This is page 6

Printer: Opaque this

2
Basics of Affine Geometry

L’algèbre n’est qu’une géométrie écrite; la géométrie n’est qu’une

algèbre figurée.
—Sophie Germain

2.1 Affine Spaces

Geometrically, curves and surfaces are usually considered to be sets of
points with some special properties, living in a space consisting of “points.”
Typically, one is also interested in geometric properties invariant under cer-
tain transformations, for example, translations, rotations, projections, etc.
One could model the space of points as a vector space, but this is not very
satisfactory for a number of reasons. One reason is that the point corre-
sponding to the zero vector (0), called the origin, plays a special role, when
there is really no reason to have a privileged origin. Another reason is that
certain notions, such as parallelism, are handled in an awkward manner.
But the deeper reason is that vector spaces and affine spaces really have dif-
ferent geometries. The geometric properties of a vector space are invariant
under the group of bijective linear maps, whereas the geometric properties
of an affine space are invariant under the group of bijective affine maps,
and these two groups are not isomorphic. Roughly speaking, there are more
affine maps than linear maps.
Affine spaces provide a better framework for doing geometry. In particu-
lar, it is possible to deal with points, curves, surfaces, etc., in an intrinsic
 2.1. Affine Spaces 7

manner, that is, independently of any specific choice of a coordinate sys-

tem. As in physics, this is highly desirable to really understand what is
going on. Of course, coordinate systems have to be chosen to finally carry
out computations, but one should learn to resist the temptation to resort
to coordinate systems until it is really necessary.
Affine spaces are the right framework for dealing with motions, trajec-
tories, and physical forces, among other things. Thus, affine geometry is
crucial to a clean presentation of kinematics, dynamics, and other parts of
physics (for example, elasticity). After all, a rigid motion is an affine map,
but not a linear map in general. Also, given an m×n matrix A and a vector
b ∈ Rm , the set U = {x ∈ Rn | Ax = b} of solutions of the system Ax = b
is an affine space, but not a vector space (linear space) in general.
Use coordinate systems only when needed!
This chapter proceeds as follows. We take advantage of the fact that
almost every affine concept is the counterpart of some concept in linear
algebra. We begin by defining affine spaces, stressing the physical interpre-
tation of the definition in terms of points (particles) and vectors (forces).
Corresponding to linear combinations of vectors, we define affine combina-
tions of points (barycenters), realizing that we are forced to restrict our
attention to families of scalars adding up to 1. Corresponding to linear
subspaces, we introduce affine subspaces as subsets closed under affine
combinations. Then, we characterize affine subspaces in terms of certain
vector spaces called their directions. This allows us to define a clean no-
tion of parallelism. Next, corresponding to linear independence and bases,
we define affine independence and affine frames. We also define convexity.
Corresponding to linear maps, we define affine maps as maps preserving
affine combinations. We show that every affine map is completely defined
by the image of one point and a linear map. Then, we investigate briefly
some simple affine maps, the translations and the central dilatations. At
this point, we give a glimpse of affine geometry. We prove the theorems of
Thales, Pappus, and Desargues. After this, the definition of affine hyper-
planes in terms of affine forms is reviewed. The section ends with a closer
look at the intersection of affine subspaces.
Our presentation of affine geometry is far from being comprehensive,
and it is biased toward the algorithmic geometry of curves and surfaces.
For more details, the reader is referred to Pedoe [136], Snapper and Troyer
[160], Berger [12, 13], Coxeter [35], Samuel [146], Tisseron [169], and Hilbert
and Cohn-Vossen [84].
Suppose we have a particle moving in 3D space and that we want to
describe the trajectory of this particle. If one looks up a good textbook
on dynamics, such as Greenwood [79], one finds out that the particle is
modeled as a point, and that the position of this point x is determined
with respect to a “frame” in R3 by a vector. Curiously, the notion of a
frame is rarely defined precisely, but it is easy to infer that a frame is a
8 2. Basics of Affine Geometry

ab

a
O

Figure 2.1. Points and free vectors

pair (O, (e1 , e2 , e3 )) consisting of an origin O (which is a point) together

with a basis of three vectors (e1 , e2 , e3 ). For example, the standard frame
in R3 has origin O = (0, 0, 0) and the basis of three vectors e1 = (1, 0, 0),
e2 = (0, 1, 0), and e3 = (0, 0, 1). The position of a point x is then defined
by the “unique vector” from O to x.
But wait a minute, this definition seems to be defining frames and the
position of a point without defining what a point is! Well, let us identify
points with elements of R3 . If so, given any two points a = (a1 , a2 , a3 ) and
b = (b1 , b2 , b3 ), there is a unique free vector , denoted by ab, from a to b,
the vector ab = (b1 − a1 , b2 − a2 , b3 − a3 ). Note that
b = a + ab,
addition being understood as addition in R3 . Then, in the standard frame,
given a point x = (x1 , x2 , x3 ), the position of x is the vector Ox =
(x1 , x2 , x3 ), which coincides with the point itself. In the standard frame,
points and vectors are identified. Points and free vectors are illustrated in
Figure 2.1.
What if we pick a frame with a different origin, say Ω = (ω1 , ω2 , ω3 ), but
the same basis vectors (e1 , e2 , e3 )? This time, the point x = (x1 , x2 , x3 ) is
defined by two position vectors:

Ox = (x1 , x2 , x3 )
in the frame (O, (e1 , e2 , e3 )) and

Ωx = (x1 − ω1 , x2 − ω2 , x3 − ω3 )
in the frame (Ω, (e1 , e2 , e3 )).
 2.1. Affine Spaces 9

This is because
Ox = OΩ + Ωx and OΩ = (ω1 , ω2 , ω3 ).
We note that in the second frame (Ω, (e1 , e2 , e3 )), points and position vec-
tors are no longer identified. This gives us evidence that points are not
vectors. It may be computationally convenient to deal with points using
position vectors, but such a treatment is not frame invariant, which has
undesirable effets.
Inspired by physics, we deem it important to define points and properties
of points that are frame invariant. An undesirable side effect of the present
approach shows up if we attempt to define linear combinations of points.
First, let us review the notion of linear combination of vectors. Given two
vectors u and v of coordinates (u1 , u2 , u3 ) and (v1 , v2 , v3 ) with respect
to the basis (e1 , e2 , e3 ), for any two scalars λ, µ, we can define the linear
combination λu + µv as the vector of coordinates
(λu1 + µv1 , λu2 + µv2 , λu3 + µv3 ).
If we choose a different basis (e01 , e02 , e03 ) and if the matrix P expressing the
vectors (e01 , e02 , e03 ) over the basis (e1 , e2 , e3 ) is
 
a 1 b1 c 1
P =  a 2 b2 c 2  ,
a 3 b3 c 3
which means that the columns of P are the coordinates of the e0j over the
basis (e1 , e2 , e3 ), since
u1 e1 + u2 e2 + u3 e3 = u01 e01 + u02 e02 + u03 e03
and
v1 e1 + v2 e2 + v3 e3 = v10 e01 + v20 e02 + v30 e03 ,
it is easy to see that the coordinates (u1 , u2 , u3 ) and (v1 , v2 , v3 ) of u and v
with respect to the basis (e1 , e2 , e3 ) are given in terms of the coordinates
(u01 , u02 , u03 ) and (v10 , v20 , v30 ) of u and v with respect to the basis (e01 , e02 , e03 )
by the matrix equations
   0    0
u1 u1 v1 v1
 u2  = P  u02  and  v2  = P  v20  .
u3 u03 v3 v30
From the above, we get
 0
v10
     
u1 u1 v1
 u02  = P −1  u2  and  v20  = P −1  v2  ,
u03 u3 v30 v3
and by linearity, the coordinates
(λu01 + µv10 , λu02 + µv20 , λu03 + µv30 )
10 2. Basics of Affine Geometry

of λu + µv with respect to the basis (e01 , e02 , e03 ) are given by

 0
λu1 + µv10
      
u1 v1 λu1 + µv1
 λu02 + µv20  = λP −1  u2  + µP −1  v2  = P −1  λu2 + µv2  .
λu03 + µv30 u3 v3 λu3 + µv3
Everything worked out because the change of basis does not involve a
change of origin. On the other hand, if we consider the change of frame
from the frame (O, (e1 , e2 , e3 )) to the frame (Ω, (e1 , e2 , e3 )), where OΩ =
(ω1 , ω2 , ω3 ), given two points a, b of coordinates (a1 , a2 , a3 ) and (b1 , b2 , b3 )
with respect to the frame (O, (e1 , e2 , e3 )) and of coordinates (a01 , a02 , a03 ) and
(b01 , b02 , b03 ) with respect to the frame (Ω, (e1 , e2 , e3 )), since
(a01 , a02 , a03 ) = (a1 − ω1 , a2 − ω2 , a3 − ω3 )
and
(b01 , b02 , b03 ) = (b1 − ω1 , b2 − ω2 , b3 − ω3 ),
the coordinates of λa + µb with respect to the frame (O, (e1 , e2 , e3 )) are
(λa1 + µb1 , λa2 + µb2 , λa3 + µb3 ),
but the coordinates
(λa01 + µb01 , λa02 + µb02 , λa03 + µb03 )
of λa + µb with respect to the frame (Ω, (e1 , e2 , e3 )) are
(λa1 + µb1 − (λ + µ)ω1 , λa2 + µb2 − (λ + µ)ω2 , λa3 + µb3 − (λ + µ)ω3 ),
which are different from
(λa1 + µb1 − ω1 , λa2 + µb2 − ω2 , λa3 + µb3 − ω3 ),
unless λ + µ = 1.
Thus, we have discovered a major difference between vectors and points:
The notion of linear combination of vectors is basis independent, but the
notion of linear combination of points is frame dependent. In order to sal-
vage the notion of linear combination of points, some restriction is needed:
The scalar coefficients must add up to 1.
A clean way to handle the problem of frame invariance and to deal
with points in a more intrinsic manner is to make a clearer distinction
between points and vectors. We duplicate R3 into two copies, the first
copy corresponding to points, where we forget the vector space structure,
and the second copy corresponding to free vectors, where the vector space
structure is important. Furthermore, we make explicit the important fact
that the vector space R3 acts on the set of points R3 : Given any point
a = (a1 , a2 , a3 ) and any vector v = (v1 , v2 , v3 ), we obtain the point
a + v = (a1 + v1 , a2 + v2 , a3 + v3 ),
 2.1. Affine Spaces 11

which can be thought of as the result of translating a to b using the vector

v. We can imagine that v is placed such that its origin coincides with a and
that its tip coincides with b. This action +: R3 × R3 → R3 satisfies some
crucial properties. For example,
a + 0 = a,
(a + u) + v = a + (u + v),
and for any two points a, b, there is a unique free vector ab such that
b = a + ab.
It turns out that the above properties, although trivial in the case of R3 ,
are all that is needed to define the abstract notion of affine space (or affine
−
→
structure). The basic idea is to consider two (distinct) sets E and E , where
−
→
E is a set of points (with no structure) and E is a vector space (of free
vectors) acting on the set E.
Did you say “A fine space”?
−→
Intuitively, we can think of the elements of E as forces moving the points
in E, considered as physical particles. The effect of applying a force (free
−→
vector) u ∈ E to a point a ∈ E is a translation. By this, we mean that
−
→
for every force u ∈ E , the action of the force u is to “move” every point
a ∈ E to the point a + u ∈ E obtained by the translation corresponding
to u viewed as a vector. Since translations can be composed, it is natural
−
→
that E is a vector space.
For simplicity, it is assumed that all vector spaces under consideration
are defined over the field R of real numbers. Most of the definitions and
results also hold for an arbitrary field K, although some care is needed when
dealing with fields of characteristic different from zero (see the problems).
It is also assumed that all families (λi )i∈I of scalars have finite support.
Recall that a family (λi )i∈I of scalars has finite support if λi = 0 for all
i ∈ I − J, where J is a finite subset of I. Obviously, finite families of
scalars have finite support, and for simplicity, the reader may assume that
all families of scalars are finite. The formal definition of an affine space is
as follows.
Definition 2.1.1 An affine space is either the degenerate space reduced
­ − → ®
to the empty set, or a triple E, E , + consisting of a nonempty set E (of
−
→
points), a vector space E (of translations, or free vectors), and an action
−
→
+: E × E → E, satisfying the following conditions.
(A1) a + 0 = a, for every a ∈ E.
−
→
(A2) (a + u) + v = a + (u + v), for every a ∈ E, and every u, v ∈ E .
12 2. Basics of Affine Geometry
−
→
E E

b=a+u
u
a c=a+w w

Figure 2.2. Intuitive picture of an affine space

−
→
(A3) For any two points a, b ∈ E, there is a unique u ∈ E such that
a + u = b.
−
→
The unique vector u ∈ E such that a + u = b is denoted by ab, or
→
−
sometimes by ab, or even by b − a. Thus, we also write
b = a + ab
→
−
(or b = a + ab, or even b = a + (b − a)).
­ − → ® −→
The dimension of the affine space E, E , + is the dimension dim( E )
−
→
of the vector space E . For simplicity, it is denoted by dim(E).
−
→
Conditions (A1) and (A2) say that the (abelian) group E acts on E,
−
→
and condition (A3) says that E acts transitively and faithfully on E. Note
that
a(a + v) = v
−
→
for all a ∈ E and all v ∈ E , since a(a + v) is the unique vector such that
a + v = a + a(a + v). Thus, b = a + v is equivalent to ab = v. Figure
2.2 gives an intuitive picture of an affine space. It is natural to think of all
vectors as having the same origin, the null vector.
­ − → ®
The axioms defining an affine space E, E , + can be interpreted intu-
−
→
itively as saying that E and E are two different ways of looking at the
same object, but wearing different sets of glasses, the second set of glasses
depending on the choice of an “origin” in E. Indeed, we can choose to
look at the points in E, forgetting that every pair (a, b) of points defines
−→
a unique vector ab in E , or we can choose to look at the vectors u in
−
→
E , forgetting the points in E. Furthermore, if we also pick any point a in
E, a point that can be viewed as an origin in E, then we can recover all
 2.1. Affine Spaces 13

−
→
the points in E as the translated points a + u for all u ∈ E . This can be
−
→
formalized by defining two maps between E and E .
−→
For every a ∈ E, consider the mapping from E to E given by
u 7→ a + u,
−
→ −
→
where u ∈ E , and consider the mapping from E to E given by
b 7→ ab,
where b ∈ E. The composition of the first mapping with the second is
u 7→ a + u 7→ a(a + u),
which, in view of (A3), yields u. The composition of the second with the
first mapping is
b 7→ ab 7→ a + ab,
which, in view of (A3), yields b. Thus, these compositions are the identity
−→ −→
from E to E and the identity from E to E, and the mappings are both
bijections.
−
→
When we identify E with E via the mapping b 7→ ab, we say that we
consider E as the vector space obtained by taking a as the origin in E, and
­ − → ®
we denote it by Ea . Thus, an affine space E, E , + is a way of defining a
vector space structure on a set of points E, without making a commitment
to a fixed origin in E. Nevertheless, as soon as we commit to an origin a in
E, we can view E as the vector space Ea . However, we urge the reader to
−
→
think of E as a physical set of points and of E as a set of forces acting on
E, rather than reducing E to some isomorphic copy of Rn . After all, points
are points, and not vectors! For notational simplicity, we will often denote
­ − → ® −→ −
→
an affine space E, E , + by (E, E ), or even by E. The vector space E is
called the vector space associated with E.
Ä One should be careful about the overloading of the addition sym-
bol +. Addition is well-defined on vectors, as in u+v; the translate
−→
a + u of a point a ∈ E by a vector u ∈ E is also well-defined, but addition
of points a + b does not make sense. In this respect, the notation b − a
for the unique vector u such that b = a + u is somewhat confusing, since it
suggests that points can be subtracted (but not added!). Yet, we will see
in Section 4.1 that it is possible to make sense of linear combinations of
points, and even mixed linear combinations of points and vectors.

−
→
Any vector space E has an affine space structure specified by choosing
−→ −
→
E = E , and letting + be addition in the vector space E . We will refer
­−
→ − → ® −
→
to the affine structure E , E , + on a vector space E as the canonical
14 2. Basics of Affine Geometry

−
→
(or natural) affine structure on­ E . In particular,
® the vector space Rn can
n n n
be viewed as the affine space R ­ ,nR , n+ , ®denoted by A . In general, if
K is any field, the affine space K , K , + is denoted by AnK . In order
to distinguish between the double role played by members of Rn , points
and vectors, we will denote points by row vectors, and vectors by column
vectors. Thus, the action of the vector space Rn over the set Rn simply
viewed as a set of points is given by
u1
 
.. 
(a1 , . . . , an ) +  . = (a1 + u1 , . . . , an + un ).
un
We will also use the convention that if x = (x1 , . . . , xn ) ∈ Rn , then the
column vector associated with x is denoted by x (in boldface notation).
Abusing the notation slightly, if a ∈ Rn is a point, we also write a ∈ An .
The affine space An is called the real affine space of dimension n. In most
cases, we will consider n = 1, 2, 3.

2.2 Examples of Affine Spaces

Let us now give an example of an affine space that is not given as a vector
space (at least, not in an obvious fashion). Consider the subset L of A 2
consisting of all points (x, y) satisfying the equation
x + y − 1 = 0.
The set L is the line of slope −1 passing through the points (1, 0) and (0, 1)
shown in Figure 2.3.
The line L can be made into an official affine space by defining the action
+: L × R → L of R on L defined such that for every point (x, 1 − x) on L
and any u ∈ R,
(x, 1 − x) + u = (x + u, 1 − x − u).
It is immediately verified that this action makes L into an affine space. For
example, for any two points a = (a1 , 1 − a1 ) and b = (b1 , 1 − b1 ) on L,
the unique (vector) u ∈ R such that b = a + u is u = b1 − a1 . Note that
the vector space R is isomorphic to the line of equation x + y = 0 passing
through the origin.
Similarly, consider the subset H of A3 consisting of all points (x, y, z)
satisfying the equation
x + y + z − 1 = 0.
The set H is the plane passing through the points (1, 0, 0), (0, 1, 0), and
(0, 0, 1). The plane H can be made into an official affine space by defining
 2.2. Examples of Affine Spaces 15

Figure 2.3. An affine space: the line of equation x + y − 1 = 0

the action +: H × R2 → H of R 2
µ on¶ H defined such that for every point
u
(x, y, 1 − x − y) on H and any ∈ R2 ,
v
µ ¶
u
(x, y, 1 − x − y) + = (x + u, y + v, 1 − x − u − y − v).
v

For a slightly wilder example, consider the subset P of A3 consisting of all

points (x, y, z) satisfying the equation

x2 + y 2 − z = 0.

The set P is a paraboloid of revolution, with axis Oz. The surface P can
be made into an official affine space by defining the action +: P × R2 → P
of R¶2 on P defined such that for every point (x, y, x2 + y 2 ) on P and any
µ
u
∈ R2 ,
v
µ ¶
2 2 u
(x, y, x + y ) + = (x + u, y + v, (x + u)2 + (y + v)2 ).
v

This should dispell any idea that affine spaces are dull. Affine spaces not
already equipped with an obvious vector space structure arise in projective
geometry. Indeed, we will see in Section 5.1 that the complement of a
hyperplane in a projective space has an affine structure.
16 2. Basics of Affine Geometry
−
→
E E

b
ab
a c ac

bc

Figure 2.4. Points and corresponding vectors in affine geometry

2.3 Chasles’s Identity

Given any three points a, b, c ∈ E, since c = a + ac, b = a + ab, and
c = b + bc, we get
c = b + bc = (a + ab) + bc = a + (ab + bc)
by (A2), and thus, by (A3),
ab + bc = ac,
which is known as Chasles’s identity, and illustrated in Figure 2.4.
Since a = a + aa and by (A1) a = a + 0, by (A3) we get
aa = 0.
Thus, letting a = c in Chasles’s identity, we get
ba = −ab.
Given any four points a, b, c, d ∈ E, since by Chasles’s identity
ab + bc = ad + dc = ac,
we have the parallelogram law
ab = dc iff bc = ad.

2.4 Affine Combinations, Barycenters

A fundamental concept in linear algebra is that of a linear combination.
The corresponding concept in affine geometry is that of an affine com-
bination, also called a barycenter . However, there is a problem with the
naive approach involving a coordinate system, as we saw in Section 2.1.
Since this problem is the reason for introducing affine combinations, at the
 2.4. Affine Combinations, Barycenters 17

risk of boring certain readers, we give another example showing what goes
wrong if we are not careful in defining linear combinations of points.
Consider R2 as an affine space, under µ ¶its natural
µ ¶coordinate system with
1 0
origin O = (0, 0) and basis vectors and . Given any two points
0 1
a = (a1 , a2 ) and b = (b1 , b2 ), it is natural to define the affine combination
λa + µb as the point of coordinates
(λa1 + µb1 , λa2 + µb2 ).
Thus, when a = (−1, −1) and b = (2, 2), the point a + b is the point
c = (1, 1).
Let us now consider the new coordinate system with respect to the origin
c = (1, 1) (and the same basis vectors). This time, the coordinates of a are
(−2, −2), the coordinates of b are (1, 1), and the point a + b is the point d
of coordinates (−1, −1). However, it is clear that the point d is identical to
the origin O = (0, 0) of the first coordinate system.
Thus, a + b corresponds to two different points depending on which
coordinate system is used for its computation!
This shows that some extra condition is needed in order for affine com-
binations to make sense. It turns out that if the scalars sum up to 1, the
definition is intrinsic, as the following lemma shows.
Lemma 2.4.1 Given an affine space E, let (ai )i∈I be a family of points in
E, and let (λi )i∈I be a family of scalars. For any two points a, b ∈ E, the
following properties hold:
P
(1) If i∈I λi = 1, then
X X
a+ λi aai = b + λi bai .
i∈I i∈I
P
(2) If i∈I λi = 0, then
X X
λi aai = λi bai .
i∈I i∈I

Proof . (1) By Chasles’s identity (see Section 2.3), we have

X X
a+ λi aai = a + λi (ab + bai )
i∈I i∈I
µX ¶ X
=a+ λi ab + λi bai
i∈I i∈I
X P
= a + ab + λi bai since i∈I λi = 1
i∈I
X
=b+ λi bai since b = a + ab.
i∈I
18 2. Basics of Affine Geometry

(2) We also have

X X
λi aai = λi (ab + bai )
i∈I i∈I
µX ¶ X
= λi ab + λi bai
i∈I i∈I
X
= λi bai ,
i∈I
P
since i∈I λi = 0.

Thus, by Lemma 2.4.1, for any P family of points (ai )i∈I in E, for any
family (λi )i∈I of scalars such that i∈I λi = 1, the point
X
x=a+ λi aai
i∈I

is independent of the choice of the origin a ∈ E. This property motivates

the following definition.

Definition 2.4.2 For any P family of points (ai )i∈I in E, for any family
(λi )i∈I of scalars such that i∈I λi = 1, and for any a ∈ E, the point
X
a+ λi aai
i∈I

(which is independent of a ∈ E, by Lemma 2.4.1) is called the barycenter (or

barycentric combination, or affine combination) of the points ai assigned
the weights λi , and it is denoted by
X
λi ai .
i∈I

In dealing with barycenters, it is convenient to introduce the notion of

a weighted point, which is just a pair (a, λ), where a ∈ E is a point, and
λ ∈ RP is a scalar. Then, given a family of weighted
P points ((ai , λi ))i∈I ,
where i∈I λi = 1, we also say that the point i∈I λi ai is the barycenter
of the family of weighted points ((ai , λi ))i∈I .
Note that the barycenter x of the family of weighted points ((ai , λi ))i∈I
is the unique point such that
X
ax = λi aai for every a ∈ E,
i∈I

and setting a = x, the point x is the unique point such that

X
λi xai = 0.
i∈I
 2.4. Affine Combinations, Barycenters 19

In physical terms, the barycenter is the center of mass of the family of

weighted
P points ((ai , λi ))i∈I (where the masses have been normalized, so
that i∈I λi = 1, and negative masses are allowed).

Remarks:

(1) Since the barycenter of a family ((ai , λi ))i∈I of weighted points is

defined
P for families (λi )i∈I of scalars with finite support (and such
that i∈I λi = 1), we might as well assume that I is finite. Then,
for all m ≥ 2, it is easy to prove that the barycenter of m weighted
points can be obtained by repeated computations of barycenters of
two weighted points.

(2) This result still holds, provided that the field K has at least three
distinct elements, but the proof is trickier!
P P
(3) When i∈I λi = 0, the vector P i∈I λi aai does not depend on the
point a, and we may denote it by i∈I λi ai . This observation will be
used in Section 4.1 to define a vector space in which linear combina-
tions
P of both points and vectors make sense, regardless of the value
of i∈I λi .

Figure 2.5 illustrates the geometric

¢ ¡ construction¡ of¢ the barycenters g1
and g2 of the weighted points a, 41 , b, 41 , and c, 21 , and (a, −1), (b, 1),
¡ ¢

and (c, 1).

The point g1 can be constructed geometrically as the middle of the
segment joining c to the middle 21 a + 21 b of the segment (a, b), since
µ ¶
1 1 1 1
g1 = a + b + c.
2 2 2 2
The point g2 can be constructed geometrically as the point such that the
middle 21 b + 12 c of the segment (b, c) is the middle of the segment (a, g2 ),
since
µ ¶
1 1
g2 = −a + 2 b + c .
2 2

Later on, we will see that a polynomial curve can be defined as a set of
barycenters of a fixed number of points. For example, let (a, b, c, d) be a
sequence of points in A2 . Observe that

(1 − t)3 + 3t(1 − t)2 + 3t2 (1 − t) + t3 = 1,

since the sum on the left-hand side is obtained by expanding (t+(1−t)) 3 = 1

using the binomial formula. Thus,

(1 − t)3 a + 3t(1 − t)2 b + 3t2 (1 − t) c + t3 d

20 2. Basics of Affine Geometry
c

g1

a b

c
g2

a b

Figure 2.5. Barycenters, g1 = 14 a + 14 b + 12 c, g2 = −a + b + c

is a well-defined affine combination. Then, we can define the curve F : A →

A2 such that
F (t) = (1 − t)3 a + 3t(1 − t)2 b + 3t2 (1 − t) c + t3 d.
Such a curve is called a Bézier curve, and (a, b, c, d) are called its control
points. Note that the curve passes through a and d, but generally not
through b and c. We will see in Chapter 18 how any point F (t) on the curve
can be constructed using an algorithm performing affine interpolation steps
(the de Casteljau algorithm).

2.5 Affine Subspaces

In linear algebra, a (linear) subspace can be characterized as a nonempty
subset of a vector space closed under linear combinations. In affine spaces,
the notion corresponding to the notion of (linear) subspace is the notion of
affine subspace. It is natural to define an affine subspace as a subset of an
affine space closed under affine combinations.
­ − → ®
Definition 2.5.1 Given an affine space E, E , + , a subset V of E is
­ −→ ®
an affine subspace (of E,P E , + ) if for every family of
Pweighted points
((ai , λi ))i∈I in V such that i∈I λi = 1, the barycenter i∈I λi ai belongs
to V .
 2.5. Affine Subspaces 21

An affine subspace is also called a flat by some authors. According to

Definition 2.5.1, the empty set is trivially an affine subspace, and every
intersection of affine subspaces is an affine subspace.
As an example, consider the subset U of R2 defined by
U = (x, y) ∈ R2 | ax + by = c ,
© ª

i.e., the set of solutions of the equation

ax + by = c,
where it is assumed that a 6= 0 or b 6= 0. Given any m points (xi , yi ) ∈ U
and any m scalars λi such that λ1 + · · · + λm = 1, we claim that
m
X
λi (xi , yi ) ∈ U.
i=1

Indeed, (xi , yi ) ∈ U means that

axi + byi = c,
and if we multiply both sides of this equation by λi and add up the resulting
m equations, we get
m
X m
X
(λi axi + λi byi ) = λi c,
i=1 i=1

and since λ1 + · · · + λm = 1, we get

Ãm ! Ãm ! Ãm !
X X X
a λi x i + b λi yi = λi c = c,
i=1 i=1 i=1

which shows that

à m m
! m
X X X
λi x i , λi yi = λi (xi , yi ) ∈ U.
i=1 i=1 i=1

Thus, U is an affine subspace of A2 . In fact, it is just a usual line in A2 .

It turns out that U is closely related to the subset of R2 defined by
−→ ©
U = (x, y) ∈ R2 | ax + by = 0 ,
ª

i.e., the set of solutions of the homogeneous equation

ax + by = 0
obtained by setting the right-hand side of ax + by = c to zero. Indeed, for
any m scalars λi , the same calculation as above yields that
m
X −
→
λi (xi , yi ) ∈ U ,
i=1
22 2. Basics of Affine Geometry

−
→
U

Figure 2.6. An affine line U and its direction

this time without any restriction on the λi , since the right-hand side
−
→ −
→
of the equation is null. Thus, U is a subspace of R2 . In fact, U is one-
dimensional, and it is just a usual line in R2 . This line can be identified
with a line passing through the origin of A2 , a line that is parallel to the
line U of equation ax + by = c, as illustrated in Figure 2.6.
Now, if (x0 , y0 ) is any point in U , we claim that
−
→
U = (x0 , y0 ) + U ,

where
−
→ n −
→o
(x0 , y0 ) + U = (x0 + u1 , y0 + u2 ) | (u1 , u2 ) ∈ U .

−
→
First, (x0 , y0 ) + U ⊆ U , since ax0 + by0 = c and au1 + bu2 = 0 for all
−
→
(u1 , u2 ) ∈ U . Second, if (x, y) ∈ U , then ax + by = c, and since we also
have ax0 + by0 = c, by subtraction, we get

a(x − x0 ) + b(y − y0 ) = 0,
−
→ −
→
which shows that (x − x0 , y − y0 ) ∈ U , and thus (x, y) ∈ (x0 , y0 ) + U .
−
→ −
→
Hence, we also have U ⊆ (x0 , y0 ) + U , and U = (x0 , y0 ) + U .
The above example shows that the affine line U defined by the equation

ax + by = c
 2.5. Affine Subspaces 23

−
→
is obtained by “translating” the parallel line U of equation
ax + by = 0
passing through the origin. In fact, given any point (x0 , y0 ) ∈ U ,
−
→
U = (x0 , y0 ) + U .
More generally, it is easy to prove the following fact. Given any m × n
matrix A and any vector b ∈ Rm , the subset U of Rn defined by
U = {x ∈ Rn | Ax = b}
is an affine subspace of An .
Actually, observe that Ax = b should really be written as Ax> = b, to be
consistent with our convention that points are represented by row vectors.
We can also use the boldface notation for column vectors, in which case
the equation is written as Ax = b. For the sake of minimizing the amount
of notation, we stick to the simpler (yet incorrect) notation Ax = b. If we
consider the corresponding homogeneous equation Ax = 0, the set
−
→
U = {x ∈ Rn | Ax = 0}
is a subspace of Rn , and for any x0 ∈ U , we have
−
→
U = x0 + U .
This is a general situation. Affine subspaces can be characterized in terms
−→
of subspaces of E . Let V be a nonempty subset of E. For every family
(a1 , . . . , an ) in V , for any family (λ1 , . . . , λn ) of scalars, and for every point
a ∈ V , observe that for every x ∈ E,
n
X
x=a+ λi aai
i=1

is the barycenter of the family of weighted points

µ ³ n
X ´¶
(a1 , λ1 ), . . . , (an , λn ), a, 1 − λi ,
i=1

since
n
X ³ n
X ´
λi + 1 − λi = 1.
i=1 i=1

−
→ −
→ −
→
Given any point a ∈ E and any subset V of E , let a + V denote the
following subset of E:
−
→ n −
→o
a+ V = a+v | v ∈ V .
24 2. Basics of Affine Geometry
−
→
E E

−
→
a V
−
→
V =a+ V

−
→
Figure 2.7. An affine subspace V and its direction V

­ − → ®
Lemma 2.5.2 Let E, E , + be an affine space.
(1) A nonempty subset V of E is an affine subspace iff for every point
a ∈ V , the set
−
→
Va = {ax | x ∈ V }
−
→ −
→
is a subspace of E . Consequently, V = a + Va . Furthermore,
−
→
V = {xy | x, y ∈ V }
−
→ −
→ − → −
→
is a subspace of E and Va = V for all a ∈ E. Thus, V = a + V .
−
→ −
→ −
→
(2) For any subspace V of E and for any a ∈ E, the set V = a + V is
an affine subspace.
Proof . The proof is straightforward, and is omitted. It is also given in
Gallier [70].

In particular, when E is the natural affine space associated with a vector

−
→
space E , Lemma 2.5.2 shows that every affine subspace of E is of the form
−
→ −
→ − → −
→
u+ U , for a subspace U of E . The subspaces of E are the affine subspaces
of E that contain 0.
−
→
The subspace V associated with an affine subspace V is called the
−
→
direction of V . It is also clear that the map +: V × V → V induced
−
→ ­ −→ ®
by +: E × E → E confers to V, V , + an affine structure. Figure 2.7
illustrates the notion of affine subspace.
−
→
By the dimension of the subspace V , we mean the dimension of V .
An affine subspace of dimension 1 is called a line, and an affine subspace
of dimension 2 is called a plane.
 2.5. Affine Subspaces 25

An affine subspace of codimension 1 is called a hyperplane (recall that a

subspace F of a vector space E has codimension 1 iff there is some subspace
G of dimension 1 such that E = F ⊕ G, the direct sum of F and G, see
Strang [166] or Lang [107]).
We say that two affine subspaces U and V are parallel if their directions
−
→ − → −
→
are identical. Equivalently, since U = V , we have U = a + U and V =
−
→
b + U for any a ∈ U and any b ∈ V , and thus V is obtained from U by the
translation ab.
In general, when we talk about n points a1 , . . . , an , we mean the sequence
(a1 , . . . , an ), and not the set {a1 , . . . , an } (the ai ’s need not be distinct).
By Lemma 2.5.2, a line is specified by a point a ∈ E and a nonzero vector
−
→
v ∈ E , i.e., a line is the set of all points of the form a + λv, for λ ∈ R.
We say that three points a, b, c are collinear if the vectors ab and ac
are linearly dependent. If two of the points a, b, c are distinct, say a 6= b,
then there is a unique λ ∈ R such that ac = λab, and we define the ratio
ac
ab = λ.
A plane is specified by a point a ∈ E and two linearly independent vectors
−→
u, v ∈ E , i.e., a plane is the set of all points of the form a + λu + µv, for
λ, µ ∈ R.
We say that four points a, b, c, d are coplanar if the vectors ab, ac, and
ad are linearly dependent. Hyperplanes will be characterized a little later.
­ − → ®
Lemma 2.5.3 Given an affine spacePE, E , + , for any P family (ai )i∈I of
points in E, the set V of barycenters i∈I λi ai (where i∈I λi = 1) is the
smallest affine subspace containing (ai )i∈I .
P
Proof . If (ai )i∈I is empty, then V = ∅, because of the condition i∈I λi =
1. If (ai )i∈I is nonempty, then the smallest affine P subspace containing
(ai )i∈I must contain the set V of barycenters i∈I λi ai , and thus, it
is enough to show that V is closed under affine combinations, which is
immediately verified.

Given a nonempty subset S of E, the smallest affine subspace of E gen-

erated by S is often denoted by hSi. For example, a line specified by two
distinct points a and b is denoted by ha, bi, or even (a, b), and similarly for
planes, etc.

Remarks:
(1) Since it can be shown that the barycenter of n weighted points can
be obtained by repeated computations of barycenters of two weighted
points, a nonempty subset V of E is an affine subspace iff for every
two points a, b ∈ V , the set V contains all barycentric combinations of
a and b. If V contains at least two points, then V is an affine subspace
iff for any two distinct points a, b ∈ V , the set V contains the line
26 2. Basics of Affine Geometry

determined by a and b, that is, the set of all points (1 − λ)a + λb,
λ ∈ R.
(2) This result still holds if the field K has at least three distinct elements,
but the proof is trickier!

2.6 Affine Independence and Affine Frames

Corresponding to the notion of linear independence in vector spaces, we
have the notion of affine independence. Given a family (ai )i∈I of points
in an affine space E, we will reduce the notion of (affine) independence of
these points to the (linear) independence of the families (ai aj )j∈(I−{i}) of
vectors obtained by choosing any ai as an origin. First, the following lemma
shows that it is sufficient to consider only one of these families.
­ − → ®
Lemma 2.6.1 Given an affine space E, E , + , let (ai )i∈I be a family of
points in E. If the family (ai aj )j∈(I−{i}) is linearly independent for some
i ∈ I, then (ai aj )j∈(I−{i}) is linearly independent for every i ∈ I.
Proof . Assume that the family (ai aj )j∈(I−{i}) is linearly independent for
some specific i ∈ I. Let k ∈ I with k 6= i, and assume that there are some
scalars (λj )j∈(I−{k}) such that
X
λj ak aj = 0.
j∈(I−{k})

Since
ak aj = a k ai + a i aj ,
we have
X X X
λ j ak aj = λ j ak ai + λ j ai aj ,
j∈(I−{k}) j∈(I−{k}) j∈(I−{k})
X X
= λ j ak ai + λ j ai aj ,
j∈(I−{k}) j∈(I−{i,k})
X µ X ¶
= λ j ai aj − λ j ai ak ,
j∈(I−{i,k}) j∈(I−{k})

and thus
X µ X ¶
λ j ai aj − λj ai ak = 0.
j∈(I−{i,k}) j∈(I−{k})

Since the family (ai aj )j∈(I−{i})

P is linearly independent, we must have λj =
0 for all j ∈ (I − {i, k}) and j∈(I−{k}) λj = 0, which implies that λj = 0
for all j ∈ (I − {k}).
 2.6. Affine Independence and Affine Frames 27

We define affine independence as follows.

­ −→ ®
Definition 2.6.2 Given an affine space E, E , + , a family (ai )i∈I of
points in E is affinely independent if the family (ai aj )j∈(I−{i}) is linearly
independent for some i ∈ I.

Definition 2.6.2 is reasonable, since by Lemma 2.6.1, the independence

of the family (ai aj )j∈(I−{i}) does not depend on the choice of ai . A crucial
property of linearly independent vectors (u1 , . . . , um ) is that if a vector v
is a linear combination
m
X
v= λi ui
i=1

of the ui , then the λi are unique. A similar result holds for affinely
independent points.
­ − → ®
Lemma 2.6.3 Given an affine space E, E , + , let (a0 , .P . . , am ) be a fam-
m
ily
Pm of m + 1 points in E. Let x ∈ E, and assume that x = i=0Pλmi ai , where
λ
i=0 i = 1. Then, the family (λ 0 , . . . , λ m ) such that x = i=0 λi ai is
unique iff the family (a0 a1 , . . . , a0 am ) is linearly independent.

Proof . The proof is straightforward and is omitted. It is also given in

Gallier [70].

Lemma 2.6.3 suggests the notion of affine frame. Affine frames are the
­ − → ®
affine analogues of bases in vector spaces. Let E, E , + be a nonempty
affine space, and let (a0 , . . . , am ) be a family of m + 1 points in E. The
family (a0 , . . . , am ) determines the family of m vectors (a0 a1 , . . . , a0 am ) in
−
→
E . Conversely, given a point a0 in E and a family of m vectors (u1 , . . . , um )
−→
in E , we obtain the family of m + 1 points (a0 , . . . , am ) in E, where ai =
a0 + ui , 1 ≤ i ≤ m.
Thus, for any m ≥ 1, it is equivalent to consider a family of m + 1 points
(a0 , . . . , am ) in E, and a pair (a0 , (u1 , . . . , um )), where the ui are vectors
−→
in E . Figure 2.8 illustrates the notion of affine independence.

Remark: The above observation also applies to infinite families (ai )i∈I of
−
→
points in E and families (−
→)
u i i∈I−{0} of vectors in E , provided that the
index set I contains 0.

−
→
When (a0 a1 , . . . , a0 am ) is a basis of E then, for every x ∈ E, since
x = a0 + a0 x, there is a unique family (x1 , . . . , xm ) of scalars such that

x = a 0 + x 1 a0 a1 + · · · + x m a0 am .
28 2. Basics of Affine Geometry
−
→
E E

a2

a0 a2

a0 a1 a0 a1

Figure 2.8. Affine independence and linear independence

The scalars (x1 , . . . , xm ) may be considered as coordinates with respect to

(a0 , (a0 a1 , . . . , a0 am )). Since
m
à m
! m
X X X
x = a0 + xi a0 ai iff x = 1 − x i a0 + x i ai ,
i=1 i=1 i=1

x ∈ E can also be expressed uniquely as

m
X
x= λi ai
i=0
Pm Pm
with i=0 λi = 1, and where λ0 = 1 − i=1 xi , and λi = xi for 1 ≤ i ≤ m.
The scalars (λ0 , . . . , λm ) are also certain kinds of coordinates with respect
to (a0 , . . . , am ). All this is summarized in the following definition.
­ −→ ®
Definition 2.6.4 Given an affine space E, E , + , an affine frame with
origin a0 is a family (a0 , . . . , am ) of m + 1 points in E such that the list of
−
→
vectors (a0 a1 , . . . , a0 am ) is a basis of E . The pair (a0 , (a0 a1 , . . . , a0 am ))
is also called an affine frame with origin a0 . Then, every x ∈ E can be
expressed as
x = a 0 + x 1 a0 a1 + · · · + x m a0 am
for a unique family (x1 , . . . , xm ) of scalars, called the coordinates of x w.r.t.
the affine frame (a0 , (a0 a1 , . . ., a0 am )). Furthermore, every x ∈ E can be
written as
x = λ 0 a0 + · · · + λ m am
for some unique family (λ0 , . . . , λm ) of scalars such that λ0 + · · · + λm =
1 called the barycentric coordinates of x with respect to the affine frame
(a0 , . . . , am ).
 2.6. Affine Independence and Affine Frames 29

The coordinates (x1 , . . . , xm ) and the barycentric

Pm coordinates (λ0 , . . .,
λm ) are related by the equations λ0 = 1 − i=1 xi and λi = xi , for 1 ≤
i ≤ m. An affine frame is called an affine basis by some authors. A family
(ai )i∈I of points in E is affinely dependent if it is not affinely independent.
We can also characterize affinely dependent families as follows.
­ − → ®
Lemma 2.6.5 Given an affine space E, E , + , let (ai )i∈I be a family of
points in E. The family (ai )i∈I is affinely
P dependent iff there
P is a family
(λi )i∈I such that λj 6= 0 for some j ∈ I, i∈I λi = 0, and i∈I λi xai = 0
for every x ∈ E.
Proof . By Lemma 2.6.3, the family (ai )i∈I is affinely dependent iff the
family of vectors (ai aj )j∈(I−{i}) is linearly dependent for some i ∈ I. For
any i ∈ I, the family (ai aj )j∈(I−{i}) is linearly dependent iff there is a
family (λj )j∈(I−{i}) such that λj 6= 0 for some j, and such that
X
λj ai aj = 0.
j∈(I−{i})

Then, for any x ∈ E, we have

X X
λ j ai aj = λj (xaj − xai )
j∈(I−{i}) j∈(I−{i})
X µ X ¶
= λj xaj − λj xai ,
j∈(I−{i}) j∈(I−{i})
³P ´ P
and letting λi = − j∈(I−{i}) λj , we get i∈I λi xai = 0, with
P
i∈I λi = 0 and λj 6= 0 for some j ∈ I. ThePconverse is obvious by
setting x = ai for some i such that λi 6= 0, since i∈I λi = 0 implies that
λj 6= 0, for some j 6= i.

Even though Lemma 2.6.5 is rather dull, it is one of the key ingredi-
ents in the proof of beautiful and deep theorems about convex sets, such
as Carathéodory’s theorem, Radon’s theorem, and Helly’s theorem (see
Section 3.1).
A family of two points (a, b) in E is affinely independent iff ab 6= 0, iff
a 6= b. If a 6= b, the affine subspace generated by a and b is the set of all
points (1 − λ)a + λb, which is the unique line passing through a and b. A
family of three points (a, b, c) in E is affinely independent iff ab and ac
are linearly independent, which means that a, b, and c are not on the same
line (they are not collinear). In this case, the affine subspace generated by
(a, b, c) is the set of all points (1 − λ − µ)a + λb + µc, which is the unique
plane containing a, b, and c. A family of four points (a, b, c, d) in E is affinely
independent iff ab, ac, and ad are linearly independent, which means that
a, b, c, and d are not in the same plane (they are not coplanar). In this
30 2. Basics of Affine Geometry
a2

a0 a0 a1

a3

a0 a1 a0 a2

a1
Figure 2.9. Examples of affine frames and their convex hulls

case, a, b, c, and d are the vertices of a tetrahedron. Figure 2.9 shows affine
frames and their convex hulls for |I| = 0, 1, 2, 3.
Given n+1 affinely independent points (a0 , . . . , an ) in E, we can consider
the set of points λ0 a0 +· · · +λn an , where λ0 +· · ·+λn = 1 and λi ≥ 0 (λi ∈
R). Such affine combinations are called convex combinations. This set is
called the convex hull of (a0 , . . . , an ) (or n-simplex spanned by (a0 , . . . , an )).
When n = 1, we get the segment between a0 and a1 , including a0 and a1 .
When n = 2, we get the interior of the triangle whose vertices are a0 , a1 , a2 ,
including boundary points (the edges). When n = 3, we get the interior of
the tetrahedron whose vertices are a0 , a1 , a2 , a3 , including boundary points
(faces and edges). The set
{a0 + λ1 a0 a1 + · · · + λn a0 an | where 0 ≤ λi ≤ 1 (λi ∈ R)}
is called the parallelotope spanned by (a0 , . . . , an ). When E has dimension
2, a parallelotope is also called a parallelogram, and when E has dimension
3, a parallelepiped .
More generally, we say that a subset V of E is convex if for any two
points a, b ∈ V , we have c ∈ V for every point c = (1 − λ)a + λb, with
0 ≤ λ ≤ 1 (λ ∈ R).
Ä Points are not vectors! The following example illustrates why
treating points as vectors may cause problems. Let a, b, c be three
 2.7. Affine Maps 31

affinely independent points in A3 . Any point x in the plane (a, b, c) can be

expressed as
x = λ0 a + λ1 b + λ2 c,
where λ0 + λ1 + λ2 = 1. How can we compute λ0 , λ1 , λ2 ? Letting a =
(a1 , a2 , a3 ), b = (b1 , b2 , b3 ), c = (c1 , c2 , c3 ), and x = (x1 , x2 , x3 ) be the
coordinates of a, b, c, x in the standard frame of A3 , it is tempting to solve
the system of equations
    
a1 b1 c1 λ0 x1
 a2 b2 c2   λ1  =  x 2  .
a3 b3 c3 λ2 x3
However, there is a problem when the origin of the coordinate system be-
longs to the plane (a, b, c), since in this case, the matrix is not invertible!
What we should really be doing is to solve the system
λ0 Oa + λ1 Ob + λ2 Oc = Ox,
where O is any point not in the plane (a, b, c). An alternative is to use
certain well-chosen cross products.
It can be shown that barycentric coordinates correspond to various ratios
of areas and volumes; see the problems.

2.7 Affine Maps

Corresponding to linear maps we have the notion of an affine map. An
affine map is defined as a map preserving affine combinations.
­ − → ® −
→
Definition 2.7.1 Given two affine spaces E, E , + and E 0 , E 0 , +0 , a
­ ®

function f : E → E 0 is an affinePmap iff for every family ((ai , λi ))i∈I of

weighted points in E such that i∈I λi = 1, we have
à !
X X
f λi ai = λi f (ai ).
i∈I i∈I

In other words, f preserves barycenters.

Affine maps can be obtained from linear maps as follows. For simplicity
of notation, the same symbol + is used for both affine spaces (instead of
using both + and +0 ).
−
→ −
→
Given any point a ∈ E, any point b ∈ E 0 , and any linear map h: E → E 0 ,
we claim that the map f : E → E 0 defined such that
f (a + v) = b + h(v)
32 2. Basics of Affine Geometry
P
is an affine map. Indeed, for any family (λi )i∈I of scalars with i∈I λi = 1
and any family (− →
vi )i∈I , since
X X X
λi (a + vi ) = a + λi a(a + vi ) = a + λ i vi
i∈I i∈I i∈I

and
X X X
λi (b + h(vi )) = b + λi b(b + h(vi )) = b + λi h(vi ),
i∈I i∈I i∈I

we have
à ! à !
X X
f λi (a + vi ) =f a+ λ i vi
i∈I i∈I
à !
X
=b+h λ i vi
i∈I
X
=b+ λi h(vi )
i∈I
X
= λi (b + h(vi ))
i∈I
X
= λi f (a + vi ).
i∈I

P
Note that the condition i∈I λi = 1 was implicitly used (in a hidden
call to Lemma 2.4.1) in deriving that
X X
λi (a + vi ) = a + λ i vi
i∈I i∈I

and
X X
λi (b + h(vi )) = b + λi h(vi ).
i∈I i∈I

As a more concrete example, the map

µ ¶ µ ¶µ ¶ µ ¶
x1 1 2 x1 3
7→ +
x2 0 1 x2 1
defines an affine map in A2 . It is a “shear” followed by a translation. The
effect of this shear on the square (a, b, c, d) is shown in Figure 2.10. The
image of the square (a, b, c, d) is the parallelogram (a0 , b0 , c0 , d0 ).
Let us consider one more example. The map
µ ¶ µ ¶µ ¶ µ ¶
x1 1 1 x1 3
7→ +
x2 1 3 x2 0
 2.7. Affine Maps 33
0 0
d c

d c

a0 b0

a b
Figure 2.10. The effect of a shear
c0

d0

d c
b0

a b a0
Figure 2.11. The effect of an affine map

is an affine map. Since we can write

µ√ √
√
µ ¶ ¶µ ¶
1 1 2/2 −√ 2/2 1 2
= 2 ,
1 3 2/2 2/2 0 1
this affine map is the composition of a shear,√ followed by a rotation of angle
π/4, followed by a magnification of ratio 2, followed by a translation. The
effect of this map on the square (a, b, c, d) is shown in Figure 2.11. The image
of the square (a, b, c, d) is the parallelogram (a0 , b0 , c0 , d0 ).
The following lemma shows the converse of what we just showed. Every
affine map is determined by the image of any point and a linear map.
Lemma 2.7.2 Given an affine map f : E → E 0 , there is a unique linear
−→ −→ −
→
map f : E → E 0 such that
−→
f (a + v) = f (a) + f (v),
−
→
for every a ∈ E and every v ∈ E .
Proof . Let a ∈ E be any point in E. We claim that the map defined such
that
−
→
f (v) = f (a)f (a + v)
34 2. Basics of Affine Geometry

−
→ −
→ − → −
→
for every v ∈ E is a linear map f : E → E 0 . Indeed, we can write
a + λv = λ(a + v) + (1 − λ)a,
since a + λv = a + λa(a + v) + (1 − λ)aa, and also
a + u + v = (a + u) + (a + v) − a,
since a+u+v = a+a(a + u)+a(a + v)−aa. Since f preserves barycenters,
we get
f (a + λv) = λf (a + v) + (1 − λ)f (a).
P
If we recall that x = P i∈I λi ai is the barycenter of a family ((ai , λi ))i∈I of
weighted points (with i∈I λi = 1) iff
X
bx = λi bai for every b ∈ E,
i∈I

we get
f (a)f (a + λv) = λf (a)f (a + v) + (1 − λ)f (a)f (a) = λf (a)f (a + v),
−
→ −
→
showing that f (λv) = λ f (v). We also have
f (a + u + v) = f (a + u) + f (a + v) − f (a),
from which we get
f (a)f (a + u + v) = f (a)f (a + u) + f (a)f (a + v),
−
→ −
→ −
→ −
→
showing that f (u + v) = f (u) + f (v). Consequently, f is a linear map.
For any other point b ∈ E, since
b + v = a + ab + v = a + a(a + v) − aa + ab,
b + v = (a + v) − a + b, and since f preserves barycenters, we get
f (b + v) = f (a + v) − f (a) + f (b),
which implies that
f (b)f (b + v) = f (b)f (a + v) − f (b)f (a) + f (b)f (b),
= f (a)f (b) + f (b)f (a + v),
= f (a)f (a + v).
−
→
Thus, f (b)f (b + v) = f (a)f (a + v), which shows that the definition of f
−
→
does not depend on the choice of a ∈ E. The fact that f is unique is
−→
obvious: We must have f (v) = f (a)f (a + v).

−
→ − → −→
The unique linear map f : E → E 0 given by Lemma 2.7.2 is called the
linear map associated with the affine map f .
 2.7. Affine Maps 35

Note that the condition

−→
f (a + v) = f (a) + f (v),
−→
for every a ∈ E and every v ∈ E , can be stated equivalently as
−
→ −
→
f (x) = f (a) + f (ax), or f (a)f (x) = f (ax),
for all a, x ∈ E. Lemma 2.7.2 shows that for any affine map f : E → E 0 ,
−→ −→ −
→
there are points a ∈ E, b ∈ E 0 , and a unique linear map f : E → E 0 , such
that
−
→
f (a + v) = b + f (v),
−→ −
→
for all v ∈ E (just let b = f (a), for any a ∈ E). Affine maps for which f
−
→
is the identity map are called translations. Indeed, if f = id,
−→
f (x) = f (a) + f (ax) = f (a) + ax = x + xa + af (a) + ax
= x + xa + af (a) − xa = x + af (a),
and so
xf (x) = af (a),
which shows that f is the translation induced by the vector af (a) (which
does not depend on a).
Since an affine map preserves barycenters, and since an affine subspace V
is closed under barycentric combinations, the image f (V ) of V is an affine
subspace in E 0 . So, for example, the image of a line is a point or a line, and
the image of a plane is either a point, a line, or a plane.
It is easily verified that the composition of two affine maps is an affine
map. Also, given affine maps f : E → E 0 and g: E 0 → E 00 , we have
−
→ ´ ³−
→ ´
g(f (a + v)) = g f (a) + f (v) = g(f (a)) + − →
³
g f (v) ,
−−→ −→
which shows that g ◦ f = − →
g ◦ f . It is easy to show that an affine map
−→ − → −→
f : E → E 0 is injective iff f : E → E 0 is injective, and that f : E → E 0 is
−→ − → −
→
surjective iff f : E → E 0 is surjective. An affine map f : E → E 0 is constant
−
→ − → −
→ −→
iff f : E → E 0 is the null (constant) linear map equal to 0 for all v ∈ E .
If E is an affine space of dimension m and (a0 , a1 , . . . , am ) is an affine
frame for E, then for any other affine space F and for any sequence
(b0 , b1 , . . . , bm ) of m + 1 points in F , there is a unique affine map f : E → F
such that f (ai ) = bi , for 0 ≤ i ≤ m. Indeed, f must be such that
f (λ0 a0 + · · · + λm am ) = λ0 b0 + · · · + λm bm ,
where λ0 + · · · + λm = 1, and this defines a unique affine map on all of E,
since (a0 , a1 , . . . , am ) is an affine frame for E.
36 2. Basics of Affine Geometry

Using affine frames, affine maps can be represented in terms of matrices.

We explain how an affine map f : E → E is represented with respect to a
frame (a0 , . . . , an ) in E, the more general case where an affine map f : E →
F is represented with respect to two affine frames (a0 , . . . , an ) in E and
(b0 , . . . , bm ) in F being analogous. Since
−→
f (a0 + x) = f (a0 ) + f (x)
−
→
for all x ∈ E , we have
−→
a0 f (a0 + x) = a0 f (a0 ) + f (x).
Since x, a0 f (a0 ), and a0 f (a0 + x), can be expressed as
x = x 1 a0 a1 + · · · + x n a0 an ,
a0 f (a0 ) = b1 a0 a1 + · · · + bn a0 an ,
a0 f (a0 + x) = y1 a0 a1 + · · · + yn a0 an ,
−
→
if A = (ai j ) is the n × n matrix of the linear map f over the ba-
sis (a0 a1 , . . . , a0 an ), letting x, y, and b denote the column vectors of
components (x1 , . . . , xn ), (y1 , . . . , yn ), and (b1 , . . . , bn ),
−→
a0 f (a0 + x) = a0 f (a0 ) + f (x)
is equivalent to
y = Ax + b.
Note that b 6= 0 unless f (a0 ) = a0 . Thus, f is generally not a linear
transformation, unless it has a fixed point, i.e., there is a point a0 such that
f (a0 ) = a0 . The vector b is the “translation part” of the affine map. Affine
maps do not always have a fixed point. Obviously, nonnull translations have
no fixed point. A less trivial example is given by the affine map
µ ¶ µ ¶µ ¶ µ ¶
x1 1 0 x1 1
7→ + .
x2 0 −1 x2 0
This map is a reflection about the x-axis followed by a translation along
the x-axis. The affine map
µ ¶ µ √ ¶µ ¶ µ ¶
x1 √ 1 − 3 x1 1
7→ +
x2 3/4 1/4 x2 1
can also be written as
µ ¶ µ ¶µ √ ¶µ ¶ µ ¶
x1 2 0 √1/2 − 3/2 x1 1
7→ +
x2 0 1/2 3/2 1/2 x2 1
which shows that it is the composition of a rotation of angle π/3, followed
by a stretch (by a factor of 2 along the x-axis, and by a factor of 12 along
 2.7. Affine Maps 37

the y-axis), followed by a translation. It is easy to show that this affine

map has a unique fixed point. On the other hand, the affine map
µ ¶ µ ¶µ ¶ µ ¶
x1 8/5 −6/5 x1 1
7→ +
x2 3/10 2/5 x2 1

has no fixed point, even though

µ ¶ µ ¶µ ¶
8/5 −6/5 2 0 4/5 −3/5
= ,
3/10 2/5 0 1/2 3/5 4/5
4
and the second matrix is a rotation of angle θ such that cos θ = 5 and
3
sin θ = 5. For more on fixed points of affine maps, see the problems.
There is a useful trick to convert the equation y = Ax+b into what looks
like a linear equation. The trick is to consider an (n + 1) × (n + 1) matrix.
We add 1 as the (n + 1)th component to the vectors x, y, and b, and form
the (n + 1) × (n + 1) matrix
µ ¶
A b
0 1

so that y = Ax + b is equivalent to
µ ¶ µ ¶µ ¶
y A b x
= .
1 0 1 1

This trick is very useful in kinematics and dynamics, where A is a rotation

matrix. Such affine maps are called rigid motions.
If f : E → E 0 is a bijective affine map, given any three collinear points
a, b, c in E, with a 6= b, where, say, c = (1 − λ)a + λb, since f pre-
serves barycenters, we have f (c) = (1 − λ)f (a) + λf (b), which shows that
f (a), f (b), f (c) are collinear in E 0 . There is a converse to this property,
which is simpler to state when the ground field is K = R. The converse
states that given any bijective function f : E → E 0 between two real affine
spaces of the same dimension n ≥ 2, if f maps any three collinear points to
collinear points, then f is affine. The proof is rather long (see Berger [12]
or Samuel [146]).
Given three collinear points a, b, c, where a 6= c, we have b = (1−β)a+βc
for some unique β, and we define the ratio of the sequence a, b, c, as
β ab
ratio(a, b, c) = = ,
(1 − β) bc
provided that β 6= 1, i.e., b 6= c. When b = c, we agree that ratio(a, b, c) =
∞. We warn our readers that other authors define the ratio of a, b, c as
−ratio(a, b, c) = ba
bc . Since affine maps preserve barycenters, it is clear that
affine maps preserve the ratio of three points.
38 2. Basics of Affine Geometry

2.8 Affine Groups

We now take a quick look at the bijective affine maps. Given an affine space
E, the set of affine bijections f : E → E is clearly a group, called the affine
group of E, and denoted by GA(E). Recall that the group of bijective
−
→ −→
linear maps of the vector space E is denoted by GL( E ). Then, the map
−→ −
→
f 7→ f defines a group homomorphism L: GA(E) → GL( E ). The kernel
of this map is the set of translations on E.
The subset of all linear maps of the form λ id− →E
, where λ ∈ R − {0}, is
−
→ ∗
a subgroup of GL( E ), and is denoted by R id− →E
(where λ id−
→E
(u) = λu,
∗ −1 ∗
and R = R − {0}). The subgroup DIL(E) = L (R id− →E
) of GA(E)
is particularly interesting. It turns out that it is the disjoint union of the
translations and of the dilatations of ratio λ 6= 1. The elements of DIL(E)
are called affine dilatations.
Given any point a ∈ E, and any scalar λ ∈ R, a dilatation or central
dilatation (or homothety) of center a and ratio λ is a map Ha,λ defined
such that
Ha,λ (x) = a + λax,
for every x ∈ E.

Remark: The terminology does not seem to be universally agreed upon.

The terms affine dilatation and central dilatation are used by Pedoe [136].
Snapper and Troyer use the term dilation for an affine dilatation and mag-
nification for a central dilatation [160]. Samuel uses homothety for a central
dilatation, a direct translation of the French “homothétie” [146]. Since dila-
tion is shorter than dilatation and somewhat easier to pronounce, perhaps
we should use that!

Observe that Ha,λ (a) = a, and when λ 6= 0 and x 6= a, Ha,λ (x) is on the
line defined by a and x, and is obtained by “scaling” ax by λ.
Figure 2.12 shows the effect of a central dilatation of center d. The tri-
angle (a, b, c) is magnified to the triangle (a0 , b0 , c0 ). Note how every line is
mapped to a parallel line.
−−→
When λ = 1, Ha,1 is the identity. Note that Ha,λ = λ id− →
E
. When λ 6= 0,
it is clear that Ha,λ is an affine bijection. It is immediately verified that
Ha,λ ◦ Ha,µ = Ha,λµ .
We have the following useful result.
Lemma 2.8.1 Given any affine space E, for any affine bijection f ∈
−
→
GA(E), if f = λ id− →E
, for some λ ∈ R∗ with λ 6= 1, then there is a
unique point c ∈ E such that f = Hc,λ .
 2.8. Affine Groups 39
0
a

b b0
d

c0

Figure 2.12. The effect of a central dilatation

Proof . The proof is straightforward, and is omitted. It is also given in

Gallier [70].

−
→
Clearly, if f = id− →E
, the affine map f is a translation. Thus, the group
of affine dilatations DIL(E) is the disjoint union of the translations and
of the dilatations of ratio λ 6= 0, 1. Affine dilatations can be given a purely
geometric characterization.
Another point worth mentioning is that affine bijections preserve the ra-
tio of volumes of parallelotopes. Indeed, given any basis B = (u1 , . . . , um )
−→
of the vector space E associated with the affine space E, given any m + 1
affinely independent points (a0 , . . . , am ), we can compute the determi-
nant detB (a0 a1 , . . . , a0 am ) w.r.t. the basis B. For any bijective affine map
f : E → E, since
³− → −
→ ´ ¡−
→¢
detB f (a0 a1 ), . . . , f (a0 am ) = det f detB (a0 a1 , . . . , a0 am )
−
→
and the determinant of a linear map is intrinsic (i.e., depends only on f ,
and not on the particular basis B), we conclude that the ratio
³−→ −
→ ´
detB f (a0 a1 ), . . . , f (a0 am ) ¡−
→¢
= det f
detB (a0 a1 , . . . , a0 am )
is independent of the basis B. Since detB (a0 a1 , . . . , a0 am ) is the volume of
the parallelotope spanned by (a0 , . . . , am ), where the parallelotope spanned
by any point a and the vectors (u1 , . . . , um ) has unit volume (see Berger
[12], Section 9.12), we see that affine bijections preserve the ratio of volumes
40 2. Basics of Affine Geometry

of parallelotopes. In fact, this ratio is independent of the choice of the par-

allelotopes of unit volume. In particular, the affine bijections f ∈ GA(E)
¡−
→¢
such that det f = 1 preserve volumes. These affine maps form a sub-
group SA(E) of GA(E) called the special affine group of E. We now take
a glimpse at affine geometry.

2.9 Affine Geometry: A Glimpse

In this section we state and prove three fundamental results of affine geom-
etry. Roughly speaking, affine geometry is the study of properties invariant
under affine bijections. We now prove one of the oldest and most basic
results of affine geometry, the theorem of Thales.
Lemma 2.9.1 Given any affine space E, if H1 , H2 , H3 are any three dis-
tinct parallel hyperplanes, and A and B are any two lines not parallel to
Hi , letting ai = Hi ∩ A and bi = Hi ∩ B, then the following ratios are equal:
a1 a3 b1 b3
= = ρ.
a1 a2 b1 b2
a1 d
Conversely, for any point d on the line A, if a1 a2 = ρ, then d = a3 .
Proof . Figure 2.13 illustrates the theorem of Thales. We sketch a proof,
leaving the details as an exercise. Since H1 , H2 , H3 are parallel, they have
−
→ −
→ −
→ − →
the same direction H , a hyperplane in E . Let u ∈ E − H be any nonnull
vector such that A = a1 + Ru. Since A is not parallel to H, we have
−
→ −
→ −→
E = H ⊕ Ru, and thus we can define the linear map p: E → Ru, the
−
→
projection on Ru parallel to H . This linear map induces an affine map
f : E → A, by defining f such that
f (b1 + w) = a1 + p(w),
−
→
for all w ∈ E . Clearly, f (b1 ) = a1 , and since H1 , H2 , H3 all have direction
−
→
H , we also have f (b2 ) = a2 and f (b3 ) = a3 . Since f is affine, it preserves
ratios, and thus
a1 a3 b1 b3
= .
a1 a2 b1 b2
The converse is immediate.

We also have the following simple lemma, whose proof is left as an easy
exercise.
Lemma 2.9.2 Given any affine space E, given any two distinct points
a, b ∈ E, and for any affine dilatation f different from the identity, if
 2.9. Affine Geometry: A Glimpse 41

a1 b1

H1

H2
a2 b2

a3 b3

H3

A B

Figure 2.13. The theorem of Thales

a0 = f (a), D = ha, bi is the line passing through a and b, and D 0 is the line
parallel to D and passing through a0 , the following are equivalent:
(i) b0 = f (b);
(ii) If f is a translation, then b0 is the intersection of D 0 with the line
parallel to ha, a0 i passing through b;
If f is a dilatation of center c, then b0 = D0 ∩ hc, bi.
The first case is the parallelogram law, and the second case follows easily
from Thales’ theorem.
We are now ready to prove two classical results of affine geometry,
Pappus’s theorem and Desargues’s theorem. Actually, these results are the-
orems of projective geometry, and we are stating affine versions of these
important results. There are stronger versions that are best proved using
projective geometry.
Lemma 2.9.3 Given any affine plane E, any two distinct lines D and D 0 ,
then for any distinct points a, b, c on D and a0 , b0 , c0 on D0 , if a, b, c, a0 , b0 ,
c0 are distinct from the intersection of D and D 0 (if D and D 0 intersect)
42 2. Basics of Affine Geometry

c
D

c0

b0

D0
a0
Figure 2.14. Pappus’s theorem (affine version)

and if the lines ha, b0 i and ha0 , bi are parallel, and the lines hb, c0 i and hb0 , ci
are parallel, then the lines ha, c0 i and ha0 , ci are parallel.

Proof . Pappus’s theorem is illustrated in Figure 2.14. If D and D 0 are

not parallel, let d be their intersection. Let f be the dilatation of center
d such that f (a) = b, and let g be the dilatation of center d such that
g(b) = c. Since the lines ha, b0 i and ha0 , bi are parallel, and the lines hb, c0 i
and hb0 , ci are parallel, by Lemma 2.9.2 we have a0 = f (b0 ) and b0 = g(c0 ).
However, we observed that dilatations with the same center commute, and
thus f ◦g = g ◦f , and thus, letting h = g ◦f , we get c = h(a) and a0 = h(c0 ).
Again, by Lemma 2.9.2, the lines ha, c0 i and ha0 , ci are parallel. If D and
D0 are parallel, we use translations instead of dilatations.

There is a converse to Pappus’s theorem, which yields a fancier version

of Pappus’s theorem, but it is easier to prove it using projective geometry.
It should be noted that in axiomatic presentations of projective geometry,
Pappus’s theorem is equivalent to the commutativity of the ground field K
(in the present case, K = R). We now prove an affine version of Desargues’s
theorem.

Lemma 2.9.4 Given any affine space E, and given any two triangles
(a, b, c) and (a0 , b0 , c0 ), where a, b, c, a0 , b0 , c0 are all distinct, if ha, bi and
ha0 , b0 i are parallel and hb, ci and hb0 , c0 i are parallel, then ha, ci and ha0 , c0 i
 2.9. Affine Geometry: A Glimpse 43
0
a

b b0
d

c0

Figure 2.15. Desargues’s theorem (affine version)

are parallel iff the lines ha, a0 i, hb, b0 i, and hc, c0 i are either parallel or
concurrent (i.e., intersect in a common point).
Proof . We prove half of the lemma, the direction in which it is assumed
that ha, ci and ha0 , c0 i are parallel, leaving the converse as an exercise. Since
the lines ha, bi and ha0 , b0 i are parallel, the points a, b, a0 , b0 are coplanar.
Thus, either ha, a0 i and hb, b0 i are parallel, or they have some intersection d.
We consider the second case where they intersect, leaving the other case as
an easy exercise. Let f be the dilatation of center d such that f (a) = a0 . By
Lemma 2.9.2, we get f (b) = b0 . If f (c) = c00 , again by Lemma 2.9.2 twice,
the lines hb, ci and hb0 , c00 i are parallel, and the lines ha, ci and ha0 , c00 i are
parallel. From this it follows that c00 = c0 . Indeed, recall that hb, ci and
hb0 , c0 i are parallel, and similarly ha, ci and ha0 , c0 i are parallel. Thus, the
lines hb0 , c00 i and hb0 , c0 i are identical, and similarly the lines ha0 , c00 i and
ha0 , c0 i are identical. Since a0 c0 and b0 c0 are linearly independent, these
lines have a unique intersection, which must be c00 = c0 .
The direction where it is assumed that the lines ha, a0 i, hb, b0 i and hc, c0 i,
are either parallel or concurrent is left as an exercise (in fact, the proof is
quite similar).

Desargues’s theorem is illustrated in Figure 2.15.

There is a fancier version of Desargues’s theorem, but it is easier to
prove it using projective geometry. It should be noted that in axiomatic
presentations of projective geometry, Desargues’s theorem is related to the
associativity of the ground field K (in the present case, K = R). Also,
Desargues’s theorem yields a geometric characterization of the affine di-
latations. An affine dilatation f on an affine space E is a bijection that
44 2. Basics of Affine Geometry

maps every line D to a line f (D) parallel to D. We leave the proof as an

exercise.

2.10 Affine Hyperplanes

We now consider affine forms and affine hyperplanes. In Section 2.5 we
observed that the set L of solutions of an equation
ax + by = c
is an affine subspace of A2 of dimension 1, in fact, a line (provided that a
and b are not both null). It would be equally easy to show that the set P
of solutions of an equation
ax + by + cz = d
is an affine subspace of A3 of dimension 2, in fact, a plane (provided that
a, b, c are not all null). More generally, the set H of solutions of an equation
λ1 x 1 + · · · + λ m x m = µ
is an affine subspace of Am , and if λ1 , . . . , λm are not all null, it turns out
that it is a subspace of dimension m − 1 called a hyperplane.
We can interpret the equation
λ1 x 1 + · · · + λ m x m = µ
in terms of the map f : Rm → R defined such that
f (x1 , . . . , xm ) = λ1 x1 + · · · + λm xm − µ
for all (x1 , . . . , xm ) ∈ Rm . It is immediately verified that this map is affine,
and the set H of solutions of the equation
λ1 x 1 + · · · + λ m x m = µ
is the null set, or kernel, of the affine map f : Am → R, in the sense that
H = f −1 (0) = {x ∈ Am | f (x) = 0},
where x = (x1 , . . . , xm ).
Thus, it is interesting to consider affine forms, which are just affine
maps f : E → R from an affine space to R. Unlike linear forms f ∗ , for which
Ker f ∗ is never empty (since it always contains the vector 0), it is possible
that f −1 (0) = ∅ for an affine form f . Given an affine map f : E → R, we
also denote f −1 (0) by Ker f , and we call it the kernel of f . Recall that an
(affine) hyperplane is an affine subspace of codimension 1. The relationship
between affine hyperplanes and affine forms is given by the following lemma.

Lemma 2.10.1 Let E be an affine space. The following properties hold:

 2.11. Intersection of Affine Spaces 45

(a) Given any nonconstant affine form f : E → R, its kernel H = Ker f

is a hyperplane.
(b) For any hyperplane H in E, there is a nonconstant affine form f : E →
R such that H = Ker f . For any other affine form g: E → R such that
H = Ker g, there is some λ ∈ R such that g = λf (with λ 6= 0).
(c) Given any hyperplane H in E and any (nonconstant) affine form
f : E → R such that H = Ker f , every hyperplane H 0 parallel to H is
defined by a nonconstant affine form g such that g(a) = f (a) − λ, for
all a ∈ E and some λ ∈ R.
Proof . The proof is straightforward, and is omitted. It is also given in
Gallier [70].

When E is of dimension n, given an affine frame (a0 , (u1 , . . . , un )) of E

with origin a0 , recall from Definition 2.6.4 that every point of E can be
expressed uniquely as x = a0 + x1 u1 + · · · + xn un , where (x1 , . . . , xn ) are
the coordinates of x with respect to the affine frame (a0 , (u1 , . . . , un )).
Also recall that every linear form f ∗ is such that f ∗ (x) = λ1 x1 + · · · +
λn xn , for every x = x1 u1 + · · · + xn un and some λ1 , . . . , λn ∈ R. Since
−→
an affine form f : E → R satisfies the property f (a0 + x) = f (a0 ) + f (x),
denoting f (a0 + x) by f (x1 , . . . , xn ), we see that we have
f (x1 , . . . , xn ) = λ1 x1 + · · · + λn xn + µ,
where µ = f (a0 ) ∈ R and λ1 , . . . , λn ∈ R. Thus, a hyperplane is the set of
points whose coordinates (x1 , . . . , xn ) satisfy the (affine) equation
λ1 x1 + · · · + λn xn + µ = 0.

2.11 Intersection of Affine Spaces

In this section we take a closer look at the intersection of affine subspaces.
This subsection can be omitted at first reading.
First, we need a result of linear algebra. Given a vector space E and any
two subspaces M and N , there are several interesting linear maps. We have
the canonical injections i: M → M + N and j: N → M + N , the canonical
injections in1 : M → M ⊕ N and in2 : N → M ⊕ N , and thus, injections
f : M ∩ N → M ⊕ N and g: M ∩ N → M ⊕ N , where f is the composition
of the inclusion map from M ∩ N to M with in1 , and g is the composition
of the inclusion map from M ∩ N to N with in2 . Then, we have the maps
f + g: M ∩ N → M ⊕ N , and i − j: M ⊕ N → M + N .
Lemma 2.11.1 Given a vector space E and any two subspaces M and N ,
with the definitions above,
f +g i−j
0 −→ M ∩ N −→ M ⊕ N −→ M + N −→ 0
46 2. Basics of Affine Geometry

is a short exact sequence, which means that f + g is injective, i − j is

surjective, and that Im (f + g) = Ker (i − j). As a consequence, we have
the Grassmann relation

dim(M ) + dim(N ) = dim(M + N ) + dim (M ∩ N ).

Proof . It is obvious that i − j is surjective and that f + g is injective.

Assume that (i−j)(u+v) = 0, where u ∈ M , and v ∈ N . Then, i(u) = j(v),
and thus, by definition of i and j, there is some w ∈ M ∩ N , such that
i(u) = j(v) = w ∈ M ∩N . By definition of f and g, u = f (w) and v = g(w),
and thus Im (f + g) = Ker (i − j), as desired. The second part of the lemma
follows from standard results of linear algebra (see Artin [5], Strang [166],
or Lang [107]).

We now prove a simple lemma about the intersection of affine subspaces.

Lemma 2.11.2 Given any affine space E, for any two nonempty affine
subspaces M and N , the following facts hold:
−
→ − →
(1) M ∩ N 6= ∅ iff ab ∈ M + N for some a ∈ M and some b ∈ N .
−
→ − →
(2) M ∩ N consists of a single point iff ab ∈ M + N for some a ∈ M
−
→ − →
and some b ∈ N , and M ∩ N = {0}.
−
→
(3) If S is the least affine subspace containing M and N , then S =
−
→ − → −
→
M + N + Kab (the vector space E is defined over the field K).

Proof . (1) Pick any a ∈ M and any b ∈ N , which is possible, since M and
−
→ −
→
N are nonempty. Since M = {ax | x ∈ M } and N = {by | y ∈ N }, if
−
→
M ∩ N 6= ∅, for any c ∈ M ∩ N we have ab = ac − bc, with ac ∈ M and
−
→ −
→ − → −
→ − →
bc ∈ N , and thus, ab ∈ M + N . Conversely, assume that ab ∈ M + N
for some a ∈ M and some b ∈ N . Then ab = ax + by, for some x ∈ M
and some y ∈ N . But we also have

ab = ax + xy + yb,

and thus we get 0 = xy + yb − by, that is, xy = 2by. Thus, b is the middle
of the segment [x, y], and since yx = 2yb, x = 2b − y is the barycenter of
the weighted points (b, 2) and (y, −1). Thus x also belongs to N , since N
being an affine subspace, it is closed under barycenters. Thus, x ∈ M ∩ N ,
and M ∩ N 6= ∅.
(2) Note that in general, if M ∩ N 6= ∅, then
−−−−→ −→ − →
M ∩N = M ∩ N,
 2.12. Problems 47

because
−−−−→ −
→ − →
M ∩ N = {ab | a, b ∈ M ∩N } = {ab | a, b ∈ M }∩{ab | a, b ∈ N } = M ∩ N .
−−−−→
Since M ∩ N = c + M ∩ N for any c ∈ M ∩ N , we have
−
→ − →
M ∩N =c+M ∩ N for any c ∈ M ∩ N .
From this it follows that if M ∩ N 6= ∅, then M ∩ N consists of a single
−
→ − →
point iff M ∩ N = {0}. This fact together with what we proved in (1)
proves (2).
(3) This is left as an easy exercise.

Remarks:
(1) The proof of Lemma 2.11.2 shows that if M ∩ N 6= ∅, then ab ∈
−
→ − →
M + N for all a ∈ M and all b ∈ N .
(2) Lemma 2.11.2 implies that for any two nonempty affine subspaces M
−
→ − → − →
and N , if E = M ⊕ N , then M ∩N consists of a single point. Indeed,
−
→ − → − → −→
if E = M ⊕ N , then ab ∈ E for all a ∈ M and all b ∈ N , and since
−
→ − →
M ∩ N = {0}, the result follows from part (2) of the lemma.
We can now state the following lemma.
Lemma 2.11.3 Given an affine space E and any two nonempty affine
subspaces M and N , if S is the least affine subspace containing M and N ,
then the following properties hold:
(1) If M ∩ N = ∅, then
−
→ − →
dim(M ) + dim(N ) < dim(E) + dim(M + N )
and
−→ − →
dim(S) = dim(M ) + dim(N ) + 1 − dim( M ∩ N ).
(2) If M ∩ N 6= ∅, then
dim(S) = dim(M ) + dim(N ) − dim(M ∩ N ).
Proof . The proof is not difficult, using Lemma 2.11.2 and Lemma 2.11.1,
but we leave it as an exercise.

2.12 Problems
Problem 2.1 Given a triangle (a, b, c), give a geometric construction of
the barycenter of the weighted points (a, 14 ), (b, 41 ), and (c, 21 ). Give a geo-
48 2. Basics of Affine Geometry

metric construction of the barycenter of the weighted points (a, 32 ), (b, 32 ),

and (c, −2).
Problem 2.2 Given a tetrahedron (a, b, c, d) and any two distinct points
x, y ∈ {a, b, c, d}, let let mx,y be the middle of the edge (x, y). Prove that
the barycenter g of the weighted points (a, 41 ), (b, 14 ), (c, 14 ), and (d, 41 ) is
the common intersection of the line segments (ma,b , mc,d ), (ma,c , mb,d ),
and (ma,d , mb,c ). Show that if gd is the barycenter of the weighted points
(a, 31 ), (b, 31 ), (c, 13 ), then g is the barycenter of (d, 41 ) and (gd , 43 ).
−
→
Problem 2.3 Let E be a nonempty set, and E a vector space and assume
−
→
that there is a function Φ: E × E → E , such that if we denote Φ(a, b) by
ab, the following properties hold:
(1) ab + bc = ac, for all a, b, c ∈ E;
−
→
(2) For every a ∈ E, the map Φa : E → E defined such that for every
b ∈ E, Φa (b) = ab, is a bijection.
−
→ −
→
Let Ψa : E → E be the inverse of Φa : E → E .
−
→
Prove that the function +: E × E → E defined such that
a + u = Ψa (u)
−
→ −
→
for all a ∈ E and all u ∈ E makes (E, E , +) into an affine space.
−
→
Note. We showed in the text that an affine space (E, E , +) satisfies the
properties stated above. Thus, we obtain an equivalent characterization of
affine spaces.
Problem 2.4 Given any three points a, b, c in the affine plane A2 , letting
(a1 , a2 ), (b1 , b2 ), and (c1 , c2 ) be the coordinates of a, b, c, with respect to
the standard affine frame for A2 , prove that a, b, c are collinear iff
¯ ¯
¯ a 1 b1 c 1 ¯
¯ ¯
¯ a2 b2 c2 ¯ = 0,
¯ ¯
¯ 1 1 1¯

i.e., the determinant is null.

Letting (a0 , a1 , a2 ), (b0 , b1 , b2 ), and (c0 , c1 , c2 ) be the barycentric coordi-
nates of a, b, c with respect to the standard affine frame for A2 , prove that
a, b, c are collinear iff
¯ ¯
¯ a 0 b0 c 0 ¯
¯ ¯
¯ a1 b1 c1 ¯ = 0.
¯ ¯
¯ a 2 b2 c 2 ¯

Given any four points a, b, c, d in the affine space A3 , letting (a1 , a2 , a3 ),

(b1 , b2 , b3 ), (c1 , c2 , c3 ), and (d1 , d2 , d3 ) be the coordinates of a, b, c, d, with
 2.12. Problems 49

respect to the standard affine frame for A3 , prove that a, b, c, d are coplanar
iff
¯ a 1 b1 c 1 d 1 ¯
¯ ¯
¯ a 2 b2 c 2 d 2 ¯
¯ ¯
¯ = 0,
¯ a 3 b3 c 3 d 3 ¯
¯
1 1 1 1
¯ ¯

i.e., the determinant is null.

Letting (a0 , a1 , a2 , a3 ), (b0 , b1 , b2 , b3 ), (c0 , c1 , c2 , c3 ), and (d0 , d1 , d2 , d3 ) be
the barycentric coordinates of a, b, c, d, with respect to the standard affine
frame for A3 , prove that a, b, c, d are coplanar iff
¯ a 0 b0 c 0 d 0 ¯
¯ ¯
¯ a 1 b1 c 1 d 1 ¯
¯ ¯
¯ = 0.
¯ a 2 b2 c 2 d 2 ¯
¯
a 3 b3 c 3 d 3
¯ ¯

Problem 2.5 The function f : A → A3 given by

t 7→ (t, t2 , t3 )
defines what is called a twisted cubic curve. Given any four pairwise distinct
values t1 , t2 , t3 , t4 , prove that the points f (t1 ), f (t2 ), f (t3 ), and f (t4 ) are not
coplanar.
Hint. Have you heard of the Vandermonde determinant?
Problem 2.6 For any two distinct points a, b ∈ A2 of barycentric coor-
dinates (a0 , a1 , a2 ) and (b0 , b1 , b2 ) with respect to any given affine frame
(O, i, j), show that the equation of the line ha, bi determined by a and b is
¯ ¯
¯ a 0 b0 x ¯
¯ ¯
¯ a1 b1 y ¯ = 0,
¯ ¯
¯ a 2 b2 z ¯

or, equivalently,
(a1 b2 − a2 b1 )x + (a2 b0 − a0 b2 )y + (a0 b1 − a1 b0 )z = 0,
where (x, y, z) are the barycentric coordinates of the generic point on the
line ha, bi.
Prove that the equation of a line in barycentric coordinates is of the form
ux + vy + wz = 0,
where u 6= v or v 6= w or u 6= w. Show that two equations
ux + vy + wz = 0 and u0 x + v 0 y + w 0 z = 0
represent the same line in barycentric coordinates iff (u0 , v 0 , w0 ) = λ(u, v, w)
for some λ ∈ R (with λ 6= 0).
50 2. Basics of Affine Geometry

A triple (u, v, w) where u 6= v or v 6= w or u =6 w is called a system of

tangential coordinates of the line defined by the equation
ux + vy + wz = 0.
Problem 2.7 Given two lines D and D 0 in A2 defined by tangential
coordinates (u, v, w) and (u0 , v 0 , w0 ) (as defined in Problem 2.6), let
¯ ¯
¯u v w¯
d = ¯¯ u v 0 w0 ¯¯ = vw0 − wv 0 + wu0 − uw0 + uv 0 − vu0 .
¯ 0 ¯
¯1 1 1 ¯
(a) Prove that D and D 0 have a unique intersection point iff d 6= 0, and
that when it exists, the barycentric coordinates of this intersection point
are
1
(vw0 − wv 0 , wu0 − uw0 , uv 0 − vu0 ).
d
(b) Letting (O, i, j) be any affine frame for A2 , recall that when x+y+z =
0, for any point a, the vector
xaO + yai + zaj
is independent of a and equal to
yOi + zOj = (y, z).
The triple (x, y, z) such that x + y + z = 0 is called the barycentric
coordinates of the vector yOi + zOj w.r.t. the affine frame (O, i, j).
Given any affine frame (O, i, j), prove that for u 6= v or v 6= w or u 6= w,
the line of equation
ux + vy + wz = 0
in barycentric coordinates (x, y, z) (where x + y + z = 1) has for direction
the set of vectors of barycentric coordinates (x, y, z) such that
ux + vy + wz = 0
(where x + y + z = 0).
Prove that D and D 0 are parallel iff d = 0. In this case, if D 6= D 0 ,
show that the common direction of D and D 0 is defined by the vector of
barycentric coordinates
(vw0 − wv 0 , wu0 − uw0 , uv 0 − vu0 ).
(c) Given three lines D, D 0 , and D 00 , at least two of which are distinct
and defined by tangential coordinates (u, v, w), (u0 , v 0 , w0 ), and (u00 , v 00 , w00 ),
prove that D, D 0 , and D 00 are parallel or have a unique intersection point
iff
¯ ¯
¯ u
¯ 0 v0 w0 ¯
¯
¯u
¯ 00 v00 w00 ¯ = 0.
¯
¯u v w ¯
 2.12. Problems 51

Problem 2.8 Let (A, B, C) be a triangle in A2 . Let M, N, P be three

points respectively on the lines BC, CA, and AB, of barycentric co-
ordinates (0, m0 , m00 ), (n, 0, n00 ), and (p, p0 , 0), w.r.t. the affine frame
(A, B, C).
(a) Assuming that M 6= C, N 6= A, and P 6= B, i.e., m0 n00 p 6= 0, show
that
MB NC PA m00 np0
= − 0 00 .
MC NA PB mn p
(b) Prove Menelaus’s theorem: The points M, N, P are collinear iff
m00 np0 + m0 n00 p = 0.
When M 6= C, N 6= A, and P 6= B, this is equivalent to
MB NC PA
= 1.
MC NA PB
(c) Prove Ceva’s theorem: The lines AM, BN, CP have a unique
intersection point or are parallel iff
m00 np0 − m0 n00 p = 0.
When M 6= C, N 6= A, and P 6= B, this is equivalent to
MB NC PA
= −1.
MC NA PB
Problem 2.9 This problem uses notions and results from Problems 2.6
and 2.7. In view of (a) and (b) of Problem 2.7, it is natural to extend the
notion of barycentric coordinates of a point in A2 as follows. Given any
affine frame (a, b, c) in A2 , we will say that the barycentric coordinates
(x, y, z) of a point M , where x + y + z = 1, are the normalized barycentric
coordinates of M . Then, any triple (x, y, z) such that x + y + z 6= 0 is
also called a system of barycentric coordinates for the point of normalized
barycentric coordinates
1
(x, y, z).
x+y+z
With this convention, the intersection of the two lines D and D 0 is either
a point or a vector, in both cases of barycentric coordinates
(vw0 − wv 0 , wu0 − uw0 , uv 0 − vu0 ).
When the above is a vector, we can think of it as a point at infinity (in the
direction of the line defined by that vector).
Let (D0 , D00 ), (D1 , D10 ), and (D2 , D20 ) be three pairs of six distinct lines,
such that the four lines belonging to any union of two of the above pairs
are neither parallel nor concurrent (have a common intersection point). If
D0 and D00 have a unique intersection point, let M be this point, and if D0
and D00 are parallel, let M denote a nonnull vector defining the common
52 2. Basics of Affine Geometry

direction of D0 and D00 . In either case, let (m, m0 , m00 ) be the barycentric
coordinates of M , as explained at the beginning of the problem. We call
M the intersection of D0 and D00 . Similarly, define N = (n, n0 , n00 ) as the
intersection of D1 and D10 , and P = (p, p0 , p00 ) as the intersection of D2 and
D20 .
Prove that
¯ ¯
¯ m
¯ 0 n0 p0 ¯
¯
¯m
¯ 00 n00 p00 ¯ = 0
¯
¯m n p ¯

iff either

(i) (D0 , D00 ), (D1 , D10 ), and (D2 , D20 ) are pairs of parallel lines; or

(ii) the lines of some pair (Di , Di0 ) are parallel, each pair (Dj , Dj0 ) (with
j 6= i) has a unique intersection point, and these two intersection
points are distinct and determine a line parallel to the lines of the
pair (Di , Di0 ); or

(iii) each pair (Di , Di0 ) (i = 0, 1, 2) has a unique intersection point, and
these points M, N, P are distinct and collinear.

Problem 2.10 Prove the following version of Desargues’s theorem. Let

A, B, C, A0 , B 0 , C 0 be six distinct points of A2 . If no three of these points
are collinear, then the lines AA0 , BB 0 , and CC 0 are parallel or collinear iff
the intersection points M, N, P (in the sense of Problem 2.7) of the pairs
of lines (BC, B 0 C 0 ), (CA, C 0 A0 ), and (AB, A0 B 0 ) are collinear in the sense
of Problem 2.9.

Problem 2.11 Prove the following version of Pappus’s theorem. Let D

and D 0 be distinct lines, and let A, B, C and A0 , B 0 , C 0 be distinct points
respectively on D and D 0 . If these points are all distinct from the intersec-
tion of D and D 0 (if it exists), then the intersection points (in the sense of
Problem 2.7) of the pairs of lines (BC 0 , CB 0 ), (CA0 , AC 0 ), and (AB 0 , BA0 )
are collinear in the sense of Problem 2.9.

Problem 2.12 The purpose of this problem is to prove Pascal’s theorem

for the nondegenerate conics. In the affine plane A2 , a conic is the set of
points of coordinates (x, y) such that

αx2 + βy 2 + 2γxy + 2δx + 2λy + µ = 0,

where α 6= 0 or β 6= 0 or γ 6= 0. We can write the equation of the conic as

  
α γ δ x
(x, y, 1)  γ β λ   y  = 0.
δ λ µ 1
 2.12. Problems 53

If we now use barycentric coordinates (x, y, z) (where x + y + z = 1), we

can write
    
x 1 0 0 x
y  = 0 1 0y .
1 1 1 1 z
Let
     
α γ δ 1 0 0 x
B = γ β λ, C = 0 1 0, X = y .
δ λ µ 1 1 1 z
(a) Letting A = C > BC, prove that the equation of the conic becomes
X > AX = 0.
Prove that A is symmetric, that det(A) = det(B), and that X > AX is ho-
mogeneous of degree 2. The equation X > AX = 0 is called the homogeneous
equation of the conic.
We say that a conic of homogeneous equation X > AX = 0 is nondegen-
erate if det(A) 6= 0, and degenerate if det(A) = 0. Show that this condition
does not depend on the choice of the affine frame.
(b) Given an affine frame (A, B, C), prove that any conic passing through
A, B, C has an equation of the form
ayz + bxz + cxy = 0.
Prove that a conic containing more than one point is degenerate iff it con-
tains three distinct collinear points. In this case, the conic is the union of
two lines.
(c) Prove Pascal’s theorem. Given any six distinct points A, B, C, A0 , B 0 ,
0
C , if no three of the above points are collinear, then a nondegenerate conic
passes through these six points iff the intersection points M, N, P (in the
sense of Problem 2.7) of the pairs of lines (BC 0 , CB 0 ), (CA0 , AC 0 ) and
(AB 0 , BA0 ) are collinear in the sense of Problem 2.9.
Hint. Use the affine frame (A, B, C), and let (a, a0 , a00 ), (b, b0 , b00 ), and
(c, c0 , c00 ) be the barycentric coordinates of A0 , B 0 , C 0 respectively, and show
that M, N, P have barycentric coordinates
(bc, cb0 , c00 b), (c0 a, c0 a0 , c00 a0 ), (ab00 , a00 b0 , a00 b00 ).
Problem 2.13 The centroid of a triangle (a, b, c) is the barycenter of
(a, 31 ), (b, 13 ), (c, 13 ). If an affine map takes the vertices of triangle ∆1 =
{(0, 0), (6, 0), (0, 9)} to the vertices of triangle ∆2 = {(1, 1), (5, 4), (3, 1)},
does it also take the centroid of ∆1 to the centroid of ∆2 ? Justify your
answer.
Problem 2.14 Let E be an affine space over R, and let (a1 , . . . , an ) be
any n ≥ 3 points in E. Let (λ1 , . . . , λn ) be any n scalars in R, with λ1 +
· · · + λn = 1. Show that there must be some i, 1 ≤ i ≤ n, such that λi 6= 1.
54 2. Basics of Affine Geometry

To simplify the notation, assume that λ1 6= 1. Show that the barycenter

λ1 a1 + · · · + λn an can be obtained by first determining the barycenter b of
the n−1 points a2 , . . . , an assigned some appropriate weights, and then the
barycenter of a1 and b assigned the weights λ1 and λ2 + · · · + λn . From this,
show that the barycenter of any n ≥ 3 points can be determined by repeated
computations of barycenters of two points. Deduce from the above that a
nonempty subset V of E is an affine subspace iff whenever V contains any
two points x, y ∈ V , then V contains the entire line (1 − λ)x + λy, λ ∈ R.
Problem 2.15 Assume that K is a field such that 2 = 1 + 1 6= 0, and let
E be an affine space over K. In the case where λ1 +· · ·+λn = 1 and λi = 1,
for 1 ≤ i ≤ n and n ≥ 3, show that the barycenter a1 + a2 + · · · + an can
still be computed by repeated computations of barycenters of two points.
Finally, assume that the field K contains at least three elements (thus,
there is some µ ∈ K such that µ 6= 0 and µ 6= 1, but 2 = 1 + 1 = 0
is possible). Prove that the barycenter of any n ≥ 3 points can be deter-
mined by repeated computations of barycenters of two points. Prove that a
nonempty subset V of E is an affine subspace iff whenever V contains any
two points x, y ∈ V , then V contains the entire line (1 − λ)x + λy, λ ∈ K.
Hint. When 2 = 0, λ1 + · · · + λn = 1 and λi = 1, for 1 ≤ i ≤ n, show that
n must be odd, and that the problem reduces to computing the barycenter
of three points in two steps involving two barycenters. Since there is some
µ ∈ K such that µ 6= 0 and µ 6= 1, note that µ−1 and (1 − µ)−1 both exist,
and use the fact that
−µ 1
+ = 1.
1−µ 1−µ
Problem 2.16 (i) Let (a, b, c) be three points in A2 , and assume that
(a, b, c) are not collinear. For any point x ∈ A2 , if x = λ0 a+λ1 b+λ2 c, where
(λ0 , λ1 , λ2 ) are the barycentric coordinates of x with respect to (a, b, c),
show that
det(xb, bc) det(ax, ac) det(ab, ax)
λ0 = , λ1 = , λ2 = .
det(ab, ac) det(ab, ac) det(ab, ac)
Conclude that λ0 , λ1 , λ2 are certain signed ratios of the areas of the
triangles (a, b, c), (x, a, b), (x, a, c), and (x, b, c).
(ii) Let (a, b, c) be three points in A3 , and assume that (a, b, c) are
not collinear. For any point x in the plane determined by (a, b, c), if
x = λ0 a + λ1 b + λ2 c, where (λ0 , λ1 , λ2 ) are the barycentric coordinates
of x with respect to (a, b, c), show that
xb × bc ax × ac ab × ax
λ0 = , λ1 = , λ2 = .
ab × ac ab × ac ab × ac
Given any point O not in the plane of the triangle (a, b, c), prove that
det(Oa, Ox, Oc) det(Oa, Ob, Ox)
λ1 = , λ2 = ,
det(Oa, Ob, Oc) det(Oa, Ob, Oc)
 2.12. Problems 55

and
det(Ox, Ob, Oc)
λ0 = .
det(Oa, Ob, Oc)
(iii) Let (a, b, c, d) be four points in A3 , and assume that (a, b, c, d) are
not coplanar. For any point x ∈ A3 , if x = λ0 a + λ1 b + λ2 c + λ3 d,
where (λ0 , λ1 , λ2 , λ3 ) are the barycentric coordinates of x with respect to
(a, b, c, d), show that
det(ax, ac, ad) det(ab, ax, ad) det(ab, ac, ax)
λ1 = , λ2 = , λ3 = ,
det(ab, ac, ad) det(ab, ac, ad) det(ab, ac, ad)
and
det(xb, bc, bd)
λ0 = .
det(ab, ac, ad)
Conclude that λ0 , λ1 , λ2 , λ3 are certain signed ratios of the volumes of the
five tetrahedra (a, b, c, d), (x, a, b, c), (x, a, b, d), (x, a, c, d), and (x, b, c, d).
(iv) Let (a0 , . . . , am ) be m + 1 points in Am , and assume that they are
affinely independent. For any point x ∈ Am , if x = λ0 a0 + · · · + λm am ,
where (λ0 , . . . , λm ) are the barycentric coordinates of x with respect to
(a0 , . . . , am ), show that
det(a0 a1 , . . . , a0 ai−1 , a0 x, a0 ai+1 , . . . , a0 am )
λi =
det(a0 a1 , . . . , a0 ai−1 , a0 ai , a0 ai+1 , . . . , a0 am )
for every i, 1 ≤ i ≤ m, and
det(xa1 , a1 a2 , . . . , a1 am )
λ0 = .
det(a0 a1 , . . . , a0 ai , . . . , a0 am )
Conclude that λi is the signed ratio of the volumes of the simplexes (a0 , . . .,
x, . . . am ) and (a0 , . . . , ai , . . . am ), where 0 ≤ i ≤ m.
Problem 2.17 With respect to the standard affine frame for the plane
A2 , consider the three geometric transformations f1 , f2 , f3 defined by
√
1 3 3
x0 = − x − y+ ,
√ 4 4 √4
3 1 3
y0 = x− y+ ,
4 4
√ 4
1 3 3
x0 = − x + y− ,
4√ 4 4
√
3 1 3
y0 = − x− y+ ,
4 4 4
1
x0 = x,
2 √
0 1 3
y = y+ .
2 2
56 2. Basics of Affine Geometry

(a) Prove that these maps are affine. Can you describe geometrically
what their action is (rotation, translation, scaling)?
(b) Given any polygonal line L, define the following sequence of polygonal
lines:
S0 = L,
Sn+1 = f1 (Sn ) ∪ f2 (Sn ) ∪ f3 (Sn ).
Construct S1 starting from the line segment L = ((−1, 0), (1, 0)).
Can you figure out what Sn looks like in general? (You may want to
write a computer program.) Do you think that Sn has a limit?
Problem 2.18 In the plane A2 , with respect to the standard affine frame,
a point of coordinates (x, y) can be represented as the complex number
z = x + iy. Consider the set of geometric transformations of the form
z 7→ az + b,
where a, b are complex numbers such that a 6= 0.
(a) Prove that these maps are affine. Describe what these maps do
geometrically.
(b) Prove that the above set of maps is a group under composition.
(c) Consider the set of geometric transformations of the form
z 7→ az + b or z 7→ az + b,
where a, b are complex numbers such that a 6= 0, and where z = x − iy if
z = x + iy. Describe what these maps do geometrically. Prove that these
maps are affine and that this set of maps is a group under composition.
Problem 2.19 Given a group G, a subgroup H of G is called a normal
subgroup of G iff xHx−1 = H for all x ∈ G (where xHx−1 = {xhx−1 | h ∈
H}).
(i) Given any two subgroups H and K of a group G, let
HK = {hk | h ∈ H, k ∈ K}.
Prove that every x ∈ HK can be written in a unique way as x = hk for
h ∈ H and k ∈ K iff H ∩ K = {1}, where 1 is the identity element of G.
(ii) If H and K are subgroups of G, and H is a normal subgroup of
G, prove that HK is a subgroup of G. Furthermore, if G = HK and
H ∩K = {1}, prove that G is isomorphic to H ×K under the multiplication
operation
(h1 , k1 ) · (h2 , k2 ) = (h1 k1 h2 k1−1 , k1 k2 ).
When G = HK, where H, K are subgroups of G, H is a normal subgroup
of G, and H ∩ K = {1}, we say that G is the semidirect product of H and
K.
−
→
(iii) Let (E, E ) be an affine space. Recall that the affine group of E,
denoted by GA(E), is the set of affine bijections of E, and that the linear
 2.12. Problems 57

−
→ −→ −
→
group of E , denoted by GL( E ), is the group of bijective linear maps of E .
−
→ −→
The map f 7→ f defines a group homomorphism L: GA(E) → GL( E ),
and the kernel of this map is the set of translations on E, denoted as T (E).
Prove that T (E) is a normal subgroup of GA(E).
(iv) For any a ∈ E, let
GAa (E) = {f ∈ GA(E) | f (a) = a},
the set of affine bijections leaving a fixed. Prove that that GAa (E) is a
−
→
subgroup of GA(E), and that GAa (E) is isomorphic to GL( E ). Prove
that GA(E) is isomorphic to the direct product of T (E) and GAa (E).
Hint. Note that if u = f (a)a and tu is the translation associated with the
vector u, then tu ◦ f ∈ GAa (E) (where the translation tu is defined such
that tu (a) = a + u for every a ∈ E).
(v) Given a group G, let Aut(G) denote the set of homomorphisms
f : G → G. Prove that the set Aut(G) is a group under composition (called
the group of automorphisms of G). Given any two groups H and K and
a homomorphism θ: K → Aut(H), we define H ×θ K as the set H × K
under the multiplication operation
(h1 , k1 ) · (h2 , k2 ) = (h1 θ(k1 )(h2 ), k1 k2 ).
Prove that H ×θ K is a group.
Hint. The inverse of (h, k) is (θ(k −1 )(h−1 ), k −1 ).
Prove that the group H ×θ K is the semidirect product of the subgroups
{(h, 1) | h ∈ H} and {(1, k) | k ∈ K}. The group H ×θ K is also called the
semidirect product of H and K relative to θ.
Note. It is natural to identify {(h, 1) | h ∈ H} with H and {(1, k) | k ∈ K}
with K.
If G is the semidirect product of two subgroups H and K as defined in
(ii), prove that the map γ: K → Aut(H) defined by conjugation such that
γ(k)(h) = khk −1
is a homomorphism, and that G is isomorphic to H ×γ K.
−→ −→
(vi) Define the map θ: GL( E ) → Aut( E ) as follows:
θ(f ) = f,
−→
where f ∈ GL( E ) (note that θ can be viewed as an inclusion map). Prove
−
→ −
→
that GA(E) is isomorphic to the semidirect product E ×θ GL( E ).
−
→ −
→
(vii) Let SL( E ) be the subgroup of GL( E ) consisting of the linear
−
→
maps such that det(f ) = 1 (the special linear group of E ), and let SA(E)
be the subgroup of GA(E) (the special affine group of E) consisting of
−
→ −→
the affine maps f such that f ∈ SL( E ). Prove that SA(E) is isomorphic
58 2. Basics of Affine Geometry

−
→ −
→ −
→ −
→
to the semidirect product E ×θ SL( E ), where θ: SL( E ) → Aut( E ) is
defined as in (vi).
−
→
(viii) Assume that (E, E ) is a Euclidean affine space, as defined in Chap-
−→ −→
ter 6. Let SO( E ) be the special orthogonal group of E , as defined in
Definition 6.4.3 (the isometries with determinant +1), and let SE(E) be
the subgroup of SA(E) (the special Euclidean group of E) consisting of the
−
→ −→
affine isometries f such that f ∈ SO( E ). Prove that SE(E) is isomorphic
−
→ −
→ −→ −
→
to the semidirect product E ×θ SO( E ), where θ: SO( E ) → Aut( E ) is
defined as in (vi).
Problem 2.20 The purpose of this problem is to study certain affine maps
of A2 .
(1) Consider affine maps of the form
µ ¶ µ ¶µ ¶ µ ¶
x1 cos θ − sin θ x1 b1
7→ + .
x2 sin θ cos θ x2 b2
Prove that such maps have a unique fixed point c if θ 6= 2kπ, for all integers
k. Show that these are rotations of center c, which means that with respect
to a frame with origin c (the unique fixed point), these affine maps are
represented by rotation matrices.
(2) Consider affine maps of the form
µ ¶ µ ¶µ ¶ µ ¶
x1 λ cos θ −λ sin θ x1 b1
7→ + .
x2 µ sin θ µ cos θ x2 b2
Prove that such maps have a unique fixed point iff (λ + µ) cos θ 6= 1 + λµ.
Prove that if λµ = 1 and λ > 0, there is some angle θ for which either there
is no fixed point, or there are infinitely many fixed points.
(3) Prove that the affine map
µ ¶ µ ¶µ ¶ µ ¶
x1 8/5 −6/5 x1 1
7→ +
x2 3/10 2/5 x2 1
has no fixed point.
(4) Prove that an arbitrary affine map
µ ¶ µ ¶µ ¶ µ ¶
x1 a1 a2 x1 b1
7→ +
x2 a3 a4 x2 b2
has a unique fixed point iff the matrix
µ ¶
a1 − 1 a2
a3 a4 − 1
is invertible.
−
→
Problem 2.21 Let (E, E ) be any affine space of finite dimension. For
every affine map f : E → E, let Fix(f ) = {a ∈ E | f (a) = a} be the set of
fixed points of f .
 2.12. Problems 59

(i) Prove that if Fix(f ) 6= ∅, then Fix(f ) is an affine subspace of E such

that for every b ∈ Fix(f ),
−
→
Fix(f ) = b + Ker ( f − id).
(ii) Prove that Fix(f ) contains a unique fixed point iff
−
→
Ker ( f − id) = {0},
−
→
i.e., f (u) = u iff u = 0.
Hint. Show that
−→
Ωf (a) − Ωa = Ωf (Ω) + f (Ωa) − Ωa,
for any two points Ω, a ∈ E.
−
→ −
→
Problem 2.22 Given two affine spaces (E, E ) and (F, F ), let A(E, F )
be the set of all affine maps f : E → F .
−
→ −
→
(i) Prove that the set A(E, F ) (viewing F as an affine space) is a vector
space under the operations f + g and λf defined such that
(f + g)(a) = f (a) + g(a),
(λf )(a) = λf (a),
for all a ∈ E.
(ii) Define an action
−
→
+: A(E, F ) × A(E, F ) → A(E, F )
−
→
of A(E, F ) on A(E, F ) as follows: For every a ∈ E, every f ∈ A(E, F ),
−
→
and every h ∈ A(E, F ),
(f + h)(a) = f (a) + h(a).
−→
Prove that (A(E, F ), A(E, F ), +) is an affine space.
Hint. Show that for any two affine maps f, g ∈ A(E, F ), the map fg defined
such that
fg(a) = f (a)g(a)
−
→
(for every a ∈ E) is affine, and thus fg ∈ A(E, F ). Furthermore, fg is the
−→
unique map in A(E, F ) such that
f + fg = g.
−
→ −
→ −
→
(iii) If E has dimension m and F has dimension n, prove that A(E, F )
has dimension n + mn = n(m + 1).
60 2. Basics of Affine Geometry

Problem 2.23 Let (c1 , . . . , cn ) be n ≥ 3 points in Am (where m ≥ 2).

Investigate whether there is a closed polygon with n vertices (a1 , . . . , an )
such that ci is the middle of the edge (ai , ai+1 ) for every i with 1 ≤ i ≤ n−1,
and cn is the middle of the edge (an , a0 ).
Hint. The parity (odd or even) of n plays an important role. When n is
odd, there is a unique solution, and when n is even, there are no solutions or
infinitely many solutions. Clarify under which conditions there are infinitely
many solutions.
Problem 2.24 Given an affine space E of dimension n and an affine
frame (a0 , . . . , an ) for E, let f : E → E and g: E → E be two affine maps
represented by the two (n + 1) × (n + 1) matrices
µ ¶ µ ¶
A b B c
and
0 1 0 1
w.r.t. the frame (a0 , . . . , an ). We also say that f and g are represented by
(A, b) and (B, c).
(1) Prove that the composition f ◦ g is represented by the matrix
µ ¶
AB Ac + b
.
0 1
We also say that f ◦ g is represented by (A, b)(B, c) = (AB, Ac + b).
(2) Prove that f is invertible iff A is invertible and that the matrix
representing f −1 is
µ −1
−A−1 b
¶
A
.
0 1
We also say that f −1 is represented by (A, b)−1 = (A−1 , −A−1 b). Prove
that if A is an orthogonal matrix, the matrix associated with f −1 is
µ >
−A> b
¶
A
.
0 1
Furthermore, denoting the columns of A by A1 , . . . , An , prove that the
vector A> b is the column vector of components
(A1 · b, . . . , An · b)
(where · denotes the standard inner product of vectors).
(3) Given two affine frames (a0 , . . . , an ) and (a00 , . . . , a0n ) for E, any affine
map f : E → E has a matrix representation (A, b) w.r.t. (a0 , . . . , an ) and
(a00 , . . . , a0n ) defined such that b = a00 f (a0 ) is expressed over the basis
(a00 a01 , . . . , a00 a0n ), and ai j is the ith coefficient of f (a0 aj ) over the basis
(a00 a01 , . . . , a00 a0n ). Given any three frames (a0 , . . . , an ), (a00 , . . . , a0n ), and
(a000 , . . . , a00n ), for any two affine maps f : E → E and g: E → E, if f has the
matrix representation (A, b) w.r.t. (a0 , . . . , an ) and (a00 , . . . , a0n ) and g has
the matrix representation (B, c) w.r.t. (a00 , . . . , a0n ) and (a000 , . . . , a00n ), prove
 2.12. Problems 61

that g ◦ f has the matrix representation (B, c)(A, b) w.r.t. (a0 , . . . , an ) and
(a000 , . . . , a00n ).
(4) Given two affine frames (a0 , . . . , an ) and (a00 , . . . , a0n ) for E, there is
a unique affine map h: E → E such that h(ai ) = a0i for i = 0, . . . , n, and
we let (P, ω) be its associated matrix representation with respect to the
frame (a0 , . . . , an ). Note that ω = a0 a00 , and that pi j is the ith coefficient
of a00 a0j over the basis (a0 a1 , . . . , a0 an ). Observe that (P, ω) is also the ma-
trix representation of idE w.r.t. the frames (a00 , . . . , a0n ) and (a0 , . . . , an ),
in that order. For any affine map f : E → E, if f has the matrix repre-
sentation (A, b) over the frame (a0 , . . . , an ) and the matrix representation
(A0 , b0 ) over the frame (a00 , . . . , a0n ), prove that
(A0 , b0 ) = (P, ω)−1 (A, b)(P, ω).
Given any two affine maps f : E → E and g: E → E, where f is invertible,
for any affine frame (a0 , . . . , an ) for E, if (a00 , . . . , a0n ) is the affine frame
image of (a0 , . . . , an ) under f (i.e., f (ai ) = a0i for i = 0, . . . , n), letting (A, b)
be the matrix representation of f w.r.t. the frame (a0 , . . . , an ) and (B, c) be
the matrix representation of g w.r.t. the frame (a00 , . . . , a0n ) (not the frame
(a0 , . . . , an )), prove that g◦f is represented by the matrix (A, b)(B, c) w.r.t.
the frame (a0 , . . . , an ).

Remark: Note that this is the opposite of what happens if f and g are
both represented by matrices w.r.t. the “fixed” frame (a0 , . . . , an ), where
g ◦ f is represented by the matrix (B, c)(A, b). The frame (a00 , . . . , a0n ) can
be viewed as a “moving” frame. The above has applications in robotics, for
example to rotation matrices expressed in terms of Euler angles, or “roll,
pitch, and yaw.”