
 
df =   0 0 b/c  , det(df ) = 1

c/a 0 0

f Exercise 3.62, p = (a, 0, 0) a, b, c > 0

39
 4

γ(t) = (a cos t, b sin t, 0), γ̃(t) = (a cos t, 0, c sin t), γ(0) = γ̃(0) = p, γ ′ (0) = (0, b, 0), γ̃ ′ (0) =
(0, 0, c), γ ′ (0)×γ̃ ′ (0) = (bc, 0, 0). γ ′ (0), γ̃ ′ (0) f (p) = (0, 0, c), df (γ ′ (0)) = (a, 0, 0), df (γ̃ ′ (0)) =
′ ′
(0, b, 0), df (γ (0)) × df (γ̃ (0)) = (0, 0, ab) dfp f

2.
df (e1 ) = df (1, 0, 0) = (0, 0, c/a), df (e2 ) = df (0, 1, 0) = (a/b, 0, 0), df (e3 ) = df (0, 0, 1) = (0, b/c, 0),

 1/2
|df (u1 , u2 , u3 ) × df (v1 , v2 , v3 )| [b/a(u2 v3 − u3 v2 )]2 + [c/b(u3 v1 − u1 v3 )]2 + [a/c(u1 v2 − u2 v1 )]2
=
|(u1 , u2 , u3 ) × (u1 , u2 , u3 )| (u2 v3 − u3 v2 )2 + (u3 v1 − u1 v3 )2 + (u1 v2 − u2 v1 )2
∈ [min{a/b, b/c, c/a}, max{a/b, b/c, c/a}].

kdf k max{a/b, b/c, c/a}, min{a/b, b/c, c/a}

Remark 10

f dfR3 R3

f : R2x,y ⊂ R3x,y,z → R2x,y ⊂ R3 , f (x, y, z) = (x, −y, −z).

S R3 f : R3 → R3
S

Exercise: 3.66.

Construct an explicit parametrization of a ”generalized Möbius strip” that returns with n half-twists. Under
what condition on the integer n is the strip orientable? Under what condition on n1 , n2 is the strip with n1
half-twists diffeomorphic to the strip with n2 half-twists?

. 1. A Möbius strip is a classic example of a non-orientable surface with one half-twist. To generalize this
to a strip with n half-twists, we can construct a surface embedded in R3 as follows.
Let the generalized Möbius strip be parametrized by two variables: the parameter t ∈ [0, 2π] along the central
circle and a radial coordinate s ∈ [−1, 1] representing the distance from the center of the strip. The central
circle lies in the xy-plane, and the strip extends a distance s from this central circle, with a twist as we go
around the circle.
The parametric equations for a strip with n half-twists can be written as:

       
nt nt nt
σ(t, s) = R + s cos cos(t), R + s cos sin(t), s sin
2 2 2
where:

• t ∈ [0, 2π] parametrizes the position along the central circle,

• s ∈ [−1, 1] parametrizes the displacement from the central circle to the boundary of the strip,
• R is the radius of the central circle,

40
 4

• n is the number of half-twists applied to the strip.

 
nt
For n = 1, this corresponds to the classic Möbius strip with a single half-twist. The term cos controls
  2
nt
the twisting of the strip as we move around the central circle, and sin controls the out-of-plane twisting.
2
2. Condition for Orientability.

σt (0, s) = (0, R + s, 0) = σt (2π, s), σs (0, s) = (1, 0, 0), σs (2π, s) = ((−1)n , 0, 0),
(
n+1 N (0, s) n = 2k;
N (2π, s) = (0, 0, (−1) )=
−N (0, s) n = 2k + 1.

• If n is even, the surface is orientable. This is because after one full revolution ( t = 0 to t = 2π), the
direction of the strip is the same as when we started, so the surface has no self-intersections or flipping
of orientation.
• If n is odd, the surface is non-orientable. In this case, after one full revolution, the strip has undergone
an odd number of half-twists, meaning that the normal vector flips, leading to a non-orientable surface.

Thus, the strip is orientable if and only if n is even.

3. Diffeomorphism Condition

• If n1 and n2 have different parity (i.e., one is odd and the other is even), the two strips are not diffeo-
morphic. This is because one surface is orientable, and the other is not, and an orientable surface cannot
be smoothly deformed into a non-orientable one.
• If n1 and n2 have the same parity, the two strips are diffeomorphic. This means that if both n1 and n2
are either both odd or both even, we can deform one into the other by gradually increasing or decreasing
the number of twists without breaking the smooth structure.

f (σ(t, s)) = σ̃(t, s), (t, s) ∈ (0, 2π) × [−1, 1].

σ, σ̃ f

σ(0, s) = (R + s, 0, 0), σ(2π, s) = (R + (−1)n s, 0, 0),

n − ñ , (2π, 0) f (0, s)

Thus, the generalized Möbius strips with n1 and n2 half-twists are diffeomorphic if and only if n1 ≡ n2 ( mod 2)
(i.e., they have the same parity). This means that the number of halftwists must either both be odd or both
be even.
Summary: 1. The parametric equations for a generalized Möbius strip with n half-twists are given by:

41
 4

       
nt nt nt
σ(t, s) = R + s cos cos(t), R + s cos sin(t), s sin
2 2 2

2. The strip is orientable if and only if n is even.

3. Two strips with n1 and n2 half-twists are diffeomorphic if and only if n1 ≡ n2 (mod2), meaning they have
the same parity.

Exercise: 3.67.

(Normal Neighborhood of a Surface). Let S be a compact regular surface. Let N be an orientation for S. For
small ϵ > 0, define f : S × (−ϵ, ϵ) → R3 as f (p, t) = p + t · N (p).
(1) Prove that for sufficiently small ϵ, f is a diffeomorphism onto its image, which makes sense if you
regard its domain as a subset of R4 .
HINT: Modify the hint for Exercise 3.11 on page 124.
(2) Let ϵ > 0 be as in part (1). Prove that for every r ∈ (−ϵ, ϵ), the set Sr = {f (p, r) | p ∈ S} is a regular

surface diffeomorphic to S, and that Sr ∪ S−r = p ∈ R3 | dist(p, S) = r .

. 1 (p, t) ∈ S × (−ϵ, ϵ), p S U (p, εp ) := {q ∈ S|dist(p, q) < εp }

R2 V (p), φ : V (p) → U (p) f˜ := f ◦ (φ, id) : V × R → R3 as f˜(u, v, t) =
f (φ(u, v), t) = φ(u, v) + t · N (φ(u, v))
h i h i h i
df˜ = dφ + td(N ◦ φ) N ◦ φ , df˜(u,v,0) = dφ N ◦ φ = φu φv N ◦φ .

N ◦ φ//φu × φv , df(u,v,0) ϵu,v > 0, f˜ V (p) × (−ϵp , ϵp )

V (p) × (−ϵp , ϵp ) U (p, εp ) × (−ϵp , ϵp ) f U (p, εp ) × (−ϵp , ϵp )
S {p1 , · · · , pn } {Wi := f (U (pi , εpi ) × (−ϵpi , ϵpi )), i = 0, · · · , n} S
Proposition A.23 δ>0 Lebesgue
 
1 δ
ϵ= min ϵp1 , . . . , ϵpn , ,
2 2
φ S × (−ϵ, ϵ) φ S × (−ϵ, ϵ) φ S × (−ϵ, ϵ)

φ S × (−ϵ, ϵ) (p, s) 6= (p̃, s̃) ∈ S × (−ϵ, ϵ), φ(p, s) = φ(p̃, s̃).

|p − p̃| < 2ϵ |f (p̃, s̃) − f (p̃, s̃)| < 4ϵ < δ f (p̃, s̃), f (p̃, s̃) Wi f U (pi , εpi ) ×
(−ϵpi , ϵpi ) |p − p̃| ≥ 2ϵ

0 = |f (p, s) − f (p̃, s̃)| ≥ ||p − p̃| − |sN (p, s)| − |s̃N (p̃, s̃)|| > 2ϵ − 2ϵ = 0,

φ S × (−ϵ, ϵ)
2 φ : f (p, r) 7→ f (p, 0). f (p, 0) = f (q, 0), p = q, f (p, r) = f (q, r),
−1
φ varphi φ , varphi
dist(p, S) = r q ∈ S, ⃗ ⊥ Tq S, p = q ± rN (q), p ∈ Sr ∪ S−r .
qp

42
 4

p ∈ Sr ∪ S−r , p = q + rN (q), r ≥ 0 dist(p, S) = r. dits(p, q) = r,

dist(p, x) ≥ r, ∀x ∈ S.
x ∈ S, dist(x, p) < r, p p0 ∈ S p⃗0 p ⊥ Tp0 S, p = p0 + sN (p0 ) = f (p0 , s) =
q + rN (q) = f (q, r), p 6= q, f

Exercise: 3.68.

Let S be an orientable regular surface and let N be an orientation for S. Let φ : S → R be a smooth function.
Assume that φ has compact support, which means that there exists a compact subset K ⊂ S such that φ
equals zero at every point of S − K. For t ∈ R, define

St = {p + tφ(p)N (p) | p ∈ S}

Prove that for sufficiently small t, St is a regular surface.

. p ∈ K ⊂ S, p S U (p) ⊂ K ⊂ S U (p) φ = 0, p∈K

\
p∈K σ : V (p) ⊂ R2 → U (p) = O(p) S O(p) R3 f : V (p) →
\
O(p) St , (u, v) 7→ (σ(u, v) + tφ(σ(u, v))N (σ(u, v))).
t=0 dfu,v dσu,v 2 df t tp t < tp dfu,v 2 f
V (p) V (p, ϵp )
K pi , i = 1, · · · , n f (V (pi , ϵpi )) St t0 = min ti , i = 1, · · · , n,
t < t 0 , St

Exercise: 3.69.

Let S be a regular surface, p ∈ S, v ∈ Tp S a nonzero vector, and N a unit normal vector to S at p. Prove that
there exists a neighborhood, V , of p in S such that the intersection of V with the plane p + span{v, N } is the
trace of a regular curve.

. u ∈ Tp S, ⃗u ⊥ ⃗v , p ⃗
⃗u, ⃗v , N R3 p S V, Π := span{⃗u, ⃗v }
U, f : U → V, f (u, v) = (u, v, g(u, v))

p = (0, 0, 0) = f (0, 0) = (0, 0, g(0, 0)), g(0, 0) = 0,

fu (0, 0), fv (0, 0) ∈ Tp S ⇒ gu (0, 0) = gv (0, 0) = 0.

\
V S Π = {γ(t) := f (0, t) = (0, t, g(0, t))|t ∈ (−ϵ, ϵ)}.
′
γ (0) = (0, 1, gv (0, 0)) 6= ⃗0, γ 0

43
 4

Exercise: 3.74.

Let S be the surface of revolution obtained when the graph of y = e−x , x ∈ (0, ∞), is revolved about the
x-axis. Use an improper integral to decide whether S has finite or infinite surface area.

. S = {σ(t, x) = (x, e−x cos t, e−x sin t)|t ∈ [0, 2π), x ∈ (0, ∞).

kdσk = |σt × σx | = |(0, −e−x sin t, e−x cos t) × (1, −e−x cos t, −e−x sin t)|
p
= |(e−2x , e−x cos t, e−x sin t)| = e−x 1 + e−2x .

Z 2π Z ∞ Z ∞ p Z 1 p
Area(S) = kdσkdxdt = 2π e−x 1 + e−2x dx = 2π 1 + y 2 dy
0 0 0 0
 q q  1 √ √
= π x (1 + x²) − ln x + (1 + x²) = π[ 2 − ln(1 + 2)].
0

Exercise: 3.75.

For fixed ϕ0 ∈ (0, π), derive the following formula for the area of the spherical cap consisting of all points of
S 2 whose spherical coordinates (θ, ϕ) (as in Example 3.24 on page 129) satisfy ϕ ≤ ϕ0 :

Area = 2π (1 − cos ϕ0 )

.
σ(θ, φ) = (sin φ cos θ, sin φ sin θ, cos φ), θ ∈ (0, 2π), φ ∈ (0, φ0 ].

kdσk = |σθ × σφ | = |(− sin φ sin θ, sin φ cos θ, 0) × (cos φ cos θ, cos φ sin θ, − sin φ)|
= |(− sin2 φ cos θ, − sin2 φ sin θ, − sin φ cos φ)| = sin φ

Z 2π Z φ0 Z φ0
Area = sin φdφdθ = 2π sin φdφ = 2π (1 − cos ϕ0 ) .
0 0 0

Exercise: 3.77.

What is the area of the Möbius strip (defined in Eq. 3.14 on page 156) ?

σ(u, v) = (cos u(2 + v sin(u/2)), sin u(2 + v sin(u/2)), v cos(u/2)), (u, v) ∈ [0, 4π) × (−1/2, 1/2).

44
 4

kdσk = |σu × σv |
= |(− sin u(2 + v sin(u/2)) + v/2 cos u cos(u/2), cos u(2 + v sin(u/2)) + v/2 sin u cos(u/2), −v/2 sin(u/2))
×(cos u sin(u/2), sin u sin(u/2), cos(u/2))|
= |(cos u cos(u/2)(2 + v sin(u/2)) + v/2 sin u, −v/2 cos u + sin u cos(u/2)(2 + v sin(u/2)),
p
− sin(u/2)(2 + v sin(u/2)))| = v 2 /4 + (2 + v sin(u/2))2

Z 4π Z 1/2 p
Area = v 2 /4 + (2 + v sin(u/2))2 dudv =
0 −1/2

Exercise: 3.78.

Let S be a regular surface. Every pair of overlapping coordinate charts from a given atlas will determine a
transition function Ψ, defined as in the proof of Proposition 3.56. Prove that S is orientable if and only if
there exists an atlas for S such for each transition function Ψ satisfies det (dΨq ) > 0 for all q in its domain.

. 1. fi : Ui → V i det (dψq ) = (fu11 ψv21 − ψv11 ψu2 1 ) > 0,

fui i × fvii
atles N i :=
|fui i × fvii |
f i : U i → Vi
fu11 × fv11 (f 2 ◦ ψ)u1 × (f 2 ◦ ψ)v1 [fu22 ψu1 1 + fv22 ψu2 1 ] × [fu22 ψv11 + fv22 ψv21 ]
N1 = = =
|fu11 × fv11 | |(f 2 ◦ ψ)u1 × (f 2 ◦ ψ)v1 | |[fu22 ψu1 1 + fv22 ψu2 1 ] × [fu22 ψv11 + fv22 ψv21 ]|
(ψu1 1 ψv21 − ψv11 ψu2 1 )fu22 × fv22 fu22 × fv22
= = = N 2.
|(ψu1 1 ψv21 − ψv11 ψu2 1 )fu22 × fv22 | |fu22 × fv22 |

2. S N atles det (dψq ) = (fu11 ψv21 −

fui i × fvii \
ψv11 ψu2 1 ) < 0 Ui N= , U1 U2 fu11 = fu22 ψu1 1 + fv22 ψu2 1 , fv11 = fu22 ψv11 + fv22 ψv21 ,
|fui i × fvii |
N U2
(f 2 ◦ ψ)u1 × (f 2 ◦ ψ)v1 [fu22 ψu1 1 + fv22 ψu2 1 ] × [fu22 ψv11 + fv22 ψv21 ]
N2 = =
|(f 2 ◦ ψ)u1 × (f 2 ◦ ψ)v1 | |[fu22 ψu1 1 + fv22 ψu2 1 ] × [fu22 ψv11 + fv22 ψv21 ]|
(ψu1 1 ψv21 − ψv11 ψu2 1 )fu22 × fv22 f 2 × fv22
= = − u22 ,
|(ψu1 ψv1 − ψv1 ψu1 )fu2 × fv2 |
1 2 1 2 2 2 |fu2 × fv22 |

Remark 11:

V {ej }, {e′j } A (e1 , e2 )A = (e′1 , e′2 )

det A > 0
S coordinate chart (U1 , ψ1 ), (U2 , ψ2 ), p ∈
U1 ∩ U2 ψj (Uj ) = Vj ∈ R 2
R 2
e = {ex , ey }

45
 4

p
Ep = (ψ1−1 )ψ1 (p) (e), Ep′ = (ψ2−1 )ψ2 (p) (e)

E, E ′ E E′ A
Tp S det A > 0
E = d(ψ1 ◦ ψ2−1 )E ′ dψq

S S
S R3 dimS = dimR3 − 1
R3 (ex , ey , ez ) S ez Tp S
Np Tp S

Exercise: 3.81.

Let f : S1 → S2 be a diffeomorphism between regular surfaces. Prove that f is an isometry if and only if for
every regular curve γ : [a, b] → S1 , the length of γ equals the length of f ◦ γ.

Z b Z b
′
. ∀ γ : [a, b] → S1 , |γ (t)|dt = |(f ◦ γ)′ (t)|dt. ∀v ∈ Tp S1 , ∃γ : [a, b] → S1 , γ(0) =
Z s Z s a a

p, γ ′ (0) = v. |γ ′ (t)|dt = |(f ◦ γ)′ (t)|dt s

a a

|γ ′ (s)| = |(f ◦ γ)′ (s)| = |dfγ(s) (γ ′ (s))|, |v| = |γ ′ (0)| = |(f ◦ γ)′ (0)| = |dfp (v)|.

∀v ∈ Tp S1
|v| = |γ ′ (0)| = |(f ◦ γ)′ (0)| = |dfp (v)|.
Z b Z b Z b
′ ′
∀ γ : [a, b] → S1 , |γ (t)|dt = |dfγ(t) (γ (t))|dt = |(f ◦ γ)′ (t)|dt.
a a a

Exercise: 3.82.

A function f : S1 → S2 between regular surfaces is called a local diffeomorphism if for all p ∈ S1 , there exists
a neighborhood U of p in S1 such that the restriction of f to U is a diffeomorphism onto its image f (U ). If
additionally hdfp (x), dfp (y)if (p) = hx, yip for all p ∈ S1 and all x, y ∈ Tp S1 , then f is called a local isometry.

(1) Show that the map from R2 to the cylinder C = (x, y, z) ∈ R3 | x2 + y 2 = 1 defined as f (x, y) =
(cos x, sin x, y) is a local isometry. It is not a (global) diffeomorphism because it is not injective.
(2) Construct a local diffeomorphism from the cylinder


C = (x, y, z) ∈ R3 | x2 + y 2 = 1 and − 1/2 < z < 1/2

to the Möbius strip defined in Sect.5. Does there exist a (global) diffeomorphism between these surfaces?

46
 4

. 1. f : S1 → S2 isometry < df, df >S2 S2

f S1 f S1

< df, df >=< fx , fx > dx2 + 2 < fx , fy > dxdy+ < fy , fy > dy 2 = dx2 + dy 2 .
isometry f df

2. Möbius strip

M = {σ(u, v) = (cos(u)(2 + v sin(u/2)), sin(u)(2 + v sin(u/2)), v cos(u/2)), (u, v) ∈ [0, 4π) × (−1/2, 1/2)}.

g(cos t, sin t, s) = (cos(2t)(2 + s sin(t)), sin(2t)(2 + s sin(t)), s cos(t)), (t, s) ∈ [0, 2π) × R, f
[0, 2π) × R σ g = σ ◦ f −1
C
t- γ(t) = (cos(t0 + t), sin(t0 + t), s0 ), t ∈ R s- γ(s) = (cos t0 , sin t0 , s0 + s), s ∈ (−1/2, 1/2),
C ft , f s

ft (cos t, sin t, s) = (− sin t, cos t, 0), fs (cos t, sin t, s) = (0, 0, 1),

N (cos t, sin t, s) = ft × fs = (cos t, sin t, 0).

h:C→M p ∈ M , dhp Tp C Th(p) M ft , f s

M dh(ft ) = (h ◦ f )t , dh(fs ) = (h ◦ f )s dh(ft )
dh(fs ) M dh(ft ) × dh(fs )
dh(ft ) × dh(fs )
|dh(ft ) × dh(fs )|
M M

Exercise: 3.90.

Prove Proposition 3.75 A diffeomorphism f : S1 → S2 is an isometry if and only if it is equiareal and

conformal.

. equiarea conformal I1 , I 2
∗
f equiarea kdf k = 1 f conformal f I2 :=< df, df >2 = λ2 I1
T p S1 e1 , e 2
p
kdf k = |df (e1 ) ∧ df (e2 )|2 = | det(df (e1 ), df (e2 ))| = |f11 f22 − f12 f21 | = | det(f ∗ I2 )|.

fij df T p S1 (e1 , e2 ) Tf (p) S2 det(f ∗ I2 ) (e1 , e2 )

abuse of notation
T p S1 (v1 , v2 )
p p
|v1 ∧ v2 | = det I1 = EG − F 2 ,

47
 4

I = E(v 1 )2 +2F (v 1 v 2 )+G(v 2 )2 v j vj (e1 , e2 )A = (v1 , v2 ) (e1 , e2 )T = A(v 1 , v 2 )T

I = (e1 )2 + (e2 )2 vj I = AT A det I = det A2 = |v11 v22 − v12 v21 |2 = |v1 ∧ v2 |2
s
| det(f ∗ I2 )|
kdf k = .
| det I1 |

T p S1 df

f isometry f ∗ I2 = I1 f ∗ I 2 = λ2 I 1 |λ| = 1
∗
f I2 = λ I1 2
kdf k = |λ| = 1

Exercise: 3.91.

Verify the following formulas that express the stereographic projection function f and its inverse in rectangular
coordinates:


2 −1 4x, 4y, x2 + y 2 − 4
f (x, y, z) = (x, y), f (x, y) = .
1−z x2 + y 2 + 4

.
2x 2y
f = (u, v), u(x, y, z) = , v(x, y, z) = ;
1−z 1−z


4u, 4v, u2 + v 2 − 4
g(u, v) = (x, y, z)(u, v) := .
u2 + v 2 + 4
   
x2 +y 2
2x 2y
1−z , 1−z , (1−z)2 −1 2x 2y 1+z
1−z , 1−z , 1−z −1
g ◦ f (x, y, z) = x2 +y 2
= 1+z = (x, y, z), (x, y, z) ∈ S 2 \ {(0, 0, 1)};
(1−z)2 +1 1−z + 1
 
8u/(u2 + v 2 + 4) 8v/(u2 + v 2 + 4)
f ◦ g(u, v) = , = (u, v), (u, v) ∈ R2 .
8/(u2 + v 2 + 4) 8/(u2 + v 2 + 4)
f S 2 \ {(0, 0, 1)} R2 g f, g f

Remark 12

(0, 0, 1) − (x, y, z) = λ[(0, 0, 1) − (u, v, z0 )],

(−x, −y, 1 − z) = λ(−2x/(1 − z), −2y/(1 − z), 1 − z0 ),

λ = (1 − z)/2, z0 = −1. f S 2 \ {(0, 0, 1)} z = −1

48
 4

Exercise: 3.92.

A ”circle in S 2 ” means the intersection of S 2 with a plane (not necessarily passing through the origin). Prove



that stereographic projection matches ( a circle in S 2 ↔ ( a circle or a line in R2 .

. S2 N = (0, 0, 1) S 2 \ {N } z=0
1.
S2
x2 + y 2 + z 2 = 1,

ax + by + cz = d.

d=0 d 6= 0
2.
S2 P = (x, y, z) N = (0, 0, 1) z=0 f (P ) = (X, Y )
x y
X= , Y = .
1−z 1−z
3.
(X, Y ) x y X Y

x = X(1 − z), y = Y (1 − z).

x2 + y 2 + z 2 = 1

X2 + Y 2 − 1
z= .
X2 + Y 2 + 1
z ax + by + cz = d X Y

AX 2 + BY 2 + CX + DY + E = 0.

R2 A = B 6= 0 A=B=0

Exercise: 3.100.

If σ : U ⊂ R2 → R3 is a parametrized surface (rather than a surface patch of a regular surface), it still makes
sense to define E, F, G : U → R and to define F1 exactly as in this section. For each function, describe the


largest domain on which it is a parametrized surface, and compute F1 : (1) σ(u, v) = u2 , v 2 , u2 + v 2 + u + v ,



(2) σ(u, v) = u, u2 , v 3 , (3) σ(u, v) = u − v, u + v, u2 + v 2 .

. σ E, F, G, F1
2 2 2 2


(1) σ(u, v) = u , v , u + v + u + v

σu = (2u, 0, 2u + 1) , σv = (0, 2v, 2v + 1) .

49
 4

   
2u 0 2u 0
  r3 −r1 −r2  
dσ = 
 0 2v   −→  0
 2v 

2u + 1 2v + 1 1 1

R2 \ {(0, 0)} rank(dσ) = 2, σ R2 \ {(0, 0)}

E = hσu , σu i = (2u)2 + (2u + 1)2 = 4u2 + (2u + 1)2 , F = hσu , σv i = 0, G = hσv , σv i = (2v)2 + (2v + 1)2 .

F1 = Edu2 + 2F dudv + Gdv 2 = (4u2 + (2u + 1))du2 + (4v 2 + (2v + 1)2 )dv 2 .



(2) σ(u, v) = u, u2 , v 3 , σu = (1, 2u, 0) , σv = 0, 0, 3v 2 .
   
1 0 1 0
  r2 −2ur1  
dσ =  
2u 0  −→ = 0 0 
 
2 2
0 3v 0 3v

R2 \ {(u, 0)|u ∈ R} rank(dσ) = 2, σ R2 \ {(u, 0)|u ∈ R}

E = hσu , σu i = 12 + (2u)2 = 1 + 4u2 , F = hσu , σv i = 0, G = hσv , σv i = (3v 2 )2 = 9v 4 .

F1 = Edu2 + 2F dudv + Gdv 2 = (1 + 4u2 )du2 + 9v 4 dv 2 .

(3) σ(u, v) = u − v, u + v, u2 + v 2 , σu = (1, 1, 2u) , σv = (−1, 1, 2v) .
 
1 −1
 
dσ = 
1 1 , rank(dσ) = 2,
2u 2v

σ R2

E = hσu , σu i = 12 + 12 + (2u)2 = 2 + 4u2 , F = hσu , σv i = (−1)(1) + (1)(1) + (2u)(2v) = 4uv,

G = hσv , σv i = (−1)2 + 12 + (2v)2 = 2 + 4v 2 .

F1 = Edu2 + 2F dudv + Gdv 2 = (2 + 4u2 )du2 + 8uvdudv + (2 + 4v 2 )dv 2 .

Exercise: 3.101.

Prove that the following are equivalent for a diffeomorphism f : S → S̃ between regular surfaces:
(1) f is an isometry.
(2) For every surface patch σ : U ⊂ R2 → V ⊂ S, the first fundamental form of σ equals the first
fundamental form of f ◦ σ.
(3) Every p ∈ S is covered by a surface patch σ such that the first fundamental form of σ equals the first
fundamental form of f ◦ σ.

50
 4

. (1) ⇒ (2)
f , f ◦ σ : U → S̃ S̃ surface patch,
1 1
hdf (v), df (w)i = {kdf (v + w)k2 − kdf (v)k2 − kdf (w)k2 } = {kv + wk2 − kvk2 − kwk2 } = hv, wi, ∀v, w ∈ Tp S.
2 2

F1 (f ◦ σ) = h(f ◦ σ)u , (f ◦ σ)u idu2 + 2h(f ◦ σ)u , (f ◦ σ)v idudv + h(f ◦ σ)v , (f ◦ σ)v idv 2
= hdf (σu ), df (σu )idu2 + 2hdf (σu ), df (σv )idudv + hdf (σv ), df (σv )idv 2
= hσu , σu idu2 + 2hσu , σv idudv + hσv , σv idv 2 = F1 (σ).

(2) ⇒ (3) p ∈ S, surface patch σ : U ⊂ R2 → V ⊂ S, p∈V

(3) ⇒ (1) (3) v ∈ Tp S, surface patch σ : U ⊂ R → V ⊂ S,
2
p∈V σ
U γ : I → U,

p = σ ◦ γ(0), v = dσ(γ ′ (0)) = (σ ◦ γ)′ (0).

3 F1 (σ) = F1 (f ◦ σ),

k(f ◦ σ)u k2 (γu′ (0))2 + 2h(f ◦ σ)u , (f ◦ σ)v iγu′ (0)γv′ (0) + k(f ◦ σ)v k2 (γv′ (0))2
= kσu k2 (γu′ (0))2 + 2hσu , σv iγu′ (0)γv′ (0) + kσv k2 (γv′ (0))2 ,
k(f ◦ σ)u γu′ (0)k2 + 2h(f ◦ σ)u γu′ (0), (f ◦ σ)v γv′ (0)i + k(f ◦ σ)v γv′ (0)k2
= kσu γu′ (0)k2 2hσu γu′ (0), σv γv′ (0)i + kσv γv′ (0)k2 ,
k(f ◦ σ)u γu′ (0) + (f ◦ σ)v γv′ (0)k2
= kσu γu′ (0) + σv γv′ (0)k2 ,
k(f ◦ σ ◦ γ)′ (0)k2 = k(σ ◦ γ)′ (0)k2 .

kdf (v)k = kvk. f

Exercise: 3.102.

If σ : U → V ⊂ S and σ̃ : U → Ṽ ⊂ S̃ are surface patches for regular surfaces (with the same domain
U ⊂ R2 ), and they have the same first fundamental form (the same E, F, G functions), prove that V and Ṽ
are isometric.

. σ, σ̃ f := σ̃ ◦ σ −1 : V → Ṽ . Exercise3.101. f

51
 4

Exercise: 3.103.

Let γ : R → R3 be a helix of the form γ(θ) = (cos θ, sin θ, cθ), where c 6= 0 is a constant, shown green in Fig.
3.39. For each value of θ, consider the infinite line (shown red) through γ(θ) that is parallel to the xy-plane
and intersects the z-axis. The union of all these lines is called a helicoid, visualized as the surface swept out
by the propeller of a rising helicopter (or lowering if c < 0 ). It can be covered by the single surface patch

σ(θ, t) = (t cos θ, t sin θ, cθ), t, θ ∈ (−∞, ∞)

(1) Describe the first fundamental form in these coordinates.

(2) What is the area of the portion of the helicoid corresponding to 0 < t < 1 and 0 < θ < 4π?
(3) At a point p of the helicoid, how does the angle that a unit normal vector at p makes with the z-axis
depend on the distance of p to the z-axis?

. (1)
σθ = (−t sin θ, t cos θ, c) , σt = (cos θ, sin θ, 0) .
E = kσθ k2 = t2 + c2 , F = hσθ , σt i = 0, G = kσt k2 = 1,
F1 = Edu2 + 2F dudv + Gdv 2 = (t2 + c2 )du2 + dv 2 .
(2)
Z 1 Z 4π p Z 1 Z 4πp Z 1p
Area = EG − F dθdt =
2 2 2
t + c dθdt = 4π t2 + c2 dt
0 0
√
0 0
!! 0
p 1+ 1 + c2
= 2π 1 + c2 + c2 ln .
c
σθ × σt (−c sin θ, c cos θ, −t) (−c sin θ, c cos θ, −t)
(3) N (p) = = = √ .
|σθ × σt | | (−c sin θ, c cos θ, −t) | t2 + c 2
p −t
dist(p, z − axis) = x(p)2 + y(p)2 = |t|, cos(α) = hN (p), (0, 0, 1)i = √ .
t2 + c 2

Exercise: 3.107.

Find the first fundamental form of S 2 in the local coordinates determined by stereographic projection.


4x, 4y, x2 + y 2 − 4
. σ(x, y) = .
x2 + y 2 + 4
4(−x2 + y 2 + 4) −8xy 16x
σx = ( 2 2 2
, 2 , 2 ,
(x + y + 4) (x + y + 4) (x + y 2 + 4)2
2 2

−8xy 4(x2 − y 2 + 4) 16y

σy = ( , , .
(x2 + y 2 + 4)2 (x2 + y 2 + 4)2 (x2 + y 2 + 4)2

E, F, G, F1

52
 4

Exercise: 3.108.

Let S be a regular surface, and σ : U ⊂ R2 → V ⊂ S a surface patch with first fundamental form Edu2 +
2F dudv + Gdv2 . (1) Prove that σ is an isometry if and only if E = G = 1 and F = 0.
p
(2) Prove that σ is equiareal if and only if EG − F 2 = 1.
(3) Prove that σ is conformal if and only if E = G and F = 0.

. 3.90

Exercise: 3.112.

Compute the first fundamental form of the Möbius strip with respect to the coordinate chart given in this
chapter. Verify that this coordinate chart is not an isometry (your experience constructing a paper model
might have misled your intuition, but we will later prove that this Möbius strip cannot be covered by surface
patches that are isometries).

. Exercise 3.108.

Exercise: 4.1.

Prove that the graph of f = 2xy is obtained by rotating the graph of f = x2 − y 2 counterclockwise 45◦ about
the z-axis.

. f = 2xy f = x2 − y 2

S1 = {(r cos θ, r sin θ, r2 sin(2θ))|(r, θ) ∈ [0, ∞) × R}, S2 = {(r cos t, r sin t, r2 cos(2t))|(r, t) ∈ [0, ∞) × R},

p ∈ S1 p = (r cos θ, r sin θ, r2 sin(2θ)) = (r cos θ, r sin θ, r2 cos(2θ − π/2)) (r cos(θ −

π/4), r sin(θ − π/4), r cos(2(θ − π/4))) ∈ S2
2
z- π/4

Exercise: 4.2.

The level sets of the function f (x, y, z) = x2 + y 2 − z 2 corresponding to the values f = −1, f = 0, and f = 1
were illustrated in Fig. 3.14 on page 134. Describe the image of the Gauss map for each of these level sets (for
f = 0, the origin must first be removed in order for the set to be a regular surface).

. xz- z- xz- g(x, z) = x2 − z 2

f = −1, 0, 1 p γ q Gauss = “p xz-
Gauss z-

53