You are on page 1of 28

Chapter 1

Affine space and affine sets

1.1 Affine space

1.1.1 Affine space

Definition 1. Let V be a vector space over a field K and A be a non-empty set whose elements
−→ −−→
are called points. As usual, vectors are often denoted by → −
x ,→

y ,...,→
−u ,→
−v . . . , AB, M N . . . ; while
points are denoted by A, B, C . . . , M, N, P, . . . . Suppose that there exists a map

Φ : A × A −→ V
(M, N ) 7−→ Φ(M, N )

satisfying the following two conditions:

(A1) for a point M ∈ A and a vector →



v ∈ V, there exists one and only one point N ∈ A such


that Φ(M, N ) = v ;

(A2) for any three points M, N, P in A we have

Φ(M, N ) + Φ(N, P ) = Φ(M, P ).


(Chasles’ relation)

Then we call A an affine space over K associate with V by Φ or just affine space for short.



V is called the associated vector space (or the based space) of A, denoted by A , and the map Φ
−−→
is called the associated map. For convenience of writing, we replace Φ(M, N ) by M N . Therefore,
the conditions in the definition can be rewritten as follows:


− −−→ −
1. ∀M ∈ A, ∀→

v ∈ A ; ∃! N ∈ A, M N = →
v;

1
Affine and Euclidean Geometry

−−→ −−→ −−→


2. ∀M, N, P ∈ A; M N + N P = M P .

When K = R, A is called a real affine space and when K = C, A is called a complex affine space.

To emphasize the field K, we say that A is a K-affine space.




To denote an affine space completly, we write (A, A , Φ). When no confusion can arise, we denote
an affine space by A(K) or just by A.


If A is a vector space of dimension n, then we say A is an affine space of dimension n and write
An to emphasize the dimension of the space. The dimension of A is denoted by dim A. Thus,


dim A = dim A .

In this lecture note, an affine space A is of dimension n and the field K will be R or C, unless
otherwise stated.

1.1.2 Examples


Example 1. Denote by E3 the usual space of points and E 3 the space of vectors. Drawing a
vector by joining two given points A and B is just the associated map Φ. The space E3 is an affine


space associated with E 3 , because we can check easily that the map


Φ : E3 × E3 −→ E 3
−→
(A, B) 7−→ AB

satisfies Conditions (A1) and (A2).


Example 2. Let V be a vector space over a field K. The map

Φ : V × V −→ V
(→
−u ,→

v ) 7−→ Φ(→

u ,→

v ) := →

v −→

u

satisfies Conditions (A1) and (A2). Thus, V is an affine space associated with itself. We say that
Φ defines the canonical affine structure on vector space V or V is an affine space with the canonical
affine structure.

As a special case, V = Kn = K
| × K{z
· · · × K} is an n-dimensional affine space with the canonical
n
affine structure.

By this example, we can see that a vector space is an affine space (with the canonical affine
structure). Conversely, we can bring a vector structure into an affine space A by choosing a fixed
−−→ → −
point O ∈ A and identifying each point M ∈ A with the vector OM ∈ A (see Exercise 1.5). Thus,
the difference between an affine space and a vector space that have the same dimension (e.g. an
affine space and its associated space) is just “a fixed point”.

2
Affine and Euclidean Geometry

1.1.3 Some simple properties

Below are some simple properties deduced from the definition.

For every M, N, P, Q ∈ A, we have

−−→ → −
1. M N = 0 if and only if M = N,
−−→ −−→
2. M N = −N M ,
−−→ −→ −−→ −−→
3. M N = P Q if and only if M P = N Q,
−−→ −−→ −−→
4. M N = P N − P M .

Proof. 1. Suppose that M = N. By Chasles’ relation we have


−−→ −−→ −−→
MM + MM = MM.

Therefore,
−−→ → −
MM = 0 .
−−→ →
− −−→ →

If M N = 0 , then by the above proof M M = 0 . Therefore, by the first condition in
Definition 1, we have M = N.

2. By Chasles’ relation
−−→ −−→ −−→ → −
MN + NM = MM = O .
Therefore,
−−→ −−→
M N = −N M .

3. We have
−−→ −→ −−→ −−→ −−→ −→ −−→ −−→
M N = P Q ⇔ M N + N P = N P + P Q ⇔ M P = N Q.

4. By Chasles’ relation and the second property.

1.2 Affine sets-Affine independence and affine dependence-


Affine hull

1.2.1 Affine sets

This is a generalized notion of points (0-dimensional case), lines (1-dimensional case) or plane
(2-dimensional case). In E3 , a line d is completely determined by a point P ∈ d and a vector → −v
parallel to d, called a directional vector of d. A plane α is completely determined by a point P ∈ α

3
Affine and Euclidean Geometry

v
u
v

M
d

M P
P 

Figure 1.1: A line is determined by a point and a Figure 1.2: A plane is determined by a point and
directional vector a pair of directional vectors.

and a pair of independent vectors {→−


u ,→
−v } that are both parallel to α, called pair of directional
vectors of the plane. We can describe the line d and the plane α as follows
−−→
d = {M ∈ E3 : P M = a→

v ; a ∈ R},
−−→
α = {M ∈ E3 : P M = a→

u + b→

v ; a, b ∈ R}.
By this description, the following definition is natural.


Definition 2. Let (A, A , Φ) be an affine space, P be a point in A and →−
α be a vector subspace


of A . The set
−−→ −
α = {M ∈ A : P M ∈ → α}
is called an affine set passing through P with the directional space →

α.

When dim →−
α = m, we say that α is an affine set of dimension m or an m-plane and write
dim α = m. Thus,
dim α = dim →−
α.
Usually, a 1-plane is called a line while a 2-plane is called a plane. Hyperplane is the name of an
affine set of codimension one, i.e. of (n − 1)-dimension.

The directional space of an affine set α is always denoted by →



α , unless otherwise stated.
−−→
Remark 1. 1. If α is an affine set passing through P and ∀M, N ∈ α; the vector M N =
−−→ −−→ →
PN − PM ∈ − α.

2. We can see a point as a 0-plane.

3. The point P in the definition of an affine set does not play any special role. This means that
−−→ −
∀Q ∈ α, α = {M ∈ A : QM ∈ →
α }.

4
Affine and Euclidean Geometry

4. Suppose that α is an affine set passing through P and β is an affine set passing through Q.


Then α ⊂ β if and only if P ∈ β and → −
α ⊂ β.


Therefore, α ≡ β if and only if P ∈ β (or Q ∈ α) and →

α ≡ β.

5. If α is an affine set, then α is an affine space associated with →



α by the associated map

Φ|α×α : α × α −→ →

α.

Because of that we can say that an affine set is an affine subspace.

1.2.2 Affine independence and affine dependence

Definition 3. A family of m + 1 points {A0 , A1 , . . . , Am } (m ≥ 1) in an affine space A is called


−−−→ −−−→ −−−→ →

affinely independent if m vectors {A0 A1 , A0 A2 , . . . , A0 Am } in A are linearly independent.

A family of points which is not affinely independent is called affinely dependent.


Remark 2. 1. For short, when saying about a family of points, sometime we just say inde-
pendent and dependent in stead of affinely independent and affinely dependent. Note that,
when saying about a family of vectors, the terminologies independent and dependent must
be understood that linearly independent and linearly dependent.

2. One point A0 is independent.

3. We can prove that the family


−−−→ −−−→ −−−→
{A0 A1 , A0 A2 , . . . , A0 Am }

is linearly independent if and only if the family


−−−→ −−−−→ −−−−→ −−−→
{Ai A0 , . . . , Ai Ai−1 , Ai Ai+1 , . . . , Ai Am }, i ∈ {1, 2, ..., m}

is. Therfore, Definition 3 does not depend on the point A0 .


−−−→ −−−→ −−−→
4. The family {A0 , A1 , . . . , Am } is affinely dependent if and only if the family {A0 A1 , A0 A2 , . . . , A0 Am }
is linearly dependent.

5. A sub-family of an affinely independent family is independent, but a sub-family of an affinely


dependent family may be not dependent.
Example 3. 1. A family of two points {P, Q} in A is independent if and only if P 6= Q.

2. A family of three points {P, Q, R} in A is independent if and only if they are not collinear.

3. A family of four points {P, Q, R, S} in A is independent if and only if they are not belonging
to a plane (non-planar).

4. Generally, a family of m + 1 points {A0 , A1 , ..., Am } in A is independent if and only if they


are not belonging to an m − 1-plane.

5
Affine and Euclidean Geometry

Theorem 1.2.1. In an n-dimensional affine space An ,there always exists independent families of
m points, where 0 < m ≤ n + 1. A family of k points, where k > n + 1, is always dependent.



Proof. Suppose that {→

e1 , →

e2 , . . . , →

en } is a basis of An and let A0 ∈ An . Then, there exist uniquely
points Ai such that
−−−→ →
A0 Ai = − ei , i = 1, 2, . . . , n.
By the definition, {A0 , A1 , A2 , . . . , An } is an independent family of n + 1 points. Of course,
{A0 , A1 , A2 , . . . , Am−1 }, where 0 < m ≤ n + 1, is an independent family of m points.
−−−→ −−−→ −−−→
Consider the family {B0 , B1 , B2 , . . . , Bp }, p > n. Then {B0 B1 , B0 B2 , . . . , B0 Bp } is a family of more
than n vectors and therefore it is dependent. By the definition, the family {B0 , B1 , B2 , . . . , Bp } is
dependent.

1.2.3 Intersections-Affine hulls

Let {αi : i ∈ I} be a non-empty family of affine sets in an affine space A.

Theorem 1.2.2. If i∈I αi 6= ∅, then i∈I αi is an affine set with the directional space i∈I →
T T T −
α i.

T T T
Proof. Because i∈I αi 6= ∅, there exists a point P ∈ i∈I αi . A point M ∈ i∈I αi if and only if
−−→ − −−→ T −
M ∈ αi , ∀i ∈ I; i.e. if and only if P M ∈ →α i , ∀i ∈ I. This is equivalent to P M ∈ i∈I →
α i . In other
words, \ −−→ \ → −
αi = {M ∈ A : P M ∈ α i },
i∈I i∈I

i.e.
T T →

i∈I αi is an affine set passing through P with the directional space i∈I α i.

T
We can see that i∈I αi is the biggest affine set that is contained in αi , for all i ∈ I.

Definition 4. Let X is a non-empty subset of an affine space A. Then the intersection of all affine
sets containing X is an affine set, called the affine hull of X, and denoted by aff(X).

The affine hull aff(X) of the set X is the smallest affine set containing X.
S
Definition 5. Let {αi : i ∈ I} be a non-emptyP family of affine sets. The affine hull of i∈I αi is
called the sum of affine sets αi , denoted by i∈I αi .

P
Thus, the sum of affine sets i∈I αi is the smallest affine set (of least dimension) containing all
affine sets αi , i ∈ I.
= {1, 2, . . . , m}, we write α1 + α2 + . . . + αm or m
P
When I is finite, e.g. I P i=1 αi for the sum of
affine sets αi , instead of i∈I αi .

If X consists of finitely many points, X = {P0 , P1 , . . . , Pm }, then the sum P0 + P1 + . . . + Pm (Pi is


viewed as a 0-plane) is the affine set that has smallest dimension containing these points. Moreover,

6
Affine and Euclidean Geometry

−−→ −−→ −−−→


dim(P0 + P1 + . . . + Pm ) = rank{P0 P1 , P0 P1 , . . . P0 Pm }. Therefore, if the family {P0 , P1 , . . . , Pm }
is independent, then dim(P0 + P1 + . . . + Pm ) = m.

We leave the proof of this remark to the readers.

Theorem 1.2.3. Let α and β be two affine sets. If α ∩ β 6= ∅, then for every point P ∈ α and for
−→ − → −
every point Q ∈ β we have P Q ∈ → α + β.
−→ − → −
If P ∈ α and there exists a point Q ∈ β such that P Q ∈ →
α + β , then α ∩ β 6= ∅.

Proof. Suppose that α ∩ β 6= ∅. Let M ∈ α ∩ β. Then for every point P ∈ α and for every point
−−→ − −−→ → −
Q ∈ β, we have P M ∈ →α and M Q ∈ β . Therefore
−→ −−→ −−→ → →

P Q = P M + MQ ∈ −
α + β.

−→ →

Suppose that there exists a point P ∈ α and a point Q ∈ β such that P Q ∈ →

α + β . Since
−→ → →

PQ ∈ −α + β,
−→ → →

PQ = − u +→
−v ; where →

u ∈→−α,→

v ∈ β.
Then there exists uniquely a point M ∈ α and there exists uniquely a point N ∈ β such that
−−→ −−→ −→ −−→ −−→ −→ −−→ −−→ −−→
PM = → −
u and QN = −→ −v . Therefore, P Q = P M − QN or P Q + QN = P N = P M . Thus,
N ≡ M, i.e. α ∩ β 6= ∅.

Theorem 1.2.4. Suppose that α and β are two affine sets. We have:

1. if α ∩ β 6= ∅, then
dim(α + β) = dim α + dim β − dim(α ∩ β);

2. if α ∩ β = ∅, then


dim(α + β) = dim α + dim β − dim(→

α ∩ β ) + 1.

Proof. 1. If α ∩ β 6= ∅, then by Theorem 1.2.2, α ∩ β is an affine set with the directional space

− →

α ∩ β . Let P ∈ α ∩ β and let γ be the affine set passing through P with the directional


space →−γ =→ −α + β . It is obviously that α ⊂ γ and β ⊂ γ. Moreover, if γ 0 is an affine set
that contains both α and β, then P ∈ γ 0 and the directional space of γ 0 must contain →−α and

− 0
β . In other words, γ ⊂ γ . Thus, γ is the smallest affine set containing both α and β, i.e.
γ = α + β. Therefore


dim(α + β) = dim γ = dim →

γ = dim(→−
α + β)

− →

= dim →

α + dim β − dim(→ −
α ∩ β)
= dim α + dim β − dim(α ∩ β).

−→ − → −
2. Suppose that α∩β = ∅. Let P ∈ α and Q ∈ β, then by Theorem 1.2.3 we have P Q 6∈ → α+β.
−→ →
− − →

Let →

γ be the (1-dimensional) vector subspace generated by P Q, we have (→

α + β )∩→
γ = { 0 }.

7
Affine and Euclidean Geometry


− −
Let η be the affine set passing through P with the directional space → −
α + β +→ γ , it is clear
that α ⊂ η and β ⊂ η. Therefore, α + β ⊂ η.


Moreover, if η 0 is an affine set containing α and β, then P ∈ η 0 and the directional space η 0


of η 0 must contain →−
α , β and → −
γ . Therefore, η ⊂ η 0 . This implies that η is the smallest affine
set containing both α and β. In other words, η = α + β.


Since dim((→ −
α + β)∩→ −γ ) = 0, we have
dim(α + β) = dim η

− −
= dim(→
−α + β +→ γ)

− →

= dim →

α + dim β + dim → −
γ − dim(→

α ∩ β)


= dim α + dim β + 1 − dim(→

α ∩ β ).

1.3 Relative position

Definition 6. Two affine sets α and β are called intersecting or concurrent of order r if α ∩ β is


an r-plane. They are called skew of order r if α ∩ β = ∅ and dim(→ −
α ∩ β ) = r. They are called

− →

parallel if →

α ⊂ β or β ⊂ →−
α.
Example 4. In the usual space E3 .

1. Two intersecting lines (by high school math meaning) is the case of two 1-planes intersecting
of order 0. The sum (of them) is the 2-plane containing them.
2. Two intersecting plane (by high school math meaning) is the case of two 2-planes intersect
of order 1. The sum (of them) is nothing but E3 .
3. Two parallel lines (by high school math meaning) is the case of two 1-planes that are parallel.
They are also two 1-planes that are skew of order 1. The sum (of them) is the 2-plane
containing them.
4. Two parallel planes (by high school math meaning) is the case of 2-planes that are parallel.
They are also two 2-planes that are skew of order 2.
5. Two skew lines (by high school math meaning) is the case of two 1-planes that are skew of
order 0. The sum (of them) is E3 .
6. By Theorem 1.2.4, two planes in E3 can not be skew of order 0 or of order 1.
Theorem 1.3.1. Given two parallel affine sets α and β. If α ∩ β 6= ∅, then α ⊂ β or β ⊂ α.

Proof. Since α and β have a common point, their intersection α ∩ β is an affine set with the

− →
− →
− →

directional space →
−α ∩ β . Since α and β are parallel, →−
α ⊂ β or β ⊂ → −
α . If →

α ⊂ β , then


α ∩ β = α, i.e. α ⊂ β. If β ⊂ →
−α then α ∩ β = β, i.e β ⊂ α.

8
Affine and Euclidean Geometry

Figure 1.3: Two parallel planes or two skew Figure 1.4: A line is parallel to a plane or
(of order 2) planes. they are skew of order 1.

Figure 1.5: Two intersecting (of order 1) Figure 1.6: A line belongs to a plane or they
planes . are intersect of order 1.

Theorem 1.3.2. There exists one and only one an m-plane passing through a given point A and
parallel to a given m-plane α.

Proof. Let β be an m-plane passing through A with the directional space →−


α . Then β is an affine
0
set of dimension m parallel to α. If β is another m-plane passing through A and parallel to α,

− →

then β 0 = β (= →
−α ). Since β and β 0 have a common point, by Theorem 1.3.1 we have β ≡ β 0 .

Theorem 1.3.3. In An , let α be a hyperplane and β be an m-plane (1 ≤ m ≤ n − 1). Then α


and β are either parallel or intersecting of order (m − 1).

Proof. If β ⊂ α, then α and β parallel by the definition.

If β 6⊂ α, then α + β = A. We have two cases:

9
Affine and Euclidean Geometry

1. Case 1: α ∩ β 6= ∅. Applying the first formula in Theorem 1.2.4 we have

dim A = dim α + dim β − dim(α ∩ β),

or
n = n − 1 + m − dim(α ∩ β).
Therefore dim(α ∩ β) = m − 1.
Thus, α and β are intersecting of order (m − 1).
2. Case 2: α ∩ β = ∅. Applying the second formula in Theorem 1.2.4 we have


n = n − 1 + m + 1 − dim(→ −
α ∩ β ).

− →
− →
− →

Therefore, dim(→

α ∩ β ) = m. This implies →
−α ∩ β = β , or β ⊂ →−
α , i.e. α and β are paralel.

1.4 Affine frames-Equations of affine sets

1.4.1 Affine frames and coordinates

Definition 7. Let An be an affine space of dimension n. The system {O; → −e1 , →



e2 , . . . , →

en }, consisting

− →
− −
→n
a point O ∈ A and a basis { e1 , . . . , en } of A , is called an affine frame (or simply a frame) of An .
n

The point O is called the origin, the vector → −


ei is called the i-th basis vector, i = 1, 2, ..., n.

For short, we often write {O; →



ei }i=1,2,...,n or {O; →

ei } instead of {O; →

e1 , →

e2 , . . . , →

en }.
−−→ −

Suppose that {O; →

ei } is a frame of an affine space An . Then for every M ∈ An , OM ∈ An .
−−→
Therefore, OM can be expressed as a linear combination of vectors →

ei :
n
−−→ X →
OM = xi −
ei .
i=1

−−→
This means that the vector OM has coordinates (x1 , x2 , . . . , xn ) in the basis {→

ei }, xi ∈ K, i =
1, 2, . . . , n.

The number xi is called the i-th coordinate of M while the sequence of coordinates (x1 , x2 , . . . , xn ) ∈
Kn , or (xi ) for short, is called coordinates of M in (or relative to) the frame {O; →

ei }. To indicate a


point M that has coordinates (xi ) in the frame {O; ei }, we often use one of the following notations

M (x1 , x2 , . . . , xn )/{O;−

ei } or M (xi )/{O;−
ei } .

If there is no risk of confusion, we will write

M (x1 , x2 , . . . , xn ) or M (xi ).

10
Affine and Euclidean Geometry

e2 e2

O x
e1 e1

Figure 1.7: An affine frame and an affine coordinate system in A2 .

z
e3 e3
e2 e2

O x
e1 e1

Figure 1.8: An affine frame and an affine coordinate system in A3 .

Suppose that M has coordinates (xi ) and N has coordinates (yi ) in an affine frame {O; →

ei }. We
have n n n
−−→ −−→ −−→ X → − →
− (yi − xi )→

X X
M N = ON − OM = yi ei − xi ei = ei .
i=1 i=1 i=1
−−→
In other words, the vector M N has coordinates (yi − xi ) in the basis {→

ei }.

Thus, “coordinates of a vector are the differences between coordinates of the tail (initial point) and
coordinates of the head (terminal point)”.
Remark 3. 1. Let {O; →
−e } be a given affine frame in An . Consider the map
i

ϕ : A −→ Kn
M 7−→ (xi )

where (xi ) are coordinates of the point M in the affine frame {O; →

ei }. The map ϕ is bijective.
By this map, each point in A is identified with an element of Kn and therefore geometric
objects will be identified with algebraic ones.
2. Let {O; →

ei } be a frame in An and let Ei ∈ An , i = 1, ..., n be points such that
−−→
OE i = →−
ei .
Then the family of points {O, E1 , E2 , . . . , En } is affinely independent. Of course, a family
of n + 1 points {O, E1 , E2 , . . . , En } which is affinely independent determines an affine frame

11
Affine and Euclidean Geometry

−−→
{O; →

ei } where →

ei = OEi . Therefore we also call an independent family of n + 1 points in An
an affine frame. The notations {O; E1 , E2 , . . . , En } or {O; Ei }i=1,2,...,n or {O; Ei } are used to


indicate a frame with the origin O. By the definition, in the affine frame {O; Ei }, the origin
O has coordinates (0, 0, . . . , 0) and the point Ei has coordinates (0, . . . , 0, 1, 0, . . . , 0), where
the number 1 is located at i-th position.
3. A hyperplane containing n independent points O, E1 , E2 , . . . , Ei−1 , Ei+1 , . . . , En is called the
i-th coordinate plane . It is easy to see that a point M belongs to i-th coordinate plane if
and only if xi = 0, here xi stands for the i-th coordinate of M.

1.4.2 Change of coordinates

In affine space An , let {O; →



ei } and {O0 , →

ei 0 } be two distinct affine frames. A point M ∈ An will
have two different (tuple of) coordinates (xi ) and (x0i ), respectively. Suppose that
n
−−→0 X
OO = bi →

ei ,
i=1

and →−
e1 0 = c11 →

e1 + c21 →

e2 + . . . + cn1 →

en

→− →
− →
− →


e2 0

= c12 e1 + c22 e2 + . . . + cn2 en

 ...
→− = c1n →

e1 + c2n →

e2 + . . . + cnn →


e 0
n en
or →−
ej 0 = i=1 cij →

Pn
ei , j = 1, 2, . . . , n. The coordinates of the point M in these frames are (xi ) and
0
(xi ), respectively, i.e.
n n
−−→ X → −−→ X
OM = xi −
ei , O0 M = x0j →

ej 0 .
i=1 j=1
We have n n n

− −−→ −−→0 −− → X →
− x0j →

X X
0
xi ei = OM = OO + O M = bi ei + ej 0
i=1 i=1 j=1
n n n
bi →
− cij →

X X X
= ei + x0j ei
i=1 j=1 i=1
n n
cij x0j + bi )→

X X
= ( ei .
i=1 j=1
Therefore,
n
X
xi = cij x0j + bi , i = 1, 2, . . . , n. (1.1)
j=1
The above formula can be written in more detail as bellows
x1 = c11 x01 + c12 x02 + . . . + c1n x0n + b1



x = c x 0 + c x 0 + . . . + c x 0 + b

2 21 1 22 2 2n n 2
, (1.2)


 ...
xn = cn1 x01 + cn2 x02 + . . . + cnn x0n + bn

12
Affine and Euclidean Geometry

or in matrix form
[x] = C[x0 ] + [b], (1.3)
where      
x1 x01 b1
 x2   x0   b2 
0  2
[x] =  ..  , [x ] =  ..  , [b] =  .. 
   
. . .
xn x0n bn
and  
c11 c12. . . c1n
 c21 c22. . . c2n 
C =  .. ,
 
... . . .. 
 . . . 
cn1 cn2. . . cnn
the matrix representing the change of basis from { ei } to {→

− −
ei 0 } (C is invertible, i.e. det C 6= 0).

The formulars (1.1) (or (1.2), (1.3)) and the matrix C are called the change of coordinate formular
(or the change of frame) and the matrix of the change of coordinate (or the matrix of the change
of frame) from {O; →−
ei } to {O0 ; →

ei 0 }.
Example 5. Given an affine frame {O; → −
ei } in affine space A. Suppose that O0 is the point that
has coordinates (b ) in the frame {O; e } and →
i


i
−e 0=→ −
e +→
i

e + ... + →

e , i = 1, 2, . . . , n.
1 2 i

1. Since the matrix of the change of basis from {→−ei } to {→



ei } is the unit matrix, the change of

− 0 →

coordinate formula from {O; ei } to {O ; ei } has the form
xi = x0i + bi , i = 1, 2, . . . , n.

2. Since the matrix of the change of basis from {→−


ei } to {→

ei 0 } is the matrix
 
1 1 ... 1
0 1 . . . 1
C =  .. .. . . ..  ,
 
. . . .
0 0 ... 1
the change of coordinate formula from {O; →−e } to {O; →−
e 0 } has the form
i i

xi = x0i + x0i+1
+ ... + x0n ,
i = 1, 2, . . . , n,
the change of coordinate formula from {O; →

ei } to {O0 ; →

ei 0 } has the form
xi = x0i + x0i+1 + . . . + x0n + bi , i = 1, 2, . . . , n.

1.4.3 Equation of an m-plane

Parametric equations. Let An be an affine space of dimension n, {O; →



ei } be a given affine frame
and α be an m-plane passing through a point P, 0 < m < n. Suppose that
n
−→ X →
OP = bi −
ei ,
i=1

13
Affine and Euclidean Geometry

and {→

a1 , →

a2 , . . . , −
a→ →

m } is a basis of α , where

n

− aip →

X
ap = ei ; p = 1, 2, . . . , m.
i=1

−−→ −
The point M that has coordinates (xi ) in the frame {O; → −
ei } belongs to α if and only if P M ∈ →
α,
i.e. if and only if there exist tp ∈ K, p = 1, 2, . . . , m such that
m
−−→ X →
PM = tp −
ap .
p=1

We have n n n
−−→ −−→ −→ X →
xi − bi →
− (xi − bi )→

X X
P M = OM − OP = ei − ei = ei , (1.4)
i=1 i=1 i=1

and m m n n X m
−−→ X →
tp − aip →
− aip tp )→

X X X
PM = ap = tp ei = ( ei . (1.5)
p=1 p=1 i=1 i=1 p=1

Therefore by (1.4) and (1.5) we get


m
X
xi = aip tp + bi , i = 1, 2, . . . , n. (1.6)
p=1

The system (1.6) can be written in the explicit form




 x1 = a11 t1 + a12 t2 + . . . + a1m tm + b1

x
2 = a21 t1 + a22 t2 + . . . + a2m tm + b2
, (1.7)


 ...

xn = an1 t1 + an2 t2 + . . . + anm tm + bn

or in the matrix form


[x] = A[t] + [b], (1.8)
where      
x1 t1 b1
 x2   t2   b2 
[x] =  ..  , [t] =  ..  , [b] =  .. 
     
. . .
xn tm bn
and  
a11 a12 . . . a1m
 a21 a22 . . . a2m 
A =  .. ..  ,
 
.. ...
 . . . 
an1 an2 . . . anm
here A is a matrix of rank m.

14
Affine and Euclidean Geometry

The system (1.6), (1.7) and (1.8) can be written in the vector form

− →

x = t1 →

a1 + t 2 →

a2 + . . . + tm −
a→
m+ b (1.9)


where → −x is the vector with coordinates (x1 , x2 , . . . , xn ) and b is the vector with coordinates
(b1 , b2 , . . . , bn ).

The system (1.6) (or (1.7), (1.8), (1.9)) is called parametric equations of the m-plane α while the
elements tp , p = 1, 2, ..., m are called parameters.

Remark 4. 1. The map M ∈ α 7→ (t1 , t2 , ..., tm ) ∈ Km is a bijection from α to Km . Thus, each


point M ∈ α can be identified with a tuple of m numbers.

2. Since the family of vectors {→



a1 , →

a2 , . . . , →

ap } is linearly independent, the matrix A = (aip )n×m
is of rank m.

3. It is easy to see that a system of form (1.6) (or (1.7), (1.8), (1.9)) where the rank of the
matrix A = (aip )n×m is m, is an equation of an m-plane Pn passing through the point P that

− →
− →

has coordinates (b1 , b2 , . . . , bn ). Obviously, { ap ap = i=1 aip ei , p = 1, 2, . . . , m} is a basis of
the directional space of the m-plane → −
α.

Example 6. Parametric equations of a line α in an affine space of dimension n is

xi = ai t + bi , i = 1, 2, . . . , n.

Where t is the parameter, a1 , a2 , . . . , an , that are not all equal to zero, (a − 1, a − 2, . . . , an ) are
coordinates of the directional vector → −a of α, (b1 , b2 , . . . , bn ) are coordinates of the given point
P ∈ α while (xi ) are coordinates of an arbitrary point M ∈ α.

General equations. In an n-dimensional affine space An , let α be an m-plane and (1.7) be its
parametric equations. We can consider the system (1.7) as a system of n equations, m unknowns
t1 , t2 , . . . , tm and xi , i = 1, 2, . . . , n, being constants. Since the rank of the coefficient matrix
A = (aip )n×m is m, from (1.7) we can choose a subsystem consisting independent m equations
whose determinant is not zero. Without loss of generality, we can assume that they are the first
m equations. Solving this subsystem (Cramer system) we obtain solutions t1 , t2 , . . . , tm that are
expressed uniquely (therefore ti , i = 1, . . . , tm ) are unique) in terms of x1 , x2 , . . . , xm . Substituting
these into the remaining n − m equations we obtain a system of form
m
X
cij xj + xm+i + ci = 0, i = 1, 2, . . . , n − m. (1.10)
j=1

This system can be written as follows




 c11 x1 + . . . + c1m xm + xm+1 + c1 = 0

c x + . . . + c x + x
21 1 2m m m+2 + c2 = 0
. (1.11)


 ...

c(n−m)1 x1 + . . . + c(n−m)m xm + xn + cn−m = 0

15
Affine and Euclidean Geometry

The coefficient matrix of (1.10) is of rank n − m because the sub-determinant of order n − m


corresponding to xm+1 , xm+2 , . . . , xn are

1 0 0 ... 0

0 1 0 ... 0
.. = 1 6= 0.

.. .. .. ..
. . . . .

0 0 0 . . . 1

A point M is in α if and only if its has coordinates satisfying the above system of equations.
In brief, every m-plane in an affine space An can be characterized by a system of linear equations
of rank n − m.
We will prove the converse statement, every system of linear equations


 c11 x1 + . . . + c1n xn + c1 = 0

c x + . . . + c x + c = 0
21 1 2n n 2
, (1.12)


 . . .

c(n−m)1 x1 + . . . + c(n−m)n xn + cn−m = 0

whose coefficient matrix is of rank n − m will define an m-plane in An .


Indeed, since the coefficient matrix is of rank n − m, the system (1.12) has at least one so-
lution by Theorem Kronecker-Capelli. Let (b1 , b2 , . . . , bn ) be a solution of the system and let
(a1j , a2j , . . . , anj ), j = 1, 2, . . . , m be a basis for the solution set of the corresponding homogeneous
−→
system. Set P (b1 , b2 , . . . , bn ) ∈ An and → −
aj (a1j , a2j , . . . , anj ) ∈ An ; j = 1, 2, . . . , m. The family of
vectors {→ −
aj } is linearly independent and therefore generates an m-dimensional vector subspace → −α

→n →
− →

of A . Note that, coordinates of any vector u ∈ α is a solution of the corresponding homogeneous
system. Let α be the m-plane passing through P with the directional space → −
α . Since any solution
of a system of linear equations is the sum of a fixed solution of the system and a solution of the
corresponding homogeneous system, we can see that a point M (xi ) ∈ α if and only if (xi ) is a
−−→ −
solution of the system. Indeed, M ∈ α if and only if P M = → u (ui ) ∈ →

α . In terms of coordinates
we have
(xi ) − (bi ) = (ui ),
or
(xi ) = (bi ) + (ui ),
i.e. (xi ) is a solution of the system.
Thus, an m-plane is defined by a system of form (1.12) whose coefficient matrix is of rank n − m.
We call the system (1.12) a general equation of the m-plane.

Example 7. 1. A general equation of a hyperplane in An in a given affine frame is


n
X
ai xi + b = 0, (1.13)
i=1

where ai ∈ K, i = 1, 2, . . . , n are not all zero.


Thus, from a general equation of an m-plane, we se that an m-plane is just the intersection
of some n − m independent hyperplanes.

16
Affine and Euclidean Geometry

2. A general equation of the hyperplane passing through the point P (b1 , b2 , . . . , bn ) is


n
X
ai (xi − bi ) = 0, (1.14)
i=1

where ai ∈ K, i = 1, 2, . . . , n are not all zero.

3. In An with a given frame {O; → −


ei }, let A1 , . . . , An be n independent points. Suppose that
the coordinates of Ai is (a1i , . . . , ani ) in the given frame. Let α be the hyperplane passing
through A1 , A2 , . . . , An . Then a point M with coordinates (x1 , x2 , . . . , xn ) in the given frame
−−−→ −−−→
belongs to α if and only if the vector A1 M together with the vectors A1 Ai , i = 2, 3, . . . , n
form a linearly dependent family, i.e. if and only if

x1 − a11 x2 − a21 . . . xn − an1

a12 − a11 a22 − a21 . . . an2 − an1
= 0, (1.15)

.. .. . . ..
. . . .

a1n − a11 a2n − a21 . . . ann − an1

or
1 x1 x2 . . . xn

1 a11 a21 . . . an1
. = 0. (1.16)

.. .. .. ..
. . . . ..

1 a1n a2n . . . ann
Developing (1.16), we get a linear equation of order 1 with n unknowns x1 , x2 , ..., xn . This is
a general equation of the hyperplane α.

1.5 Barycenter. Single ratio

1.5.1 Barycenter

Definition 8. Let {P1 , P2 , . . . , Pm } ⊂ An be a family of distinct points on an affine space and let
{λ1 , λ2 , . . . , λm }, λi ∈ R, be a family of scalars that satisfies the following condition

λ := λ1 + λ2 + . . . + λm 6= 0.

Let I be an arbitrary point, then


1 −→ −−→
(λ1 IP1 + . . . + λm IPm )
λ
is a fixed vector. Therefore, there exists uniquely a point G such that
−→ 1 −→ −−→
IG = (λ1 IP1 + . . . + λm IPm ). (1.17)
λ
We call G the barycenter of the family of points {P1 , P2 , . . . , Pm } with respect to (w.r.t.) the family
of scalars {λ1 , λ2 , . . . , λm }.

17
Affine and Euclidean Geometry

Theorem 1.5.1. The point G is the barycenter of the family of points {P1 , P2 , . . . , Pm } w.r.t. the
family of scalars {λ1 , λ2 , . . . , λm } if and only if G satisfies the following equation
−−→ −−→ −−→ → −
λ1 GP1 + λ2 GP2 + . . . + λm GPm = 0 . (1.18)

Proof. Indeed,
−→ 1 −→ −−→
IG = (λ1 IP1 + . . . + λm IPm )
λ
−→ −→ −−→
⇔(λ1 + λ2 + . . . + λm )IG = λ1 IP1 + . . . + λm IPm
−→ −→ −→ −−→ →

⇔λ1 (GI + IP1 ) + . . . + λm (GI + IPm ) = 0
−−→ −−→ −−→ → −
⇔λ1 GP1 + λ2 GP2 + . . . + λm GPm = 0 .

By Theorem 1.5.1, we have two following corollaries.

Corollary 1.5.2. The definition of barycenter does not depend on the point I.

Corollary 1.5.3. The barycenter of the family of points {P1 , P2 , . . . , Pm } w.r.t. the family of
scalars {λ1 , λ2 , . . . , λm } is coincident with the barycenter of the family of points {P1 , P2 , . . . , Pm }
w.r.t. the family of scalars {kλ1 , kλ2 , . . . , kλm }, where k 6= 0.

Definition 9. The barycenter G of the family of points {P1 , P2 , . . . , Pm } w.r.t. the family of
scalars {λ1 , λ2 , . . . , λm }, where λ1 = λ2 = . . . = λm (by Corollary 1.5.3, we can choose λ1 = λ2 =
. . . = λm = 1) is called the center of mass of the given family of points.

Thus, if I is an arbitrary point, the center of mass G is determined by


m
−→ 1 X −→
IG = IPi , (1.19)
m i=1

or m
X −−→ → −
GPi = 0 . (1.20)
i=1

Remark 5. 1. The center of mass of two distinct points {P, Q} is the midpoint of the line
segment P Q.

2. The center of mass of three non-collinear points {A, B, C} is the centroid of the triangle
ABC.

We note that every family of m points has the center of mass if and only if the characteristic of
the field K is not m. If K = R or C, then every family of finite points has the center of mass.

The following theorem shows that we can use the notion barycenter to characterize affine sets.

Theorem 1.5.4. The set of all barycenters of the family of points {P0 , P1 , . . . , Pm } w.r.t. different
families of scalars is just the affine set α = P0 + P1 + . . . + Pm .

18
Affine and Euclidean Geometry

−−→ −−→ −−−→


Proof. We can see that → −
α is generated by the family of vectors {P0 P1 , P0 P2 , . . . , P0 Pm }. Assuming
−−→ −−→ −−→
that {P0 P1 , P0 P2 , . . . , P0 Pk }, is a basis of α, i.e. α is of dimension k. Suppose G ∈ α. This is
−−→ − −−→ P −−→
equivalent to P0 G ∈ → α or P0 G = ki=1 λi P0 Pi . We have
k k k
−−→ X −−→ −−→ X −−→ X −−→ → −
P0 G = λi (GPi − GP0 ) ⇔ (1 − λi )GP0 + λi GPi = 0 .
i=1 i=1 i=1

This equality proves that G is the barycenter of the family {P0 , P1 , . . . , Pm } w.r.t. the family of
scalars
Xk
{1 − λi , λ1 , λ2 , . . . , λk , 0, . . . , 0}.
i=1

Conversely, Suppose that G is the barycenter of the family {P0 , P1 , . . . , Pm } w.r.t. the family of
scalars {λ0 , λ1 , . . . , λm }. Then
m m
X −−→ → − X −−→ −−→ →

λi GPi = 0 ⇔ λi (GP0 + P0 Pi ) = 0
i=0 i=0
m m
X −−→ X −−→ → −
⇔ λi GP0 + λi P0 Pi = 0
i=0 i=1
m
−−→ 1 X −−→
⇔ P0 G = Pm λ i P0 Pi .
i=1 λi i=1

−−→ −
This equality proves that P0 G ∈ →
α , or G ∈ α.

Corollary 1.5.5. Given m + 1 points {P0 , P1 , . . . , Pm }. Let α = P0 + P1 + . . . + Pm P and I be an


m
arbitrary point. Then the point M ∈ α if and only if there exist µi , i = 0, 1, 2, . . . , m; i=0 µi = 1
such that m
−−→ X −−→
OM = µi OPi .
i=0

Moreover, if the family {P0 , P1 , . . . , Pm } is linearly independent, then the existences of µi are
unique.

Proof. By Theorem 1.5.4, M ∈ α if and only if M is the barycenter of the family {P0 , P1 , . . . , Pm }
λi
w.r.t some family of scalars {λ0 , λ1 , . . . , λm }. Set µi = Pm . By the definition of barycenter we
i=0 λi
have the first statement.

Suppose that the family {P0 , P1 , . . . , Pm } is linearly independent and


m m
−−→ X −−→ X
OM = µi OPi , where µi = 1.
i=0 i=0

−−→ −−→ −−→ −−→


We have OM = m
P Pm
i=0 µi OPi ⇔ P0 M = i=1 µi P0 Pi . This implies that µi , i = 1, 2, . . . , m are
uniquely determined and therefore µ0 is uniquely determined, too.

19
Affine and Euclidean Geometry

1.5.2 Single ratio

Definition 10. Let P, Q be two distinct points in A. A point M 6= Q belongs to the line P Q (the
−−→ −−→
affine set P + Q) if and only if there exists a real number k 6= 1 such that M P = k M Q. We call k
the single ratio of three points {P, Q, M }, and write k = (P QM ).
Remark 6. It is easy to see that if k = (P QM ) then M is the barycenter of {P, Q} w.r.t. {1, −k}.
If k = −1, then M is the midpoint of the segment P, Q.

Below are some properties of single ratio.


Proposition 1.5.6. In An , let A, B, C be three collinear distinct points. We have

1. (ABC) + (ACB) = 1,
2. (ABC).(BAC) = 1,
1
3. (BCA) = 1 − (ABC)
.

−→ −−→ −−→ −→ −−→


Proof. Suppose that (ABC) = k, i.e. CA = k CB or CB + BA = k CB. Therefore,
−→ −−→
BA = (1 − k)BC or (ACB) = 1 − k.
The readers can prove the remains easily. Note that because A, B, C are three distinct points,
k 6= 0, 1.

1.6 Convex sets in a real affine space

1.6.1 Convex sets

Segments. In a real affine space A, given two points P and Q. The set of all point M such that
−−→ −→ −→
IM = (1 − t)IP + tIQ, where I is an arbitrary poiny and 0 ≤ t ≤ 1, is call the segment P Q. Two
points P and Q are called the endpoints of the segment P Q.
Remark 7. 1. If P 6= Q, the by Corollary 1.5.5, the line P + Q is the set of all points M
satisfying:
−−→ −→ −→
OM = λOP + µOQ where λ + µ = 1.
This implies that the segment P Q is a subset of the line P + Q.
The point P is corresponding to λ = 1) while the point Q is corresponding to λ = 0.
2. If P ≡ Q, then the segment P Q is just one point P ≡ Q.
Definition 11. A set X in a real affine space A is called convex if for every P, Q ∈ X, the segment
P Q is contained in X.

Convex hull of a set. It is easy to see that the intersection of a non-empty family of convex sets
is a convex set. We define the convex hull of a set X is the intersection of all convex sets containing
X, i.e. the smallest convex set contains X.

20
Affine and Euclidean Geometry

Figure 1.9: Convex sets.

Figure 1.10: Non-convex sets.

1.6.2 Examples of convex sets

It is easy to see that every m-plane in a real affine space is a convex set. The followings, especially
in a space of dimension 1, 2, 3; are well known in high school mathematics.

A half space. A hyperplane α in a real affine space A divides A − α into two convex sets.

− −
Considering 1-dimensional quotient space A /→
α . Let [→

u ] be a non-zero vector in this quotient
space and let

− →
− −
p : A −→ A /→ α
be the canonical projection. Let I be a point in α, we have two sets
−−→
X = {M ∈ A : p(OM ) = λ[→

u ], λ > 0};
−−→
Y = {M ∈ A : p(OM ) = λ[→

u ], λ < 0}.
−−→
We will prove they are convex. Indeed, suppose that M, N ∈ X; p(OM ) = λ[→ −u ], λ > 0 and
−−→ →
− −→ −−→ −−→
p(ON ) = µ[ u ], µ > 0. Then for every P in the segment M N, OP = (1 − t)OM + tON where
0 ≤ t ≤ 1, we have
−→ −−→ −−→
p(OP ) = (1 − t)p(OM ) + tp(ON ) = ((1 − t)λ + tµ)[→

u ].

If 0 ≤ t ≤ 1, then (1 − t)λ + tµ > 0, i.e. P ∈ X. The proof for the case M, N ∈ Y is quite similar.
Remark 8. The proofs for following remarks are left for the readers.

1. The sets X and Y do not depend on the choice of the point I and the vector [→

u ].

21
Affine and Euclidean Geometry

2. Let P ∈ X. We can describe X and Y as follows: X is the set of all points M ∈ A − α α


while Y is the set of all points M ∈ A − α such that the segment P M intersects α at one
point.

m-dimensional parallelepiped. In An , given a point I and a linearly independent family of


vectors {→

e1 , →

e2 , ..., −
e→
m }. The set

m
−−→
xi →

X
n
H = {M ∈ A : OM = ei , 0 ≤ xi ≤ 1, i = 1, 2, ..., m }
i=1

is called an m-dimensional parallelepiped or m-parallelepiped. A points in H w.r.t. xi , where


xi = 0 or xi = 1, i = 1, 2, . . . , m, is called a vertex of the m-parallelepiped. There are 2m vertices.

By the definitions, 0-parallelepiped is a point, 1-parallelepiped is a segment. By the usual calling,


2-parallelepiped is called a parallelogram, 3-parallelepiped is called simply a parallelepiped.

We can define the notions of k-dimensional faces, of edges of an m-parallelepiped.

Theorem 1.6.1. Every m-parallelepiped is convex.

−−→ Pm − −−→
→ Pm → −
Proof. Suppose that M, N ∈ H, OM = i=1 xi ei , ON = i=1 yi ei ; 0 ≤ xi , yi ≤ 1, i =
1, 2, ..., m. Then for every P in the segment M N, we have
−→ −−→ −−→
OP = (1 − t)OM + tON
m m
xi →
− yi →

X X
= (1 − t) ei + t ei
i=1 i=1
m
((1 − t)xi + tyi )→

X
= ei .
i=1

It is easy to see that 0 ≤ (1 − t)xi + tyi ≤ 1, i.e. P ∈ H.

m-dimensional simplices. In An , given an affinely independent family of points {A0 , A1 , ..., Am }.


Let I ∈ An , then the set
m
−−→ X −→
n
C = { M ∈ A : IM = xi IAi , xi ≥ 0, x0 + x1 + x2 + · · · + xm = 1}
i=0

Is callled an m-dimensional simplex or m-simplex. The points A0 , A1 , ..., Am are called vertices of
C.

By the definition, 0-simplex is a point, 1-simplex is a segment. By the usual calling, 2-simplex is
called a triangle, 3-simplex is called a tetrahedron.

In an m-simplex C, every family of (k + 1) vertices (0 ≤ k ≤ m − 1) determines an k-simplex called


a k-dimensional face (or k-face) of C. The other (m − k) vertices determines an (m − k − 1)-simplex
called the opposite face of the above k-dimensional face.

22
Affine and Euclidean Geometry

Figure 1.11: m-simplex and m-parallelepiped

A 0-dimensional face is a vertex, a 1-dimensional face is called an edge of the simplex C.


From now on, the center of mass of the family {A0 , A1 , ..., Am } is also callled the center of mass of
the simplex.

Remark 9. 1. The definition of a simplex does not depend on the point I. Indeed, suppose
that xi ≥ 0, x0 + x1 + x2 + · · · + xm = 1, we have
m
−−→ X −→
IM = xi IAi
i=0
m

→0 −− 0
→ X −
→ −−→
⇔ II + I M = xi (II 0 + I 0 Ai )
i=0
m
−−→ X −−→
⇔ I 0M = xi I 0 Ai .
i=0

2. If I = A0 , then we have
m
−−−→ X −−−→
C = { M ∈ An : A0 M = xi A0 Ai , xi ≥ 0, x1 + x2 + · · · + xm ≤ 1}.
i=1

Theorem 1.6.2. Every m-simplex is convex

−−→ Pm −→ −→ Pm −→
Proof.
Pm Suppose
Pmthat M, N ∈ C, IM = i=0 x i IAi , IN = i=0 yi IAi ; xi , yi ≥ 0, i = 0, 1, 2, . . . , m;
x
i=0 i = 1, y
i=0 i = 1. Then for every P in the segment M N, we have
−→ −−→ −→
IP = (1 − t)IM + tIN
m m
X −→ X −→
= (1 − t) xi IAi + t yi IAi
i=0 i=0
m
X −→
= ((1 − t)xi + tyi )IAi .
i=0

23
Affine and Euclidean Geometry

Pm Pm Pm
It is easy to see that (1 − t)xi + tyi ≥ 0 and i=0 ((1 − t)xi + tyi ) = (1 − t) i=0 xi + t i=0 yi = 1,
i.e. P ∈ C. 2

EXERCISES CHAPTER 1

Exercise 1.1. Prove that C is a real affine space of dimension two.




Exercise 1.2. Let (A, A , Φ) be an n-dimensional affine space and B is a non-empty set. Prove
that if there exists a bijective map f : A −→ B, then one can define an affine structure on B, i.e.
B become an n-dimensional affine space (transfer affine structure from A to B by f ).

− →

Exercise 1.3. Let (A, A , Φ) and (A0 , A0 , Φ0 ) be two affine spaces and consider the map defined as
follows
− →
→ −
Φ × Φ0 : (A × A0 ) × (A × A0 ) −→ A × A0
−−→ −−−→
((M, M 0 ), (N, N 0 )) 7−→ (M N , M 0 N 0 ).
− →
→ −
Prove that (A × A0 , A × A0 , Φ × Φ0 ) is an affine space.

− →

Exercise 1.4. Let (A, A , Φ) be an affine space and →

α be a vector subspace of A . Two points
−−→ −
M, N ∈ A are called →

α -equivalent if M N ∈ →α.

1. Prove that the above relation is an equivalence relation.

2. Denote by A/−α the set of all equivalence classes and by [M ] the equivalence class containing

M. Consider the map


Φ−
→α : A/−α × A/−
→ →α −→ A /− →
α
−−→ .
([M ], [N ]) 7−→ [M N ]


Prove that (A/−
→α , A /−
α , Φ−
→ α ) is an affine space.

Exercise 1.5. Let A be an affine space and O be a point in A. Then the map that maps a point
−−→ →

M ∈ A to the vector OM ∈ A is bijective. Thank to this map, one can transfer the vector


structure on A to A. Construct explicit operators on A such that A is a vector space.

Exercise 1.6. In An , let α and α0 be two (different) hyperplanes that are parallel. Let β be an
m-plane not containing in α (β 6⊂ α). Prove that if β intersects α then β intersects α0 . Is the
above result also true in the case of α and α0 are parallel m-planes? Think about similar results
that you’ve studied in high school.


Exercise 1.7. Let A be a vector space with canonical affine structure. Prove that every vector


subspace of A is an m-plane. Is the contrary also true? Give some examples.

Exercise 1.8. Prove that there exists one and only one (m + 1)-plane passing through a given
point A and a given m-plane α that does not contain the point. Think about similar results that
you’ve studied in high school.

24
Affine and Euclidean Geometry

Exercise 1.9. Prove that if the affine sets α and β are parallel to the affine set γ, then α ∩ β, if
non-empty, is an affine set parallel to γ. Think about similar results that you’ve studied in high
school.
Exercise 1.10. Prove that, if two distinct hyperplanes α and β intersect, the hyperplane γ is
parallel to α ∩ β such that the intersections α ∩ γ and β ∩ γ are non-empty, then α ∩ γ is parallel
to β ∩ γ. Think about similar results that you’ve studied in high school.
Exercise 1.11. In An , consider the relative position between a line and an m-plane. Consider the
cases when n = 2, 3, 4.
Exercise 1.12. Let α be an m-plane, A be a point that is not in α. Then, there exists a unique
m-plane β passing through A, parallel to α and α ∩ β = ∅.

1. How many are there l-planes β, l < m, passing through A and parallel to α. Give some
remarks on α ∩ β.

2. How many are there l-planes β, l > m, passing through A and parallel to α. Give some
remarks on α ∩ β.
Exercise 1.13. Let α and β be two affine sets of dimensions p and q, respectively. Prove that α
and β are parallel if and only if their intersection is of order r or they are skew of order r, where
r = min{p, q}.
Exercise 1.14. Given a system of linear equations with m equations and n unknowns, where
m < n. 

 a1 =a11 x1 + a12 x2 + · · · + a1n xn

 a =a x + a x + · · · + a x
2 21 1 22 2 2n n
.

 ······

am =am1 x1 + am2 x2 + · · · + amn xn

Suppose that rank(aij ) = m, prove that:

1. the solution set of the corresponding homogeneous system is a vector subspace of Rn ;

2. the solution set of the above system is an (m − n)-plane in the affine space Rn (endowed the
canonical affine structure) with directional space →

α.
Exercise 1.15. Given p + 1 points {M0 , M1 , . . . , Mp } in an affine space An . Prove that

1. dim(M0 + M1 + . . . + Mp ) ≤ p,

2. {M0 , M1 , . . . , Mp } is affine independent if and only if dim(M0 + M1 + . . . + Mp ) = p,

3. if {M0 , M1 , . . . , Mp } is affine independent, then

M0 + M1 + . . . + Mp = (M0 + M1 + . . . + Mk ) + (Mk+1 + Mk+2 + . . . + Mp )

and
(M0 + M1 + . . . + Mk ) ∩ (Mk+1 + Mk+2 + . . . + Mp ) = ∅.

25
Affine and Euclidean Geometry

Exercise 1.16. In An with a given affine frame, given parametric equations of an m-plane α. Find
parametric equations of an m-plane β parallel to α?
Exercise 1.17. Given a general equation of an m-plane α. Prove that the solution set of the
corresponding homogeneous system determines the directional space →

α of α.
Exercise 1.18. Write parametric equations and a general equation of the m-plane E0 +E1 +. . .+Em
−−−→
and of the hyperplane E1 + E2 + . . . + En , where {E0 ; E0 Ei } is a given affine frame of An .
Exercise 1.19. In An , n is odd, let {O; →

ei } be an affine frame. Let E ∈ An be a point such that
−−→ →
OE = −
e1 + →

e2 + . . . + →

en .

Find the change of coordinates formula from the given frame to the frame

{E; →

e1 + →

e2 , →

e2 + →

e3 , . . . , →

en + →

e1 }.

Exercise 1.20. In A3 with an affine frame {O; →



ei } (Frame (1)), consider the following points

A0 (1, 1, 1), A1 (2, 0, 0), A2 (1, 0, 0), A3 (1, 1, 0);

A00 (0, 0, 0), A01 (1, 1, 0), A02 (2, 0, 1), A03 (1, 0, 1).

−−−→ −−−→ −−−→ −−−→ −−−→ −−−→


1. Prove that {A0 ; A0 A1 , A0 A2 , A0 A3 } and {A00 ; A00 A01 , A00 A02 , A00 A03 } are affine frames in A3
(Frame (2) and Frame (3)).

2. Find the change of coordinates formula from Frame (1) to Frame (2) and from Frame (2) to
Frame (3).
Exercise 1.21. In an affine space An with a given affine frame, describe the intersection between
the line and the plane determined by the following equations
x 1 − b1 x 2 − b2 xn − b n
= = ... =
a1 a2 an
and n
X
ci xi + d = 0.
i=1

Exercise 1.22. Let α be an m-plane and A be a point not belonging to α. Prove that there exists
one and only one (m + 1)-plane passing through A and containing α.
Exercise 1.23. In A4 , write general equations of the affine sets whose dimensions are smallest (in
each cace)



1. passing through A(1, 2, 1, 1) and parallel to two vectors →

a (0, 1, 2, 0), b (1, 1, 0, 0);


2. passing through M (1, 0, 1, 0) and parallel to three vectors →

a (1, 0, 1, 0), b (2, 1, 2, 1), →
−c (4, 1, 4, 1);

3. passing through two points A(1, 1, 1, 1), B(2, 3, 1, 0) and parallel to two vectors →

a (0, 1, 1, 1)


and b (1, 2, 0, 0);

26
Affine and Euclidean Geometry

4. passing through two points A(1, 1, 1, 1), B(2, 3, 1, 0) and parallel to two vectors →

a (3, 4, 2, 1)


and b (2, 2, −2, 2);

5. passing through three points A(2, 1, 2, 1), B(1, 1, 1, 1), C(2, 0, 2, 0) and parallel to two vectors

− →

a (2, 3, 1, 4) and b (0, 0, 0, 1).

Write equations of affine sets (having the same dimensions) passing through O(0, 0, 0, 0) and par-
allel to the above affine sets, respectively.

Exercise 1.24. In A4 , write a parametric equation of the affine set that has the following general
equation 
x1 +x2 +2x3 +x4 −1 = 0
−x1 +2x2 −x4 +2 = 0.

Exercise 1.25. In A4 , given 4 points A(3, 1, 1, 2), B(0, 1, 0, 0), C(3, 2, 3, 2), D(1, 0, 0, 1).

1. Find intersection points between AB and the coordinate hyperplanes.

2. Consider the relative position between two lines AB and CD.

Exercise 1.26. In A4 , consider the relative position between two affine sets α and β, whose
equations are as follows:

1. 
2x1 +x2 +x3 +x4 −1 = 0
α: ,
−x1 +x2 −x4 +2 = 0

x1 −x2 −x4 +1 = 0
β: .
−2x1 +x2 +2x3 +x4 +3 = 0

2.
α: 2x1 − x2 − x3 + x4 − 2 = 0,


 x1 = 1
x2 = 1 + t

β: , t ∈ R.

 x3 = 1 + t
x4 = 1 + 2t

3. 
3x1 +x2 +x3 +2x4 −6 = 0
α: ,
x1 +x4 +1 = 0


 x1 = 2 + t
x2 = 1 − t

β: , t ∈ R.

 x3 = 1 − t
x4 = −1 + 2t

Exercise 1.27. Given a family of points {Pi : i ∈ I} in an affine space A (I 6= ∅). Let α be the
affine hull of the family of points.

1. Prove that α is just the set of barycenters of non-empty sub-families of the given family.

27
Affine and Euclidean Geometry

2. Suppose that I is finite. Prove that the family {Pi : i ∈ I} is affinely independent if and
only if for every M ∈ α, there exists uniquely (up to a non-zero constant) a family of scalars
{λi : i ∈ I} such that M is the barycenter of the family {Pi : i ∈ I} with respect to (w.r.t.)
that family of scalars. The such family {λi : i ∈ I} is called barycenter coordinates of M
(w.r.t. the family {Pi : i ∈ I}).
Exercise 1.28 (Thales Theorem). In An given three distinct m-planes that are parallel. Two lines
d and d0 intersect, respectively, these m-planes at A, B, C and A0 , B 0 , C 0 . Prove that

(ABC) = (A0 B 0 C 0 ).

Exercise 1.29. In An given three distinct hyperplanes that all contain an (n − 2)-plane. Two
parallel lines d and d0 intersect, respectively, these hyperplanes at A, B, C and A0 , B 0 , C 0 . Prove
that (ABC) = (A0 B 0 C 0 )
Exercise 1.30 (Ceva Theorem, Menelaüs Theorem). In an affine space A, let A, B, C be three
points that are not collinear and three points P, Q, R that satisfy: P ∈ BC, Q ∈ CA, R ∈ AB,
and A 6= P, B 6= Q, C 6= R. Prove that:

1. Three lines AP, BQ, CR are parallel or concurrent if and only if

(BCP ).(CAQ).(ABR) = −1.

2. Three points P, Q, R are collinear if and only if

(BCP ).(CAQ).(ABR) = 1.
Exercise 1.31. Prove that a simplex is just the convex hull of the set of all its vertices and
contained in the affine hull (of the set of all its vertices).
Exercise 1.32. Given an m-simplex whose vertices are {P0 , P1 , . . . , Pm }.

1. Prove that the affine hulls of two opposite faces are skew.

2. Consider the line connecting a vertex and the center of mass of the opposite (m − 1)-face.
Prove that these lines are concurrent at a point G.
Consider some special cases: 2-simplex, 3-simplex.

3. Let G0 and G00 be centers of mass of a pair of opposite faces. Calculate (G0 G00 G).
Consider some special cases: m = 2, 3.
Exercise 1.33. 1. Prove that the convex hull of a family of finite points {P1 , . . . , Pm } in a real
affine space An is the set of all barycenters of that family w.r.t. the family of same-sign
scalars {λ1 , . . . , λm }.

2. Prove that the convex hull of a set S in a real affine space An is the set of all barycenters of
finite sub-families of points {P1 , . . . , Pm } ⊂ S w.r.t. the family of same-sign scalars
{λ1 , . . . , λm }.

28

You might also like