Differential Geometry

Lecture notes on MTH352: Differential Geometry
MTH352: Differential Geometry

Module Handbook
Credit Hours: 3(3,0)
For
Master of Mathematics
By
Dr. SOHAIL IQBAL

Assistant Professor
Department of Mathematics,
CIIT Islamabad, Pakistan.
1 of 161
The following lecture notes are designed for the course of

MTH-352
Differential Geometry
The notes are not intended to be an independent script and
are designed to go together with the videos and the
recommended book.
2 of 161
Chapter 1
Calculus On Euclidean Space
3 of 161
Lecture # 1
History
From the very beginning of the evolution human beings started observing
geometry of objects around.
The first written work is by Euclid. He compiled of his and others work
into volume form known as Euclid's Elements. His work is one of the most
influential works in the history of mathematics. The “Elements” have been
serving as the main textbook for teaching geometry from the 300 B.C. until
the 20th century. For his contributions he is often referred as “Father of
geometry”.
The next name we came across in the history of geometry is Archimedes.

He understood the geometry of different objects about him, for example,
he calculated the area and volumes of different objects, and the
techniques were later formulated in calculus. He understood geometry to
such an extent that once he said give me a place to stand on and I will
move the Earth.
The next major development in geometry was done in Muslim Era. The
need to predict the phases of the Moon for Ramadan and other religious
festivals led to great steps forward in geometry of celestial objects
(astronomy).
Introduction of “Cartesian coordinates” marked a new stage for geometry,

since geometric figures, could now be represented analytically, that is, with
functions and equations. It is said that Cartesian coordinates were invented
by René Descartes and Pierre de Fermat independently. In Cartesian
coordinates we can associate a point in plane with an ordered pair. In a
coordinate plane we can associate a curve starting from an equation. Hence
4 of 161
the Cartesian coordinates built a bridge between algebra and geometry. The following example
reminds us the procedure.
Example: Graph of equation 𝑦 = 𝑥 2 in Euclidean plane. We can sketch the graph by first finding
the images of different values of 𝑥 under the equation 𝑦 = 𝑥 2 . Then we can sketch the ordered pair
in Cartesian plane to get the graph of the equation 𝑦 = 𝑥 2 .
X 0 1 -1 2 -2
Y 0 1 1 4 4
On the same lines we can have Cartesian coordinates in three dimensional space. On the same
lines we can associate a point in space with an ordered triple.
Using these coordinates we can associate curves and surfaces in
space with an equation.
There is a particular emphasis on surfaces in geometry. Mainly because there are many examples
of surfaces around us, for example, surface of earth and the geometry of space due to some heavy
object.
5 of 161
Aim of the course:

In this course we are mainly interested in doing calculus on surfaces. For this we aim at doing the
followings.
 Review of differential calculus.
 Develop tools to study curves and surfaces in space.
 Proper definition of surface. How to do calculus on surface.
 A detailed study of geometry of surface.
Let us start the course with some review of basics.
Some definitions
Set
A set S is a collection of objects that are called the elements of S.
Example1: S = {1,2,3,4,…} is a set of natural numbers.

Example2: S = { x : x is an integer ^ x is divisible by 2} = Set of even integers.
Subset
A set A is a subset of S provided each element of A is also an element of S.
Example1: A = {1,2,3} is a subset of S = {1,2,3,4,…}.

Example2: A = { x : x is an integer ^ x is divisible by 2 } is a subset of integers.
Function
A function from set 𝐷 to set 𝑅, written as 𝑓 : 𝐷 → 𝑅, is a rule that assigns each element element
of 𝐷 to a unique element of 𝑅
6 of 161
Example1: Let 𝐷 = {𝑎, 𝑏, 𝑐}, 𝑅 = {1,2,3,4,5,6}
Domain
𝐷 is called domain of 𝑓.
Range
𝑅 is called range of 𝑓.
Image
For x ∈ 𝐷, 𝑓(𝑥) is called image of 𝑥 under 𝑓. In above example 𝑓(𝑎) = 1. Image of the function 𝑓
is the set {𝑓(𝑥): 𝑥 ∈ 𝐷}. In above example image of 𝑓 is {1,2,4}.
Example: 𝑓: 𝑹 → 𝑹, 𝑓(𝑥) = 1 + 𝑥, Image of 𝑓 = 𝑹.
Composite Function
Let 𝑓: 𝐷 → 𝑅 and 𝑔: 𝐸 → 𝑆 such that g(𝐸) ⊂ 𝐷 then 𝑓 ∘ 𝑔(𝑥) = 𝑓(𝑔(𝑥)).
Example: 𝑓(𝑥) = 1 + 𝑥, 𝑔(𝑥) = 𝑥 2 , then

𝑓 ∘ 𝑔(𝑥) = 𝑓(𝑔(𝑥)) = 𝑓(𝑥 2 ) = 1 + 𝑥 2 .
One-to-one
𝑓: 𝐷 → 𝑅, 𝑓(𝑥) = 𝑓(𝑦) )𝑥 = 𝑦.
Example: 𝑓: 𝐴 → 𝐵, for 𝐴 = {1,2,3,4}, 𝐵 = {𝑎, 𝑏, 𝑐, 𝑑}.
Example: 𝑓(𝑥) = 1 + 𝑥.
7 of 161
Onto functions
A function 𝑓: 𝐴 → 𝐵, is onto if 𝑓(𝐴) = 𝐵.
Example: 𝑓: 𝐴 → 𝐵, for 𝐴 = {1,2,3,4}, 𝐵 = {𝑎, 𝑏, 𝑐, 𝑑}
Inverse
𝑓: 𝐷 → 𝑅 is one-to-one and onto, then 𝑓 has inverse 𝑓 −1 : 𝑅 → 𝐷, such that 𝑦 → 𝑥 where 𝑦 = 𝑓(𝑥).
Or
𝑔 is inverse of 𝑓 if 𝑓 ∘ 𝑔 = 𝑔 ∘ 𝑓 = 𝐼.
Example: 𝑓(𝑥) = 1 + 𝑥, has inverse 𝑔(𝑥) = 𝑥 − 1.
Euclidean 3-Space
Euclidean 3-space 𝑹 is the set of all ordered triples of real
numbers. Such a triple 𝑝 = (𝑝1 , 𝑝2 , 𝑝3 ) is called a point of 𝑹𝟑 .
For 𝑝 = (𝑝1 , 𝑝2 , 𝑝3 ), 𝑞 = (𝑞1 , 𝑞2 , 𝑞3 ) then
𝑝 + 𝑞 = (𝑝1 + 𝑞1 , 𝑝2 + 𝑞2 , 𝑝3 + 𝑞3 )
For 𝑝 = (𝑝1 , 𝑝2 , 𝑝3 ), and 𝛼 a scalar then
𝛼𝑝 = (𝛼𝑝1 , 𝛼𝑝2 , 𝛼𝑝3 )
𝑹𝟑 is a vector space .
The point 𝑶 = (0,0,0) is called the origin of 𝑹𝟑 .
Review of Fields and vector spaces

Field
A Field is a set 𝐹 such that for any 𝛼, 𝛽, 𝛾 in 𝐹 the following holds:
 Closure of 𝐹 under addition and multiplication:
8 of 161
𝛼 + 𝛽 ∈ 𝐹, and 𝛼. 𝛽 ∈ 𝐹
 Associativity of addition and multiplication:
𝛼 + (𝛽 + 𝛾) = (𝛼 + 𝛽) + 𝛾 and 𝛼. (𝛽. 𝛾) = (𝛼. 𝛽). 𝛾
 Commutativity of addition and multiplication:
𝛼 + 𝛽 = 𝛽 + 𝛼 and 𝛼. 𝛽 = 𝛽. 𝛼
 Existence of additive and multiplicative identity elements:
𝛼 + 0 = 0 + 𝛼 = 𝛼, 𝛼. 1 = 𝛼
 Existence of additive inverses and multiplicative inverses:
−𝛼 ∈ 𝐹 , 𝛼 + (−𝛼) = (−𝛼) + 𝛼 = 0
for 𝛼 ≠ 0, there exists an element 𝛼 −1 ∈ 𝐹 , such that 𝛼. 𝛼 −1 = 1
 Distributivity of multiplication over addition:
𝛼. (𝛽 + 𝛾) = 𝛼. 𝛽 + 𝛼𝛾
Example: Set of real numbers and set of rational numbers are examples of field.
Vector Space
A vector space over a field is a set 𝑉 such that: for any 𝛼, 𝛽 ∈ 𝐹 and 𝑣, 𝑤 ∈ 𝑉, the following hold.
 Closed under addition and scalar

u + v ∈ V , αv ∈ V
multiplication
 Associativity of addition 𝑢 + (𝑣 + 𝑤) = (𝑢 + 𝑣) + 𝑤
 Commutativity of addition 𝑢+𝑣 =𝑣+𝑢
 Identity element of addition 0 ∈ 𝑉 such that 𝑣 + 0 = 0 + 𝑣 = 𝑣
For every 𝑣 ∈ 𝑉 there exist −𝑣 ∈ 𝑉 such that 𝑣 +

 Inverse elements of addition
(−𝑣) = (−𝑣) + 𝑣 = 0
 Distributivity of scalar multiplication with

𝛼(𝑢 + 𝑣) = 𝛼𝑢 + 𝛼𝑣
respect to vector addition
9 of 161
 Distributivity of scalar multiplication with

(𝛼 + 𝛽)𝑣 = 𝛼𝑣 + 𝛽𝑣
respect to field addition
 Compatibility of scalar multiplication with

𝛼(𝛽𝑣) = (𝛼𝛽)𝑣
field multiplication
 Identity element of scalar multiplication 1𝑣 = 𝑣, where 1 is the multiplicative identity in 𝐹.
Example: The Euclidean 3-space 𝑅3 is a vector space.
Coordinate functions
Let 𝑥, 𝑦, and 𝑧 be real-valued functions on 𝑅 3 such the for each point 𝑝 = (𝑝1 , 𝑝2 , 𝑝3 ) we have
𝑥(𝑝) = 𝑝1 , 𝑦(𝑝) = 𝑝2 , 𝑧(𝑝) = 𝑝3
We can use
𝑥1 = 𝑥, 𝑥2 = 𝑦, 𝑥3 = 𝑧
𝒑 = (𝑝1, 𝑝2 , 𝑝3 ) = (𝑥1 (𝒑), 𝑥2 (𝒑), 𝑥3 (𝒑))
Differentiable OR 𝑪∞ functions
A real-valued function 𝑓 on 𝑹𝟑 is differentiable (or infinitely differentiable, or smooth, or of class
𝐶 ∞ ) provided all partial derivatives of 𝑓, of all orders, exist and are continuous.
Example: Consider a function 𝑓: 𝑅2 → 𝑅, defined as 𝑓(𝑥, 𝑦) = 𝑥 2 𝑦. Is 𝑓 a smooth function?

Solution: Since 𝑓 is a polynomial function, hence 𝑓 is differentiable. Now the first partial
derivatives of 𝑓 are :
𝜕𝑓 𝜕𝑓
= 2𝑥𝑦, = 𝑥2.
𝜕𝑥 𝜕𝑦
Which are continuous. Now the second partial derivatives are:
𝜕 2𝑓 𝜕 2𝑓 𝜕 2𝑓
= 2𝑦, = 0, = 2𝑥.
𝜕𝑥 2 𝜕𝑦 2 𝜕𝑥𝜕𝑦
Which are continuous again. Similarly we can see that all the partial derivatives of 𝑓(𝑥, 𝑦) are
continuous so 𝑓(𝑥, 𝑦) is a smooth function.
Arithmetic of differentiable functions
10 of 161
Differentiable real-valued functions 𝑓 and 𝑔 may be added and multiplied in the following way
 (𝑓 + 𝑔)(𝒑) = 𝑓(𝒑) + 𝑔(𝒑),

 (𝑓𝑔)(𝒑) = 𝑓(𝒑)𝑔(𝒑)
Q1(a) (Exercise 1.1):

If 𝑓 = 𝑥 2 𝑦 and 𝑔 = 𝑦 𝑠𝑖𝑛 𝑧 then find 𝑓𝑔2 .
Solution: Here 𝑓𝑔2 represents the point wise multiplication. So
𝑓𝑔2 = (𝑥 2 𝑦)(𝑦 sin 𝑧) = 𝑥 2 𝑦𝑧 sin 𝑧 .
Chain rule
If g is differentiable at x and f is differentiable at g(x) then the composition 𝑓 ∘ 𝑔 is differentiable at
x. Moreover, if
y  f ( g ( x)) and u  g ( x)
Then 𝑦 = 𝑓(𝑢) and

dy dy du
 
dx du dx
Alternatively
d 
 f  g  x    f  g   x   f  g  x g  x 
dx
Derivative of outside function
Derivative of
inside function
Example:
𝑑𝑦
Find 𝑑𝑥 if 𝑦 = 𝑐𝑜𝑠 (𝑥 3 ).
Solution: Let
𝑢 = 𝑥3
Then
11 of 161
𝑦 = cos 𝑢 .
Now
𝑑𝑦
= − sin 𝑢.
𝑑𝑢
And
𝑑𝑢
= 3𝑥 2
𝑑𝑥
So by using chain rule
Rates of
𝑑𝑦 𝑑𝑦 𝑑𝑢
= × change
𝑑𝑥 𝑑𝑢 𝑑𝑥 multiply
𝑑𝑦
= (− sin 𝑢)(3𝑥 2 )
𝑑𝑥
𝑑𝑦
= −3𝑥 2 sin 𝑥 3
𝑑𝑥
Q1(d) (Exercise 1.1):

𝜕
Find (sin 𝑓) where 𝑓 = 𝑥 2 𝑦.
𝜕𝑦
Solution: Using chain rule
𝜕 𝜕 𝜕𝑓
sin 𝑓 = ( sin 𝑓 ) ( ).
𝜕𝑦 𝜕𝑓 𝜕𝑦
𝜕
sin 𝑓 = (cos 𝑓 )(𝑥 2 ).
𝜕𝑦
12 of 161
Lecture # 2
The lecture contents

 Vectors in 𝑅 3
 Tangent vectors
 Vector field
 Natural frame field on 𝑅 3
 Differentiable vector fields
 Some questions from Exercise 1.2
Vectors in 𝑹𝟑
Intuitively a vector in 𝑅 3 is an oriented line segment or “arrow”. Vectors are used to describe
vector quantities such as force, velocities, angular momenta etc.
For practical purposes there is a problem with this definition and needs to be improved. According
to above definition two translated vectors are same, but in real life two vectors can have different
effect if there initial points are different (see picture below). So we need to incorporate the initial
point in the definition of vector.
13 of 161
To define a vector v precisely we mention the starting point p of the vector as well.
Definition: A vector in R3 is p + v, where p is initial point of vector v.

Strictly speaking v is a point in R3 .
Tangent vectors
Definition: A tangent vector 𝑣𝑝 to 𝑅3 consists of two points of 𝑅3; its vector part 𝑣 and its
point of application 𝑝.
A tangent vector 𝑣𝑝 is drawn as an arrow from the point 𝑝 to the point 𝑝 +

𝑣.
Example: If 𝑣 = (2,3,2), and 𝑝 = (1,1,3) then the tangent

vector
𝑣𝑝 = (2,3,2)(1,1,3)
Starts in (1,1,3) and ends in (3,4,5).
Equality of tangent vectors

 Two tangent vectors 𝑣𝑝 and 𝑤𝑞 are parallel if and only in 𝑣 = 𝑤.
14 of 161
 Two tangent vectors 𝑣𝑝 and 𝑤𝑞 are equal if and only in 𝑣 = 𝑤 and 𝑝 = 𝑞.
 𝑣𝑝 and 𝑣𝑞 are different tangent vectors if 𝑝 ≠ 𝑞.
Definition: Let 𝒑 be a point of 𝑹𝟑. The set 𝑇𝑝 (𝑹𝟑) consisting of

all tangent vectors that have 𝒑 as point of application is called the
tangent space of 𝑹𝟑 at 𝒑.
Note: Each point of 𝑹𝟑 has its own tangent space. They are all
different from each other.
Here we give a recall of the method of parallelogram law, scalar multiple of a vector, linear
transformation, and isomorphism of vector spaces.
 Given two vectors 𝑉 and 𝑊 in 𝑅 3 we add them by parallelogram law.
 Let 𝑐 be scalar, then for vector 𝐴, 𝑐𝐴 stretches or shrinks 𝐴 by the factor 𝑐, and reverse the
direction if 𝑐 < 0.
For 𝑣𝑝 , 𝑤𝑝 ∈ 𝑇𝑝 (𝑅 3 ) then we can define:
𝑣𝑝 + 𝑤𝑝 = (𝑣 + 𝑤)𝑝
And for any scalar c ∈ 𝑅 we define
𝑐𝑣𝑝 = (𝑐𝑣)𝑝
Fact : 𝑇𝑝 (𝑅3) is vector space under the above addition and scalar multiplication.
Linear Transformation: A linear transformation L: 𝑉 → 𝑊 from a vector space 𝑉 to

vector space 𝑊 is a function that satisfy the following conditions: for any vectors 𝑣, 𝑤 ∈ 𝑉 and any
scalar 𝑐
𝐿(𝑣 + 𝑤) = 𝐿(𝑣) + 𝐿(𝑤)
L(𝑐𝑣) = 𝑐𝐿(𝑣).
15 of 161
Isomorphism: Two vector spaces 𝑉 and 𝑊 are isomorphic if there exist a bijective linear
transformation between them L: 𝑉 → 𝑊. A bijective linear transformation 𝐿 is also called an
isomorphism.
Fact: Tp (R3) is isomorphic to R3. The isomorphism L: Tp (R3 ) → R3 between them such that for
vp ∈ Tp (R3 )
Using above definition we can conclude that the map L: Tp (R3 ) → R3 defined as
L(vp ) = v.
is an isomorphism.
Vector field
There are many examples of vector field around us, for example, the gravitational field etc. In such
vector fields each point of the domain space represents a vector, for example, in case of gravitation
vector field each point of the space represents a vector, directed towards the center of the earth
and the magnitude representing the amount of force with which earth attracts the object towards
itself. As the force of attraction of earth reduced as the object moves away from the center of earth
so there will be vectors of different length in the gravitational vector field.
Hence we define a vector field as.
Definition: A vector field 𝑉 on 𝑅3 is a function that assigns to each point 𝑝 of 𝑅3 a tangent

vector 𝑉(𝑝) to 𝑅 3 at 𝑝.
16 of 161
Since a vector field is a function so we can add two vector fields just like we add functions. So if 𝑉
and 𝑊 are two vector fields on 𝑅 3 then
(𝑉 + 𝑊)(𝑝) = 𝑉(𝑝) + 𝑊(𝑝).
For any real-valued function 𝑓: 𝑅 3 → 𝑅 we define
(𝑓𝑉)(𝑝) = 𝑓(𝑝)𝑉(𝑝).
Natural frame field on 𝐑𝟑

To investigate vector fields on 𝑅 3 we introduce a set of basic vector fields on 𝑅 3 .
Definition: Let 𝑈1 , 𝑈2, 𝑈3 be the vector fields on 𝑅3 such

that
𝑈1 (𝑝) = (1,0,0)𝑝
𝑈2 (𝑝) = (0,1,0)𝑝
𝑈3 (𝑝) = (0,0,1)𝑝
For each point 𝑝 of 𝑅 3 . We call 𝑈1 , 𝑈2 , 𝑈3 -collectively-the natural frame field on 𝑅 3 .
The following result shows how to factorize given vector field into combination of natural frame:
Lemma: If 𝑉 is a vector field on 𝑅3, there are three uniquely determined real-valued functions,
𝑣1 , 𝑣2 , 𝑣3 on 𝑅 3 such that
𝑉 = 𝑣1 𝑈1 + 𝑣2 𝑈2 + 𝑣3 𝑈3
The functions 𝑣1 , 𝑣2 , 𝑣3 are called the Euclidean coordinate functions of 𝑉.
17 of 161
Differentiable vector field

Definition: A vector field 𝑉 is differentiable if each of its three Euclidean coordinate function
𝑣1 , 𝑣2 , 𝑣3 , is differentiable function (smooth, of class 𝐶 ∞ ).
We always assume vector fields are differentiable. So from now on vector field means
differentiable vector field
End of the lecture
18 of 161
Lecture # 3
Contents:
 Directional derivatives
 How to differentiate composite functions (Chain rule)
 How to compute directional derivatives more efficiently
 The main properties of directional derivatives
 Operation of a vector field
 Basic properties of operations of vector fields
The main of the lecture is to discuss directional derivative of real-valued functions on 𝑅 3 . Recall
that a real-valued function on 𝑅 3 , written as 𝑓: 𝑅 3 → 𝑅, has domain 𝑅 3 and range 𝑅. Graph of such
a function is a surface, for example, the graph in the following is a graph of a real-valued function
on 𝑅 3 .
To define directional derivative we first need the following.
Associated with each tangent vector 𝑣𝑝 to 𝑹𝟑 is the straight line
𝒕 → 𝒑 + 𝒕𝒗
The line is parallel to vector 𝑣 and passes through point 𝑝.
19 of 161
Definition: Let 𝑓 be a differentiable real-valued function on 𝑹𝟑, and let 𝑣𝑝 be a tangent vector
to 𝑹𝟑 . Then the number
𝑑
𝑣𝑝 [𝑓] = (𝑓(𝑝 + 𝑡𝑣))|𝑡=0
𝑑𝑡
Is called the derivative of 𝑓 with respect to 𝑣𝑝 .
From the picture shows the geometrically

interpretation of the directional derivative. The
surface is the graph of the function 𝑓: 𝑅 3 → 𝑅.
The directional derivative of 𝑓 corresponding to
the tangent vector is calculated in the following
mannar:
 First calculate the line corresponding to

the tangent vector 𝑣𝑝 .
 The line corresponds to a curve on the
surface.
 The directional derivative 𝑣𝑝 [𝑓] is then
the derivative of the curve at point 𝑝.
Here the derivative of the curve means the same thing as we did in calculus of one variable (In fact
the green curve in above graph is contained in a plane which is not shown in the graph).
The following are worth noting:
 The number 𝑣𝑝 [𝑓] is called the derivative of 𝑓 with respect to 𝑣𝑝
 We also say that 𝑣𝑝 [𝑓] is a directional derivative.
 The directional derivative 𝑣𝑝 [𝑓] informs us about the change in the value of 𝑓 as we move
away from 𝑝 in the direction 𝑣𝑝 .
Example: Compute 𝑣𝑝 [𝑓] for the function 𝑓 = 𝑥 2 𝑦𝑧, with 𝑝 = (1,1,0) and 𝑣 = (1,0, −3).
Solution: First we compute the line defined by 𝑝 and 𝑣

𝑝 + 𝑡𝑣 = (1,1,0) + 𝑡(1,0, −3) = (1 + 𝑡, 1, −3𝑡)
Now evaluating 𝑓 along the line.
𝑓(𝑝 + 𝑡𝑣) = 𝑓(1 + 𝑡, 1, −3𝑡) = (1 + 𝑡)2 (1)(−3𝑡) = −3𝑡 − 6𝑡 2 − 3𝑡 3
Now
𝑑 𝑑
(𝑓(𝑝 + 𝑡𝑣)) = (−3𝑡 − 6𝑡 2 − 3𝑡 3 ) = −3 − 12𝑡 − 9𝑡 2
𝑑𝑡 𝑑𝑡
20 of 161
Now applying definition of𝑣𝑝 [𝑓] we get

𝑑
𝑣𝑝 [𝑓] = 𝑑𝑡 (𝑓(𝑝 + 𝑡𝑣))|𝑡=0 = (−3 − 12𝑡 − 9𝑡 2 )|𝑡=0 = −3.
For next discussion we need to recall the following.
The following results can be used to calculate directional derivatives more efficiently.
Lemma: If 𝑣𝑝 = (𝑣1, 𝑣2 , 𝑣3 ) is a tangent vector to 𝑹𝟑, then

𝜕𝑓
𝑣𝑝 [𝑓] = ∑𝑣𝑖 (𝑝)
𝜕𝑥𝑖
Proof: Let 𝑝 = (𝑝1, 𝑝2, 𝑝3); then

𝑝 + 𝑡𝑣 = (𝑝1 + 𝑡𝑣1 , 𝑝2 + 𝑡𝑣2 , 𝑝3 + 𝑡𝑣3 )
Now
𝑓(𝑝 + 𝑡𝑣) = 𝑓((𝑝1 + 𝑡𝑣1 , 𝑝2 + 𝑡𝑣2 , 𝑝3 + 𝑡𝑣3 ))
Since
𝑑
(𝑝 + 𝑡𝑣𝑖 ) = 𝑣𝑖
𝑑𝑡 𝑖
By using chain rule, we get
𝑑 𝜕𝑓 𝑑 𝜕𝑓
𝑣𝑝 [𝑓] = 𝑓(𝑝 + 𝑡𝑣 )|𝑡=0 = ∑ (𝑝𝑖 + 𝑡𝑣𝑖 )|𝑡=0 = ∑𝑣𝑖 (𝑝)
𝑑𝑡 𝜕𝑥𝑖 𝑑𝑡 𝜕𝑥𝑖
The same example as before:
Example: Compute 𝑣𝑝 [𝑓] for the function 𝑓 = 𝑥 2𝑦𝑧, with 𝑝 = (1,1,0) and 𝑣 = (1,0, −3).
21 of 161
Solution: We know that

If 𝑣𝑝 = (𝑣1 , 𝑣2 , 𝑣3 ) is a tangent vector to 𝑹𝟑 , then
𝜕𝑓
𝑣𝑝 [𝑓] = ∑𝑣𝑖 (𝑝).
𝜕𝑥𝑖
So we calculate
𝜕𝑓 𝜕𝑓 𝜕𝑓
= 2𝑥𝑦𝑧, = 𝑥 2 𝑧, = 𝑥2𝑦
𝜕𝑥 𝜕𝑦 𝜕𝑧
Thus at point 𝑝 = (1,1,0)
𝜕𝑓 𝜕𝑓 𝜕𝑓
= 0, = 0, =1
𝜕𝑥 𝜕𝑦 𝜕𝑧
So
𝜕𝑓
𝑣𝑝 [𝑓] = ∑𝑣𝑖 (𝑝) = 1(0) + 0(0) − 3(1) = −3
𝜕𝑥𝑖
Theorem: Let 𝑓 and 𝑔 be functions on 𝑹𝟑, 𝑣𝑝 and 𝑤𝑝 tangent vectors, 𝑎 and 𝑏 numbers. Then
1)- (𝑎𝑣𝑝 + 𝑏𝑤𝑝 )[𝑓] = 𝑎𝑣𝑝 [𝑓] + 𝑏𝑤𝑝 [𝑓].
2)- 𝑣𝑝 [𝑎𝑓 + 𝑏𝑔] = 𝑎𝑣𝑝 [𝑓] + 𝑏𝑣𝑝 [𝑔].
3)- 𝑣𝑝 [𝑓𝑔] = 𝑣𝑝 [𝑓]. 𝑔(𝑝) + 𝑓(𝑝). 𝑣𝑝 [𝑔]
Proof (3):
22 of 161
Operation of a vector field

Operation of a vector field 𝑉 on a function 𝑓 is defined as follows:
Definition: Let 𝑉 be a vector field, and let 𝑓: 𝑹𝟑 → 𝑹 be a function. Then 𝑉[𝑓] be the function
defined by
𝑉[𝑓]: 𝑹𝟑 → 𝑹, 𝑉[𝑓](𝑝) = 𝑉(𝑝)[𝑓]
So 𝑉[𝑓] is a function which to each point 𝑝 in 𝑹𝟑 gives us the derivative of 𝑓 with respect to the
tangent vector 𝑉(𝑝).
Example: For the natural fame field 𝑈1, 𝑈2, 𝑈3, we get
𝑈1 [𝑓](𝑝) = 𝑈1 (𝑝)[𝑓] = (1,0,0)𝑝 [𝑓]
Lemma: If 𝑣𝑝 = (𝑣1, 𝑣2 , 𝑣3 ) is a tangent vector to 𝑹𝟑, then

𝜕𝑓
𝑣𝑝 [𝑓] = ∑𝑣𝑖 (𝑝)
𝜕𝑥𝑖
So we get
𝜕𝑓 𝜕𝑓 𝜕𝑓 𝜕𝑓
𝑈1 [𝑓](𝑝) = 𝑈1 (𝑝)[𝑓] = (1,0,0)𝑝 [𝑓] = 1 (𝑝) + 0. (𝑝) + 0 (𝑝) = (𝑝).
𝜕𝑥 𝜕𝑦 𝜕𝑥 𝜕𝑥
Basic properties of operations of vector fields
Corollary: Let 𝑉 and 𝑊 be vector fields on 𝑹𝟑 and 𝑓, 𝑔 and ℎ are real-valued function, then
(1)- (𝑓𝑉 + 𝑔𝑊)[ℎ] = 𝑓𝑉[ℎ] + 𝑔𝑊[ℎ]
(2)- 𝑉[𝑎𝑓 + 𝑏𝑔] = 𝑎𝑉[𝑓] + 𝑏𝑉[𝑔]
(3)- 𝑉[𝑓𝑔] = 𝑉[𝑓]. 𝑔 + 𝑓𝑉[𝑔]
Proof (2): 𝑉[𝑎𝑓 + 𝑏𝑔] = 𝑎𝑉[𝑓] + 𝑏𝑉[𝑔]

For the proof we need to recall the following:
1)- (𝑎𝑣𝑝 + 𝑏𝑤𝑝 )[𝑓] = 𝑎𝑣𝑝 [𝑓] + 𝑏𝑤𝑝 [𝑓].
2)- 𝑣𝑝 [𝑎𝑓 + 𝑏𝑔] = 𝑎𝑣𝑝 [𝑓] + 𝑏𝑣𝑝 [𝑔].
3)- 𝑣𝑝 [𝑓𝑔] = 𝑣𝑝 [𝑓]. 𝑔(𝑝) + 𝑓(𝑝). 𝑣𝑝 [𝑔] 23 of 161

Proof (3): 𝑉[𝑓𝑔] = 𝑉[𝑓]. 𝑔 + 𝑓𝑉[𝑔]
Example: Let 𝑉 be a vector field 𝑉 = 𝑥 2 𝑈1 + 𝑦𝑈3 and let 𝑓be the function 𝑓 = 𝑥 2 𝑦 − 𝑦 2𝑧.
Calculate 𝑉[𝑓].
Solution:
𝑉[𝑓] = (𝑥 2 𝑈1 + 𝑦𝑈3 )(𝑥 2 𝑦 − 𝑦 2 𝑧)
Using the following two properties:
1)- (𝑓𝑉 + 𝑔𝑊)[ℎ] = 𝑓𝑉[ℎ] + 𝑔𝑊[ℎ]
2)- 𝑉[𝑎𝑓 + 𝑏𝑔] = 𝑎𝑉[𝑓] + 𝑏𝑉[𝑔]
We get
𝑉[𝑓] = (𝑥 2 𝑈1 + 𝑦𝑈3 )(𝑥 2 𝑦 − 𝑦 2 𝑧)
= (𝑥 2 𝑈1 + 𝑦𝑈3 )𝑥 2 𝑦 − (𝑥 2 𝑈1 + 𝑦𝑈3 )𝑦 2 𝑧
= 𝑥 2 𝑈1 [𝑥 2 𝑦] + 𝑦𝑈3 [𝑥 2 𝑦]−𝑥 2 𝑈1 [𝑦 2 𝑧] − 𝑦𝑈3 [𝑦 2 𝑧]
= 𝑥 2 (2𝑥𝑦) + 𝑦. 0 − 𝑥 2 . 0 − 𝑦(𝑦 2 )
= 2𝑥 3 𝑦 − 𝑦 3 .
End of the lecture
24 of 161
Lecture # 4
Contents:
 Curves in 𝑅 3
 Velocity Vectors
 Reparametrization of Curves in 𝑅 3
 Derivative with respect to Velocity
 Properties of Curves
 Conclusion
Curves in 𝑹𝟑
Open intervals in 𝑅
An open interval in 𝑅 is a set 𝐼 of real numbers of one of the four forms
• {𝑡| 𝑎 < 𝑡 < 𝑏}
• {𝑡| 𝑏 < 𝑡}
• {𝑡|𝑡 < 𝑎}
• 𝑅
Where 𝑎 and 𝑏 are real numbers.
25 of 161
What is a curve in space intuitively?
The path followed by the bird in 𝑅 3 gives

a trajectory in space.
Let for time 𝑡 the bird is located at
𝛼(𝑡) = (𝛼1 (𝑡), 𝛼2 (𝑡), 𝛼3 (𝑡)),
in 𝑹𝟑 for 𝑡 ∈ (0,8).
This path together with some extra

conditions gives us a curve.
In rigorous terms:
𝛼 is a function from 𝐼 to 𝑹𝟑 ,
where 𝐼 is an open interval. Thus we
write
𝛼(𝑡) = (𝛼1 (𝑡), 𝛼2 (𝑡), 𝛼2 (𝑡)) 𝑓𝑜𝑟 𝑎𝑙𝑙 𝑡 𝑖𝑛 𝐼.
The real-valued functions 𝛼1 (𝑡), 𝛼2 (𝑡), and 𝛼3 (𝑡) are called Euclidean coordinate functions.
Differentiable functions: We define the function 𝛼 to be differentiable provided its coordinate

functions 𝛼1 (𝑡), 𝛼2 (𝑡), and 𝛼3 (𝑡) are differentiable.
Definition: A curve in 𝑅 3 is a differentiable function 𝜶: 𝑰 → 𝑹𝟑 from an open interval 𝐼 to 𝑹𝟑 .
Example 1: Straight line

A line through point 𝑝 and in the direction 𝑞 is a curve
𝛼: 𝑹 → 𝑹𝟑
defined as
𝛼(𝑡) = 𝑝 + 𝑡𝑞 = (𝑝1 + 𝑡𝑞1 , 𝑝2 + 𝑡𝑞2 , 𝑝3 + 𝑡𝑞3 ).
26 of 161
Circle:
The curve 𝑡 → (acos 𝑡, 𝑎𝑠𝑖𝑛𝑡, 0) travels around a circle of radius a > 0 in

the 𝑥𝑦-plane of 𝑅 3 .
Helix:
If we allow this curve to rise (or fall) at a constant rate, we obtain a helix
𝛼: 𝑹 → 𝑹𝟑
defined as
𝛼(𝑡) = (𝑎 cos 𝑡 , asin 𝑡 , 𝑏𝑡)
where 𝑎 > 0, 𝑏 ≠ 0.
Example: The curve 𝛼: 𝑹 → 𝑹𝟑 defined by
𝑡
𝛼(𝑡) = (1 + cos 𝑡 , sin 𝑡 , 2 sin )
2
is a curve. The green line in the figure shows the curve.
27 of 161
Example:
The curve 𝛼: 𝑹 → 𝑹𝟑 defined as
𝜶(𝒕) = (𝒆𝒕 , 𝒆−𝒕 , √𝟐 𝒕 )
is shown in the picture.
Example:
The curve 𝛼: 𝑹 → 𝑹𝟑 defined by
𝛼(𝑡) = (3𝑡 − 𝑡 3 , 3𝑡 2 , 3𝑡 + 𝑡 3 )
is shown in the figure.
This particular curve is called 3-curve.
Velocity vector
Definition: Let 𝛼: 𝐼 → 𝑹𝟑 be a curve in 𝑹𝟑 with 𝛼 = (𝛼1, 𝛼2, 𝛼3). For each number 𝑡 in 𝐼, the
velocity vector of 𝛼 at 𝑡 is the tangent vector
𝑑𝛼1 𝑑𝛼2 𝑑𝛼3

𝛼 ′ (𝑡) = ( (𝑡), (𝑡), (𝑡))
𝑑𝑡 𝑑𝑡 𝑑𝑡
𝛼(𝑡)
at the point 𝛼(𝑡) in 𝑹𝟑 .
28 of 161
Example: Let 𝛼: 𝑹 → 𝑹𝟑 is defined by

𝛼(𝑡) = (3𝑡 − 𝑡 3 , 3𝑡 2 , 3𝑡 + 𝑡 3 ).
Geometric interpretation:
29 of 161
Q1 (Exercise 1.4): Compute the velocity vector of the curve in Example 4.2(3) for arbitrary
𝜋 𝜋
𝑡 and for 𝑡 = 0, 𝑡 = 4 , 𝑡 = 2 visualizing these on figure 1.8.
Solution: Here 𝛼: 𝑹 → 𝑹𝟑 𝑖𝑠 𝑔𝑖𝑣𝑒𝑛 𝑎𝑠
𝑡
𝛼(𝑡) = (1 + cos 𝑡 , sin 𝑡 , 2 sin 2).
Now by definition
𝑑𝛼 𝑑𝛼2 𝑑𝛼3
𝛼 ′ (𝑡) = ( 𝑑𝑡1 (𝑡), (𝑡), (𝑡))
𝑑𝑡 𝑑𝑡
𝛼(𝑡)
which in this case is

𝑡
𝛼 ′ (𝑡) = (− sin 𝑡, cos 𝑡 , cos 2)
𝛼(𝑡)
Now at 𝒕 = 𝟎
𝛼(0) = (2,0 , 0 ).
and hence
𝛼 ′ (0) = (0,1,1 )(2,0 ,0 ) .
30 of 161
𝝅
Now at 𝒕 = 𝟒
𝜋 1+√2 1
𝛼 (4 ) = ( , , 0.76537).
2 √2
and hence
𝜋 1 1
𝛼 ′ (4 ) = (− , , 0.92388 ) 1+√2 1
.
√2 √2 ( , ,0.76537)
2 √2
𝝅
Now at 𝒕 = 𝟐
𝜋
𝛼 (2 ) = (1 , 1 , √2 )
and hence
𝜋
𝛼 ′ (2 ) = (−1,0,
1
√2
) .
(1 ,1 ,√2 )
Reparametrization of Curves in 𝑹𝟑
Definition: Let 𝛼: 𝐼 → 𝑹𝟑 be a curve. If ℎ: 𝐽 → 𝐼 is a differentiable function on an open interval
𝐽, then the composite function
𝛽 = 𝛼(ℎ): 𝐽 → 𝑹𝟑
Is a curve called a reparametrization of 𝛼 by ℎ.
Example: Let the curve 𝛼 is defined as

𝛼(𝑡) = (√𝑡, 𝑡√𝑡, 1 − 𝑡) 𝑜𝑛 𝐼 = (0,4)
If ℎ: 𝐽 → 𝐼 be the function given as ℎ(𝑠) = 𝑠 2 , where 𝐽 = (0,2). Then the reparametrization is
𝛽(𝑠) = 𝛼(ℎ(𝑠)) = 𝛼(𝑠 2 ) = (𝑠, 𝑠 3 , 1 − 𝑠 2 )
31 of 161
Lemma: If 𝛽 is the reparametrization of 𝛼 by ℎ, then

𝑑ℎ
𝛽 ′ (𝑠) = (𝑠)𝛼 ′ (ℎ(𝑠)).
𝑑𝑠
Proof: If 𝛼 = (𝛼1, 𝛼2, 𝛼3 ), then

𝛽(𝑠) = 𝛼(ℎ(𝑠)) = (𝛼1 (ℎ(𝑠)), 𝛼2 (ℎ(𝑠)), 𝛼3 (ℎ(𝑠))).
Now
𝛽′(𝑠) = 𝛼′(ℎ(𝑠)) = (𝛼1 (ℎ(𝑠))′, 𝛼2 (ℎ(𝑠))′, 𝛼3 (ℎ(𝑠))′).
By applying chain rule on 𝛼𝑖 ′(ℎ(𝑠)), for 𝑖 = 1,2,3 we get

′
𝛼𝑖 (ℎ(𝑠)) = 𝛼𝑖′ (ℎ(𝑠))ℎ′ (𝑠) 𝑓𝑜𝑟 𝑖 = 1,2,3.
Which yields
𝛽′(𝑡) = 𝛼′( ℎ(𝑠)) = (𝛼1′ (ℎ(𝑠))ℎ′ (𝑠), 𝛼2′ (ℎ(𝑠))ℎ′ (𝑠), 𝛼3′ (ℎ(𝑠))ℎ′ (𝑠))
= ℎ′ (𝑠) (𝛼1′ (ℎ(𝑠)), 𝛼2′ (ℎ(𝑠)), 𝛼3′ (ℎ(𝑠)))
= ℎ′ (𝑠)𝛼 ′ (𝑠) .
Q3 (Exercise 1.4): Find the coordinate function of the curve 𝛽 = 𝛼(ℎ), where 𝛼 is the curve
in example 4.2(3) and ℎ(𝑠) = cos−1 (𝑠) on 𝐽: 0 < 𝑠 < 1.
Solution: The curve 𝛼 from 4.2(3) is given by:
𝑡
𝛼(𝑡) = (1 + cos 𝑡 , sin 𝑡 , 2 sin ).
2
Lemma: Let 𝛼 be a curve in 𝑹𝟑 and let 𝑓 be a differentiable function on 𝑹𝟑. Then

𝑑(𝑓(𝛼))
𝛼 ′ (𝑡)[𝑓] = (𝑡)
𝑑𝑡
Proof:
32 of 161
In this case
𝑑𝛼1 𝑑𝛼2 𝑑𝛼3

𝛼′ = ( , , )
So
𝜕𝑓 𝑑𝛼𝑖
𝛼 ′ (𝑡)[𝑓] = ∑ (𝛼(𝑡)) (𝑡)
𝜕𝑥𝑖 𝑑𝑡
But by chain rule.
𝑑(𝑓(𝛼))
𝛼 ′ (𝑡)[𝑓] = (𝑡).
𝑑𝑡
Properties of Curves
One-to-one curves: The function 𝛼: 𝐼 → 𝑹𝟑 is one-to one if
𝛼(𝑡1 ) ≠ 𝛼(𝑡2 )𝑓𝑜𝑟 𝑡1 ≠ 𝑡2
If 𝛼 is one-to-one then the curve does not intersect itself, or stay in the same point without
moving on, or repeat itself over after some time.
Periodic curves:
A curve 𝛼: 𝑹 → 𝑹𝟑 is called periodic if there is a positive number 𝑝 such that
𝛼(𝑡 + 𝑝) = 𝛼(𝑡) 𝑓𝑜𝑟 𝑎𝑙𝑙 𝑡 ∈ 𝑹
The smallest such number 𝑝 is the period of 𝛼.
Regular curves: A curve 𝛼: 𝐼 → 𝑹𝟑 is called regular if

𝛼 ′ (𝑡) ≠ (0,0,0)𝛼(𝑡) 𝑓𝑜𝑟 𝑎𝑙𝑙 𝑡 ∈ 𝐼
Example: The curve 𝛼: 𝑹 → 𝑹𝟑 given as 𝛼(𝑡) = (𝑡 2 , 𝑡 3 , 0) is not regular.
End of the lecture
33 of 161
Lecture # 5
Contents:
 1-forms
 Differentials
 Properties of differentials
Classical differentials:
If 𝑓 is a real-valued function on 𝑹𝟑 , then in elementary calculus the differential of 𝑓 is usually
defined as
𝜕𝑓 𝜕𝑓 𝜕𝑓
𝑑𝑓 = 𝑑𝑥 + 𝑑𝑦 + 𝑑𝑧.
𝜕𝑥 𝜕𝑦 𝜕𝑧
𝑑𝑓 here calculates the small change in the value of 𝑓 when there is small change (dx, 𝑑𝑦, 𝑑𝑧) in 𝑥,
𝑦 and 𝑧.
𝝏𝒇 𝝏𝒇 𝝏𝒇
What does 𝒅𝒇 = 𝝏𝒙 𝒅𝒙 + 𝝏𝒚 𝒅𝒚 + 𝝏𝒛 𝒅𝒛 means?
We will see the meaning of this expression using the notion of 1-forms.
Definition:
A 1 −form 𝜙 on 𝑹𝟑 is a real-valued function on the set of all tangent vectors to 𝑹𝟑 such that 𝜙 is
linear at each point, that is,
𝜙(𝑎𝒗 + 𝑏𝒘) = 𝑎 𝜙(𝒗) + 𝑏𝜙(𝒘)
For any numbers 𝑎, 𝑏 and tangent vectors 𝒗 𝒘 at the same point of 𝑹𝟑 .
Sometimes we write 𝑣𝑝 instead of 𝑣 for tangent vector to 𝑹𝟑 at 𝑝.
34 of 161
Fact:
For fix point p in 𝑹𝟑 the resulting fucntion 𝜙𝑝 : 𝑇𝑝 (𝑹𝟑 ) → 𝑹 is linear.
Sum of two 𝟏-forms:
Let 𝜙 and 𝜓 be two 1 −forms then their sum is defined as:
(𝜙 + 𝜓)(𝒗) = 𝜙(𝒗) + 𝜓 (𝒗)
for all tangent vectors 𝒗.
Product of a 𝟏-form with a function 𝒇: 𝑹𝟑 → 𝑹
Let 𝑓: 𝑹𝟑 → 𝑹 be a real valued function and 𝜙 be a one form. Then for any tangent vector 𝑣𝑝 to 𝑹𝟑
we define:
(𝑓𝜙)(𝑣𝑝 ) = 𝑓(𝑝)𝜙(𝑣𝑝 )
Evaluating a 𝟏-form on a vector form

Let 𝑉 be a vector field, then for any point ∈ 𝑹𝟑 , 𝑉(𝑝) is a tangent vector to 𝑹𝟑 at point 𝑝.
So we can evaluate a 1-form 𝜙 on a vector field 𝑉 in the following way
At each point 𝑝 ∈ 𝑹𝟑 the value of 𝜙(𝑉) is the number 𝜙(𝑉(𝑝)).
Differentiable 𝟏-form
A 1-form 𝜙 is differentiable if 𝜙(𝑉) is differentiable whenever 𝑉 is differentiable then 𝜙 is
differentiable .
Properties of 𝝓(𝑽)
The linearity of 𝜙 and vector field properties imply that:
𝜙(𝑓𝑉 + 𝑔𝑊) = 𝑓𝜙(𝑉) + 𝑔𝜙(𝑊)
and
(𝑓𝜙 + 𝑔𝜓)(𝑉) = 𝑓𝜙(𝑉) + 𝑔𝜓(𝑉).
Where 𝑓 and 𝑔 are functions from 𝑹𝟑 to 𝑹.
35 of 161
Definition:
If 𝑓 is a differentiable real-valued function on 𝑹𝟑 , the differential 𝑑𝑓 of 𝑓 is the
1-form such that
𝑑𝑓(𝑣𝑝 ) = 𝑣𝑝 [𝑓]
For all tangent vectors 𝑣𝑝 .
The 1-form 𝑑𝑓 satisfies all the conditions of 1-form:
 𝑑𝑓 maps every tangent vector to a real number.
 𝑑𝑓 is linear because of the properties of directional derivative.
Fact:
For any real-valued function 𝑓 on 𝑹𝟑 the differential of 𝑓 knows directional derivative of 𝑓 in
every direction 𝑣𝑝 .
Let us consider some examples of 1-forms on 𝑅 3
Example 1 ( Differential of Natural coordinate functions )

Natural coordinate functions: Natural coordinate functions 𝑥1 , 𝑥2 and 𝑥3 are functions from 𝑹𝟑
to 𝑹 are defined as:
𝑥1 (𝑝1 , 𝑝2 , 𝑝3 ) = 𝑝1
𝑥2 (𝑝1 , 𝑝2 , 𝑝3 ) = 𝑝2
𝑥3 (𝑝1 , 𝑝2 , 𝑝3 ) = 𝑝3 .
Now we calculate differentials of these natural coordinate functions. Using definition of 𝑑𝑓 we get,
for any tangent vector 𝑣𝑝 :
𝜕𝑥𝑖
𝑑𝑥𝑖 (𝑣𝑝 ) = 𝑣𝑝 [𝑥𝑖 ] = ∑ (𝑝)𝑣𝑗 = ∑ 𝛿𝑖𝑗 𝑣𝑗 = 𝑣𝑖
𝜕𝑥𝑗
𝑗
Where 𝛿𝑖𝑗 is Kronecker delta defined as:
1 𝑤ℎ𝑒𝑛 𝑖 = 𝑗
𝛿𝑖𝑗 = {
0, 𝑤ℎ𝑒𝑛 𝑖 ≠ 𝑗
36 of 161
Thus 𝑑𝑥𝑖 only depends on 𝑣 and not on the point of application 𝑝.
Example 2 (Linear combination of 1-forms)

Let 𝑓1 , 𝑓2 , 𝑓3 : 𝑹𝟑 → 𝑹 be three functions. Let
𝜓 = 𝑓1 𝑑𝑥1 + 𝑓2 𝑑𝑥2 + 𝑓3 𝑑𝑥3
Then 𝜓 is a 1-form, and for any vector 𝑣𝑝 we have
𝜓(𝑣𝑝 ) = 𝑓1 (𝑝) 𝑑𝑥1 (𝑣) + 𝑓2 (𝑝) 𝑑𝑥2 (𝑣) + 𝑓3 (𝑝) 𝑑𝑥3 (𝑣)
= 𝑓1 (𝑝) 𝑣1 + 𝑓2 (𝑝) 𝑣2 + 𝑓3 (𝑝) 𝑣3
= ∑ 𝑓𝑖 (𝑝)𝑣𝑖
𝑖
Next: we show that in fact each 1-form can be written in the form of 𝜓 above.
Euclidean coordinate functions of a 𝟏-form

Lemma: If 𝜙 is a 1-form on 𝑹𝟑 , then 𝜙 = ∑ 𝑓𝑖 𝑑𝑥𝑖 , where 𝑓𝑖 = 𝜙(𝑈𝑖 ).
These functions 𝑓1 , 𝑓2 , and 𝑓3 are called the Euclidean coordinate functions of 𝜙.
Proof:
The classical differentials:

Corollary: Let 𝑓: 𝑅 3 → 𝑅 be a differentiable function. Then
𝜕𝑓 𝜕𝑓 𝜕𝑓
𝜕𝑥 𝜕𝑦 𝜕𝑧
37 of 161
Proof:
Product rule:
Lemma: Let 𝑓𝑔 be the product of differentiable functions 𝑓 and 𝑔 on 𝑹𝟑. Then
𝑑(𝑓𝑔) = 𝑔𝑑𝑓 + 𝑓𝑑𝑔
Proof:
Chain rule:
Lemma: Let 𝑓: 𝑹𝟑 → 𝑹 and ℎ: 𝑹 → 𝑹 be differentiable functions, so the composite function
ℎ(𝑓): 𝑹𝟑 → 𝑹 is also differentiable. Then
𝑑(ℎ(𝑓)) = ℎ′ (𝑓)𝑑𝑓.
Proof: We know that chain rule for ℎ(𝑓) is
Now
38 of 161
Example: Calculate 𝑑𝑓, where 𝑓 = (𝑥 2 − 1)𝑦 + (𝑦 2 + 2)𝑧.

Solution:
𝑑𝑓 = (2𝑥𝑑𝑥)𝑦 + (𝑥 2 − 1)𝑑𝑦 + (2𝑦𝑑𝑦)𝑧 + (𝑦 2 + 2)𝑑𝑧
= 2𝑥𝑦 𝑑𝑥 + (𝑥 2 − 1 + 2𝑦𝑧)𝑑𝑦 + (𝑦 2 + 2)𝑑𝑧
Now we evaluate 𝑑𝑓 on the tangent vector 𝑣𝑝 where 𝑝 = (𝑝1 , 𝑝2 , 𝑝3 ) and 𝑣 = (𝑣1 , 𝑣2 , 𝑣3 ).
𝑑𝑓(𝑣𝑝 ) = 𝑣𝑝 [𝑓] = 2𝑝1 𝑝2 𝑣1 + (𝑝12 − 1 + 2𝑝2 𝑝3 )𝑣2 + (𝑝22 + 2 )𝑣3
End of the lecture
39 of 161
Lecture # 6
Contents
 Differential Forms
 𝑝-forms
 Exterior derivatives
 Conclusion
Differential Forms
 1 −forms (we discussed in last lecture) are part of a larger system called differential forms.
 Will discuss some properties of differential forms rather than full description.
To construct a differential form we follow the following procedure:
 Take real-valued functions 𝑓1 , 𝑓2 , 𝑓3 , … , 𝑓𝑛 : 𝑅 3 → 𝑅
 Take differential 1-forms of the coordinate functions 𝑑𝑥, 𝑑𝑦, 𝑑𝑧.
 Take sum and product of above.
Examples:
1)- 𝑥 2 𝑑𝑥𝑑𝑦 + 𝑥𝑦𝑑𝑦𝑑𝑧 + (𝑥𝑧 + 𝑦 2 )𝑑𝑧𝑑𝑦
2)- 𝑥𝑦𝑧 𝑑𝑥 + (𝑥 2 𝑧 2 + 𝑦)𝑑𝑦 + 𝑥𝑦𝑧𝑑𝑧
3)- 𝑥 2 𝑦 2 𝑑𝑥𝑑𝑦𝑑𝑧 .
40 of 161
 The multiplication of forms 𝑑𝑥 𝑑𝑦 is not the usual multiplication and is not commutative. In
fact
𝑑𝑥 𝑑𝑦 = −𝑑𝑦 𝑑𝑥.
called the alternation rule.
This is a special kind of multiplication called “Wedge product”, we denote it by " ∧ " (called
wedge). Hence the above expression becomes:
𝑑𝑥 ∧ 𝑑𝑦 = −𝑑𝑦 ∧ 𝑑𝑥.
Other properties of wedge product are as follows. For differential forms 𝛼, 𝛽 and 𝛾 we have:
1) 𝛼 ∧ (𝛽 ∧ 𝛾) = (𝛼 ∧ 𝛽) ∧ 𝛾
2) 𝛼 ∧ (𝛽 + 𝛾) = 𝛼 ∧ 𝛽 + 𝛼 ∧ 𝛾
We will ignore the " ∧ " sign where ever it is clear that what kind of multiplication we are
considering.
An important consequence of the wedge product is that “repeats are zero”, due to “alternation
rule”. Which means
𝑑𝑥 ∧ 𝑑𝑥 = 0.
Similarly
𝑑𝑦 ∧ 𝑑𝑦 = 0 and 𝑑𝑧 ∧ 𝑑𝑧 = 0.
In fact
𝑑𝑥 ∧ 𝑑𝑥 = −𝑑𝑥 ∧ 𝑑𝑥 ⇒ 2 𝑑𝑥 ∧ 𝑑𝑥 = 0 ⇒ 𝑑𝑥 ∧ 𝑑𝑥 = 0
Similarly we can show
𝑑𝑦 ∧ 𝑑𝑦 = 0 and 𝑑𝑧 ∧ 𝑑𝑧 = 0.
Exercise: Calculate (𝑥 2 𝑑𝑥𝑑𝑦 + 𝑥𝑦 𝑑𝑦𝑑𝑧) ∧ (𝑥𝑧 𝑑𝑥 + (𝑧 2 + 𝑦)𝑑𝑥𝑑𝑦).

𝑝-forms: A 𝑝-form, for 𝑝 one of 0,1,2,3, is defined as:
1) A 0-form is just a differentiable function 𝑓.
2) A 1-form is an expression 𝑥 + 𝑔𝑑𝑦 + ℎ𝑑𝑧 .
3) A 2-form is an expression 𝑓𝑑𝑥 ∧ 𝑑𝑦 + 𝑔𝑑𝑥 ∧ 𝑑𝑧 + ℎ𝑑𝑦 ∧ 𝑑𝑧
(or simply 𝑓𝑑𝑥𝑑𝑦 + 𝑔𝑑𝑥𝑑𝑧 + ℎ𝑑𝑦𝑑𝑧) .
41 of 161
4) A 3-form is an expression 𝑓𝑑𝑥 ∧ 𝑑𝑦 ∧ 𝑑𝑧 ( or simply 𝑓𝑑𝑥 ∧ 𝑑𝑦 ∧ 𝑑𝑧).
Notice that on 𝑅 3 , where we have three coordinates namely 𝑥, 𝑦, 𝑧 the 4-forms and higher 𝑝-forms
are zero due to the “alternation rule”. For instance
𝑑𝑥 ∧ 𝑑𝑦 ∧ 𝑑𝑧 ∧ 𝑑𝑦 = 0.
“Product of a 𝒑-form and a 𝒒-form is a (𝒑 + 𝒒)-form”.
Lemma: If 𝜙 and 𝜓 are 1-forms, then

𝜙 ∧ 𝜓 = −𝜓 ∧ 𝜙.
Proof: Write
𝜙 = ∑ 𝑓𝑖 𝑑𝑥𝑖
and
𝜓 = ∑ 𝑔𝑖 𝑑𝑥𝑖 .
Then
42 of 161
Exterior derivative
Recall:
Definition:
If 𝑓 is a differentiable real-valued function on 𝑅 3 , the differential 𝑑𝑓 of 𝑓 is the 1-form such that
𝑑𝑓(𝑣𝑝 ) = 𝑣𝑝 [𝑓]
For all tangent vectors 𝑣𝑝 .
We can generalize this concept to define an apply it on a 𝑝-form and get a (𝑝 + 1)-form.
For example for 𝑝 = 1 we define
Definition: If 𝜙 = ∑ 𝑓𝑖 𝑑𝑥𝑖 is a 1-form on 𝑅 3 , the exterior derivative of 𝜙 is the 2-form 𝑑𝜙 =

∑ 𝑑𝑓𝑖 ∧ 𝑑𝑥𝑖 .
We know that:
Corollary: Let 𝑓: 𝑹𝟑 → 𝑹 be a differentiable function. Then

𝜕𝑓 𝜕𝑓 𝜕𝑓
𝜕𝑥 𝜕𝑦 𝜕𝑧
So if we expand 𝑑𝜙 = ∑ 𝑑𝑓𝑖 ∧ 𝑑𝑥𝑖 using above then using:
𝜕𝑓1 𝜕𝑓1 𝜕𝑓1

𝑑𝑓1 = 𝑑𝑥1 + 𝑑𝑥2 + 𝑑𝑥3 .
𝜕𝑥1 𝜕𝑥2 𝜕𝑥3
𝜕𝑓2 𝜕𝑓2 𝜕𝑓2

𝑑𝑓2 = 𝑑𝑥1 + 𝑑𝑥2 + 𝑑𝑥3 .
𝜕𝑥1 𝜕𝑥2 𝜕𝑥3
𝜕𝑓3 𝜕𝑓3 𝜕𝑓3

𝑑𝑓3 = 𝑑𝑥1 + 𝑑𝑥2 + 𝑑𝑥3 .
𝜕𝑥1 𝜕𝑥2 𝜕𝑥3
𝑑𝜙 = ∑ 𝑑𝑓𝑖 ∧ 𝑑𝑥𝑖 using above then using:
𝜕𝑓1 𝜕𝑓1 𝜕𝑓1

𝑑𝑓1 = 𝑑𝑥1 + 𝑑𝑥2 + 𝑑𝑥3 .
𝜕𝑥1 𝜕𝑥2 𝜕𝑥3
𝜕𝑓2 𝜕𝑓2 𝜕𝑓2

𝑑𝑓2 = 𝑑𝑥1 + 𝑑𝑥2 + 𝑑𝑥3 .
𝜕𝑥1 𝜕𝑥2 𝜕𝑥3
𝜕𝑓3 𝜕𝑓3 𝜕𝑓3

𝑑𝑓3 = 𝑑𝑥1 + 𝑑𝑥2 + 𝑑𝑥3 .
𝜕𝑥1 𝜕𝑥2 𝜕𝑥3
43 of 161
Definition: Let 𝜓 = 𝑓𝑑𝑥𝑑𝑦 + 𝑔𝑑𝑥𝑑𝑧 + ℎ𝑑𝑦𝑑𝑧 be a 2-form. Then the exterior derivative of 𝜓 is
the 3-form
𝑑𝜓 = 𝑑𝑓 ∧ 𝑑𝑥𝑑𝑦 + 𝑑𝑔 ∧ 𝑑𝑥𝑑𝑧 + 𝑑ℎ ∧ 𝑑𝑦𝑑𝑧.
It is easy to check the following
𝑑 (𝑎𝜙 + 𝑏𝜓) = 𝑎𝑑𝜙 + 𝑏𝑑𝜓.
Where 𝜙 and 𝜓 are arbitrary forms and a and b are numbers.
Theorem: Let 𝑓 and 𝑔 be functions, 𝜙 and 𝜓 be 1-forms. Then

1) 𝑑(𝑓𝑔) = 𝑑𝑓 𝑔 + 𝑓𝑑𝑔
2) 𝑑(𝑓𝜙) = 𝑑𝑓 ∧ 𝜙 + 𝑓 𝑑𝜙
3) 𝑑(𝜙 ∧ 𝜓) = 𝑑𝜙 ∧ 𝜓 − 𝜙 ∧ 𝑑𝜓
Proof (1): 𝒅(𝒇𝒈) = 𝒅𝒇 𝒈 + 𝒇𝒅𝒈
Proof (3): 𝑑(𝜙 ∧ 𝜓) = 𝑑𝜙 ∧ 𝜓 − 𝜙 ∧ 𝑑𝜓
Lets first consider the case where
𝜙 = 𝑓 𝑑𝑥 and 𝜓 = 𝑔 𝑑𝑦 then:
44 of 161
and
and
Lets first consider now
𝜙 = ∑ 𝑓𝑖 𝑑𝑥𝑖
and
𝜓 = ∑ 𝑔𝑖 𝑑𝑥𝑖 .
End of the lecture
45 of 161
Lecture # 7
Contents:
 Introduction to Mappings
 Tangent Maps
Introduction to Mappings
Our aim:
To understand the mappings
𝑓: 𝑅 𝑚 → 𝑅 𝑛
for different values of 𝑚 and 𝑛.
In particular we are interested in the following cases
𝑓: 𝑅 2 → 𝑅 2 , 𝑓: 𝑅 2 → 𝑅 3, 𝑓: 𝑅 3 → 𝑅 3
Definition: Given a function 𝐹: 𝑅 𝑛 → 𝑅 𝑚 , let 𝑓1 , 𝑓2 , … , 𝑓𝑚 denote the real-valued functions on 𝑅 𝑛

such that
𝐹(𝑝) = (𝑓1 (𝑝), 𝑓2 (𝑝), … , 𝑓𝑚 (𝑝))
for all points 𝑝 in 𝑅 𝑛 .
These function are called the Euclidean coordinate functions of 𝐹, and we write 𝐹 = (𝑓1 , 𝑓2 , … , 𝑓𝑚 ).
Mappings: A function 𝐹: 𝑅𝑛 → 𝑅𝑚 is differentiable if all its Euclidean coordinate functions are

differentiable. A differentiable function from 𝑅 𝑛 to 𝑅 𝑚 is called a mapping.
46 of 161
Example: Consider 𝐹: 𝑅3 → 𝑅3 defined by

𝐹(𝑥, 𝑦, 𝑧) = (𝑥 2 , 𝑦𝑧, 𝑥𝑦)
is differentiable and hence a mapping.
Definition: If 𝛼: 𝐼 → 𝑅𝑛 is a curve in 𝑅𝑛 and 𝐹: 𝑅𝑛 → 𝑅𝑚 is a mapping then the composite

function 𝛽 = 𝐹(𝛼): 𝐼 → 𝑅 𝑚 is a curve in 𝑅 𝑚 called the image of 𝛼 under 𝐹.
Example: Consider the mapping 𝐹: 𝑅2 → 𝑅2 define by:

𝐹(𝑢, 𝑣) = (𝑢2 − 𝑣 2 , 2𝑢𝑣)
To see the effect of this mapping we consider images of a special curve 𝛼 in 𝑅 2 under this mapping,
namely circle of radius r, given as
𝛼(𝑡) = (𝑟 cos 𝑡 , 𝑟 sin 𝑡) where 0 ≤ 𝑡 ≤ 2𝜋
Example: Consider the mapping 𝐹: 𝑅3 → 𝑅3 such that

F(𝑥, 𝑦, 𝑧) = (𝑥 − 𝑦, 𝑥 + 𝑦, 2𝑧)
Observe that 𝐹 is a linear transformation.
Tangent Maps
47 of 161
Our next goal is to find analogous linear approximation for the mapping 𝐹: 𝑅 𝑛 → 𝑅 𝑚 near a
point 𝑝 ∈ 𝑅 𝑛 .
For this we once again used special curves in 𝑅 𝑛 , namely lines associated to tangent vectors 𝑣𝑝 . We
approximate 𝐹 near 𝑝 by the map 𝐹∗ .
Definition: Let 𝐹: 𝑅𝑛 → 𝑅𝑚 be a mapping. If 𝑣 is a tangent vector to 𝑅𝑛 at 𝑝, let 𝐹∗ (𝑣) be the

initial velocity vector of the curve 𝑡 → 𝐹(𝑝 + 𝑡𝑣). The resulting function 𝐹∗ sends tangent vectors
to 𝑅 𝑛 to tangent vectors to 𝑅 𝑚 , and is called the tangent map of 𝐹.
How to calculate tangent maps:

Proposition: Let 𝐹 = (𝑓1, 𝑓2, … , 𝑓𝑚 ) be a mapping from 𝑅𝑛 to 𝑅𝑚 . If 𝑣 is a tangent vector to 𝑅𝑛
at 𝑝, then
𝐹∗ (𝑣) = (𝑣[𝑓1 ], 𝑣[𝑓2 ], … , 𝑣[𝑓𝑚 ]) at 𝐹(𝑝)
See book for proof.
Tangent map of a mapping at a point 𝒑

The tangent map 𝐹∗ sends tangent vectors at 𝒑 to tangent vectors at 𝐹(𝒑). Thus for each 𝒑 in 𝑹𝒏 ,
the map 𝐹∗ give rise to a function
𝐹∗ : 𝑇𝑝 (𝑹𝒏 ) → 𝑇𝐹(𝑝) (𝑹𝒎 )
Called the tangent map of 𝐹 at 𝒑.
Tangent maps at a point
48 of 161
are linear
An important property of tangent maps which is a consequence of proposition is:
Corollary: If 𝐹: 𝑅𝑛 → 𝑅𝑚 is a mapping, then at each point 𝑝 of 𝑅3 the tangent map 𝐹∗: 𝑇𝑝 (𝑅𝑛 ) →
𝑇𝐹(𝑝) (𝑅 𝑚 ) is a linear transformation.
See book for proof.
The velocity of an image curve

Another important consequence of the proposition is that mappings preserved velocities.
Corollary: Let 𝐹: 𝑅𝑛 → 𝑅𝑚 be a mapping. If 𝛽 = 𝐹(𝛼) is the image of a curve 𝛼 in 𝑅𝑛 , then 𝛽′ =

𝐹∗ (𝛼′).
Proof: We take 𝑚 = 3
By proposition above we have:
we have
Tangent maps on vector fields

̅𝑖 } (1 ≤ 𝑖 ≤ 𝑚) be the natural frame fields of 𝑅 𝑛 and 𝑅 𝑚 respectively.
Let {𝑈𝑗 } (1 ≤ 𝑗 ≤ 𝑛) and {𝑈
Then:
Corollary: If 𝐹 = (𝑓1, 𝑓2 , … , 𝑓𝑚 ) is a mapping from 𝑅𝑛 to 𝑅𝑚 , then

𝑚
𝜕𝑓𝑖
𝐹∗ (𝑈𝑗 (𝑝)) = ∑ ̅𝑖 (𝑝)
(𝑝) 𝑈
𝜕𝑥𝑗
𝑖=1
for 1 ≤ 𝑗 ≤ 𝑛 .
See book for proof.
49 of 161
The matrix appearing in the preceding formula,
𝜕𝑓𝑖
( (𝑝))
𝜕𝑓𝑗
1≤𝑗≤𝑛,1≤𝑖≤𝑚
Is called the Jacobian matrix of 𝐹 at point 𝑝.
In expanded form it is given as:
We study mappings 𝐹 using the tangent map.
Definition: A mapping 𝐹: 𝑅 𝑛 → 𝑅 𝑚 is regular provided that at every point 𝑝 of 𝑅 𝑛 the tangent map
𝐹∗𝑝 is one-to-one.
For 𝑝 ∈ 𝑅 𝑛 the following are equivalent:
 𝐹∗ 𝑝 is one-to-one
 𝐹∗𝑝 (vp ) = 0 implies 𝑣𝑝 = 0
 The rank of the Jacobian matrix of 𝐹 at 𝑝 is 𝑛.
Diffeomorphism
Definition: if 𝐹: 𝑅𝑛 → 𝑅𝑚 has an inverse mapping, then 𝐹 is called diffeomorphism.
Inverse function theorem.
Theorem: Let 𝐹: 𝑅𝑛 → 𝑅𝑚 be a mapping. If 𝐹∗𝑝 is one-to-one at 𝑝, then there is a restriction of
𝐹 to an open set containing 𝑝 which is a diffeomorphism.
End of the Lecture
50 of 161
Chapter 2
Lecture #8
Contents:
 The Dot Product
 Frames
The Dot Product
We study distances and angles in geometry. We will see that geometry of Euclidean spaces can be
derived from dot product.
Definition: Let
𝑝 = (𝑝1 , 𝑝2 , 𝑝3 ) and 𝑞 = (𝑞1 , 𝑞2 , 𝑞3 )
be two points in 𝑅 3 . The dot product of 𝑝 and 𝑞 is the real number
𝑝 ∙ 𝑞 = 𝑝1 𝑞1 + 𝑝2 𝑞2 + 𝑝3 𝑞3
Dot product has the following properties
If 𝑝, q and 𝑟 are points in 𝑹𝟑 and 𝑎, 𝑏 are real numbers, then
1. Bilinearity:
51 of 161
(𝑎𝑝 + 𝑏𝑞). 𝑟 = 𝑎(𝑝. 𝑟) + 𝑏(𝑞. 𝑟)
𝑟. (𝑎𝑝 + 𝑏𝑞) = 𝑎(𝑟. 𝑝) + 𝑏(𝑟. 𝑞)
2. Symmetry:
𝑝. 𝑞 = 𝑞. 𝑝
3. Positive definiteness
𝑝. 𝑝 ≥ 0
and
𝑝. 𝑝 = 0 ⇒ 𝑝 = 0
The norm of a point
The norm of a point 𝑝 = (𝑝1 , 𝑝2 , 𝑝3 ) is the number
1
‖𝑝‖ = (𝑝 ∙ 𝑝)2 = √𝑝12 + 𝑝22 + 𝑝32
The norm is a real-valued function on 𝑹𝟑 . It has the following properties
 ‖𝒑 + 𝒒‖ ≤ ‖𝒑‖ + ‖𝒒‖
 ‖𝑎𝒑‖ = |𝑎|‖𝒑‖
for vectors 𝑝 and 𝑞 of 𝑅 3 and real number 𝑎.
Distances in 𝑹𝟑
If 𝑝 and 𝑞 are two points of 𝑅 3 , the Euclidean distance from 𝑝 to 𝑞 is the number
𝑑(𝑝, 𝑞) = ‖𝑝 − 𝑞‖
So we calculate
𝑑(𝑝, 𝑞) = √(𝑝1 − 𝑞1 )2 + (𝑝2 − 𝑞2 )2 + (𝑝3 − 𝑞3 )2
52 of 161
Open sets
The distance formula can be used to give a precise definition of
open sets.
Definition: If 𝑝 is a point in 𝑅 3 and 𝜖 is a positive real number,

then the 𝜖 −neighborhood N 𝜖 (𝑝) of 𝑝 in 𝑅 3 is the set of all points
𝑞 in 𝑅 3 such that 𝑑(𝑝, 𝑞) < 𝜖:
N 𝜖 (𝑝) = {𝑞 ∈ 𝑅 3 : 𝑑(𝑝, 𝑞) < 𝜖}
Definition: A subset O of 𝑅 3 is open if every point of O has some

𝜖 −neighborhood that is completely contained in O :
∀𝑝 ∈ O ∃ 𝜖 > 0: N 𝜖 (𝑝) ⊂ O
This definition of open set can be used on any 𝑅 𝑛 .
We can also replace 𝑑 with some other distance function.
Dot product on tangent vectors

We know that tangent space at any point of 𝑅 3 , that is, 𝑇𝑝 (𝑅3 ) is isomorphic to 𝑅 3 . We can use this
relation to define dot product on any tangent space 𝑇𝑝 (𝑅 3 ).
Definition: The dot product of tangent vectors 𝑣𝑝 and 𝑤𝑝 at the same point of 𝑅 3 is the number
𝑣𝑝 . 𝑤𝑝 = 𝑣. 𝑤.
Similarly we can define norm of any tangent vector in the following way.
Definition: The norm, or length, of a tangent vector 𝑣𝑝 is the norm of its vector part, that is,
‖𝑣𝑝 ‖ = ‖𝑣‖
Schwarz inequality:
A fundamental result of linear algebra is Schwarz inequality given by
|𝑣. 𝑤| ≤ ‖𝑣‖‖𝑤‖
This permits us to define the cosine of the angel between two vectors as:
53 of 161
Definition: To any non-zero vectors 𝑣 and 𝑤 define the cosine of the angle between 𝑣 and 𝑤
by
𝑣. 𝑤 = ‖𝑣‖‖𝑤‖ cos 𝜃
Definition: Two vectors are orthogonal if 𝑣. 𝑤 = 0.
Definition: A vector of length 1 is called a unit vector.
Frames
Definition: A set 𝑒1 , 𝑒2 , 𝑒3 of three mutually orthogonal unit vectors
tangent to 𝑅 3 at point 𝑝 is called a frame at the point 𝑝.
We can say that a set of three tangent vectors 𝑒1 , 𝑒2 , 𝑒3 in 𝑇𝑝 (𝑅 3 ) is a frame at 𝑝 if the vectors
satisfy
𝑒1 . 𝑒2 = 𝑒1 . 𝑒3 = 𝑒2 . 𝑒3 = 0
𝑒1 . 𝑒1 = 𝑒2 . 𝑒2 = 𝑒3 . 𝑒3 = 1
We can write down the same definition using notion of Kronecker delta function 𝛿𝑖𝑗 defined as
0, 𝑖𝑓 𝑖 ≠ 𝑗
𝛿𝑖𝑗 = {
1, 𝑖𝑓 𝑖 = 𝑗
So the alternate definition becomes:
Definition: The set of tangent vectors 𝑒1, 𝑒2, 𝑒3 to 𝑅3 at 𝑝 is a frame at 𝑝 if and only if
𝑒𝑖 . 𝑒𝑗 = 𝛿𝑖𝑗 for all 𝑖, 𝑗 ∈ {1,2,3}
Theorem: Let 𝑒1, 𝑒2, 𝑒3 be a frame at a point 𝑝 of 𝑅 3 . If 𝑣 is any tangent vector to 𝑅 3 at 𝑝, then
𝑣 = (𝑣. 𝑒1 )𝑒1 + (𝑣. 𝑒2 )𝑒2 + (𝑣. 𝑒3 )𝑒3
See book for proof.
54 of 161
Now we know that the tangent space 𝑇𝑝 (𝑹𝟑 ) is isomorphic to 𝑹𝟑 . So three linearly independent
vectors form a basis of 𝑇𝑝 (𝑹𝟑 ).
End of the lecture
55 of 161
Lecture # 9
Contents:
 The Attitude Matrix
 Cross Product
The Attitude Matrix

The Attitude Matrix Of A Frame
Let {𝑒1 , 𝑒2 , 𝑒3 } be a frame at point 𝑝 in 𝑹𝟑
If
𝑒1 = (𝑎11 , 𝑎12 , 𝑎13 )
𝑒2 = (𝑎21 , 𝑎22 , 𝑎23 )
𝑒3 = (𝑎31 , 𝑎32 , 𝑎33 )
Then
𝑎11 𝑎12 𝑎13
𝐴 = (𝑎21 𝑎22 𝑎23 )
𝑎31 𝑎32 𝑎33
is called the attitude matrix of the frame {𝑒1 , 𝑒2 , 𝑒3 }.
Definition:
If the rows of a square matrix are orthonormal, then the matrix is called orthogonal.
Hence the attitude matrix of any frame is an orthogonal matrix.
56 of 161
Definition:
The transpose of 3 × 3 matrix is defined by
𝑎11 𝑎12 𝑎13 𝑎11 𝑎21 𝑎31

𝐴 = (𝑎21 𝑎22 𝑎23 ) ⇒ 𝑡𝐴 = (𝑎12 𝑎22 𝑎32 )
𝑎31 𝑎32 𝑎33 𝑎13 𝑎23 𝑎33
Property of orthogonal matrices
If 𝐴 is a 3 × 3 orthogonal matrix, then the matrix product of 𝐴 with 𝐴𝑡 is equal to
1 0 0
𝑡
𝐴𝐴 = (0 1 0)
0 0 1
So 𝐴𝑡 is inverse of 𝐴.
Cross Product
Cross Product:
Let 𝑣 and 𝑤 be tangent vectors to 𝑅 3 at a point 𝑝. The cross product of 𝑣 and 𝑤 is the
determinant
𝑈1 (𝑝) 𝑈2 (𝑝) 𝑈3 (𝑝)

𝑣 × 𝑤 = | 𝑣1 𝑣2 𝑣3 |
𝑤1 𝑤2 𝑤3
Which is also element of 𝑇𝑝 (𝑅 3 ).
Properties Of Cross Product

Linearity: Let 𝑢, v and 𝑤 be tangent vectors to 𝑅 3 at the same
point 𝑝, and let 𝑎, 𝑏 be real numbers. Then
(𝑎𝑢 + 𝑏𝑣) × 𝑤 = 𝑎(𝑢 × 𝑤) + 𝑏(𝑣 × 𝑤)
and
𝑢 × (𝑎𝑣 + 𝑏𝑤) = 𝑎(𝑢 × 𝑣) + 𝑏(𝑢 × 𝑤)
Properties Of Cross Product
57 of 161
Alternation Rule:
If 𝑣, 𝑤 are tangent vectors to 𝑅 3 at a point 𝑝, then
𝑣 × 𝑤 = −𝑤 × 𝑣
It is also easy to see that 𝑣 × 𝑣 = (0,0,0).
Direction of the Cross Product
Lemma: The cross product of 𝑣 and 𝑤 is orthogonal to both 𝑣 and 𝑤.

See book for the proof.
The length of the cross product:

The cross product of 𝑣 and 𝑤 has length such that
‖𝑣 × 𝑤‖2 = (𝑣. 𝑣)(𝑤. 𝑤) − (𝑣. 𝑤)2
See book for the proof.
A more intuitive description of the length of a cross product is
‖𝑣 × 𝑤‖ = ‖𝑣‖‖𝑤‖ sin 𝜃
58 of 161
Triple Scalar Product
Let 𝑢, 𝑣, 𝑤 be three vectors. The triple scalar product of 𝑢, 𝑣, 𝑤 is the number
𝑢. 𝑣 × 𝑤
We can also write the above as

𝑢1 𝑢2 𝑢3
𝑢. 𝑣 × 𝑤 = | 𝑣1 𝑣2 𝑣3 |
𝑤1 𝑤2 𝑤3
End of the lecture
59 of 161
Lecture # 10
Contents:
 Speed Of A Curve
 Vector Fields On Curves
 Differentiation of Vector Fields
Speed Of A Curve
Speed
Let a car moves between Islamabad and Quetta and follow the path
𝛼 = (𝛼1 , 𝛼2 ): 𝐼 → 𝑅 2
given by
𝛼(𝑡) = (𝑡 2 , 𝑡 3 )
What is the speed of the car after 3 hours. To answer such questions we need the following.
Definition: Let
𝛼 = (𝛼1 , 𝛼2 , 𝛼3 ): 𝐼 → 𝑅 3
be a curve
The speed of 𝛼 is defined as the length of the velocity of :
𝑑𝛼1 2 𝑑𝛼2 2 𝑑𝛼3 2

𝑣= ‖𝛼 ′ ‖ √
= ( ) +( ) +( )
60 of 161
Arc length
Let a car moves between Islamabad and Quetta and follow the path
𝛼 = (𝛼1 , 𝛼2 , 𝛼3 ): 𝐼 → 𝑅 2
given by
𝛼(𝑡) = (2𝑡, 𝑡 2 )
Question: What is the distance travelled in 3 hours?
To answer such questions we need the following.
Definition:
The arc length of 𝛼 from 𝑡 = 𝑎 to 𝑡 = 𝑏 is equal to
𝑏 𝑏
∫ ‖𝛼′‖𝑑𝑡 = ∫ 𝑣(𝑡)𝑑𝑡
𝑎 𝑎
Definition: The restriction of a curve 𝛼: 𝐼 → 𝑅3 to a closed interval [𝑎, 𝑏] is called a curve

segment.
If 𝜎 is a curve segment of 𝛼,
𝜎: [𝑎, 𝑏] → 𝑅 3
then the length of 𝜎 is written as 𝐿(𝜎).
Definition: A curve 𝛽 has unit speed if its speed is 1 at every point, that is,
‖𝛽 ′ ‖ = 1.
Example: Let 𝛽: 𝐼 → 𝑅3 be defined by

𝛽(𝑡) = (cos 𝑡, sin 𝑡 , 𝑡)
Arc-Length parameterization
Theorem: Let 𝛼: 𝐼 → 𝑅3 be a regular curve. Then there exist a reparameterization 𝛽 of 𝛼 such
that
‖𝛽 ′ ‖ = 1.
61 of 161
See book for proof.
Orientation of a reparameterization
A reparameterization of a curve 𝛼 given by
𝛽(𝑡) = 𝛼(ℎ(𝑠))
Is orientation preserving if
ℎ′ (𝑠) ≥ 0
It is orientation reversing if
ℎ′ (𝑠) ≤ 0
FACT: Unit speed reparameterization is always orientation preserving.
Let 𝛼: 𝐼 → 𝑅 3 be a curve. We define:
Definition: A vector field 𝑌 on curve 𝛼 is a function that assigns to each number 𝑡 in 𝐼 a

tangent vector 𝑌(𝑡) to 𝑅 3 at the point 𝛼(𝑡).
Vector Fields On Curves
Example:
For curve 𝛼 the velocity 𝛼′ satisfies the definition of vector field on a
curve.
Properties of vector fields on curves
Definition:
If 𝑌 is a vector field on 𝛼: 𝐼 → 𝑹𝟑 , then for each 𝑡 in 𝐼 we can write
𝑌(𝑡) = (𝑦1 (𝑡), 𝑦2 (𝑡), 𝑦3 (𝑡))𝛼(𝑡) = ∑ 𝑦𝑖 (𝑡)𝑈𝑖 (𝛼(𝑡)).
The real-valued function 𝑦1 , 𝑦2 𝑦3 : 𝑹 → 𝑹 are called Euclidean coordinate functions.
62 of 161
Addition of vector fields
Let 𝑌(𝑡) = 𝑦1 𝑈1 + 𝑦2 𝑈2 + 𝑦3 𝑈3 and 𝑌(𝑡) = 𝑧1 𝑈1 + 𝑧2 𝑈2 + 𝑧3 𝑈3 be two vector valued functions on

𝛼: 𝐼 → 𝑹𝟑 . Then the addition of 𝑌(𝑡) and 𝑍(𝑡) is defined as:
Scalar Multiplication
Let 𝑌(𝑡) = 𝑦1 𝑈1 + 𝑦2 𝑈2 + 𝑦3 𝑈3 be a vector field on 𝛼: 𝐼 → 𝑹𝟑 and 𝑓(𝑡) be a real valued functions.

Then we define 𝑓𝑌
Dot product of two vector fields

𝛼: 𝐼 → 𝑹𝟑 . Then the dot product of of 𝑌(𝑡) and 𝑍(𝑡) is defined as:
Cross product of two vector fields

𝛼: 𝐼 → 𝑹𝟑 . Then the cross product of of 𝑌(𝑡) and 𝑍(𝑡) is defined as:
Differentiation of a vector field on a curve
Definition: Let 𝑌(𝑡) be a vector field on a curve , with 𝑌(𝑡) = ∑ 𝑦𝑖 𝑈𝑖 .
Then the new vector filed on 𝛼
𝑑𝑦𝑖
𝑌 ′ (𝑡) = ∑ 𝑈
𝑑𝑡 𝑖
is the derivative of 𝑌(𝑡).
Definition: Let 𝛼: 𝐼 → 𝑹𝟑 be a curve with coordinates 𝛼(𝑡) = (𝛼1(𝑡), 𝛼2 (𝑡), 𝛼3(𝑡)) then the
acceleration of 𝜶 is defined as
′′
𝑑2 𝛼1 𝑑 2 𝛼2 𝑑 2 𝛼3
𝛼 =( 2 , , )
𝑑𝑡 𝑑𝑡 2 𝑑𝑡 2
Properties of differentiation

𝛼: 𝐼 → 𝑹𝟑 . Let 𝑓: 𝑅 → 𝑅 be a real-valued function. Let 𝑎 and 𝑏 be two real numbers then
63 of 161
Linearity:
(𝑎𝑌 + 𝑏𝑍)′ = 𝑎𝑌 ′ + 𝑏𝑍 ′
Leibnizian properties
(𝑌. 𝑍)′ = 𝑌 ′ . 𝑍 + 𝑌. 𝑍 ′
(𝑓𝑌) = 𝑓 ′ 𝑌 + 𝑓𝑌 ′
Parallel vector fields
A vector field on a curve is parallel if all its values are parallel

tangent vectors.
That is
𝑌(𝑡) = (𝑐1 , 𝑐2 , 𝑐3 )𝛼(𝑡) = ∑ 𝑐𝑖 𝑈𝑖 (𝛼(𝑡))
Vanishing derivatives:
Lemma: Let 𝛼: 𝐼 → 𝑅 3 be a curve then:
1. 𝛼 is constant if and only if 𝛼 ′ = 0.
2. If 𝛼 is not constant, then 𝛼 is a straight line if and only if 𝛼 ′′ = 0.
3. A vector field 𝑌 on a curve 𝛼 is parallel if and only if 𝑌 ′ = 0.
See book for proof.
End of the lecture
64 of 161
Lecture #11
Contents:
 Curvature
 Frenet Frame Field
 Frenet Formulas
 Unit-Speed Helix
Aim of today’s lecture is to understand the geometry of the curves. In daily life we can understand
the geometry of the curve by looking at them, for example, in the following figure there are three
curves given. We can easily see that which curves have more curvature and which curve is turning
very fast.
Today we will define these properties mathematically.
Curvature
Definition: The rate of change of tangent, that is,
𝑇 ′ = 𝛽 ′′
is called the curvature vector filed of 𝛽.
65 of 161
Since we have that
‖𝑇‖ = 1
So
𝑇. 𝑇 = 1
And this gives us
𝑇 ′. 𝑇 = 0
Definition: A vector field 𝑉 on 𝛽 is called normal to 𝛽 if 𝑉. 𝛽 ′ = 0 .
We can see that curvature vector field is normal to 𝛽.
Definition (Curvature): The function 𝜅: 𝐼 → 𝑹 defined by

𝜅(𝑡) = ‖𝑇 ′ (𝑡)‖
Is called the curvature function of 𝛽.
The curvature function satisfies 𝜅(𝑡) ≥ 0 for all 𝑠 ∈ 𝐼 .
Note:
 When 𝜅 is small then the curve is close to straight.
 When 𝜅 is large then the turn in the curve is sharp.
Principal normal
Let 𝛽 be a curve. Let > 0 , that is, curve is never straight. We
define principal normal vector 𝑁 as:
T′
𝑁=
𝜅
We can see that ‖𝑁‖ = 1, hence a unit vector.
The principal normal vector tells us the direction in which the curve turns.
66 of 161
Binormal vector field

The vector field on curve 𝛽
𝐵 = 𝑇×𝑁
is called the binormal vector field of 𝛽.
Lemma: If 𝛽 is a unit-speed curve in 𝑹𝟑 with 𝜅 > 0. Then the three vector fields 𝑇, 𝑁, 𝐵 on 𝛽
are unit vectors that are mutually orthogonal at each point.
Proof: See book.
Definition (Frenet Frame )

The triplet (𝑇, 𝐵, 𝑁) is called the Frenet frame field on 𝛽.
The derivatives of a Frenet field

We can express derivatives of 𝑇, 𝑁, 𝐵 in terms of 𝑇, 𝑁, 𝐵.
The derivative of 𝑻
We defined
𝑇′
𝑁=
𝜅
So
𝑇 ′ = 𝜅𝑁
67 of 161
The derivatives of a Frenet field

We can express derivatives of 𝑇, 𝑁, 𝐵 in terms of 𝑇, 𝑁, 𝐵.
The derivative of B
We can write 𝐵′ in the following form
𝐵 ′ = (𝐵 ′ . 𝑇)𝑇 + (𝐵′ . 𝑁)𝑁 + (𝐵 ′ . 𝐵)𝐵
The torsion of a curve

The function 𝜏: 𝐼 → 𝑹 such that
𝐵 = −𝜏𝑁
is called the torsion of the curve.
The Frenet Formulas

Theorem: If 𝛽 is a unit speed curve with curvature 𝜅 and torsion 𝜏 then
𝑇′ = 𝜅𝑁
𝑁 ′ = −𝜅𝑇 + 𝜏𝐵
𝐵′ = − 𝜏𝑁
Proof: see book.
End of the lecture
68 of 161
Lecture # 12
Contents:
 Frenet Approximation
 Plane Curves
 Planes
In this lecture we are mainly interested in approximating the curve using Frenet Frame and Frenet
formulas. So we want to approximate the curve using the geometric information, for example,
using curvature, torsion, normal, etc.
First we recall definition of a plane in 𝑅 3 .
A plane through 𝒑 and orthogonal to 𝒒 ≠ 0 consists of all points 𝒓 such that
(𝒓 − 𝒑). 𝒒 = 𝟎
forms a plane.
Next we are going to recall Taylor series.
69 of 161
Taylor Series
Let 𝑓(𝑥) be a function from 𝑅 to 𝑅. If 𝑓(𝑥) is infinitely
differentiable near 0. Then the functional values of 𝑓(𝑥)
near 𝑥 = 0 are given by the Taylor series.
𝑑𝑓 𝑑2𝑓 𝑥 2 𝑑3𝑓 𝑥3
𝑓(𝑥) = 𝑓(0) + (0) 𝑥 + 2 (0) + 3 (0) +⋯
𝑑𝑥 𝑑𝑥 2 𝑑𝑥 3!
To approximate a curve near an arbitrary point on the curve. We are going to achieve this by
using Taylor series and Frenet Frame at a point.
Frenet Approximation
Approximation of Euclidean Coordinate Functions
The Taylor series around origin is given.
Taylor series around point 𝐱 = 𝟎
𝑑𝑓 𝑑2 𝑓 𝑠 2 𝑑3𝑓 𝑠3
𝑓(𝑥) = 𝑓(0) + (0) 𝑠 + 2 (0) + 3 (0) + ⋯
𝑑𝑠 𝑑𝑠 2 𝑑𝑠 3!
Where 𝑥 is any point near 𝑥 = 0.
Now let 𝛽 = (𝛽1 , 𝛽2 , 𝛽3 ) be a unit speed curve. We can approximate each of the functions 𝛽𝑖 , for 𝑖 =
1,2,3 in the following way:
𝑑𝛽𝑖 𝑑2 𝛽 𝑠 2 𝑑3 𝛽 𝑠3
𝛽𝑖 ~ 𝛽𝑖 (0) + (0) 𝑠 + 2 (0) + 3 (0)
𝑑𝑠 𝑑𝑠 2 𝑑𝑠 3!
In coordinate form we can write.
𝑠 2 ′′ 𝑠 3 ′′′
𝛽(𝑠) = (𝛽1 (𝑠), 𝛽2 (𝑠), 𝛽3 (𝑠)) ~ 𝛽(0) + 𝑠 𝛽′(0) + 𝛽 (0) + 𝛽 (0)
2 3!
Now the above is an approximation to the curve. Now we want to see what kind of geometrical
properties can be obtained from this geometrical approximation of the curve. For this we apply
Frenet formulas.
Applying Frenet Formulas
We know that 𝑇(𝑠) = 𝛽′(𝑠) so this gives us
70 of 161
𝛽 ′ (0) = 𝑇(0) = 𝑇0
We also know that
𝑇 ′ (𝑠) = 𝜅(𝑠)𝑁(𝑠) ⇒ 𝛽 ′′ (𝑠) = 𝜅(𝑠)𝑁(𝑠)
⇒ 𝛽 ′′ (0) = 𝜅(0) 𝑁(0) ⇒ 𝛽0′′ = 𝜅0 𝑁0
To calculate 𝛽 ′′′ (0) we use Leibniz rule
𝑑(𝜅 𝑁) 𝑑𝜅 𝑑𝑁
𝛽 ′′′ = = 𝑁+𝜅
𝑑𝑠 𝑑𝑠 𝑑𝑠
Frenet formulas imply that
𝑁 ′ (𝑠) = −𝜅 𝑇 + 𝜏 𝐵
Using this in above expression we get
𝑑𝜅
𝛽 ′′′ = 𝑁 − 𝜅 2 𝑇 + 𝜏𝜅 𝐵
𝑑𝑠
So we get
𝑑𝜅
𝛽0′′′ = 𝛽 ′′′ (0) = (0) 𝑁0 − 𝜅02 𝑇0 + 𝜏0 𝜅0 𝐵0
𝑑𝑠
Definition: Let 𝛽 be a unit speed curve then

𝑠2 𝑠3
𝛽̂ (𝑠) = 𝛽(0) + 𝑠𝑇0 + 𝜅0 𝑁0 + 𝜏0 𝜅0 𝐵
2 6 0
is called Frenet approximation of 𝛽 near 𝑠 = 0.
Next question: Can we do the same process for a point 𝑠0 other than 𝑠 = 0.
The Frenet approximation at point 𝒔𝟎

For this we use Taylor series around point 𝑠 = 𝑠0 .
Taylor series around point 𝐱 = 𝒔𝟎
𝒅𝒇 𝒅𝟐 𝒇 (𝒔 − 𝒔𝟎 )𝟐 𝒅𝟑 𝒇 (𝒔 − 𝒔𝟎 )𝟑
𝒇(𝒙) = 𝒇(𝒔𝟎 ) + (𝒔𝟎 ) (𝒔 − 𝒔𝟎 ) + 𝟐 (𝒔𝟎 ) + 𝟑 (𝒔𝟎 ) +⋯
𝒅𝒔 𝒅𝒔 𝟐 𝒅𝒔 𝟑!
Where 𝑥 is any point near 𝑥 = 𝑠0.
Using above and Frenet Formulas we can show that for a general point 𝑠0 on the curve the
approximation is given by
71 of 161
(𝑠 − 𝑠0 )2 (𝑠 − 𝑠0 )3
𝛽̂𝑠0 (𝑠) = 𝛽(𝑠0 ) + (𝑠 − 𝑠0 )𝑇𝑠0 + 𝜅𝑠0 𝑁𝑠0 + 𝜅𝑠0 𝜏𝑠0 𝐵𝑠0
2 3!
Important: We have triplet (𝑇, 𝐵, 𝑁) at each and every point of the curve.
So we have at each point of curve
𝑠2 𝑠3
𝛽̂ (𝑠) = 𝛽(0) + 𝑠𝑇0 + 𝜅0 𝑁0 + 𝜏0 𝜅0 𝐵
2 6 0
Now we discuss the terms of approximation one by one.
The curve defined by 𝑠 → 𝛽(0) + 𝑠𝑇0 is the tangent line

to 𝛽 at 𝑠 = 0. It is the best linear approximation to 𝛽
near 𝛽(0).
𝑠2
The approximation 𝛽(0) + 𝑠𝑇0 + 𝜅0 𝑁0 is in fact a quadric function in 𝑠 and in particular defines
2
a parabola. The parabola lies in the plane with axis 𝑇0 and 𝑁0 . This plane is called Osculating
plane.
The approximation
𝑠2 𝑠3
𝛽(0) + 𝑠𝑇0 + 𝜅0 𝑁0 + 𝜏0 𝜅0 𝐵
2 6 0
is curve which is moving the parabola in the third dimension of 𝐵0.
72 of 161
Lets recall what is a planar curve.
Planar curve
A plane curve in 𝑹𝟑 is the set of all point which lie in the same
plane in 𝑹𝟑 .
See figure below for a geometrical idea.
Torsion Free Curves

Let 𝛽 be a curve in 𝑹𝟑 with positive curvature 𝜅 > 0. Then 𝛽 is a plane curve if and only if 𝜏 = 0.
Theorem: Let 𝛽 be a curve in 𝑹𝟑 with positive curvature 𝜅 > 0. Then 𝛽 is a plane curve if and
only if 𝜏 = 0.
See book for proof.
End of the lecture
73 of 161
Lecture #13
Contents:
 Frenet Approximation
 Conclusion
 Frenet Frame For Arbitrary Speed Curves
 Velocity And Acceleration
Curvature Of Circle
Consider the following circle
𝑠 𝑠
𝛽(𝑠) = (acos , 𝑎 sin )
𝑎 𝑎
1
Then it is a unit speed curve. And its curvature is 𝑎.
Now the question is: every curve with torsion zero (contained in plane) and constant curvature is
a circle?
Lemma: If 𝛽 is a unit speed curve with constant curvature 𝜅 and torsion zero, then 𝛽 is part of a
1
circle of radius 𝜅.
See book for proof.
Arbitrary Speed Curves

We use the following result.
Theorem: Let 𝛼: 𝐼 → 𝑹𝟑 be a regular curve. Then there exist a reparameterization 𝛽 of 𝛼 such that
‖𝛽 ′ ‖ = 1.
74 of 161
Arc length parameter

Let 𝛼: I → 𝑅 be a regular curve. We can find arc-length parameter 𝑠 in the following way
𝑡
𝑠 = ∫ ‖𝛼′(𝑢)‖𝑑𝑢
𝑎
Let 𝛼̅ be the arc-length reparametrization of 𝛼 such that
𝛼(𝑡) = 𝛼(𝑠(𝑡)) 𝑡∈𝐼
Example: Consider a helix 𝛼: 𝑹 → 𝑹 defined by

𝛼(𝑡) = (𝑎 cos 𝑡 , 𝑎 sin 𝑡, 𝑏𝑡)
Now 𝛼̅ is a unit speed curve. If 𝜅̅ > 0 then we can define the following
𝜅̅ , 𝜏̅, 𝑇̅, 𝑁
̅ , 𝐵̅
The Frenet Frame of 𝜶

Definition: Let 𝛼 be a regular curve and let 𝜶̅ be its arc-length parametrization. Then we
define
The curvature function of : 𝜅 = 𝜅̅ (𝑠)
The torsion function of : 𝜏 = 𝜏̅(𝑠)
75 of 161
Unit tangent vector field of 𝛼: 𝑇 = 𝑇̅(𝑠)
̅ (𝑠)
Principal normal vector field of : 𝑁 = 𝑁
Binormal vector field of : 𝐵 = 𝐵̅ (𝑠)
For arbitrary speed curve 𝛼 the speed 𝑣(𝑡) = ‖𝛼′(𝑡)‖ is an important factor in the Frenet
formulas for arbitrary speed curves.
Lemma: If 𝛼 is a regular curve in 𝑹𝟑 with 𝜅 > 0, then

𝑇′ = 𝜅𝑣𝑁
𝑁 ′ = −𝜅𝑣𝑇 + 𝜏𝑣𝐵
𝐵′ = − 𝜏𝑁
If a object (e.g. bird, airplane, etc ) is moving on a constant speed then velocity and acceleration
are perpendicular. If the object follow the path 𝛼: 𝐼 → 𝑹𝟑 . Since 𝛼 has constant speed so
‖𝛼′‖ = 𝑐
Definition: Let 𝛼: 𝐼 → 𝑹𝟑 be a curve with coordinates 𝛼(𝑡) = (𝛼1(𝑡), 𝛼2 (𝑡), 𝛼3(𝑡)) then the
acceleration of 𝜶 is defined as
𝑑2 𝛼1 𝑑 2 𝛼2 𝑑 2 𝛼3
𝛼 ′′ = ( 2 , , )
𝑑𝑡 𝑑𝑡 2 𝑑𝑡 2
So
𝛼.′ 𝛼 ′ = 𝑐 2
𝛼 ′ . 𝛼 ′′ = 0
So velocity and acceleration are perpendicular.
We can analyze the situation by expressing velocity and acceleration in terms of Frenet Frame
field.
Lemma: If 𝛼 is a regular curve with speed function 𝑣, then its velocity and acceleration are
expressed by
𝑑𝑣
𝛼 ′ = 𝑣𝑇, 𝛼 ′′ = 𝑇 + 𝜅𝑣 2 𝑁
𝑑𝑡
76 of 161
Definition: The term 𝑑𝑣

𝑑𝑡
𝑇 is called the tangential component of acceleration.
Definition: The term 𝜅𝑣 2 𝑁 is called normal component of 𝛼′′.
End of the lecture
77 of 161
Lecture # 14
Contents:
 Frenet Apparatus For A Regular Curve
 Computing Frenet Frame
 The Spherical Image
 Cylindrical Helix
 Conclusion
In this lecture we will learn to calculate the Frenet Frame for arbitrary speed curves. For this
purpose the following result is useful.
Theorem: Let 𝛼 be a curve in 𝑹𝟑. Then

𝛼
𝑇= ,
‖𝛼 ′ ‖
𝑁 =𝐵×𝑇
𝛼′ × 𝛼 ′′
𝐵= ,
‖𝛼 ′ × 𝛼 ′′ ‖
‖𝛼 ′ × 𝛼 ′′ ‖
𝜅=
‖𝛼 ′ ‖3
(𝛼 ′ × 𝛼 ′′ ). 𝛼 ′′′
𝜏=
‖𝛼 ′ × 𝛼 ′′ ‖2
See lecture video of the book for the proof of this.
78 of 161
Exercise: Consider 𝛼 a curve in 𝑹𝟑 given by

𝛼(𝑡) = (3𝑡 − 𝑡 3 , 3𝑡 2 . 3𝑡 + 𝑡 3 )
Find the Frenet apparatus for 𝛼.
Applications Of Frenet Formulas

Given a curve 𝛽 we construct a new curve 𝛽̂ . Such that 𝛽̂ has some properties of 𝛽 and helps us to
understand those properties in a better way.
Spherical Image:
Let 𝛽: 𝐼 → 𝑹𝟑 be a unit speed curve. The spherical image 𝛽 is the curve 𝜎 ≈ 𝑇. That is
𝜎: 𝐼 → 𝑹𝟑
is given by
𝜎(𝑡) = (𝑣1 (𝑡), 𝑣2 (𝑡), 𝑣3 (𝑡))
where
(𝑣1 (𝑡), 𝑣2 (𝑡), 𝑣3 (𝑡))𝛽(𝑡)
is tangent at point 𝛽(𝑡) of the curve.
79 of 161
Example: The unit helix is given by

𝑠 𝑠 𝑏𝑠
𝛽(𝑡) = (𝑎 𝑐𝑜𝑠 , 𝑎 𝑠𝑖𝑛 , )
𝑡 𝑡 𝑐
where 𝑐 = √𝑎2 + 𝑏 2 . Then the spherical image 𝜎 of 𝛽 is given by:
Curvature Of The Spherical Image

Let 𝜎 be the spherical image of the unit speed curve 𝛽. Let us calculate the curvature of 𝜎.
End of the lecture
80 of 161
Lecture # 15
Contents:
 Cylindrical Helix
 Covariant Derivatives
 Euclidean Coordinate Representation
 Properties Of The Covariant Derivative
 The Vector Field Of Covariant Derivatives
Cylindrical Helix
Sometimes the geometrical information like torsion and curvature are enough to tell what the
curve is exactly going to be. Cylindrical is one of those examples where, by just looking at the
𝜏
torsion and curvature, you can tell with exactness. In fact the ration 𝜅 is some times very decisive
in the case of some curves. We start with the definition of the cylindrical helix.
Definition: A regular curve 𝛼 in 𝑹𝟑 is a cylindrical helix provided the unit tangent 𝑇 of 𝛼 has
constant angle 𝜃 with some fixed unit vector 𝒖; that is, 𝑇(𝑡). 𝒖 = 𝑐𝑜𝑠 𝜃.
81 of 161
Movement of a cylindrical helix

Assume that 𝛽 is a unit-speed curve which is a cylindrical helix, that is, 𝑇. 𝑢 = cos 𝜃, for
some fixed unit vector 𝒖. Lets consider the following function:
ℎ(𝑠) = (𝛽(𝑠) − 𝛽(0)). 𝒖
𝜏
Theorem: A regular curve 𝛼 with 𝜅 > 0 is a cylindrical helix if and only if the ration 𝜅 is constant.
See book for proof.
In summary we have the following information directly about the curve by just looking at the
Frenet apparatus.
82 of 161
Covariant Derivative
We have seen the definition of vector field. For
an example of a vector field see the diagram.
As we can see that the arrows are different at

each and every point. Our next aim is
understand the rate of change of vector field at
any point. For this we define covariant
derivative.
Definition:
Let 𝑊 be a vector field on 𝑹𝟑 . Let 𝑣𝑝 be a tangent vector to 𝑹𝟑 at point 𝑝. The covriant derivative of
𝑊 with respect to 𝒗𝒑 is the tangent vector
𝛻𝑣 𝑊 = 𝑊(𝒑 + 𝑡𝒗)′ (0)
at the point 𝑝.
The symbol 𝛻 is called nabla.
In fact 𝛻𝒗 𝑊 measures the initial rate of change of 𝑊(𝒑) as 𝒑 moves in

the direction of 𝒗.
Example: Wind Vector field

One of the practical examples of a
vector fields are wind vector field,
which tells us the direction and
magnitude of the wind at any point.
Calculating the rate of change of this

vector field is a crucial part of weather
prediction. We use covariant derivative
to calculate such quantities.
83 of 161
Now lets see how to calculate the covariant derivative.
Exercise: Consider a vector field

𝑊 = 𝑥 2 𝑈1 + 𝑦𝑧𝑈2
and
𝑣 = (−1,0,2), 𝑎𝑡 𝑝 = (2,1,0).
Lemma: If 𝑊 = ∑ 𝑤𝑖 𝑈𝑖 is a vector field on 𝑹𝟑 , and 𝒗 is a tangent vector at 𝒑, then
∇v W = ∑ 𝑣[𝑤𝑖 ]𝑈𝑖 (𝑝)
See book for proof.
From the properties of directional derivative we can drive
Theorem: Let 𝒗 and 𝒘 be tangent vectors to 𝑹𝟑 at 𝒑, and let 𝑌 and 𝑍 be vector fields on 𝑹𝟑 .
Then for numbers 𝑎, 𝑏 and functions 𝑓.
(1) ∇𝑎𝑣+𝑏𝑤 𝑌 = 𝑎∇𝑣 𝑌 + 𝑏 ∇𝑤 𝑌
(2) ∇𝑣 (𝑎𝑌 + 𝑏𝑍) = 𝑎 ∇𝑣 𝑌 + 𝑏 ∇𝑣 𝑍
(3) ∇𝑣 (𝑓∇) = 𝑣[𝑓]𝑌(𝑝) + 𝑓(𝑝)∇𝑣 𝑌
(4) 𝑣[𝑌. 𝑍] = ∇𝑣 𝑌. 𝑍 (𝑝) + 𝑌(𝑝). ∇𝑣 (𝑍)
The vector field of covariant derivatives
Definition: Let 𝑉 and 𝑊 be vector fields on 𝑹𝟑.

The covariant derivative ∇𝑉 𝑊 of 𝑊 with respect to 𝑉 is a vector field defined by
(∇𝑉 𝑊)(𝑝) = ∇𝑉(𝑝) 𝑊
Corollary: If = ∑ 𝑤𝑖 𝑈𝑖 , then
∇𝑉 𝑊 = ∑ 𝑉[𝑤𝑖 ]𝑈𝑖
Example: We always use ∇𝑉 𝑊 = ∑ 𝑉[𝑤𝑖 ]𝑈𝑖 and 𝑈𝑖 [𝑓] = 𝜕𝑥

𝜕𝑓
𝑖
:
If 𝑊 = 𝑥𝑦𝑈1 − 𝑒 𝑧 𝑈3 and 𝑉 = 𝑧𝑈1 + (𝑥 − 𝑦)𝑈2 , then
𝑉[𝑥𝑦] = 𝑧𝑈1 [𝑥𝑦] + (𝑥 − 𝑦)𝑈2 [𝑥𝑦] = 𝑦𝑧 + 𝑥(𝑥 − 𝑦)
𝑉[𝑒 𝑧 ] = 𝑧𝑈1 [𝑒 𝑧 ] + (𝑥 − 𝑦)𝑈2 [𝑒 𝑧 ] = 0
84 of 161
and therefore
∇𝑉 𝑊 = (𝑦𝑧 + 𝑥(𝑥 − 𝑦))𝑈1
Corollary: Let 𝑉, 𝑊, 𝑌 and 𝑍 be vector fields on 𝑹𝟑 . Then
(1) ∇𝑓𝑉+𝑔𝑊 𝑌 = 𝑓∇𝑣 𝑌 + 𝑔 ∇𝑊 𝑌 for all functions 𝑓 and 𝑔
(2) ∇𝑉 (𝑎𝑌 + 𝑏𝑍) = 𝑎∇𝑉 𝑌 + 𝑏 ∇𝑉 𝑍 for all numbers 𝑎 and 𝑏
(3) ∇𝑉 (𝑓𝑌) = 𝑉[𝑓]𝑌 + 𝑓 ∇V Y for all functions 𝑓
(4) 𝑉[𝑌. 𝑍] = ∇𝑉 𝑌. 𝑍 + 𝑌. ∇𝑉 𝑍
See book for proof.
End of the lecture
85 of 161
Lecture #16
Contents:
 From Curves to Space
 Frame Fields
 Coordinate Functions
From Curves To Space
In 1847 Frenet formulas were discovered by Jean Frédéric

Frenet (7 February 1816 – 12 June 1900) in 1847.
In 1851 Joseph Alfred Serret (August 30, 1819 - March 2,

1885) discovered the same formulas independently.
So we can call them Frenet—Serret Formulas.
𝑇′ 0 𝜅𝑣 0 𝑇
(𝑁′) = (−𝜅𝑣 0 𝜏𝑣 ) (𝑁)
From Curves to Surfaces 𝐵′ 0 −𝜏𝑣 0 𝐵
So far we have discussed the geometry of curves in 𝑅 3 . We mainly use the Frenet frame 𝑇, 𝑁 , 𝐵.
Then we calculated the rate of change 𝑇 ′ , 𝑁 ′ , 𝐵 ′ using Frenet formulas.
Now to go from curves to surfaces we use the same idea, that is,
86 of 161
Assign a frame to each point of the surface
And then we calculate the changes in the frame in terms of the frame itself.
The following theorem helps
Theorem: Let 𝑒1 , 𝑒2 , 𝑒3 be a frame at a point 𝑝 of 𝑹𝟑 . If 𝑣 is any tangent vector to 𝑹𝟑 at 𝑝, then
𝑣 = (𝑣. 𝑒1 )𝑒1 + (𝑣. 𝑒2 )𝑒2 + (𝑣. 𝑒3 )𝑒3 .
Instead of defining frame field on the geometrical object, like sphere, we

carry on this idea for the Euclidean space 𝑹𝟑 . Then the restriction of the
idea to geometries in 𝑹𝟑 is simple.
Frame Fields
We can extend the idea of point wise operation to define the following terms for vector fields.
Definition: Let 𝑉 and 𝑊 be vector fields in 𝑹𝟑 , then we can define their dot product 𝑉. 𝑊 as
follows:
𝑉. 𝑊 (𝒑) = 𝑉(𝑝). 𝑊(𝒑)
Similarly we can define the following:
Definition: Let 𝑉 be a vector field then we define ‖𝑉‖ to be a real valued function on 𝑅3
‖𝑉‖: 𝑅 3 → 𝑅
87 of 161
given by
‖𝑉(𝑝)‖ = √𝑉(𝑝). 𝑉(𝑝)
The norm ‖𝑉‖ is not defined around 0.
Definition: Three vector fields 𝐸1, 𝐸2 , 𝐸3 on 𝑹𝟑 forms a frame field

on 𝑹𝟑 if they satisfy
𝐸𝑖 . 𝐸𝑗 = 𝛿𝑖𝑗 for 1 ≤ 𝑖, 𝑗 ≤ 3
Thus at each point 𝒑 the vectors 𝐸1 (𝑝), 𝐸2 (𝑝), 𝐸3 (𝑝) do in fact form a
frame since they have unit length and are mutually orthogonal.
Example ( Cylindrical Frame Field)

Before defining the cylindrical frame we recall what are
cylindrical coordinates.
Cylindrical Coordinates: Let 𝑟 , 𝜃, 𝑧 be the cylindrical

coordinate functions of 𝑹𝟑 .
We pick unit vector field in the direction in which each coordinate increases
𝐸1 = cos 𝜃 𝑈1 + sin 𝜃 𝑈2
𝜋 𝜋
𝐸2 = cos (𝜃 + ) 𝑈1 + sin (𝜃 + )
2 2
= − sin 𝜃 𝑈1 + cos 𝜃 𝑈2
𝐸3 = 𝑈3
It is easy to check that
𝐸𝑖 . 𝐸𝑗 = 𝛿𝑖𝑗
So
(𝐸1 , 𝐸2 , 𝐸3 ) = (cos 𝜃 𝑈1 + sin 𝜃 𝑈2 , − sin 𝜃 𝑈1 + cos 𝜃 𝑈2 , 𝑈3 )
forms a frame field on 𝑹𝟑 called cylindrical frame field.
88 of 161
Now the question is are there points where the cylindrical frame is not defined?
Yes, the points on 𝑧 −axis does not have a cylindrical frame defined on them. Cylindrical
coordinates are useful in many practical situations, like, designing of a robotic arm etc.
Spherical Coordinates:
Why spherical coordinates?
Just like cylindrical coordinates, there are situations in real like that
Could be dealt in a good way if we use the spherical coordiantes.
From Cylindrical Frame to Spherical Frame
𝐹2 = 𝐸2
89 of 161
𝐹1 = cos 𝜙 𝐸1 + sin 𝜙 𝐸3
𝜋 𝜋
𝐹3 = cos (𝜙 + 2 ) 𝐸1 + sin (𝜙 + 2 ) 𝐸3
= − sin 𝜙 𝐸1 + cos 𝜙 𝐸3
So we have
𝐹2 = 𝐸2
𝐹1 = cos 𝜙 𝐸1 + sin 𝜙 𝐸3
𝐹3 = − sin 𝜙 𝐸1 + cos 𝜙 𝐸3
Lemma: Let 𝐸1, 𝐸2, 𝐸3 be a frame field on 𝑹𝟑.

1) Let 𝑉 be a vector field, then 𝑉 = ∑ 𝑓𝑖 𝐸𝑖 . Where the functions 𝑓𝑖 are called the coordinate
functions of 𝑉 with respect to 𝐸1 , 𝐸2 , 𝐸3 .
2) If 𝑉 = ∑ 𝑓𝑖 𝐸𝑖 and 𝑊 = ∑ 𝑔𝑖 𝐸𝑖 then
𝑉. 𝑊 = ∑ 𝑓𝑖 𝑔𝑖
And
1
1
2 2
‖𝑉‖ = (𝑉. 𝑉)2 = (∑ 𝑓𝑖 )
See book for proof.
So far we have seen many frame fields.
But for a given problem we would like to choose a frame field that is especially easy to work with.
The calculations could be very easy with the right selection of frame field.
90 of 161
End of the lecture
Lecture # 17
91 of 161
Contents:
 Connection Form
 Connection Equations
 How To Calculate Connection Forms
Connection Form
Covariant derivative of a frame field
Let {𝐸1 , 𝐸2 , 𝐸3 } be a frame field on 𝑹𝟑 . Let 𝑣 be an arbitrary tangent vector at point 𝑝 of 𝑹𝟑 . Then
we can write
∇v E1 = c11 E1 + c12 E2 + c13 E3
∇v E2 = c21 E1 + c22 E2 + c23 E3
∇v E3 = c31 E1 + c32 E2 + c33 E3
We can use orthonormal expansion:
Theorem: Let 𝑒1, 𝑒2, 𝑒3 be a frame at a point 𝑝 of 𝑹𝟑 . If 𝑣 is any tangent vector to 𝑹𝟑 at 𝑝, then
𝑣 = (𝑣. 𝑒1 )𝑒1 + (𝑣. 𝑒2 )𝑒2 + (𝑣. 𝑒3 )𝑒3 .
To express each coefficient 𝑐𝑖𝑗 as
𝑐𝑖𝑗 = ∇𝑣 𝐸𝑖 . 𝐸𝑗 (𝑝) for 1 ≤ 𝑖, 𝑗 ≤ 3
Note that the coefficients 𝑐𝑖𝑗 depend only the vector 𝑣, so we can write them
𝑐𝑖𝑗 (𝑣) = 𝜔𝑖𝑗 (𝑣) = ∇𝑣 𝐸𝑖 . 𝐸𝑗 (𝑝) for 1 ≤ 𝑖, 𝑗 ≤ 3
The function 𝜔𝑖𝑗 (𝑣) given by
𝜔𝑖𝑗 (𝑣) = ∇𝑣 𝐸𝑖 . 𝐸𝑗 (𝑝) for 1 ≤ 𝑖, 𝑗 ≤ 3
92 of 161
is a function of tangents vectors on 𝑹𝟑 , and the output is a real number.
Have we met any function that satisfy this property?
Yes, we have.
Definition:
A 1 −form 𝜙 on 𝑹𝟑 is a real-valued function on the set of all tangent vectors to 𝑹𝟑 such that 𝜙 is
linear at each point, that is,
𝜙(𝑎𝒗 + 𝑏𝒘) = 𝑎 𝜙(𝒗) + 𝑏𝜙(𝒘)
For any numbers 𝑎, 𝑏 and tangent vectors 𝒗 , 𝒘 at the same point of 𝑹𝟑 .
Lemma: Let 𝐸1, 𝐸2, 𝐸3 be a frame field on 𝑹𝟑. For each tangent vector 𝑣 to 𝑹𝟑 at point 𝒑, let
𝜔𝑖𝑗 (𝑣) = ∇𝑣 𝐸𝑖 . 𝐸𝑗 (𝑝) for 1 ≤ 𝑖, 𝑗 ≤ 3
Then each 𝜔𝑖𝑗 is a one form, and 𝜔𝑖𝑗 = −𝜔𝑗𝑖 .
See book for proof.
Theorem: Let 𝜔𝑖𝑗 be the connection forms of the frame field 𝐸1, 𝐸2 , 𝐸3 on 𝑹𝟑.
Then for any vector field 𝑉 on 𝑹𝟑 ,
∇𝑉 𝐸𝑖 = ∑ 𝜔𝑖𝑗 (𝑉)𝐸𝑗
We call the equations

𝛻𝑉 𝐸𝑖 = ∑ 𝜔𝑖𝑗 (𝑉)𝐸𝑗
Connection equations.
The matrix of connection forms

We can write the connection forms 𝜔𝑖𝑗 in matrix form
93 of 161
𝜔11 𝜔12 𝜔13

𝜔 = (𝜔21 𝜔22 𝜔13 )
𝜔31 𝜔32 𝜔33
But since 𝜔𝑖𝑗 = −𝜔𝑗𝑖 , so for 𝑖 = 𝑗 we have 𝜔𝑖𝑖 = −𝜔𝑖𝑖 . Hence we have
𝜔𝑖𝑖 = 0 for 1 ≤ 𝑖 ≤ 3.
Hence the matrix of connection forms becomes
0 𝜔12 𝜔13
𝜔 = (𝜔21 0 𝜔13 )
𝜔31 𝜔32 0
Hence we can write connection equations as
𝛻V E1 = + 𝜔12 (V)E2 + 𝜔13 (𝑉)E3

𝛻V E2 = 𝜔21 (𝑉 )E1 + 𝜔23 (𝑉)E3
𝛻𝑉 E3 = 𝜔31 (𝑉 )E1 + 𝜔32 (𝑉 ) E2
How To Calculate Connection Forms

Let {𝐸1 , 𝐸2 , 𝐸3 } be an arbitrary frame field and let {𝑈1 , 𝑈2 , 𝑈3 } be natural frame field. We can write
𝐸1 = 𝑎11 𝑈1 + 𝑎12 𝑈2 + 𝑎13 𝑈3
𝐸2 = 𝑎21 𝑈1 + 𝑎22 𝑈2 + 𝑎23 𝑈3
𝐸3 = 𝑎31 𝑈1 + 𝑎32 𝑈2 + 𝑎33 𝑈3
here the coefficients can be calculated using orthonormal expansion
𝑎𝑖𝑗 = 𝐸𝑖 . 𝑈𝑗
Here 𝑎𝑖𝑗 are real valued functions on 𝑹𝟑 .
From here we can define the attitude matrix of the frame.

𝑎11 𝑎12 𝑎13
𝑎
𝐴 = ( 21 𝑎22 𝑎23 )
𝑎31 𝑎32 𝑎33
Now we define differential of 𝐴 = (𝑎𝑖𝑗 ) to be 𝑑𝐴 = (𝑑𝑎𝑖𝑗 ). So 𝑑𝐴 is a matrix whose entries are 1-

forms.
We can calculate the connection forms in terms the attitude matrix as follows:
94 of 161
Theorem: If 𝐴 = (𝑎𝑖𝑗 ) is an attitude matrix of a frame field {𝐸1, 𝐸2, 𝐸3}. Let
𝜔𝑖𝑗 be the connection forms of this frame field, then
𝜔𝑖𝑗 = 𝑑𝐴 𝑡𝐴
Or equivalently
𝜔𝑖𝑗 = ∑ 𝑑𝑎𝑖𝑘 𝑎𝑘𝑗
See book for proof.
Connection forms for the cylindrical frame

We know that
So the attitude matrix of the cylindrical frame is
cos 𝜃 sin 𝜃 0
𝐴 = (− sin 𝜃 cos 𝜃 0)
0 0 1
Hence
0 𝑑𝜃 0
𝜔 = (−𝑑 𝜃 0 0)
0 0 0
Now the connection forms for cylindrical frame are
0 𝑑𝜃 0
𝜔 = (−𝑑 𝜃 0 0)
0 0 0
Hence the connection equations become
∇𝑉 𝐸1 = 𝑑𝜃 𝑈2
∇𝑉 𝐸2 = −𝑑 𝜃 𝑈1
∇𝑉 𝐸3 = 0
Since 𝜃 is a real valued function on 𝑹𝟑 . So
𝑑𝜃(𝑉) = 𝑉[𝜃]
Hence the equations
∇𝑉 𝐸1 = 𝑑𝜃 𝑈2
∇𝑉 𝐸2 = −𝑑 𝜃 𝑈1
95 of 161
∇𝑉 𝐸3 = 0
Becomes
∇𝑉 𝐸1 = 𝑉[𝜃] 𝑈2
∇𝑉 𝐸2 = −𝑉[𝜃] 𝑈1
∇𝑉 𝐸3 = 0
End of the lecture
Lecture # 18
96 of 161
Contents:
 Dual Forms
 Cartan Structural Equations
 Structural Equations For Spherical Frame
Dual Forms
The dual forms of a frame field
Any frame field can be described in terms of 1-forms.
Definition: If 𝐸1, 𝐸2, 𝐸3 is a frame field on 𝑅3, then the dual 1-forms 𝜃1, 𝜃2, 𝜃3 of the frame
field are the 1-forms such that
𝜃𝑖 (𝒗) = 𝑣. 𝐸𝑖 (𝒑)
For each tangent vector 𝒗 to 𝑹𝟑 at point 𝒑.
Here 𝜃𝑖 are 1-forms since
1)- 𝜃𝑖 takes vectors to real numbers
2)- 𝜃𝑖 are linear
Another property of 𝜽𝒊
𝜃𝑖 (𝐸𝑗 ) = 𝛿𝑖𝑗 for 1 ≤ 𝑖, 𝑗 ≤ 3
here 𝛿𝑖𝑗 is Kronecker delta function.
Orthonormal expansion formula for vector field

Let 𝑉 be a vector field then we can write in terms of frame field {𝐸1 , 𝐸2 , 𝐸3 }
97 of 161
𝑉 = ∑ 𝑓𝑖 𝐸𝑖
In terms of dual 1-forms the orthonormal expansion can be written as
𝑉 = ∑ 𝜃𝑖 (𝑉)𝐸𝑖
Expressing 1-forms in terms of dual forms

In case of natural frame field 𝑈1 , 𝑈2 , 𝑈3 the dual 1-forms are 𝑑𝑥1 , 𝑑𝑥2 , 𝑑𝑥3 . We also know that any
1-form 𝜙 can be written as
𝜙 = ∑ 𝜙(𝑈𝑖 ) 𝑑𝑥𝑖
Question: Is it true in general?
Answer: Yes, by the following result.
Theorem: Let 𝜃1 , 𝜃2 , 𝜃3 be the dual 1-forms of a frame field 𝐸1, 𝐸2, 𝐸3 . Then any 1-form 𝜙 on
𝑅 3 has a unique expression
𝜙 = ∑ 𝜙(𝐸𝑖 )𝜃𝑖
Formula for 𝜽𝒊
We know that any frame field {𝐸1 , 𝐸2 , 𝐸3 } can be written in terms of natural frame field,
𝐸𝑖 = ∑ 𝑎𝑖𝑗 𝑈𝑗 for 1 ≤ 𝑖, 𝑗 ≤ 3
We have the following result for dual 1-forms 𝜃1 , 𝜃2 , 𝜃3 of 𝐸1 , 𝐸2 , 𝐸3 as:
Fact:
𝜃𝑖 = ∑ 𝑎𝑖𝑗 𝑑𝑥𝑗 𝑓𝑜𝑟 1 ≤ 𝑖, 𝑗 ≤ 3
Cartan Structural Equations

Exterior derivatives
Exterior derivatives of 1-forms and connection forms were given by Cartan.
98 of 161
Theorem (Cartan Structural Equations)
Let 𝐸1 , 𝐸2 , 𝐸3 be a frame field on 𝑹𝟑 with dual forms 𝜃1 , 𝜃2 , 𝜃3 and connection forms
𝜔𝑖𝑗 (1 ≤ 𝑖, 𝑗 ≤ 3).
The exterior derivatives of these forms satisfy:
(1) The first structural equations:
𝑑𝜃𝑖 = ∑ 𝜔𝑖𝑗 ∧ 𝜃𝑗 .
(2) The second structural equation:
𝑑 𝜔𝑖𝑗 = ∑ 𝜔𝑖𝑘 ∧ 𝜔𝑘𝑗 .
Proof is not included.
Matrix notation
We define some notations and remarks to make these structural equations more easy to
understand. We define
𝜔11 𝜔12 𝜔13
𝜔 = (𝜔21 𝜔22 𝜔23 )
𝜔31 𝜔32 𝜔33
and we also know the attitude matrix

𝑎11 𝑎12 𝑎13
𝑎
𝐴 = ( 21 𝑎22 𝑎23 )
𝑎31 𝑎32 𝑎33
And we can write the dual forms 𝜃1 , 𝜃2 , 𝜃3 in matrix form as
𝜃1
𝜃 = (𝜃2 )
𝜃3
And the dual of natural frame fields as:
𝑑 𝑥1
𝑑𝜉 = (𝑑 𝑥2 )
𝑑 𝑥3
Using these notations we can write the following fact
Fact:
99 of 161
𝜃𝑖 = ∑ 𝑎𝑖𝑗 𝑑𝑥𝑗 𝑓𝑜𝑟 1 ≤ 𝑖, 𝑗 ≤ 3
In the following form:
𝜃1 𝑎11 𝑎12 𝑎13 𝑑 𝑥1

𝜃 𝑎
( 2 ) = ( 21 𝑎22 𝑎23 ) (𝑑 𝑥2 )
𝜃3 𝑎31 𝑎32 𝑎33 𝑑 𝑥3
In short form it becomes:
𝜃 =𝐴𝑑𝜉
So we can write the Cartan structural equations as
First structural equation
𝑑𝜃𝑖 = ∑ 𝜔𝑖𝑗 ∧ 𝜃𝑗
can be written as
𝑑 𝜃1 𝜔11 𝜔12 𝜔13 𝜃1

𝜔
(𝑑 𝜃2 ) = 𝑑 𝜃 = ∑ 𝜔𝑖𝑗 ∧ 𝜃𝑖 = ( 21 𝜔22 𝜔23 ) (𝜃2 )
𝑑 𝜃3 𝜔31 𝜔32 𝜔33 𝜃3
Similarly
The second structural equation
can be written as
𝑑𝜔 =𝜔𝜔
Example: The Spherical Frame Field

The spherical Frame field is given by
𝐹1 = cos 𝜃 cos 𝜙 𝑈1 + sin 𝜃 sin 𝜙 𝑈2 + sin 𝜙 𝑈3
𝐹2 = − sin 𝜃 𝑈1 + cos 𝜃 𝑈2
𝐹3 = − cos 𝜃 sin 𝜙 𝑈1 − sin 𝜃 sin 𝜙 𝑈2 + cos 𝜙 𝑈3
100 of 161
So the attitude matrix is
cos 𝜃 cos 𝜙 sin 𝜃 sin 𝜙 sin 𝜙

𝐴=( − sin 𝜃 cos 𝜃 0 )
− cos 𝜃 sin 𝜙 − sin 𝜃 sin 𝜙 cos 𝜙
Now if 𝜃1 , 𝜃2 , 𝜃3 are dual 1-forms of the spherical frame field then:
𝜃 = 𝐴 𝑑𝜉
Calculation of 𝜽
Now to calculate 𝑑𝑥1 , 𝑑𝑥2 , 𝑑𝑥3 we consider
x1 = 𝜌 cos 𝜙 cos 𝜃
𝑥2 = 𝜌 cos 𝜙 sin 𝜃
𝑥3 = sin 𝜙
We calculate 𝑑𝑥1 𝑑𝑥2 , 𝑑𝑥3 use the following to get 𝜃.
𝜕𝑓 𝜕𝑓 𝜕𝑓
𝑑𝑓 = 𝑑𝑥1 + 𝑑𝑥2 + 𝑑𝑥3
𝜕𝑥1 𝜕𝑥2 𝜕𝑥3
Calculation of 𝝎
Now we will
cos 𝜃 cos 𝜙 sin 𝜃 sin 𝜙 sin 𝜙

𝐴=( − sin 𝜃 cos 𝜃 0 )
− cos 𝜃 sin 𝜙 − sin 𝜃 sin 𝜙 cos 𝜙
Now we after calculating 𝑑𝐴, the matrix of 𝜔 is given by
0 cos 𝜙 𝑑𝜃 𝑑𝜙
𝜔=( 0 sin 𝜙 𝑑𝜃 )
0
First Structural Equation for Spherical Frame
So we have
(1) The first structural equations:
𝑑𝜃𝑖 = ∑ 𝜔𝑖𝑗 ∧ 𝜃𝑗 .
(2) The second structural equation:
101 of 161
Where
𝑑𝑟 0 cos 𝜙 𝑑𝜃 𝑑𝜙
𝜃 = (𝑟 cos 𝜙) 𝑎𝑛𝑑 𝜔=( 0 sin 𝜙 𝑑𝜃)
𝑟 𝑑𝜙 0
Recipe for dealing with geometrical problems in 𝑹𝟑
Procedure
1) Select a frame field
2) Find its dual 1-forms
3) Calculate connection forms
4) Apply structural equations
5) Interpret the results
End of the lecture
Lecture # 19
102 of 161
Contents:
 Surfaces in 𝑅 3
 The Sphere
 Construction Of A Surface
Surfaces in 𝑹𝟑
Coordinate patches
A coordinate patch
𝑥 : 𝐷 → 𝑅3
is a
 Mapping
 one-to-one
 regular
of an open set 𝐷 of 𝑅 2 into 𝑅 3 .
Proper Patches
A coordinate patch 𝒙: 𝐷 → 𝑹𝟑 is called proper if its inverse function
𝒙−1 : 𝒙(𝐷) → 𝑹𝟐
is continuous.
103 of 161
What is a surface roughly?
Every small region of surface looks like a region in 𝑅 2 , or image of patch.
Definition: Let 𝒙: 𝐷 → 𝑹𝟑 be a patch with 𝒙(𝐷) ⊂ 𝑀 is a patch in 𝑀.

Definition:
A neighborhood in 𝑀 ⊂ 𝑹𝟑 of a point 𝑝 is the subset of all
points in 𝑀 at distance less that 𝜖, for some real number
𝜖 > 0.
Surface
A surface in 𝑹𝟑 is a subset 𝑀 of 𝑹𝟑 such that for each point 𝒑 of 𝑀 there exist a proper patch in 𝑀
whose image contains a neighborhood of 𝒑 in 𝑀.
Sphere:
A sphere Σ is the set of all point in 𝑹𝟑 whose distance from origin is 1, that is all points 𝒑 such that
𝟏
‖𝒑‖ = (𝒑𝟐𝟏 + 𝒑𝟐𝟐 + 𝒑𝟐𝟑 )𝟐 = 𝟏
Patch containing north pole
Consider the point (0,0,1) on sphere, known as North pole.
Let
𝐷 = {(𝑢, 𝑣): 𝑢2 + 𝑣 2 < 1}
Define
104 of 161
𝑥: 𝐷 → Σ such that x(u, v) = (u, v, √1 − 𝑢2 − 𝑣 2 )
Lets see if 𝑥 is a patch or not. For this check the following:
 Mapping: since the coordinate functions are differentiable
 One to one
To check if the patch 𝑥 is regular we need to calculate the Jacobian of this map:
Similarly we can define the patch for the south pole, that is, the point (0,0, −1).
From the sphere example we can guess the following
The map
𝑥: 𝐷 → 𝑅 3
given by
𝑥(𝑢, 𝑣) = (𝑢, 𝑣, 𝑓(𝑢, 𝑣))
for any differentiable function 𝑓(𝑢, 𝑣) is a patch. These kind of patches are called the Monge
patches.
Simple Surface
Let 𝑥: 𝐷 → 𝑹𝟑 be patch. Then the image 𝑥(𝐷) satisfies definition of

a surface. This type of surface is called simple surface.
Example:
Let 𝑓(𝑢, 𝑣) be a real valued differentiable function from 𝑅 2 to 𝑅 3 .

The graph of z = 𝑓(𝑢, 𝑣) is the set of all points in 𝑅 3 satisfying
𝑧 − 𝑓(𝑢, 𝑣) = 0.
In fact 𝑀 is the image of the Monge patch.
105 of 161
So 𝑀 is a simple surface.
End of the lecture
Lecture 20
Contents:
106 of 161
 Implicitly Defined Surfaces
 Surfaces of Revolution
 Properties Of Patches
Implicitly Defined Surfaces

Let 𝑔 be a real valued function from 𝑅 3 to 𝑅. For any real number 𝑐, the equation
𝑔(𝑥, 𝑦, 𝑧) = 𝑐
defines the set of all points 𝑝 of 𝑅 3 such that
𝑔(𝑝) = 𝑐
Example:
Let g(𝑥, 𝑦, 𝑧) = 𝑥 2 + 𝑦 2 + 𝑧 2 and let 𝑐 = 1 then the set of all points 𝒑 such that
𝑔(𝑝) = 1
forms a sphere of radius 1.
Example:
Let 𝑔(𝑥, 𝑦, 𝑧) = 𝑧 and 𝑐 = 1 then the set of all points 𝒑 in 𝑹𝟑 such that
𝑔(𝑝) = 1
is
𝑧=1
forms a plane parallel to 𝑥𝑦 −plane.
An Implicit Construction
Theorem: Let 𝑔: 𝑹𝟑 → 𝑹 be a differentiable real valued

function on 𝑹𝟑 , let 𝑐 be a number. The subset
𝑀: 𝑔(𝑥, 𝑦, 𝑧) = 𝑐 of 𝑹𝟑 is a surface if the differential 𝑑𝑔 is
not zero at any point of 𝑀.
See book for proof.
107 of 161
Monge Patches:
The map
𝑥: 𝐷 → 𝑅 3
given by
𝑥(𝑢, 𝑣) = (𝑢, 𝑣, 𝑓(𝑢, 𝑣))
for any differentiable function 𝑓(𝑢, 𝑣) is a patch. These kind of patches are called the Monge
patches.
Sphere is a surface
A sphere Σ with center 𝒄 = (𝒄𝟏 , 𝒄𝟐 , 𝒄𝟑 ) and radius r is the set of all point in 𝑹𝟑 whose distance
from 𝒄 is 𝑟, that is all points 𝒑 = (𝒙, 𝒚, 𝒛) such that
𝟏
‖𝒑 − 𝒄‖ = ((𝒙 − 𝒄𝟏 )𝟐 + (𝒚 − 𝒄𝟐 )𝟐 + (𝒛 − 𝒄𝟑 )𝟐 )𝟐 = 𝒓𝟐
Let 𝑔(𝑥, 𝑦, 𝑧) = (𝒙 − 𝒄𝟏 )𝟐 + (𝒚 − 𝒄𝟐 )𝟐 + (𝒛 − 𝒄𝟑 )𝟐
Then 𝑔(𝑥, 𝑦, 𝑧) = 𝑟 2 defines sphere with center 𝒄 and radius 𝑟.
We apply the previous theorem to shwo if this sphere is a surface or not.
Surfaces of Revolution
If 𝐶 is defined by 𝑓(𝑥, 𝑦) = 𝑐, where 𝑓: 𝑅 2 → 𝑅 is a differentiable function. Also 𝐶 is regular curve.
Then the surface of revolution is defined by
𝑔(𝑥, 𝑦, 𝑧) = 𝑓 (𝑥, √𝑦 2 + 𝑧 2 ) = 𝑐
More surfaces of revolution
108 of 161
We can generate a torus using the idea of surface of revolution.
Lets see why we need properness in defining surface patches.
Example: Let
𝐷 = {(𝑢, 𝑣): −𝜋 < 𝑢 < −𝜋, < 0𝑣 < 1}
and
𝑥: 𝐷 → 𝑅 3 𝑑𝑒𝑓𝑖𝑛𝑒𝑑 𝑏𝑦 𝑥(𝑢, 𝑣) = (sin 𝑢, sin 2𝑢 , 𝑣)
Here 𝑥 is a patch.
But the image is not a surface.
End of the lecture
Lecture 21
109 of 161
Contents:
 Parameter Curves on Surfaces
 Parametrizations
 Torus
 Ruled Surface
Parameter Curves on Surfaces

Parameter Curves
Let 𝑥: 𝐷 → 𝑅 3 be a coordinate patch. For each point (𝑢0 , 𝑣0 ) in 𝐷 the curve
𝑢 → 𝑥(𝑢, 𝑣0 )
is called the 𝑢 −parameter curve, 𝑣 = 𝑣0 , of 𝑥.
Similarly the curve
𝑣 → 𝒙(𝒖𝟎 , 𝒗)
Is called 𝑣 −parameter curve, 𝑢 = 𝑢0 , of 𝒙.
 The image 𝑥(𝐷) of 𝐷 is the union of all such curves.
 Each point of 𝑥(𝐷) lies on exactly one curve of each type.
Partial Velocities
If 𝑥: 𝐷 → 𝑅 3 is a patch, for each point (𝑢0 , 𝑣0 ) in 𝐷.
 The velocity vector at 𝑢0 of the 𝑢 −parameter curve, 𝑣 = 𝑣0 , is denoted by 𝑥𝑢 (𝑢𝑜 , 𝑣0 ).
110 of 161
 The velocity vector at 𝑣0 of the v −parameter curve, u = 𝑢0 , is denoted by 𝒙𝒗 (𝑢𝑜 , 𝑣0 ).
The vectors 𝒙𝒖 (𝑢0 , 𝑣0 ) and 𝒙𝒗 (𝑢0 , 𝑣0 ) are called the partial velocities of 𝒙 at (𝑢0 , 𝑣0 ).
Calculation of partial velocities
Given a coordinate patch 𝑥: 𝐷 → 𝑅 3
𝑥(𝑢, 𝑣) = (𝑥1 (𝑢, 𝑣), 𝑥2 (𝑢, 𝑣), 𝑥3 (𝑢, 𝑣))
Then the partial velocities functions are
𝜕𝑥1 𝜕𝑥2 𝜕𝑥3

𝑥𝑢 = ( , , )
𝜕𝑢 𝜕𝑢 𝜕𝑢 (𝑢0 ,𝑣0)
𝜕𝑥1 𝜕𝑥2 𝜕𝑥3

𝑥𝑣 = ( , , )
𝜕𝑣 𝜕𝑣 𝜕𝑣 (𝑢0 ,𝑣0 )
Where (𝑢0 , 𝑣0 ) is point of application and can be omitted.
Geographical patch in a sphere
Let Σ be a sphere of radius 𝑟 and center (0,0,0) in 𝑅 3 . Let 𝑥 be a mapping defined on

𝜋 𝜋
𝐷 = {(𝑢, 𝑣): −𝜋 < 𝑢 < 𝜋, − <𝑣< }
2 2
Defined by
𝑥(𝑢, 𝑣) = (𝑟 cos 𝑣 cos 𝑢, 𝑟 cos 𝑣 sin 𝑢 , 𝑟 sin 𝑣)
111 of 161
Parametric Curves on the sphere
 The 𝑢 −parameter curve, 𝑣 = 𝑣0 , is a circle with constant

value of 𝑧. It is the parallel at latitude 𝑣0 .
 The 𝑣 −parameter curve, 𝑢 = 𝑢0 , is a half circle. It is the

meridian of longitude 𝑢0 .
Parametrizations
Definition:
A regular mapping 𝑥: 𝐷 → 𝑀 whose image lies in a surface 𝑀 is called a parametrization of the
region 𝑥(𝐷) in 𝑀.
Properties of parametrization
 A 1-1 parametrization is a patch
 In parametrization of a curve mean a surjective map onto the whole curve. But in case of
surface parametrization it might be only the region 𝒙(𝐷) of 𝑀.
When a mapping is a parametrization?
Given a mapping 𝒙: 𝐷 → 𝑹𝟑 . To check whether it is a parametrization of a surface 𝑀 ⊂ 𝑹𝟑 we

check the following:
 𝒙(𝑫) ⊂ 𝑴 ?
First determine that 𝑥(𝐷) ⊂ 𝑀 or not. For example if surface is defined by 𝑀: 𝑔 = 𝑐. Then
112 of 161
"𝑥(𝐷) is contained in 𝑀 if and only if 𝑔(𝑥(𝐷)) = 𝑐.”
 𝒙 is regular?
The map 𝑥: 𝐷 → 𝑀 is regular if
𝑈1 𝑈2 𝑈3
𝜕𝑥1 𝜕𝑥2 𝜕𝑥3
| |
𝒙𝒖 × 𝒙𝒗 = 𝜕𝑢 𝜕𝑢 𝜕𝑢 |
|
𝜕𝑥1 𝜕𝑥2 𝜕𝑥3
𝜕𝑣 𝜕𝑣 𝜕𝑣
is non zero.
Parametrization for sphere
Given mapping
Question: Is it parametrization of sphere Σ of radius 𝑟 and center (0,0,0)?
Lets check the two conditions to see.
 𝒙(𝑫) ⊂ 𝚺 ?
The equation of sphere is
𝑔(𝑥, 𝑦, 𝑧) = 𝑥 2 + 𝑦 2 + 𝑧 2 = 𝑟 2
Then 𝑥(𝐷) ⊂ 𝑀 if and only if (𝑔(𝑥)) = 𝑟 2 .
𝜋 𝜋
Here 𝐷 = {(𝑢, 𝑣): −𝜋 < 𝑢 < 𝜋, − 2 < 𝑣 < 2 }
To check if 𝒙 is a patch we need to only check if 𝒙 is 1-1 or not.
Surface of revolution
Let the curve 𝐶 has parametrization
𝛼(𝑢) = (𝑔(𝑢), ℎ(𝑢), 0)
here ℎ > 0.
As the point on the curve is rotated then it reaches the point
𝑥(𝑢, 𝑣) = (𝑔(𝑢), ℎ(𝑢) cos 𝑣 , ℎ(𝑢) sin 𝑣)
113 of 161
Question: Is it a parametrization of the surface of revolution?
Answer: check the following two.
 𝒙(𝑫) ⊂ 𝑴?
Torus
Definition:
A torus of revolution 𝑇 is the surface of revolution of a circle 𝐶.
Example:
Suppose that 𝐶 is the circle in the 𝑥𝑧 −plane with radius 𝑟 > 0 and center (𝑅, 0,0).
We rotate the circle around 𝑧 −axis here 𝑅 > 𝑟.
Here parametrization of 𝐶 is
𝛼(𝑢) = (𝑅 + 𝑟 cos 𝑢 , 𝑟 sin 𝑢)
This yields the parametrization
𝑥(𝑢, 𝑣) = ((𝑅 + 𝑟 cos 𝑢 ) cos 𝑣 , (𝑅 + 𝑟 cos 𝑢 ) sin 𝑣 , 𝑟 sin 𝑢)
Definition:
A ruled surface is a surface
swept out by a straight line 𝐿 moving along a curve 𝛽.
The various positions of the ruling line 𝐿 are called the rulings of the surface.
Such a surface always has a ruled parametrization
𝑥(𝑢, 𝑣) = 𝛽(𝑢) + 𝑣𝛿(𝑢)
here 𝛽 is base curve and 𝛿 is the director curve.
114 of 161
We put conditions on 𝛽 and 𝛿 to make 𝑥 a parametrization.
End of the lecture
Lecture # 22
115 of 161
Contents:
 Coordinate Expressions
 Curves on a Surface
 Differentiable Functions
Definition:
A regular mapping 𝑥: 𝐷 → 𝑀 whose image lies in a surface 𝑀 is called a parametrization of the
region 𝑥(𝐷) in 𝑀.
In parametrization of a curve mean a surjective map onto the whole curve. But in case of surface
parametrization it might be only the region 𝒙(𝐷) of 𝑀.
Coordinate Expressions
Suppose that 𝑓 is a real valued function on a surface 𝑀.
If
𝑥: 𝐷 → 𝑀
is a coordinate patch of 𝑀 then the composite function 𝑓(𝑥) is called the coordinate expression
for 𝑓.
The coordinate expression 𝑓 is just a real-valued function
𝑓(𝑥): 𝐷 → 𝑅
The function 𝑓 is differentiable provided all its coordinate expressions
are differentiable.
116 of 161
For a function
𝐹: 𝑅 𝑛 → 𝑀
each patch of 𝑀
𝑥: 𝐷 → 𝑀
gives a coordinate expression of 𝐹, given by
𝑥 −1 (𝐹): 𝑂 → 𝐷
here 𝑂 is an open set containing points from the set
{𝑝 ∈ 𝑅 𝑛 : 𝐹(𝑝) ∈ 𝑥(𝐷)}
The function 𝐹 is differentiable if every coordinate expression for 𝐹 is differentiable. Every set 𝑂 is
open set.
Definition:
A curve on a surface 𝑀 is a differentiable function
𝛼: 𝐼 → 𝑀
from open interval 𝐼 to 𝑀.
Coordinate functions of a curve

Lemma: If 𝛼 is a curve 𝛼: 𝐼 → 𝑀 whose route lies in the image 𝑥(𝐷) of a single patch 𝑥, then
there exist unique differentiable functions 𝑎1 , 𝑎2 on 𝐼 such that
𝛼(𝑡) = 𝑥(𝑎1 (𝑡), 𝑎2 (𝑡)) 𝑓𝑜𝑟 𝑎𝑙𝑙 𝑡
Or 𝛼 = 𝑥(𝑎1 , 𝑎2 ).
See book for proof
Map of the region
117 of 161
Let 𝑥: 𝐷 → 𝑀 be a patch in 𝑀.
We take the domain 𝐷 as the map of the region 𝑥(𝐷).
The functions 𝑥 and 𝑥 −1 establish a one-to-one correspondence between objects in D and the
objects in 𝒙(𝐷).
Too many patches
To check the differentiability of functions 𝐹: 𝑅 𝑛 → 𝑀 we need to check differentiability at each

and every patch. The following theorem simplify the situation.
Theorem: Let 𝑀 be a surface in 𝑅3.

If 𝐹: 𝑅 𝑛 → 𝑅 3 is a (differentiable) mapping whose image lies in 𝑀,
then considered as a function 𝐹: 𝑅 𝑛 → 𝑀 into 𝑀, 𝐹 is differentiable.
Overlapping patches
Corollary: Let 𝑥: 𝐷 → 𝑀 and 𝑦: 𝐸 → 𝑀 be patches in 𝑀, where 𝑀 ⊂ 𝑅3. Assume that images of

𝑥 and 𝑦 overlap, that is, 𝑥(𝐷) ⋂ 𝑦(𝐸) ≠ 𝜙. Then each of 𝑥 −1 𝑦 and 𝑦 −1 𝑥 is a differentiable mapping
defined on an open set in 𝑅 2 .
Here the mapping 𝑥 −1 𝑦 is defined on the set 𝑥 −1 (𝑥(𝐷) ⋂ 𝑦(𝐸)).
Similarly the mapping 𝑦 −1 𝑥 is defined on the set 𝑦 −1 (𝑥(𝐷) ⋂ 𝑦(𝐸)).
118 of 161
How to show that 𝑓: 𝑀 → 𝑹 is differentiable
By definition 𝑓: 𝑀 → 𝑹 is differentiable if 𝑓(𝑥) is differentiable for all patches 𝑥: 𝐷 → 𝑀.
Assume that we show that 𝑓(𝑥) is differentiable for a patch 𝑥. Let 𝑦 be another patch. Then 𝑓(𝑥) is
differentiable and 𝑥 −1 𝑦 is also differentiable. Therefore the composition 𝑓𝑥𝑥 −1 𝑦 is differentiable.
The function 𝑓𝑥𝑥 −1 𝑦 is the restriction of 𝑓(𝑦) to only a small region of 𝑀, where 𝑥 and 𝑦 overlap.
But 𝑀 is covered by such regions, therefore 𝑓(𝑦) is differentiable in an open region around every
point of 𝑀
End of the lecture
Lecture #23
Contents:
119 of 161
 Tangents
 Tangent Vector Fields
 Gradient Vector Field
Tangents
Tangents to a surface
Definition: Given an surface 𝑀. Let 𝑝 is a point of 𝑀. Then a tangent vector to 𝑅 3 at 𝑝 is tangent

vector to 𝑀 if 𝑣 is velocity of some curve in 𝑀.
The set of all tangent vectors to 𝑀 at 𝑝 is called the tangent plane of 𝑀 at point 𝑝 and is denoted by
𝑇𝑝 (𝑀).
Fact: At each point the tangent plane 𝑇𝑝 (𝑀) is actually a 2-dimensional vector subspace of the
tangent space 𝑇𝑝 (𝑅3 ).
Lemma:
Let 𝑀 be a surface in 𝑅 3 and 𝑝 be a point on 𝑀. Let 𝑥 be a patch of 𝑀 such that 𝑥(𝑢0 , 𝑣0 ) = 𝑝. A
tangent vector to 𝑅 3 is a tangent vector to 𝑀 at point 𝑝 if and only if 𝑣 can be written as a linear
combination of 𝑥𝑢 (𝑢0 , 𝑣0 ) and 𝑥𝑣 (𝑢0 , 𝑣0 ).
See book for proof.
Tangent plane as approximation to 𝑴 at point 𝒑
We can think 𝑇𝑝 (𝑀) is linear approximation of surface near point 𝑝.
Euclidean and tangent vector fields
120 of 161
Definition:
A Euclidean vector field on 𝑀 is a function 𝑍 that maps every point 𝑝 of 𝑀 to a tangent vector 𝑍(𝑝)
to 𝑹𝟑 at point 𝑝.
Definition: A tangent vector field on 𝑀 is a Euclidean

vector field 𝑉 on 𝑀 that maps every point 𝑝 of 𝑀 to a tangent
vector V(𝑝) at 𝑀.
Most of the time a Euclidean vector field are defined only
on a small region of 𝑀.
Normal vector field

A tangent vector 𝑧 to 𝑅 3 at point 𝑝 is normal to 𝑀 if 𝑧 is orthogonal to the tangent plane 𝑇𝑝 (𝑅3 ).
A Euclidean vector field 𝑍 is normal vector field on 𝑀 if 𝑍(𝑝) is normal to 𝑀 for every point 𝑝 of 𝑀.
It is easy to deal with tangent and normal vector fields for surfaces given in implicit form. To see
this we define:
Gradient Vector Field

Gradient vector field
Let 𝑀 be a surface given in implicit form 𝑀: 𝑔 = 𝑐. The gradient vector field is the Euclidean vector
field ∇𝑔 that maps every point 𝑝 of 𝑀 to
𝜕𝑔
∇𝑔 (𝑝) = ∑ (𝑝) 𝑈𝑖 (𝑝)
𝜕𝑥𝑖
Question: What does the gradient vector field look like?
121 of 161
Answer is given by the following Lemma.
Lemma: Let 𝑀: 𝑔 = 𝑐 be a surface in implicit form. Then the gradient vector field has the following
properties:
 A normal vector field on 𝑀.
 Nonzero at every point of 𝑀.
Directional derivatives
 Let 𝑝 ∈ 𝑀 be a point on surface and 𝑣 ∈ 𝑇𝑝 (𝑀) be a tangent vector. Let 𝑓: 𝑀 → 𝑅 be a real

valued function on 𝑀.
 The directional derivative of 𝑓 with respect to 𝑣, written as 𝑣[𝑓], is defined as

𝑑𝑓(𝛼)
 𝑣[𝑓] = (0)
𝑑𝑡
 for any curve 𝛼 in 𝑀 with 𝛼(0) = 𝑝 and 𝛼 ′ (0) = 𝑣.
 Directional derivative has the same properties as we studied for 𝑅 3 .
End of the lecture
Lecture # 24
Contents:
 Differential Forms
122 of 161
 Exterior Derivatives
 Differential Forms On The Euclidean Plane
 Closed And Exact Forms
Differential Forms
Let 𝑀 be a surface in 𝑅 3
 A 0 −form on 𝑀 is a function 𝑓: 𝑀 → 𝑅
 A 1 −form on 𝑀 is a real-valued function 𝜙 on the tangent vectors on 𝑀, such that 𝜙 is

linear on 𝑇𝑝 (𝑀) for every 𝑝 ∈ 𝑀.
We defined 2-form and 3-form on 𝑅 3 in the following way.
But we did’t define them properly.
Intuitively what is a 2-form.
Definition
A 2 −form 𝜂 on a surface 𝑀 is a real-valued function on all ordered pairs of tangent vectors 𝑣, 𝑤 to
𝑀 such that
1. 𝜂(𝑣, 𝑤) is linear in 𝑣 and in 𝑤
2. 𝜂(𝑣, 𝑤) = −𝜂(𝑤, 𝑣)
Definition: A 𝑑 −form is a form of degree 𝑑.

Fact: Since surface is two dimensional hence all 𝑝 −forms with 𝑝 > 2 are zero.
Properties of forms
 Two 𝑝 −forms, for 𝑝 = 0,1,2, can be added.
 A 1 −form can be evaluated on a vector field 𝑉.
 Similarly a 2 −form can be added evaluated on pair of vector fields 𝑉, 𝑊.
 Second axiom of the definition of the 2 −form implies that
𝜂(𝑣, 𝑣) = 0.
123 of 161
𝟐 −forms and determinants
Let 𝜂 be a 2 −form on surface 𝑀. Let 𝑣, 𝑤 be two linearly independent tangent vectors at some
point of 𝑀. Then
𝑎 𝑏
𝜂(𝑎𝑣 + 𝑏𝑤, 𝑐𝑣 + 𝑑𝑤) = | |.
𝑐 𝑑
See book for proof.
Wedge product
 Product of a 𝑝 −form and a 𝑞 −form is a (𝑝 + 𝑞) −form.
 The product of a 0 −form and a 1 −form and 2 −form is usual multiplication.
 On surface the wedge product is zero if 𝑝 + 𝑞 > 2. So we need only the following definition:
Definition: If 𝜙 and 𝜓 are 1 −forms on a surface 𝑀, the wedge product 𝜙 ∧ 𝜓 is a 2 −form on

𝑀 such that
(𝜙 ∧ 𝜓)(𝑣, 𝑤) = 𝜙(𝑣)𝜓(𝑤) − 𝜙(𝑤)𝜓(𝑣)
for all pairs 𝑣, 𝑤 of tangent vectors to 𝑀.
Question : Why 𝜙 ∧ 𝜓 is a 2 −form?
Rule for interchanging the factors in a wedge product
If 𝜉 is a 𝑑1 −form and 𝜂 is a 𝑑2 −form, then
𝜉 ∧ 𝜂 = (−1)(𝑝+𝑞) 𝜂 ∧ 𝜉
In case of surfaces sign changes only occur in case of multiplication of two 1 −forms.
In particular for 1 −form 𝜙 we have
𝜙 ∧ 𝜙 = 0.
Exterior Derivatives
The differential calculus of forms is based on the exterior derivatives 𝑑.
 For 0 −form, that is a real-valued function 𝑓 , the exterior 𝑑𝑓 is defined as
𝑑𝑓 = 𝑣[𝑓].
 The exterior derivative of a 𝑝 −form is a (𝑝 + 1) −form.
 For surface we need the following:
124 of 161
Definition: Let 𝜙 be a 1 −form on a surface 𝑀. Then the exterior derivative 𝑑 𝜙 of 𝜙 is a

2 −form such that for each patch 𝑥: 𝐷 → 𝑀,
𝜕 𝜕
𝑑 𝜙(𝑥𝑢 , 𝑥𝑣 ) = (𝜙(𝑥𝑣 )) − (𝜙(𝑥𝑢 )).
𝜕𝑢 𝜕𝑣
Definition: Let 𝜙 be a 1 −form on a surface 𝑀. Then the exterior derivative 𝑑 𝜙 of 𝜙 is a

2 −form such that for each patch 𝑥: 𝐷 → 𝑀,
𝜕 𝜕
𝑑 𝜙(𝑥𝑢 , 𝑥𝑣 ) = (𝜙(𝑥𝑣 )) − (𝜙(𝑥𝑢 )).
𝜕𝑢 𝜕𝑣
Lemma: Let 𝜙 be a 1 −form on 𝑀. If 𝑥 and 𝑦 are patches in 𝑀. If 𝑥 and 𝑦 are patches in 𝑀, then
dx 𝜙 = 𝑑𝑦 𝜙 on the overlap of 𝑥(𝐷) and 𝑦(𝐸).
Differentiation Rule
Let 𝑀 be a surface in 𝑅 3 . Let 𝑓: 𝑀 → 𝑅 and 𝑔: 𝑅 → 𝑅 be a function.
Then
𝑑(𝑔(𝑓)) = 𝑔′ (𝑓) 𝑑𝑓
Differentiation of wedge product
Let 𝑓, 𝑔, ℎ be three functions on 𝑀 and 𝜙 is a 1 −form. Then
 𝑑(𝑓𝑔ℎ) = 𝑔ℎ 𝑑𝑓 + 𝑓ℎ 𝑑𝑔 + 𝑓𝑔 𝑑ℎ
 𝑑(𝑓 𝜙) = 𝑓 𝑑𝜙 − 𝜙 ∧ 𝑑𝑓
 (𝑑𝑓 ∧ 𝑑𝑔)(𝑣, 𝑤) = 𝑣[𝑓] 𝑤[𝑔] − 𝑣[𝑔] 𝑤[𝑓]
Theoerem: If 𝑓 is a real valued function on 𝑀, then 𝑑(𝑑 (𝑓)) = 0.

How to show that the forms are equal
Two criteria
125 of 161
Let 𝑥: 𝐷 → 𝑀 be a patch in 𝑀. Then
 Two 1 −forms, 𝜙, 𝜓 are equal on 𝑥(𝐷) if and only if
𝜙(𝑥𝑢 ) = 𝜓(𝑥𝑢 ) 𝑎𝑛𝑑 𝜙(𝑥𝑣 ) = 𝜓(𝑥𝑣 )
 Two 2 −forms 𝜈, 𝜇 are equal on 𝑥(𝐷) if and only if
𝜇(𝑥𝑢 , 𝑥𝑣 ) = 𝜈(𝑥𝑢 , 𝑥𝑣 ).
Closed and exact forms
Let 𝜙 be a differential form.
If the exterior derivative of 𝜙 is zero, 𝑑𝜙 = 0, then 𝜙 is closed.
If 𝜙 is the exterior derivative of some form 𝜉, so that 𝜙 can be written as 𝜙 = 𝑑 𝜉, then 𝜙 is exact.
End of the lecture
Lecture # 25
Contents:
 Mappings of Surfaces
 Tangent Maps of Mappings
126 of 161
 Diffeomorphism
Mappings of Surfaces
Differential functions
Let 𝑀 and 𝑁 be two surfaces in 𝑅 3 . Let 𝐹: 𝑀 → 𝑁 be a function. Then 𝐹 is differentiable if the
following condition is satisfied
For each patch 𝑥: 𝐷 → 𝑀 in 𝑀 and 𝑦: 𝐸 → 𝑁 in 𝑁 the composite function 𝑦 −1 𝐹𝑥 is Euclidean

differentiable (defined on an open subset of 𝑅 2 ).
𝐹 is then called a mapping.
The function 𝑦 −1 𝐹𝑥 is defined at all points (𝑢, 𝑣) of 𝐷 such that 𝐹(𝑢, 𝑣) lies in the image of 𝑦.
FACT: It suffices to check enough patches to cover both 𝑀 and 𝑁.
Cylindrical Projection
Let Σ be a unit sphere in 𝑹𝟑 with center at origin with north and south poles removed, that is,
𝚺 = {𝐩 ∈ 𝑹𝟑 : 𝒅(𝒑, 𝟎) = 𝟏\{(𝟎, 𝟎, 𝟏), (𝟎, 𝟎, −𝟏)} }.
Let 𝐶 be the cylinder bases on the unit circle in the 𝑥𝑦 plane. So 𝐶 is in contact with the sphere
along the equator, that is 𝐶 is given by
𝐶 = {𝑝 ∈ 𝑅 3 : 𝑝12 + 𝑝22 = 1}
Cylindrical projection
Define 𝐹: Σ → 𝐶 as follows.
Start from a point 𝑝 = (𝑝1 , 𝑝2 , 𝑝3 ) ∈ Σ.
127 of 161
Draw the line 𝐿 through (0,0, 𝑝3 ) and = (𝑝1 , 𝑝2 , 𝑝3 ) .
Let 𝐹(𝑝) be the intersection of 𝐿 with 𝐶 as seen in the figure.
Question: Is 𝐹 a mapping?
Consider the following patch 𝑥: 𝐷 → Σ of the sphere which we discussed earlier
where
𝜋 𝜋
𝐷 = {(𝑢, 𝑣): −𝜋 < 𝑢 < 𝜋, − < 𝑣 < }.
2 2
Now consider a patch 𝑦: 𝐸 → 𝐶 of the cylinder given by
𝑦(𝑢, 𝑣) = (cos 𝑢 , sin 𝑢 , 𝑣)
where
𝐸 = {(𝑢, 𝑣): 𝜋 < 𝑢 < −𝜋, − 1 < 𝑣 < 1}.
Now from the definition of 𝐹 we get
𝐹(𝑥(𝑢, 𝑣)) = (cos 𝑢, sin 𝑢 , sin 𝑣).
But
𝑦(𝑢, sin 𝑣) = (cos 𝑢, sin 𝑢 , sin 𝑣).
Hence
𝐹(𝑥(𝑢, 𝑣)) = 𝑦(𝑢, sin 𝑣).
Applying 𝐹 −1 we get
𝑦 −1 𝐹𝑥(𝑢, 𝑣) = (𝑢, sin 𝑣).
Hence 𝐹: Σ → 𝐶 is a mapping.
Stereographic projection
Let Σ be a unit sphere with center (0,0,1). Delete
the north pole from Σ, that is, consider
Σ = {𝑝 = (𝑝1 , 𝑝2 , 𝑝3 ) ∈ 𝑅 3 : 𝑑(𝑝, (0,0,1)) = 1}

\{(0,0,2)}.
128 of 161
Let us identify 𝑥𝑦-plane with 𝑅 2 , that is,
𝑅 2 = {(𝑝1 , 𝑝2 , 0): 𝑝1 , 𝑝2 ∈ 𝑅}.
The function 𝐹 is given by
𝑅𝑝1 𝑅𝑝2
𝐹(𝑝1 , 𝑝2 , 𝑝3 ) = ( , )
𝑟 𝑟
where 𝑟 and 𝑅 are shown in the figure below.
Using the similar triangles in the figure below (left), we get,
𝑅 𝑟
=
2 2 − 𝑝3
So
2𝑝1 2𝑝2
𝐹(𝑝1 , 𝑝2 , 𝑝3 ) = ( , ).
2 − 𝑝3 2 − 𝑝3
The function
2𝑝1 2𝑝2
𝐹(𝑝1 , 𝑝2 , 𝑝3 ) = ( , ).
2 − 𝑝3 2 − 𝑝3
is differentiable.
For any patch 𝑥: 𝐷 → Σ in Σ the function 𝐹(𝑥) is differentiable.
Let 𝑦: 𝑅 2 → 𝑅 3 given by
𝑦(𝑝1 , 𝑝2 ) = (𝑝1 , 𝑝2 , 𝑝3 )
be the patch for the 𝑥𝑦 − plane, then 𝑦 −1 is differentiable. Hence 𝑦 −1 𝐹𝑥 is differentiable, making 𝐹
a mapping.
Definition
Let 𝐹: 𝑀 → 𝑁 be a mapping of surfaces. Let 𝑣 be a tangent vector to 𝑀 such that there exist a curve
𝛼 in 𝑀 and 𝑣 = 𝛼 ′ (0). Then F(𝛼) is a curve in 𝑁.
We define the tangent map 𝐹∗ of 𝐹 as
𝐹∗ (𝑣) = 𝐹(𝛼)′ (0)
129 of 161
that takes tangent vectors of 𝑀 to the tangent vectors of 𝑁.
Fact1: The tangent map 𝐹∗ is a linear map from 𝑇𝑝 (𝑀) to 𝑇𝑇(𝑝) (𝑁).
Fact 2: Let 𝐹: 𝑀 → 𝑁 and 𝐺: 𝑁 → 𝑃 be mappings of surfaces. Then 𝐺(𝐹): 𝑀 → 𝑃 is a mapping of

surfaces and
(𝐺(𝐹))∗ = 𝐺∗ (𝐹∗ ).
Computation of tangent map
Let 𝐹: 𝑀 → 𝑁 be a mapping.
Let 𝑥: 𝐷 → 𝑀 be a parameterization in 𝑀. Let 𝑦 = 𝐹(𝑥): 𝐷 → 𝑁.
Diffeomorphisms
Definition: Let 𝐹: 𝑀 → 𝑁 be a mapping of surfaces. If 𝐹 has an inverse 𝐹 −1 : 𝑁 → 𝑀 which is also a
mapping of surfaces then 𝐹 is a diffeomorphism.
Inverse function theorem
Let 𝐹: 𝑀 → 𝑁 be a mapping of the surface.

Let 𝑝 ∈ 𝑀 be a point. Assume that
𝐹∗ : 𝑇𝑝 (𝑀) → 𝑇𝑇(𝑝) (𝑁)
is an isomorphism. Then there are neighborhoods 𝑈 ⊂ 𝑀 of 𝑝 and 𝑉 ⊂ 𝑁 of 𝐹(𝑝) such that
𝐹|𝑈 : 𝑈 → 𝑉
is a diffeomorphism.
Corollary: If 𝐹 is regular and bijective, then 𝐹 is a diffeomorphism.
Definition: Two surfaces are diffeomorphic if there is a diffeomorphism between them.
130 of 161
End of the lecture
Lecture # 26
Contents:
131 of 161
 Diffeomorphic Surfaces
 Mapping of Differential Forms
Diffeomorphic Surfaces
Examples of diffeomorphic surfaces
Example1: An open rectangle in 𝑹𝟐 is isomorphic to the entire plane 𝑹𝟐
Let 𝑅 be an open set given as

𝜋 𝜋 𝜋 𝜋
𝑅 = {(𝑢, 𝑣): 𝑢 < − < 𝑢 < , − < 𝑣 < }.
2 2 2 2
Then the function 𝐹: 𝑅 → 𝑅 2 defines as
𝐹(𝑢, 𝑣) = (tan 𝑢, tan 𝑣)
is a mapping of 𝑅 into 𝑅 2 . The inverse 𝐹 −1 is given by
𝐹 −1 (𝑢, 𝑣) = (tan−1 𝑢, tan−1 𝑣).
So 𝐹 is a diffeomorphism.
Similarly every open rectangle in 𝑅 2 is
diffeomorphic to 𝑅 2 .
Example2: The sphere minus one point to

𝑹𝟐
Let Σ0 be a sphere minus one point, namely

the north pole. A parametrization of Σ0 is
given by:
𝑥(𝑢, 𝑣) = (cos 𝑣 cos 𝑢 , cos 𝑣 sin 𝑢 , 1 + sin 𝑣).
where
132 of 161
3𝜋 𝜋
(𝑢, 𝑣) ∈ 𝑅 × (− , )
2 2
Let 𝐹 from Σ0 to 𝑹𝟐 be the projection map, so a point (𝑝1 , 𝑝2 , 𝑝3 ) of Σ0 has image:
2𝑝1 2𝑝2
𝐹(𝑝1 , 𝑝2 , 𝑝3 ) = ( , , 0).
2 − 𝑝3 2 − 𝑝3
So
2 cos 𝑣
𝑦(𝑢, 𝑣) = 𝐹(𝑥(𝑢, 𝑣)) = (cos 𝑢 , sin 𝑢 , 0).
1 − sin 𝑣
Example 3: A cylinder and 𝑹𝟐 minus a point
Let 𝐶: 𝑥 2 + 𝑦 2 = 1 be a cylinder in 𝑅 3 . Define 𝐹: 𝐶 → 𝑅 2 by
𝐹(𝑥, 𝑦, 𝑧) = 𝑒 𝑧 (𝑥, 𝑦)
Then 𝐹 maps objectively 𝐶 to 𝑅 2 \(0,0).
The inverse of 𝐹 is gives as:

𝑢 𝑣
𝐹 −1 (𝑢, 𝑣) = ( , , ln √𝑢2 + 𝑣 2 ).
√𝑢2 + 𝑣 2 √𝑢2 + 𝑣 2
This is again a mapping, therefore 𝐹 is a diffeomorphism.
Mappings of Differential Forms
Mapping of a 𝟎 −form
Let 𝐹: 𝑀 → 𝑁 be a mapping of surface.
Question: If g: M → 𝑅 is a function then how to move 𝑔 to a function defined on 𝑁?
This is not possible in general, but we take functions 𝑓: 𝑁 → 𝑅 to functions on 𝑀.
For functions 𝑓: 𝑁 → 𝑅 we define:
133 of 161
𝑓(𝐹): 𝑀 → 𝑅 , 𝑝 → 𝑓(𝐹(𝑝)).
We call 𝑓(𝐹) the pull back of 𝑓.
Pull-back of 𝟏 −forms and 𝟐 −forms
Definition: Let 𝐹: 𝑀 → 𝑁 be a mapping of surfaces.
 Let 𝜙 be a 1 −form on 𝑁, the pull-back of 𝜙, written as 𝐹 ∗ (𝜙) is defined as:
For every tangent vector 𝑣 to 𝑀 define
(𝐹 ∗ 𝜙)(𝑣) = 𝜙(𝐹∗ 𝑣).
 Let 𝜂 be a 2 −form on 𝑁. The pull-back of 𝜂, denoted by 𝐹 ∗ (𝜂), is defined as:
For every pair of tangent vectors 𝑢, 𝑣 to 𝑀
𝐹 ∗ 𝜂(𝑢, 𝑣) = 𝜂(𝐹∗ 𝑢, 𝐹∗ 𝑣).
The essential operations on forms are sum, wedge product, and exterior derivative, are all
preserved by mapping.
Theorem:
Let 𝐹: 𝑀 → 𝑁 be a mapping of surfaces, and let 𝜉 and 𝜂 be forms on 𝑁. Then
1. 𝐹 ∗ (𝜉 + 𝜂) = 𝐹 ∗ 𝜉 + 𝐹 ∗ 𝜂.
2. 𝐹 ∗ (𝜉 ∧ 𝜂) = 𝐹 ∗ 𝜉 ∧ 𝐹 ∗ 𝜂.
3. 𝐹 ∗ (𝑑𝜉) = 𝑑 (𝐹 ∗ 𝜉).
End of the lecture
Lecture # 27
Contents:
134 of 161
 Line Integrals
 Integration of 2 −forms
 Integral Over a Boundary
Line Integrals
Pull-back of 𝟏 −form on a curve
Let
𝛼: [𝑎, 𝑏] → 𝑀
be a closed curve segment on surface 𝑀.
Let 𝜙 be a 1 −form on 𝑀. The pull-back of 𝜙 is a 1 −form on [𝑎, 𝑏] is of the form
𝑓(𝑡) 𝑑𝑡 where
𝑓(𝑡) = (𝛼 ∗ 𝜙)(𝑈1 (𝑡)) = 𝜙 (𝛼∗ (𝑈1 (𝑡))) = 𝜙(𝛼 ′ (𝑡)).
Definition: Let 𝜙 be a 1 −form on 𝑀.

Let 𝛼: [𝑎, 𝑏] → 𝑀 be a curve segment. Then the integral of 𝜙 over 𝛼 is equal to
𝑏
∫ 𝜙=∫ 𝛼 ∗ 𝜙 = ∫ 𝜙(𝛼 ′ (𝑡)) 𝑑𝑡 .
𝛼 [𝑎,𝑏] 𝑎
Another name is the

integral of 𝜙 over 𝛼.
Force field: Particle moving in a force field
Example: Let 𝑉 be a vector field on a surface 𝑀. We assume that 𝑉 is a force field.
135 of 161
Assume that a particle moves on a curve 𝛼: [𝑎, 𝑏] → 𝑀. We want to know how much work is done
by the force field on the particle as it moves between points 𝛼(𝑎) and 𝛼(𝑏).
Work is done on the particle only by the component of force tangent to 𝛼, that is,
𝛼′
𝑉(𝛼). ‖𝛼′ ‖ = ‖𝑉(𝛼)‖ cos 𝜃
The work done by the force during time Δ𝑡 is approximately given as
𝛼′
(𝑉(𝛼). ′ ) ‖𝛼 ′ (𝑡)‖Δ𝑡 .
‖𝛼 ‖
Adding these on the whole time interval [𝑎, 𝑏] we get

𝑏
𝑊 = ∫ 𝑉(𝛼(𝑡)). 𝛼 ′ (𝑡) 𝑑𝑡 .
𝑎
To express the following more easily

𝑏
𝑊 = ∫ 𝑉(𝛼(𝑡)). 𝛼 ′ (𝑡) 𝑑𝑡
𝑎
we introduce the following
Definition: The dual to the vector field 𝑉 on 𝑀 is the 1 − form 𝜙 defined by

𝜙(𝑤𝑝 ) = 𝑤. 𝑉(𝑝)
for every 𝑝 ∈ 𝑀 and every 𝑤𝑝 ∈ 𝑇𝑝 (𝑀).
The total work done by the particle
becomes
𝑊=∫ 𝜙
𝛼
where 𝜙 is the dual 1 −form of 𝑉.
136 of 161
The integral of a differential
Theorem: Let 𝑓: 𝑀 → 𝑅 be a function on 𝑀. Let 𝛼: [𝑎, 𝑏] → 𝑀 be a curve segment on 𝑀, where

𝛼(𝑎) = 𝑝 and 𝛼(𝑏) = 𝑞. Then
∫ 𝑑𝑓 = 𝑓(𝑞) − 𝑓(𝑝).
𝛼
See the book for proof.
Theorem: Let 𝑓: 𝑀 → 𝑅 be a function on 𝑀. Let 𝛼: [𝑎, 𝑏] → 𝑀 be a curve segment on 𝑀, where

𝛼(𝑎) = 𝑝 and 𝛼(𝑏) = 𝑞. Then
∫ 𝑑𝑓 = 𝑓(𝑞) − 𝑓(𝑝).
𝛼
The above can also be restated as:
Let the “boundary” of the curve be 𝑞 − 𝑝, where 𝑝 is the starting point and 𝑞 is the end point. Then
the integral of 𝑑𝑓 over 𝛼 equals the “integral” of 𝑓 over the boundary of 𝛼, namely 𝑓(𝑞) − 𝑓(𝑝).
Extending to surface maps
Definition:
Let
𝑅 = [𝑎, 𝑏] × [𝑐, 𝑑] = {(𝑢, 𝑣): 𝑎 ≤ 𝑢 ≤ 𝑏 𝑎𝑛𝑑 𝑐 ≤ 𝑣 ≤ 𝑑} .
be a closed rectangle in 𝑅 2 .
A 2 −segment is a function 𝑥: 𝑅 → 𝑀 which is differentiable, that is, there exist an open set 𝐷 ⊂ 𝑅 2
such that 𝑅 ⊂ 𝐷, and 𝑥: 𝐷 → 𝑀 is differentiable.
137 of 161
Note:
We do not assume that a 2 −segment 𝑥 is regular or injective.
The partial velocities 𝑥𝑢 and 𝑥𝑣 are defined as usual.
But they may behave in some strange way if 𝑥 is not regular.
Pull-back of a 𝟐 −form
Let 𝑥: 𝑅 → 𝑀 be a 2 −segment on a surface 𝑀. Let 𝜂 be a 2 −form on 𝑀. Then the pull-back by 𝑥 of

𝜂 has the form
ℎ 𝑑𝑢 𝑑𝑣
Where
ℎ = (𝑥 ∗ 𝜂)(𝑈1 , 𝑈2 ) = 𝜂(𝑥∗ 𝑈1 , 𝑥∗ 𝑈2 ) = 𝜂(𝑥𝑢 , 𝑥𝑣 ).
Definition: The integral of 𝜂 over 𝑥 is

𝑏 𝑑
∬ 𝜂 = ∬ 𝑥 ∗ 𝜂 = ∫ ∫ 𝜂(𝑥𝑢 , 𝑥𝑣 ) 𝑑𝑢 𝑑𝑣 .
𝑥 𝑅 𝑎 𝑐
Definition: Let = [𝑎, 𝑏] × [𝑐, 𝑑] . Let 𝑥: 𝑅 → 𝑀 be a 2 −segment in 𝑀.

Let 𝛼, 𝛽, 𝛾, 𝛿 be the curve segments such that
𝛼(𝑢) = 𝑥(𝑢, 𝑐), 𝛽(𝑣) = 𝑥(𝑏, 𝑣), 𝛾(𝑢) = (𝑢, 𝑑 ), 𝛿(𝑣) = (𝑎, 𝑣 ) .
The boundary of 𝑥 is a formal expression
𝜕𝑥 = 𝛼 + 𝛽 − 𝛾 − 𝛿
138 of 161
Let 𝜙 be a 1 −form on 𝑀 then the integral of 𝜙 on 𝜕𝑥 is defined as
∫ 𝜙 =∫ 𝜙+∫ 𝜙−∫ 𝜙−∫ 𝜙 .

𝜕𝑥 𝛼 𝛽 𝛾 𝛿
End of the lecture
Lecture # 28
139 of 161
Contents:
 Stokes Theorem
 Reparametrization
Stokes Theorem
Theorem: Let 𝜙 be a 1 −form on 𝑀. Let 𝑅 be a closed rectangle 𝑅 = [𝑎, 𝑏] × [𝑐, 𝑑]. Let 𝑥: 𝑅 →
𝑀 be a 2 −segment in 𝑀. Then
∫ ∫𝑑 𝜙 = ∫ 𝜙 .
𝑥 𝜕𝑥
Proof:
Idea of the proof: We will start from the double integral and show that it turns into the integral
over the boundary of 𝑥.
We have
𝑏 𝑑
∫ ∫𝑑 𝜙 = ∫ ∫ 𝑑𝜙(𝑥𝑢 , 𝑥𝑣 )𝑑𝑢 𝑑𝑣
𝑥 𝑎 𝑐
By definition of exterior derivative:

𝑏 𝑑 𝑏 𝑑
𝜕 𝜕
∫ ∫ 𝑑𝜙(𝑥𝑢 , 𝑥𝑣 )𝑑𝑢 𝑑𝑣 = ∫ ∫ ( 𝜙(𝑥𝑣 ) − 𝜙(𝑥𝑢 )) 𝑑𝑢 𝑑𝑣
𝑎 𝑐 𝑎 𝑐 𝜕𝑢 𝜕𝑢
So
𝑏 𝑑
𝜕 𝜕
∫ ∫𝑑 𝜙 = ∫ ∫ ( 𝜙(𝑥𝑣 ) − 𝜙(𝑥𝑢 )) 𝑑𝑢 𝑑𝑣 .
𝑥 𝑎 𝑐 𝜕𝑢 𝜕𝑢
Let 𝑓 = 𝜙(𝑥𝑢 ): 𝑅 → 𝑹 and 𝑔 = 𝜙(𝑥𝑣 ): 𝑅 → 𝑹. . So

𝑏 𝑑 𝑏 𝑑
𝜕 𝜕 𝜕𝑔 𝜕𝑓
∫ ∫ ( 𝜙(𝑥𝑣 ) − 𝜙(𝑥𝑢 )) 𝑑𝑢 𝑑𝑣 = ∫ ∫ ( − ) 𝑑𝑢 𝑑𝑣
𝑎 𝑐 𝜕𝑢 𝜕𝑢 𝑎 𝑐 𝜕𝑢 𝜕𝑢
140 of 161
𝑏 𝑑 𝑏 𝑑
𝜕𝑔 𝜕𝑓
= ∫ ∫ ( ) 𝑑𝑢 𝑑𝑣 − ∫ ∫ ( ) 𝑑𝑢 𝑑𝑣
𝑎 𝑐 𝜕𝑢 𝑎 𝑐 𝜕𝑢
Now we will calculate:

𝑏 𝑑 𝑏 𝑑
𝜕𝑔 𝜕𝑓
∫ ∫ ( ) 𝑑𝑢 𝑑𝑣 𝑎𝑛𝑑 ∫ ∫ ( ) 𝑑𝑢 𝑑𝑣
𝑎 𝑐 𝜕𝑢 𝑎 𝑐 𝜕𝑢
Now we will calculate:

𝑑 𝑏 𝑑 𝑏
𝜕𝑔 𝜕𝑓
∫ ∫ ( ) 𝑑𝑢 𝑑𝑣 𝑎𝑛𝑑 ∫ ∫ ( ) 𝑑𝑢 𝑑𝑣
𝑐 𝑎 𝜕𝑢 𝑐 𝑎 𝜕𝑢
Consider
𝑑 𝑏 𝑑 𝑏 𝑑
𝜕𝑔 𝜕𝑔
∫ ∫ ( ) 𝑑𝑢 𝑑𝑣 = ∫ (∫ ( ) 𝑑𝑢) 𝑑𝑣 = ∫ (𝐼) 𝑑𝑣
𝑐 𝑎 𝜕𝑢 𝑐 𝑎 𝜕𝑢 𝑐
Where
𝑏
𝜕𝑔
𝐼 = ∫ ( ) 𝑑𝑢
𝑎 𝜕𝑢
By fundamental theorem of calculus:

𝑏
𝜕𝑔(𝑢, 𝑣)
𝐼=∫ 𝑑𝑢 = 𝑔(𝑏, 𝑣) − 𝑔(𝑎, 𝑣).
𝑎 𝜕𝑢
If 𝛽(𝑣) = 𝑥(𝑏, 𝑣) and 𝛿(𝑣) = 𝑥(𝑎, 𝑣) then
𝑑 𝑑 𝑑 𝑑
𝜕
∫ 𝑔(𝑏, 𝑣) 𝑑𝑣 = ∫ 𝜙(𝑥𝑣 (𝑏, 𝑣)) 𝑑𝑣 = ∫ 𝜙 ( 𝑥(𝑏, 𝑣)) 𝑑𝑣 = ∫ 𝜙(𝛽 ′ (𝑣)) 𝑑𝑣 = ∫ 𝜙
𝑐 𝑐 𝑐 𝜕𝑣 𝑐 𝛽
Now
𝑑 𝑑 𝑑 𝑑
∫ ∫𝒅 𝝓 = (∫ 𝜙(𝛽 ′ (𝑣)) 𝑑𝑣 − ∫ 𝜙(𝛿 ′ (𝑣)) 𝑑𝑣 ) − (∫ 𝜙(𝛾 ′ (𝑣)) 𝑑𝑣 − ∫ 𝜙(𝛼 ′ (𝑣)) 𝑑𝑣 )
𝒙 𝑐 𝑐 𝑐 𝑐
∫ ∫𝑑 𝜙 = (∫ 𝜙 − ∫𝜙 ) − (∫𝜙 − ∫ 𝜙 )
𝑥 𝛽 𝛿 𝛾 𝛼
=∫ 𝜙
𝜕𝑥
Reparametrization
141 of 161
Line integrals of reparametrized curves
Let 𝛼(ℎ): [𝑎, 𝑏] → 𝑀 be a reparametrization of a curve segment 𝛼: [𝑐, 𝑑] → 𝑀 by ℎ: [𝑎, 𝑏] → [𝑐, 𝑑].
Then for any 1 −form on 𝑀
1) If ℎ is orientation preserving that is ℎ(𝑎) = 𝑐 and ℎ(𝑏) = 𝑑 then
∫ 𝜙 = ∫𝜙
𝑎(ℎ) 𝛼
2) ℎ is orientation reversing that is ℎ(𝑎) = 𝑑 and ℎ(𝑏) = 𝑐 then
∫ 𝜙 = − ∫𝜙
𝛼(ℎ) 𝛼
Integral of the boundary without the minus sign
Now if 𝜉: [𝑡1 , 𝑡2 ] → 𝑀 be any curve segment, define
−𝜉: [𝑡1 , 𝑡2 ] → 𝑀
by
−𝜉(𝑡) = 𝜉(𝑡1 + 𝑡2 − 𝑡)
Then 𝜉 is orientation reversing reparametrization of 𝜉
So by previous result
∫ 𝜙 = − ∫𝜙
−𝜉 𝜉
We have seen that
∫ ∫𝑑 𝜙 = ∫ 𝜙 = (∫ 𝜙 − ∫𝜙 ) − (∫𝜙 − ∫ 𝜙 )
𝑥 𝜕𝑥 𝛽 𝛿 𝛾 𝛼
So we can write
∫ ∫𝑑 𝜙 = ∫ 𝜙 = ∫ 𝜙 + ∫ 𝜙 + ∫ 𝜙 + ∫ 𝜙
𝑥 𝜕𝑥 𝛼 𝛽 −𝛿 −𝛾
End of the lecture
142 of 161
Lecture # 29
Contents:
 Connectedness
143 of 161
 Compactness
 Orientability
Connectedness
Definition:
Let 𝑀 be a surface. The surface 𝑀 is connected if for any two points 𝑝
and 𝑞 of 𝑀 there exists a curve segment in 𝑀 from 𝑝 to 𝑞.
Example: Let 𝑀 be implicitly defined surface
𝑀: 𝑧 2 = 𝑥 2 + 𝑦 2 + 1
Then 𝑀 is disconnected.
There are no curves in 𝑀 between points with 𝑧 > 0 and points with 𝑍 < 0.
Compactness
Definition: Let 𝐴 be a set in a topological space. If for every covering of 𝐴 by open sets
𝐴 = ⋃ 𝑈𝛼 ,
𝛼∈𝐽
there exists 𝑘 ∈ 𝑵 and there exist 𝛼1 , 𝛼2 , … , 𝛼𝑘 ∈ 𝐽 such that

𝑘
𝐴⊂⋃ 𝑈𝛼𝑗 ,
𝑗=1
then 𝐴 is compact.
144 of 161
Examples:
 Closed and bounded intervals [𝑎, 𝑏] ⊂ 𝑹.
 Rectangles [𝑎, 𝑏] × [𝑐, 𝑑] ⊂ 𝑅 2 .
 Closed and bounded subsets of 𝑅 𝑛 .
Compactness of surfaces
Lemma: A surface 𝑀 is compact if and only if there exist a 𝑘 ∈ 𝑵 and 2 −segments
𝑥𝑖 : 𝑅𝑗 → 𝑀 𝑓𝑜𝑟 𝑗 = 1,2, … , 𝑘
such that
𝑘
𝑀=⋃ 𝑥𝑗 (𝑅𝑗 )
𝑗=1
Proof:
145 of 161
Compact regions on surfaces
If follows from the previous result that
“A region 𝑅 in 𝑀 in compact if it is composed of the images of finitely many 2 −segments in 𝑀”.
Examples:
 A sphere is compact.
 A torus is compact.
 Surface of revolution whose profile curve is closed.
How to check compactness
A continuous real valued function on a closed rectangle 𝑅 in the plane takes a maximum at some
point of 𝑅.
146 of 161
Lemma:
A continuous function 𝑓 on a compact region in a surface 𝑀 takes on a maximum at some point of
𝑀.
Examples of non-compact surfaces
 A cylinder is not compact.
 The disk in 𝑹𝟐 is not compact.
Orientability
Definition:
A surface 𝑀 is orientable if there exists a continuous 2 −form on 𝑀 that is non-zero at every point
of 𝑀.
Example:
 The plane 𝑹𝟐 is a surface.
The product 𝑑𝑢 ∧ 𝑑𝑣 is a 2 −form on 𝑅 2 that is non-zero at every point of 𝑹𝟐 .
Therefore 𝑹𝟐 is an orientable surface.
Unit normal
A unit normal 𝑈 on 𝑀 is a differentiable Euclidean vector field on 𝑀 that has unit length and is
everywhere normal to 𝑀.
Proposition:
A surface 𝑀 ⊂ 𝑅 3 is orientable if and only if there exists a unit normal vector field on 𝑀.
147 of 161
Examples of orientable surfaces
 Spheres
 Cylinder
 Surface of revolution
 Implicitly defined surfaces.
End of the lecture
Lecture # 30
Contents:
 Homotopy
148 of 161
 Simply Connectd Surfaces
 Poincare Lemma
 Conditions of Orientability
Homotopy
Definition:
149 of 161
Simply Connected Surfaces
Definition:
A surface 𝑀 is simply connected provided it is connected and every loop in 𝑀 in homotopic to a
constant.
Example:
The plane 𝑅 2 is simply connected. In general every Euclidean space is simply connected.
Example: A Sphere is simply connected
Not simply connected surfaces
150 of 161
Definition: If 𝜙 is any differential form, then 𝜙 is called closed if 𝑑 𝜙 = 0.

Lemma: Let 𝑀 be a surface and 𝜙 a closed 1 −form on 𝑀. Let 𝛼 be a loop in 𝑀. If 𝛼 is homotopic
to a constant then
∫ 𝜙 = 0.
𝛼
Proof:
Example: 𝑹𝟐 minus a point is not simply connected
Definition:
A differential form is exact if it is exterior derivative of another form.
Lemma: Every closed 1 −form on a simply connected space is exact.
151 of 161
Compactness and orientability
Theorem: Every compact surface in 𝑅3 is orientable.

The theorem is a consequence of the following result.
Jordan-Brouwe-Separation theorem
A compact surface in 𝑅 3 separates 𝑅 3 into two non-empty sets, an exterior and an interior.
Connectedness and orientability
Theorem: Every simply connected surface in 𝑅 3 is orientable.
152 of 161
End of the lecture
Lecture # 31
Contents:
153 of 161
 Abstract Surfaces
 Manifolds
Abstract Surfaces
Abstract patch
Let 𝑀 be any set.
An abstract patch in 𝑀 is a function
𝑥: 𝐷 → 𝑀
that satisfies the following
 𝑥 is injective.
 𝐷 is an open set in 𝑅 2 .
Definition
An abstract surface is a pair (𝑀, 𝑃) such that 𝑃 is a set of abstract patches in 𝑀, and such that the
following three axioms are satisfied:
 The covering axiom: The union of the images of the patches in 𝑃 is equal to 𝑀.
 The smooth overlap axiom: If 𝑥, 𝑦 are elements of 𝑃, then 𝑦 −1 (𝑥) and 𝑥 −1 (𝑦) are
differentiable functions from open sets in 𝑅 2 into 𝑅 2 .
 The Hausdorff axiom: If 𝑝 and 𝑞 are different points in 𝑀 then there are disjoint patches
𝑥, 𝑦 ∈ 𝑃 such that 𝑝 is in the image of 𝑥 and 𝑞 is in the image of 𝑦.
154 of 161
Open sets of a surface
Let (𝑀, 𝑃) be an abstract surface.
An open set of 𝑀 is any union of sets 𝑥(𝑈) such that 𝑥 ∈ 𝑃 and 𝑈 is open subset of the domain of 𝑥.
The projective plane

Let 𝑥: 𝐷 → Σ be a patch in Σ.
Call 𝑥 a small patch if 𝑥(𝐷) is contained in a hemisphere of Σ. So {𝑝, −𝑝} is not subset of 𝑥(𝐷) for all
𝑝 ∈ Σ.
If 𝑥 is small, then 𝐹(𝑥): 𝐷 → 𝑃 is injective. Therefore 𝐹(𝑥) is an abstract patch in 𝑃.
(𝑺, 𝑷)is an abstract surface
It is clear that (𝑆, 𝑃) satisfies the covering axiom (1).
Is also satisfies the Hausdorff axiom (3).
This is easy to show, because the domain 𝐷 of every patch has the Hausdorff property: for any two
given distinct points there are disjoint open sets around them.
What about the smooth overlap axiom?
(𝑆, 𝑃) satisfies the smooth overlap axiom
Assume that 𝐹(𝑥), 𝐹(𝑦) ∈ 𝑃 and there is a point {𝑝, −𝑝} which is in the image of both 𝐹(𝑥) and
𝐹(𝑦). Then either the images of 𝑥 and 𝑦 in Σ both contain 𝑝 or −𝑝 or one contains 𝑝 and the other
contains −𝑝.
If 𝑥 and 𝑦 overlap in Σ, then
𝐹(𝑦)−1 (𝐹(𝑥)) = 𝑦 −1 𝑥
and we know that 𝑦 −1 𝑥 is differentiable.
If 𝑥 and 𝑦 do not overlap, then 𝑥 and 𝐴(𝑦) overlap instead.
−1 −1
(𝐹(𝑦)) (𝐹(𝑥)) = (𝐹(𝐴(𝑦))) (𝐹(𝑥)) = 𝑦 −1 𝑥
which proves that 𝐹(𝑥) and 𝐹(𝑦) satisfies the smooth overlap
155 of 161
condition.
Surfaces in 𝑹𝟑
A surface in 𝑅 3 is an example of an abstract surface, if we explain 𝑆 as the set of patches that is
used to construct 𝑀.
By definition (𝑀, 𝑃1 ) and (𝑀, 𝑃2 ) are different surfaces if 𝑃1 ≠ 𝑃2 .
We will assume that the set 𝑃 of patches in an abstract surface (𝑀, 𝑃) is as large as possible.
So all patches in 𝑀 that satisfies the overlap condition with all elements of 𝑃 also have to be in 𝑃.
Sometimes we only speak of the surface 𝑀 when we mean the abstract surface (𝑀, 𝑃).
Curves in abstract surface
Let 𝑀 be an abstract surface. A curve in 𝑀 is a function
𝛼: 𝐼 → 𝑀
Such that 𝛼 has a velocity vector 𝛼 ′ (𝑡) for every 𝑡 ∈ 𝐼.
Definition:
A velocity vector of a function 𝛼: 𝐼 → 𝑀 is a function 𝛼 ′ (𝑡) such that for every differentiable
function 𝑓: 𝑀 → 𝑹
𝑑𝑓(𝛼)
𝛼 ′ (𝑡)(𝑓) = 𝛼 ′ (𝑡)[𝑓] = (𝑡) 𝑓𝑜𝑟 𝑎𝑙𝑙 𝑡 ∈ 𝐼 .
𝑑𝑡
We notice that differential forms and integration on surfaces can all be explained in terms of
velocity of curves.
Manifold
An 𝑛 −dimensional manifold is a pair (𝑀, 𝑃), where 𝑀 is a set of points, and 𝑃 is a set of abstract
patches 𝑥: 𝐷 → 𝑀 injective, with 𝐷 ⊂ 𝑅 𝑛 an open set, satisfying:
 The covering axiom: The union of the images of the patches in 𝑃 is equal to 𝑀.
 The smooth overlap axiom: If 𝑥, 𝑦 are elements of 𝑃, then 𝑦 −1 (𝑥) and 𝑥 −1 (𝑦) are
differentiable functions from open sets in 𝑅 2 into 𝑅 2 .
 The Hausdorff axiom: If 𝑝 and 𝑞 are different points in 𝑀 then there are disjoint patches
𝑥, 𝑦 ∈ 𝑃 such that 𝑝 is in the image of 𝑥 and 𝑞 is in the image of 𝑦.
Example:
156 of 161
 An abstract surface is precisely the same as a 2 −dimensional manifold.
 𝑅 𝑛 is an example of an 𝑛 −dimensional manifold.
Tangent bundle of a surface
Let 𝑀 be a surface in 𝑅 3 (or abstract surface, or 2 −dimensional manifold).
Let 𝑇(𝑀) be the set of all tangent vectors to 𝑀 at all points.
Then 𝑇(𝑀) is the tangent bundle of 𝑀.
Example:
An element of 𝑇(𝑀) corresponds to a pair (𝑝, 𝑣) ∈ 𝑀 × 𝑇𝑝 (𝑀). Since 𝑀 and 𝑇𝑝 (𝑀) each has
dimension 2. We expect that 𝑇(𝑀) has dimension 4.
Let 𝑥: 𝐷 → 𝑀 be any patch in 𝑀.
̌ = {(𝑝1 , 𝑝2 , 𝑝3 , 𝑝4 ): (𝑝1 , 𝑝2 ) ∈ 𝐷} = 𝐷 × 𝑅 2
Let 𝐷
̌ → 𝑇(𝑀) be defined as
Let 𝑥̃: 𝐷
(𝑝1 , 𝑝2 , 𝑝3 , 𝑝4 ) → 𝑝3 𝑥𝑢 (𝑝1 , 𝑝2 ) + 𝑝4 𝑥𝑣 (𝑝1 , 𝑝2 ).
Then 𝑥̃ is injective, and if 𝑃 is the set of all derived patches 𝑥̃ of patches in 𝑀, then (𝑇(𝑀), 𝑃)
satisfies the definition of a manifold.
If follows that 𝑇(𝑀) is a 4 −dimensional manifold.
End of the lecture
Lecture # 32
Contents:
 Geodesic Curves
157 of 161
 Examples
Geodesic Curves
Definition:
Let 𝑀 be a surface in 𝑀. The 𝛼 is a geodesic of 𝑀 if 𝛼′′ is normal to 𝑀.
Remark: Geodesics is a curve that does not turn right or left on 𝑀. It only moves up and down to
stay on 𝑀 itself.
Properties of geodesics
 A geodesic has constant speed ‖𝛼 ′ ‖
 Every straight line is a geodesic
Geodesics on spheres
 Let Σ be a sphere with center

(0,0,0) in 𝑅 3 . Let 𝑃 be a plane
in 𝑅 3 through (0,0,0). Then
𝑃 ∩ Σ is a great circle.
Great circles are geodesics
Assume that 𝛼 is a parametrization of a great circle on Σ.
We know that 𝛼 ′′ at every point of 𝛼 points towards the (0,0,0) of the circle.
This implies that 𝛼 ′′ is normal to Σ.
It follows that 𝛼 is a geodesic.
Geodesics on cylinders
158 of 161
Geodesics are helices on cylinders
Assume that 𝑀 is a circular cylinder 𝑀: 𝑥 2 + 𝑦 2 = 𝑟 2 in 𝑅 3 .
Let 𝛼(𝑡) = (𝑟 cos 𝜃(𝑡) , 𝑟 sin 𝜃(𝑡) , ℎ(𝑡)) be any curve in 𝑀.
For every 𝑝 ∈ 𝑀 we have (0,0,1) ∈ 𝑇𝑝 (𝑀).
So if 𝛼 ′′ is normal to 𝑀, then ℎ′′ (𝑡) = 0. This implies
ℎ(𝑡) = 𝑐𝑡 + 𝑑
For constants 𝑐, 𝑑 ∈ 𝑅. The speed of 𝛼 is √𝑟 2 𝜃 ′ + 𝑐 2 .
The speed is constant if and only if 𝜃 ′ is constant 𝑎.
Thus if and only in
𝜃(𝑡) = 𝑎𝑡 + 𝑏.
It follows that a geodesic is a curve of the form
𝛼(𝑡) = (𝑟 cos (𝑎𝑡 + 𝑏) , 𝑟 sin (𝑎𝑡 + 𝑏) , 𝑐𝑡 + 𝑑)
If 𝑎 and 𝑐 are non-zero then this is a helix.
If 𝑐 = 0 and 𝑎 ≠ 0 then 𝛼 is a circle around the 𝑧 −axis.
Closed geodesics
Assume that 𝛼: [𝑎, 𝑏] → 𝑀 is a segment of a geodesic.
If 𝛼(𝑎) = 𝛼(𝑏) and 𝛼 ′ (𝑎) = 𝛼 ′ (𝑏) then 𝛼 is a closed geodesic.
Example:
Every geodesic on a sphere is closed.
Example:
A geodesic on a cylinder is closed only if it is a circle around
159 of 161
the 𝑧-axis.
Geodesics in orthogonal planes
Theorem: Assume that 𝛼 is a unit speed curve in 𝑀. Let there exist a plane 𝑃 such that
𝛼 ⊂𝑃∩𝑀.
If 𝑃 is orthogonal to 𝑇(𝑀) at every point of 𝛼, then 𝛼 is a geodesic.
Proof:
Let 𝛼 has a constant speed. This implies that 𝛼 ′ . 𝛼 ′′ = 0 .
We know that
𝛼 ′ ∈ 𝑃 ∩ 𝑇(𝑀)
And 𝛼 ′′ ∈ 𝑃. Since 𝑃 is orthogonal to 𝑇(𝑀), this implies that 𝛼 ′′ is orthogonal to 𝑇(𝑀).
So 𝛼 is a geodesic.
Surface of revolution
If 𝑀 is a surface of revolution, then every meridian in 𝑀 is a geodesic.
Proof:
Every meridian is the intersection of 𝑀 with a plane 𝑃 that contains the rotation axis.
Such a plane is orthogonal to 𝑀 at every point.
It follows from the previous result that a meridian is a geodesic.
160 of 161
End of the lecture
161 of 161

Differential Geometry

Uploaded by

Copyright:

Available Formats

Differential Geometry

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Differential Geometry

Uploaded by

Copyright:

Available Formats

Lecture notes on MTH352: Differential Geometry

MTH352: Differential Geometry

Dr. SOHAIL IQBAL

The following lecture notes are designed for the course of

Calculus On Euclidean Space

The next name we came across in the history of geometry is Archimedes.

Introduction of “Cartesian coordinates” marked a new stage for geometry,

Aim of the course:

 Review of differential calculus.

 Develop tools to study curves and surfaces in space.

 Proper definition of surface. How to do calculus on surface.

 A detailed study of geometry of surface.

Let us start the course with some review of basics.

Example1: S = {1,2,3,4,…} is a set of natural numbers.

Example1: A = {1,2,3} is a subset of S = {1,2,3,4,…}.

Example1: Let 𝐷 = {𝑎, 𝑏, 𝑐}, 𝑅 = {1,2,3,4,5,6}

Example: 𝑓: 𝑹 → 𝑹, 𝑓(𝑥) = 1 + 𝑥, Image of 𝑓 = 𝑹.

Example: 𝑓(𝑥) = 1 + 𝑥, 𝑔(𝑥) = 𝑥 2 , then

Example: 𝑓: 𝐴 → 𝐵, for 𝐴 = {1,2,3,4}, 𝐵 = {𝑎, 𝑏, 𝑐, 𝑑}.

Example: 𝑓: 𝐴 → 𝐵, for 𝐴 = {1,2,3,4}, 𝐵 = {𝑎, 𝑏, 𝑐, 𝑑}

Example: 𝑓(𝑥) = 1 + 𝑥, has inverse 𝑔(𝑥) = 𝑥 − 1.

For 𝑝 = (𝑝1 , 𝑝2 , 𝑝3 ), 𝑞 = (𝑞1 , 𝑞2 , 𝑞3 ) then

For 𝑝 = (𝑝1 , 𝑝2 , 𝑝3 ), and 𝛼 a scalar then

𝛼𝑝 = (𝛼𝑝1 , 𝛼𝑝2 , 𝛼𝑝3 )

The point 𝑶 = (0,0,0) is called the origin of 𝑹𝟑 .

Review of Fields and vector spaces

 Closure of 𝐹 under addition and multiplication:

 Associativity of addition and multiplication:

𝛼 + (𝛽 + 𝛾) = (𝛼 + 𝛽) + 𝛾 and 𝛼. (𝛽. 𝛾) = (𝛼. 𝛽). 𝛾

 Commutativity of addition and multiplication:

 Existence of additive and multiplicative identity elements:

 Existence of additive inverses and multiplicative inverses:

for 𝛼 ≠ 0, there exists an element 𝛼 −1 ∈ 𝐹 , such that 𝛼. 𝛼 −1 = 1

 Distributivity of multiplication over addition:

 Closed under addition and scalar

 Commutativity of addition 𝑢+𝑣 =𝑣+𝑢

 Identity element of addition 0 ∈ 𝑉 such that 𝑣 + 0 = 0 + 𝑣 = 𝑣

For every 𝑣 ∈ 𝑉 there exist −𝑣 ∈ 𝑉 such that 𝑣 +

 Distributivity of scalar multiplication with

 Distributivity of scalar multiplication with

 Compatibility of scalar multiplication with

 Identity element of scalar multiplication 1𝑣 = 𝑣, where 1 is the multiplicative identity in 𝐹.

Example: The Euclidean 3-space 𝑅3 is a vector space.

𝑥(𝑝) = 𝑝1 , 𝑦(𝑝) = 𝑝2 , 𝑧(𝑝) = 𝑝3

𝒑 = (𝑝1, 𝑝2 , 𝑝3 ) = (𝑥1 (𝒑), 𝑥2 (𝒑), 𝑥3 (𝒑))

Example: Consider a function 𝑓: 𝑅2 → 𝑅, defined as 𝑓(𝑥, 𝑦) = 𝑥 2 𝑦. Is 𝑓 a smooth function?

Which are continuous. Now the second partial derivatives are:

Arithmetic of differentiable functions

 (𝑓 + 𝑔)(𝒑) = 𝑓(𝒑) + 𝑔(𝒑),

Q1(a) (Exercise 1.1):

Solution: Here 𝑓𝑔2 represents the point wise multiplication. So

𝑓𝑔2 = (𝑥 2 𝑦)(𝑦 sin 𝑧) = 𝑥 2 𝑦𝑧 sin 𝑧 .

Then 𝑦 = 𝑓(𝑢) and

Derivative of outside function

Q1(d) (Exercise 1.1):

Solution: Using chain rule

The lecture contents

 Natural frame field on 𝑅 3

 Differentiable vector fields