Euclidean Space

This is a brief review of some basic concepts that I hope will already be familiar to you.

There are three sets of numbers that will be especially important to us:
• The set of all real numbers, denoted by R.
• The set of all natural numbers, denoted by N — i.e., N = {1, 2, 3, . . .}.
• The set of all integers, denoted by Z — thus, Z = {x | x = 0 or x ∈ N or − x ∈ N}.

For any set X, we can form n -tuples, or lists, of members of X. We write an n-tuple as
(x1 , x2 , . . . , xn ), where each component xi is a member of X. The order of the components
matters — i.e., the 3-tuples (2, 5, 2) and (5, 2, 2) are distinct, different 3-tuples of numbers. We
therefore sometimes call them ordered lists or ordered n-tuples.

Notice that in both 3-tuples above, the number 2 appeared twice. That’s allowed: the components
of an n-tuple don’t have to be distinct. Suppose we’re using n-tuples to represent lists of the prices,
in dollars, of a gallon of gasoline, a liter of wine, and a gallon of milk, in that order. We don’t
want to rule out the possibility that both gasoline and milk have the same price, as in the 3-tuple
(2, 5, 2). And that would be different than the price-list (5, 2, 2): the order does matter.

It’s important to remember the distinction between an n-tuple and a set. The facts in the preceding
paragraph are just the opposite for sets: the order of a set’s elements doesn’t matter, and elements
can’t repeat. For example, {2, 5, 2} makes no sense, and the sets {5, 2} and {2, 5} are the same set
(the order of elements doesn’t matter).

For any given set X, we use the notation X n to denote the set of all n-tuples of elements of X. So
Rn is the set of all n-tuples (x1 , . . . , xn ), where each component xi is a real number. Similarly, Zn
is the set of all n-tuples of integers, and Nn is the set of all n-tuples of natural numbers. The two
3-tuples (2, 5, 2) and (5, 2, 2) are elements of N3 , Z3 , and R3 .

The geometry we associate with the sets Rn will be an important part of everything we do, and
should already be familiar. The set R is identified geometrically with the “real line,” a line with a
designated point that represents the number 0 and where each number x ∈ R is represented by the
point |x| units of distance to the right of 0 (if x > 0) or to the left of 0 (if x < 0); the line therefore
stretches infinitely far in both directions. The set R2 is represented geometrically as “the Euclidean
plane” and the elements of R2 as points on the plane, where the two components x1 and x2 of an
ordered pair (a 2-tuple) (x1 , x2 ) ∈ R2 are the Cartesian coordinates of the point associated with
the pair (x1 , x2 ). We often refer to a pair (x1 , x2 ) in R2 as the point (x1 , x2 ). Similarly, the set
R3 is represented as a three-dimensional space and the elements (x1 , x2 , x3 ) of R3 as points whose
Cartesian coordinates are the components x1 , x2 and x3 . As in R2 , we often refer to the elements
(x1 , x2 , x3 ) of R3 as points in R3 . Generalizing this to any natural number n, we refer to Rn as
n-dimensional space and its elements (x1 , . . . , xn ) as points in Rn , even though for n > 3 we can’t
actually draw a diagram of Rn .
Wherever possible, we’ll use boldface letters to denote n-tuples and regular font to denote the
components of the n-tuple, like this: x = (x1 , . . . , xn ) ∈ Rn . (One reason for the phrase “wherever
possible” is that I won’t be able to follow this convention on the whiteboard: I haven’t found a
reasonable way to write boldface letters on the whiteboard.)

We’ve said that the elements of Rn are called points. But we also call them vectors. Here’s why,
focusing first on R2 . While an ordered pair x = (x1 , x2 ) ∈ R2 can be interpreted as a position
or a location — i.e., a point — in the R2 -plane, we can also interpret (x1 , x2 ) as a movement or
displacement from one location on the plane to another.

Let’s consider an example, depicted in Figure 1. Starting from the location x = (3, 1), suppose we
move to the location x e = (4, 3). This is a movement, or displacement, of ∆x = (1, 2). But note
that this displacement ∆x is itself an ordered pair, an element of R2 . In fact, even the locations x
and xe can be regarded as displacements from the origin of R2 , i.e., from the point 0 = (0, 0). So
actually, all the elements of R2 can be interpreted interchangeably as points or as displacements,
and we often refer to elements of R2 generically as vectors. Of course, the geometry works the
same way in R3 ; and generalizing to any n, we continue to use the “vector” terminology in Rn .

Figure 1 Figure 2

Once we think of Rn as consisting of vectors, it’s natural to also think of adding vectors together. In
the example above, starting from the origin 0 = (0, 0) in R2 , we can add the two vectors (3, 1) and
(1, 2) to give us the vector (4, 3) which is the location (or vector) we end up at, as depicted in Figure
2. It’s also natural to think of multiplying a vector by a number — for example, 21 (4, 3) = (2, 1 12 ),
and −2(1, 2) = (−2, −4). We capture the ideas of adding vectors and multiplying them by numbers
(called scalars) in a formal definition:

Definition: Addition and scalar multiplication of n-tuples (vectors) in Rn are defined component-
wise:
∀x, y ∈ Rn : ∀λ ∈ R : x + y := (x1 + y1 , . . . , xn + yn ) and λx := (λx1 , . . . , λxn ).

2
Theorem: Addition and scalar multiplication in Rn have the following properties:
(1) ∀x, y ∈ Rn : x + y ∈ Rn . (Rn is closed under vector addition.)
(2) ∀λ ∈ R, ∀x ∈ Rn : λx ∈ Rn . (Rn is closed under scalar multiplication.)
(3) ∀x, y ∈ Rn : x + y = y + x. (Vector addition is commutative.)
(4) ∀x, y, z ∈ Rn : (x + y) + z = x + (y + z),
and ∀λ, µ ∈ R, ∀x ∈ Rn : λ(µx) = (λµ)x. (Both operations are associative.)
(5) ∀ λ ∈ R, ∀x, y ∈ Rn : λ(x + y) = λx + λy,
and ∀λ, µ ∈ R, ∀x ∈ Rn : (λ + µ)x = λx + µx. (The operations are distributive.)
(6) ∀x ∈ Rn : 0 + x = x. (0 is called the additive identity.)
(7) ∀x ∈ Rn : x + (−x) = 0. (−x is called the additive inverse of x.)
(8) ∀x ∈ Rn : λx = x for the scalar λ = 1.

The proofs of the eight properties in the theorem are all elementary, consisting of simply writing the
n-tuples component-wise and then applying the definition of addition and/or scalar multiplication.

Length, Distance, and the Dot Product

In an earlier paragraph we added the vector (3, 1) to the vector (1, 2) to give us the vector (4, 3).
In Figure 3 we obtain the same vector, (4, 3), as the sum of the vectors (4, 0) and (0, 3). Depicting
the sum (4, 0) + (0, 3) this way gives us three vectors that form a right triangle. We can use this
to determine the length of the vector (4, 3): assuming the lengths of (4, 0) and (0, 3) are 4 and 3,
respectively, the Pythagorean Theorem tells us the square of the length of (4, 3) is equal to 42 + 32 ,
√
i.e., 25, so the length of (4, 3) is 25 = 5. This observation leads us to define the length of any
p
vector x = (x1 , x2 ) ∈ R2 as x21 + x22 , and to define the length of vectors in Rn as follows:

Definition: The length of a vector x = (x1 , x2 , . . . , xn ) ∈ Rn , denoted kxk, and also called the
Euclidean norm of x, is given by the equation
q
kxk := x21 + x22 + · · · + x2n .

We call this the Euclidean norm because later on we’re going to define some other norms. These
alternative norms will give us different notions of length and distance in Rn .

Exercise: We appealed to the Pythagorean Theorem to justify, or motivate, the definition of length
for vectors: we showed that at least in R2 our definition does coincide with the usual concept of
length. We can do the same thing for R3 by invoking the Pythagorean Theorem twice in succession:
we use kxk to denote the length of a vector x, but we don’t use our definition of length; instead we
assume only that k(x1 , 0, 0)k = |x1 |, k(0, x2 , 0)k = |x2 |, and k(0, 0, x3 )k = |x3 |; this yields

k(x1 , x2 , x3 )k2 = k(x1 , x2 , 0)k2 + k(0, 0, x3 )k2 , by the Pythagorean Theorem

= k(x1 , 0, 0)k2 + k(0, x2 , 0)k2 + k(0, 0, x3 )k2 , by the Pythagorean Theorem
= x21 + x22 + x23 , by our assumption about the lengths of these three vectors,

3
 p
and therefore k(x1 , x2 , x3 )k = x21 + x22 + x23 . Draw a diagram in R3 showing how we can use the
Pythagorean Theorem twice, as we’ve just done for general vectors in R3 , to verify that the length
of the vector (4, 3, 12) is 13 — in the same way as Figure 3 showed that the length of (4, 3) is 25.

The following theorem gathers together the fundamental properties of our definition of length. Note
how these are all properties we intuitively think the concept of length should have.

Theorem: For all x, y ∈ Rn and all α ∈ R:

(N1) kxk = 0 ;
(N2) kxk = 0 ⇔ x = 0 ;
(N3) kαxk = |α|kxk ;
(N4) kx + yk 5 kxk + kyk , the triangle inequality.

Proofs of (N1), (N2), and (N3) are elementary applications of the definition of the norm. We
briefly defer the proof of (N4), until we’ve defined the dot product and proved the Cauchy-Schwarz
Inequality.

With a formal definition of the length of vectors, we have an obvious way to define distance in Rn :
the distance between two points x and x e is the length of the displacement vector ∆x between x
and x
e, where ∆x = x e − x, as depicted in Figure 4.

e in Rn is the length of the

Definition: The Euclidean distance between two vectors x and x
e − x, i.e., ke
vector x x − xk.

Another useful concept in Euclidean space is the dot product, which is closely linked to the concept
of length:

Definition: The dot product of two vectors x and x e in Rn (also called the scalar product, or
e), denoted x · x
inner product, of x and x e, is defined as

x·x
e := x1 x e2 + · · · + xn x
e1 + x2 x en .

Note that for any vector x ∈ Rn , we have x · x = kxk2 , and therefore x · x = 0 for all x ∈ Rn ; and
we have x · x = 0 if and only if x = 0.

The following theorem provides some additional important properties of the dot product. Each of
the properties follows immediately from the definition.

Theorem: For all x, y, z ∈ Rn and all α ∈ R:

(D1) x · y = y · x,
(D2) x · (y + z) = x · y + x · z,
(D3) (αx) · y = α(x · y),
(D4) (x + y) · (x + y) = x · x + 2(x · y) + y · y.

An important link between the concepts of dot product and length is the Cauchy-Schwarz Inequality

4
(it’s also called the Cauchy-Bunyakovsi-Schwarz Inequality). Note that the proof of the inequality
uses all four of the properties (D1) - (D4) of the dot product in the preceding theorem.

Theorem (The Cauchy-Schwarz Inequality): For all x, y ∈ Rn : |x · y| 5 kxkkyk.

Proof: First note that if either x = 0 or y = 0, then |x · y| = kxkkyk. So we’ll assume that
both x 6= 0 and y 6= 0.

Define the vector αx + βy, where α = kyk and β = kxk. We have

kαx + βyk2 = kykx + kxky · kykx + kxky

= kyk2 x · x + 2kxkkykx · y + kxk2 y · y

= 2kxk2 kyk2 + 2kxkkykx · y


= 2kxkkyk kxkkyk + x · y .

Since kαx + βyk2 , kxk, and kyk are all non-negative, the above equation yields kxkkyk + x · y = 0,
i.e., kxkkyk = −x · y.

Now let’s consider the vector αx + βy, where α = kyk and β = −kxk. Now we have

kαx + βyk2 =



kykx − kxky · kykx − kxky
= kyk2 x · x − 2kxkkykx · y + kxk2 y · y


= 2kxkkyk kxkkyk − x · y ,

so by the same argument as above we now have kxkkyk − x · y = 0, i.e., kxkkyk = x · y.

We’ve now established that kxkkyk = −x · y and kxkkyk = x · y — i.e.,

kxkkyk = max{x · y, −x · y} = |x · y|,

completing the proof of the Cauchy-Schwarz Inequality. 

Remark: The Cauchy-Schwarz Inequality is an equation — i.e., |x · y| = kxkkyk — if and only if

one of the vectors x or y is a scalar multiple of the other.
Proof: The equation obviously holds if either x = 0 or y = 0, so we’ll assume that both x 6= 0
and y 6= 0. First note that if |x · y| = kxkkyk, then either kxkkyk − x · y = 0 or kxkkyk + x · y = 0.
In either case this implies, according to the equations in the proof of the Cauchy-Schwarz Inequality,
that kαx + βyk2 = 0 — i.e., that αx + βy = 0, which implies that x and y are scalar multiples of
one another. Conversely, suppose y is a scalar multiple of x — i.e., y = tx for some t ∈ R. Then

|x · y| = |x · (tx)| = |t(x · x)| = |t|(x · x) = |t|kxk2 = |t|kxkkxk = kxkktxk = kxkkyk. 

Now we can return to the Triangle Inequality and provide a simple proof that uses the Cauchy-
Schwarz Inequality:

5
Theorem (The Triangle Inequality): For all x and y in Rn : kx + yk 5 kxk + kyk.
Proof:
kx + yk2 = kxk2 + 2x · y + kyk2
5 kxk2 + 2|x · y| + kyk2
5 kxk2 + 2kxkkyk + kyk2

2
= kxk + kyk ,

which, since kxk + kyk = 0, yields kx + yk 5 kxk + kyk. 

The following remark provides a necessary and sufficient condition for the Triangle Inequality to be
an equation:

Remark: kx + yk = kxk + kyk if and only if one of the vectors x or y is a non-negative scalar
multiple of the other.
Proof: The equation obviously holds if either vector is 0, so we assume that both x 6= 0 and
y 6= 0. In the proof of the Triangle Inequality it’s clear that kx + yk = kxk + kyk if and only if
both inequalities are equations — i.e., if and only if x · y = |x · y| and |x · y| = kxkkyk. The second
equation is true if and only if y = tx for some t ∈ R, as we’ve shown above; and given that y = tx,
the first equation is true if and only if t = 0. 

Figure 3 Figure 4