Math 4441 Aug 21, 20071

Differential Geometry
Fall 2007, Georgia Tech

Lecture Notes 0

Basics of Euclidean Geometry

By R we shall always mean the set of real numbers. The set of all n-tuples of
real numbers Rn := {(p1 , . . . , pn ) | pi ∈ R} is called the Euclidean n-space.
So we have
p ∈ Rn ⇐⇒ p = (p1 , . . . , pn ), pi ∈ R.
Let p and q be a pair of points (or vectors) in Rn . We define p + q := (p1 +
q 1 , . . . , pn +q n ). Further, for any scalar r ∈ R, we define rp := (rp1 , . . . , rpn ).
It is easy to show that the operations of addition and scalar multiplication
that we have defined turn Rn into a vector space over the field of real num-
bers. Next we define the standard inner product on Rn by

hp, qi = p1 q 1 + . . . + pn q n .

Note that the mapping h·, ·i : Rn × Rn → R is linear in each variable and is

symmetric. The standard inner product induces a norm on Rn defined by
1
kpk := hp, pi 2 .

If p ∈ R, we usually write |p| instead of kpk.

The first nontrivial fact in Euclidean geometry, is the following important
result which had numerous applications:

Theorem 1. (The Cauchy-Schwartz inequality) For all p and q in Rn

|hp, qi| 6 kpk kqk.

The equality holds if and only if p = λq for some λ ∈ R.

1
Last revised: August 22, 2021

1
 Since this is such a remarkable and far reaching result we will include here
three different proofs. The first proof is quite short and slick, but also highly
nontransparent, i.e., it is not easy to see how someone could come up with
that. The second proof is perhaps more reasonable, but also more advanced.
The third proof is the most elementary, but then again it is quite tricky.
Proof I (Quadratic Formula). If p = λq it is clear that equality holds. Oth-
erwise, let f (λ) := hp − λq, p − λqi. Then f (λ) > 0. Further, note that f (λ)
may be written as a quadratic equation in λ:

f (λ) = kpk2 − 2λhp, qi + λ2 kqk2 .

Hence its discriminant must be negative:

4hp, qi2 − 4kpk2 kqk2 < 0

which completes the proof.

Proof II (Lagrange Multipliers). Again suppose that p 6= λq. Then
 
p q
hp, qi = kpkkqk , .
kpk kqk

Thus it suffices to prove that for all unit vectors p and q we have

|hp, qi| ≤ 1,

and equality holds if and only if p = ±q. This may be proved by using the
method of lagrangne multipliers to find the maximum of the function hx, yi
subject to the constraints kxk = 1 and kyk = 1. More explicitly we need to
find the critical points of

f (x, y, λ1 , λ2 ) := hx, yi + λ1 (kxk2 − 1) + λ2 (kyk2 − 1)

Xn
= (xi yi + λ1 x2i + λ2 yi2 ) − λ1 − λ2 .
i=1

At a critical point we must have 0 = ∂f /∂xi = yi + 2λ1 xi , which yields that

y = ±x.

2
Proof III (Induction). First note that the case n = 1 is trivial. For n = 2,
the proof amounts to showing that
(p1 q1 + p2 q2 )2 ≤ (p21 + p22 )(q12 + q 2 )2 .
This is also easily verified by the expansion and simplification of both sides
which reduces the above inequality to (p1 q2 − q2 p1 )2 ≥ 0. Now suppose that
the inequality we like to prove holds for n. Then to prove this for n + 1 note
that
n+1
X n
X
pi qi = pi qi + pn+1 qn+1
i=1 i=1
v v
u n u n
u X uX
2t
≤ t pi qi2 + pn+1 qn+1
i=1 i=1
v v
u n u n
uX uX
2 2
≤ t pi + pn+1 t qi2 + qn+1
2

i=1 i=1
v v
u n+1 u n+1
uX uX
= t p2i t qi2 .
i=1 i=1

The first inequlity above is just the inductive step, i.e., the assumption that
the inequality we want to prove holds for n, and the second inequality above
is just an application of the case n = 2 which we established earlier.
There is yet another proof of the Cauchy-Schwartz ineqaulity which com-
bines ideas from the first and second proofs mentioned above, but avoids
using either the quadratic formula or the Lagrange multipliers:
Exercise 2. (The simplest proof of the Cauchy-Schwartz inequality)
Assume, as in the second proof above, that kpk = 1 = kqk and note that
0 < kp − hp, qiqk2 = p − hp, qiq, p − hp, qiq .
whenever p 6= λq. Expanding the right hand side yields the desired result.
The standard Euclidean distance in Rn is given by
dist(p, q) := kp − qk.
An immediate application of the Cauchy-Schwartz inequality is the following

3
Exercise 3. (The triangle inequality) Show that
dist(p, q) + dist(q, r) > dist(p, r)
for all p, q in Rn .
By a metric on a set X we mean a mapping d : X × X → R such that
1. d(p, q) > 0, with equality if and only if p = q.
2. d(p, q) = d(q, p).
3. d(p, q) + d(q, r) > d(p, r).
These properties are called, respectively, positive-definiteness, symmetry, and
the triangle inequality. The pair (X, d) is called a metric space. Using the
above exercise, one immediately checks that (Rn , dist) is a metric space. Ge-
ometry, in its broadest definition, is the study of metric spaces, and Euclidean
Geometry, in the modern sense, is the study of the metric space (Rn , dist).
Finally, we define the angle between a pair of nonzero vectors in Rn by
hp, qi
angle(p, q) := cos−1 .
kpk kqk
Note that the above is well defined by the Cauchy-Schwartz inequality. Now
we have all the necessary tools to prove the most famous result in all of
mathematics:
Exercise 4. (The Pythagorean theorem) Show that in a right triangle
the square of the length of the hypotenuse is equal to the sum of the squares
of the length of the sides.
Next let us define the angle defined by 3 points p, o, q as
hp − o, q − oi
∠poq := angle(p − o, q − o) = cos−1 .
kp − ok kq − ok
Exercise 5. Suppose that p, o, q lie on a line and o lies between p and q.
Show that then ∠poq = π.
We say that a set L ⊂ Rn is a line if there exists a pair of points p,
q ∈ Rn such that
L = { p + t(p − q) | t ∈ R }.
A pair of lines in R2 are called parallel if they do not intersect:

4
Exercise 6. (Playfair’s Axiom) Show that, in R2 , through every point p
outside a line L there passes one and only one line parallel to L.

The Playfair’s Axiom is equivalent to

Exercise 7. (Euclid’s Fifth Postulate) Suppose that two parallel lines

L, L0 in R2 are intersected by a third line L00 . Then the acute angles formed
by these three lines are equal.

The next exercise is concerned with another corner stone of Euclidean

Geometry, which may be proved using the last three exercises:

Exercise 8. (Sum of the angles in a triangle) Show that the sum of

the angles in a triangle is π (Hint: through one of the vertices draw a line
parallel to the opposite side).

The most important result in classical differential geometry is the Gauss-

Bonnet theorem, which generalizes the fact proved in the last exercises to
regions which lie on a curved surface.