AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L.

Rueda

CHAPTER II: AFFINE AND EUCLIDEAN GEOMETRY

 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

2. AFFINE TRANSFORMATIONS

2.1 Definition and first properties

Definition Let (A, V, φ) and (A0, V 0, φ0) be two real affine spaces. We will say
that a map
f : A −→ A0
is an affine transformation if there exists a linear transformation f¯: V −→ V 0
such that:
f¯(P Q) = f (P ) f (Q), ∀P, Q ∈ A.
This is equivalent to say that for every P ∈ A and every vector ū ∈ V we
have
f (P + u) = f (P ) + f¯(u).
We call f¯ the linear transformation associated to f .
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda
Proposition Let (A, V, φ) and (A0 , V 0 , φ0 ) be two affine subspaces and let f : A −→ A0 be
an affine transformation with associated linear transformation f¯: V −→ V 0 . The following
statements hold:
1. f is injective if and only if f¯ is injective.
2. f is surjective if and only if f¯ is surjective.
3. f is bijective if and only if f¯ is bijective.

Proposition Let g : A −→ A0 and f : A0 −→ A00 be two affine transformations. The composi-

tion f ◦g : A −→ A00 is also an affine transformation and its associated linear transformation
is f ◦ g = f¯ ◦ ḡ.
Proposition Let f, g : A −→ A0 be two affine transformations which coincide over a point
P , f (P ) = g(P ), and which have the same associated linear transformation f¯ = ḡ. Then
f = g.
Proof Every X ∈ A verifies:

f (X) = f (P + P X) = f (P ) + f¯(P X) = g(P ) + ḡ(P X) = g(X).

 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

2.2 Matrix associated to an affine transformation

Let (A, V, φ) and (A0, V 0, φ0) be two affine subspaces and let f : A −→ A0 be
an affine transformation with associated linear transformation f¯: V −→ V 0.
We consider affine coordinate systems R = {O; B}, B = {e1, . . . , en} and
R0 = {O0; B 0}, B 0 = {e01, . . . , e0m} of the spaces A, A0 respectively. Let us
assume that:
O0f (O) = b1e01 + · · · + bme0m,
 f¯(e1) = a11e01 + · · · + am1e0m

...
 ¯
f (en) = a1ne01 + · · · + amne0m
Let P be the point with coordinates (x1, . . . , xn) with respect to R and let
(y1, . . . , ym) be the coordinates of f (P ) ∈ A0.
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

Then:     
1 1 0 ··· 0 1
 y1   b1 a11 · · · a1n   x1
    
 =  ..

 .. ... . . . ...   ..
 .   .  .


ym bm am1 · · · amn xn
We will write
 
1 0 ··· 0
 t 
1 0 b1 a11 · · · a1n
 
0
Mf (R, R ) = =
 
... ... . . . ...
b Mf¯(B, B 0)

 
bm am1 · · · amn

where b are the coordinates of f (O) in the coordinate system R0 and Mf¯(B, B 0)
is the matrix associated to the linear transformation f¯ taking in V the basis
B and in V 0 the basis B 0.
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

Example 1
Let (A, V, φ) be an affine space with affine coordinate system R = {O; B =
{e1, e2, e3}}, and let (A0, V 0, φ0) be an affine space with affine coordinate sys-
tem R0 = {O0; B 0 = {e01, e02}}. Is the transformation f : A −→ A0, f (x, y, z) =
(x − 2y + 5, x − z + 1) an affine transformation? Give its associated linear
transformation and obtain the matrix associated to f in the coordinate sys-
tems R, R0.

Solution.
To see if f is an affine transformation we have to check if there exists a
linear transformation f¯: V −→ V 0 such that f (P )f (Q) = f¯ P Q for every

pair of points P, Q ∈ A. We take P (x1, y1, z1) and Q(x2, y2, z2) then
P Q = Q − P = (x2 − x1, y2 − y1, z2 − z1)
and
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

f (P )f (Q) = f (Q) − f (P ) = f (x2, y2, z2) − f (x1, y1, z1)

= (x2 − 2y2 + 5, x2 − z2 + 1) − (x1 − 2y1 + 5, x1 − z1 + 1)
= ((x2 − x1) − 2 (y2 − y1) , (x2 − x1) − (z2 − z1))
= f¯ (x2 − x1, y2 − y1, z2 − z1) .
Therefore, f is an affine transformation and its associated linear transfor-
mation is f¯ (x, y, z) = (x − 2y, x − z).

The coordinates of the origin O in the coordinate system R are the cooor-
dinates of the vector OO = (0, 0, 0) in the basis B where e1 = (1, 0, 0)B ,
e2 = (0, 1, 0)B and e3 = (0, 0, 1)B .
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

Then:
f (O) = f (0, 0, 0) = (5, 1),
f¯(e1) = f¯(1, 0, 0) = (1, 1),
f¯(e2) = f¯(0, 1, 0) = (−2, 0),
f¯(e3) = f¯(0, 0, 1) = (0, −1).
So,  
1 0 0 0
Mf (R, R0) =  5 1 −2 0  .
1 1 0 −1
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

Example 2
Let (A, V, φ) be an affine space with affine coordinate system R = {O; B =
{e1, e2}}, and let (A0, V 0, φ0) be an affine space with affine coordinate system
R0 = {O0; B 0 = {e01, e02, e03}}. Determine the affine transformation f : A −→
A0, such that
f (1, 2) = (1, 2, 3),
f¯(e1) = e01 + 4e02,
f¯(e2) = e01 − e02 + e03.
Find the matrix associated to f in the coordinate systems R, R0.
Solution.
We know the value of f at the point P (1, 2) and the linear transformation
associated to f , therefore we can determine f .
f¯(1, 0) = (1, 4, 0)B 0 ,
f¯(0, 1) = (1, −1, 1)B 0 ,
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

To calculate the matrix associated to f we need to compute f (O). We have:

f (P ) = f (O) + f¯(OP ) = f (O) + f¯(1e1 + 2e2)
f is affine
= f (O) + f¯(e1) + 2f¯(e2)
f¯ is linear
= f (O) + (1, 4, 0) + 2(1, −1, 1)
so
f (O) = f (1, 2) − (1, 4, 0) − 2(1, −1, 1)
= (1, 2, 3) − (1, 4, 0) − 2(1, −1, 1)
= (−2, 0, 1) .
therefore  
1 0 0
 −2 1 1 
 
Mf (R, R0) =  .
 0 4 −1 
1 0 1
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

As    
1 0 0   1
1
 −2 1 1  x + x − 2
   
1 2
 x1  = 
 
 0 4 −1   4x1 − x2 
 
x2
1 0 1 x2 + 1
we have:
f (x1, x2) = (x1 + x2 − 2, 4x1 − x2, x2 + 1) .
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

Example 3
Let (R2, R2, φ) be an affine space with affine coordinate system R = {O; B =
{e1, e2}}. Determine the affine transformation f : R2 −→ R2 such that
f (P0) = f (1, 1) = (7, 5), f (P1) = f (1, 2) = (11, 4), f (P2) = f (2, 1) = (8, 8).
Solution.
To determine an affine transformation f : R2 −→ R2 we need three points
that form an affine coordinate system and their images.
First method
Since P0P1 = (0, 1) and P0P2 = (1, 0) , then we know that
f¯(e1) = f¯(1, 0) = f¯(P0P2) = f (P2) − f (P0) = (1, 3),
f¯(e2) = f¯(0, 1) = f¯(P0P1) = f (P1) − f (P0) = (4, −1).
Also OP0 = e1 + e2 so we have:
f (P0) = f (O) + f¯(OP0) = f (O) + f¯(e1 + e2)
= f (O) + f¯(e1) + f¯(e2)
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

then
f (O) = f (P0) − f¯(e1) − f¯(e2) = (7, 5) − (1, 3) − (4, −1)
= (2, 3).
finally,  
1 0 0
Mf (R, R) =  2 1 4 
3 3 −1
and f (x1, x2) = (2 + x1 + 4x2, 3 + 3x1 − x2).
Second method
The set R0 = {P0(1, 1), P1(1, 2), P2(2, 1)} is an affine coordinate system since
P0P1 = (0, 1) and P0P2 = (1, 0) is a basis of R2. We have:
f (P0) = f (1, 1) = (7, 5),
f¯(P0P1) = f (P0)f (P1) = f (P1) − f (P0) = (11, 4) − (7, 5) = (4, −1),
f¯(P0P2) = f (P0)f (P2) = f (P2) − f (P0) = (8, 8) − (7, 5) = (1, 3).
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

Therefore,  
1 0 0
Mf (R, R0) =  7 4 1 
5 −1 3
To obtain Mf (R, R)
Mf (R, R) = Mf (R0, R)M (R, R0) = Mf (R0, R)M (R0, R)−1
  −1  
1 0 0 1 0 0 1 0 0
=  7 4 1  1 0 1  =  2 1 4 
5 −1 3 1 1 0 3 3 −1
Therefore f (x1, x2) = (2 + x1 + 4x2, 3 + 3x1 − x2).
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

Example 4
Determine the affine transformation f : A3 −→ A3 which transforms the
points P0(0, 0, 0), P1(0, 1, 0), P2(1, 1, 1) and P3(1, 1, 4) in the points Q0(2, 0, 2),
Q1(2, −1, 1), Q2(2, 1, 3) and Q3(5, 7, 6) respectively.

Solution.
To determine an affine transformation f : A3 −→ A3 we will need four points
that form an affine coordinate system and their images.
The set R0 = {P0(0, 0, 0), P1(0, 1, 0), P2(1, 1, 1), P3(1, 1, 4)} is an affine coordi-
nate system as P0P1 = (0, 1, 0), P0P2 = (1, 1, 1) and P0P3 = (1, 1, 4) is a basis
of R3 because
dim(P0P1, P0P2, P0P3) = 3.
We have:
f (P0) = Q0 = (2, 0, 0),
f¯(P0P1) = f (P0)f (P1) = f (P1) − f (P0) = Q1 − Q0 = (0, −1, −1),
f¯(P0P2) = f (P0)f (P2) = f (P2) − f (P0) = Q2 − Q0 = (0, 1, 1),
f¯(P0P3) = f (P0)f (P3) = f (P3) − f (P0) = Q3 − Q0 = (3, 7, 4).
Therefore  
1 0 0 0
2 0 0 3
 
0
Mf (R , R) = 
0 −1 1 7

0 −1 1 4
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

To obtain Mf (R, R)
Mf (R, R) = Mf (R0, R)M (R, R0) =
  −1
1 0 0 0 1 0 0 0
 2 0 0 3  0 0 1 1
  
= 
 0 −1 1 7   0 1 1 1
 
0 −1 1 4 0 0 1 4
 
1 0 0 0
 2 −1 0 1 
 
= 
 0 0 −1 2 

0 1 −1 1

Thus, f (x1, x2, x3) = (2 − x1 + x3, −x2 + 2x3, x1 − x2 + x3).

 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

2.3 Affine invariant subspaces

Proposition Let (A, V, φ) and (A0, V 0, φ0) two affine spaces and let f : A −→
A0 be an affine transformation with associated linear transformation f¯: V −→
V 0. The following statements hold:
1. If L ⊂ A is an affine subspace of A then
f (L) = {P 0 ∈ A0 | there exists P ∈ L such that f (P ) = P 0}
is an affine subspace of A0.
2. If L0 ⊂ A0 is an affine subspace of A0 then the set
L = {P ∈ A | f (P ) ∈ L0}
is an affine subspace of A, called the inverse image of L0 and denoted
f −1 (L0).

Definition Let (A, V, φ) be an affine space and f : A −→ A an affine trans-

formation. We will say that a point P ∈ A is a fixed point of f if f (P ) = P .
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

Proposition Let (A, V, φ) be an affine space and f : A −→ A an affine trans-

formation. The set of fixed points of f ; this is,
F = {X ∈ A | f (X) = X}
is an affine subspace of A with associated vector space the subspace of V
of eigenvectors of f¯ associated to the eigenvalue λ = 1.

Strategy to search for fixed points Let R = {O; B} be a coordinate system

of A. Let  t 
1 0
Mf (R) =
b A
be the matrix associated to f , where A is the matrix associated to the linear
transformation f¯ in the basis B.
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

If P is a fixed point, the following statement holds:

P = f (P ) = f (O + OP ) = f (O) + f¯(OP )
= b + A · OP
or equivalently,
0 = (A − I) OP + b
which is the equation that the fixed points of f must satisfy.

Example
Obtain the fixed points of the affine transformation
f (x, y) = (−2y + 1, x + 3y − 1).
Solution.
The matrix associated to f is
 
1 0 0
Mf (R, R) =  1 0 −2 
−1 1 3
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

and the matrix associated to the linear transformation f¯ is

 
0 −2
A= .
1 3

The fixed points of f are the solutions P (x, y) of the following matrix equa-
tion:
0 = (A − I) P + b
this is
      
0 −1 −2 x 1
= + ⇐⇒ x + 2y − 1 = 0
0 1 2 y −1
Therefore, the points of the line x + 2y − 1 = 0 are the fixed points of f .
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

Definition Let (A, V, φ) be an affine space, f : A −→ A an affine transforma-

tion and S an affine subspace of A. We will say that S is an invariant affine
subspace of f if f (S) ⊂ S.

Observation Let f : A −→ A be an affine transformation with associated

linear transformation f¯: V −→ V and S an affine subspace of A, which
contains the point P and whose associated vector space is generated by
the vectors u1, . . . , ur ; this is,
S ≡ P + hu, . . . , ur i .
Then the affine subspace f (S) contains the point f (P ) and is generated by
the vectors f¯(u1), . . . , f¯(ur ); this is,
f (S) = f (P ) + f¯(u1), . . . , f¯(ur ) .


 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

Then S is invariant under f if and only if

¯


1. f (u1), . . . , f (ur ) ⊂ hu, . . . , ur i
2. P f (P ) ∈ hu, . . . , ur i
In particular, a line r ≡ P + hui is invariant under f if and only if
1. f (u) ⊂ hui ⇐⇒ f¯(u) = λu; this is , ū is an eigenvector of the linear
¯



transformation f¯
2. P f (P ) ∈ hui
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

Example
Obtain the invariant subspaces of the transformation f of the former exam-
ple.
Solution.
To search for the invariant subspaces of f first we compute the eigenvalues
of f¯. The characteristic polynomial A is
 
 −λ −2 
det(A − λI) =   = λ2 − 3λ + 2 = (λ − 1) (λ − 2)
1 3−λ 
and, therefore, the eigenvalues of A are λ = 1, 2.
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

The corresponding eigenspaces of f¯ are


V (1) = v | (A − 1I)v = 0
      
−1 −2 x 0
= (x, y) such that =
1 2 y 0
= {(x, y) such that x + 2y = 0} = h(2, −1)i

V (2) = v | (A − 2I)v = 0
      
−2 −2 x 0
= (x, y) such that =
1 1 y 0
= {(x, y) such that x + y = 0} = h(1, −1)i
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

On the other hand

P f (P ) = f (P ) − P = (−2y + 1, x + 3y − 1) − (x, y)
= (−x − 2y + 1, x + 2y − 1) ∈ V (2)
as the components of vector P f (P ) satisfy the equation of V (2).

Therefore, the lines with associated vector space V (2) = h(1, −1)i are inva-
riant lines of f because
f¯(1, −1) = 2(1, −1)
P f (P ) ∈ V (2)
If x + 2y − 1 = 0 (this is the line of fixed points of f ) then P f (P ) = 0 ∈ V (1).

The line of fixed points, is in particular, an invariant line of f .

 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

Example
Let (A3, V, φ) be an affine space and R = {O; {e1, e2, e3}} a coordinate sys-
tem of A3. Determine the affine transformation f : A3 −→ A3 such that the
plane π ≡ x + 2y − z = 1 is a plane of fixed points of f and the vector e1 is
an eigenvector of f¯ associated to the eigenvalue 3.

Solution.
The plane π is a plane of fixed points, any point of the plane is a fixed
point of f . For example, the point P (1, 0, 0) ∈ π is a fixed point of f ; this is,
f (P ) = P . Also, we know that the vectors of the vector subspace associated
to π, this is, vectors from the plane π ≡ x + 2y − z = 0, are eigenvectors
associated to the eigenvalue 1.
For example, for
u = (1, 0, 1) ∈ π,v = (0, 1, 2) ∈ π
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

we have: f¯(u) = ū and f¯(v) = v, this is,

f¯(1, 0, 1) = (1, 0, 1) and f¯(0, 1, 2) = (0, 1, 2),
and, we also know that f¯(e1) = 3e1; this is, f¯(1, 0, 0) = 3(1, 0, 0).
As B 0 = {e1, u, v} is a basis of V , we consider the coordinate system R =
{P ; B 0}. We have:  
1 0 0 0
1 3 1 0
 
0
Mf (R , R) =  .
0 0 0 1
0 0 1 2
Also, we have:
Mf (R) = Mf (R0, R)M (R, R0) =
  −1  
1 0 0 0 1 0 0 0 1 0 0 0
 1 3 1 0  1 1 1 0  −2 3 4 −2 
    
=  = .
 0 0 0 1  0 0 0 1  0 0 1 0 
 
0 0 1 2 0 0 1 2 0 0 0 1
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

Checking the answer. Obviously f¯(e1) = 3e1 and it holds:

    
1 0 0 0 1 1
 −2 3 4 −2   1   1 
    
f (P ) = f (1, 0, 0) =    =   = P
 0 0 1 0  0   0 
0 0 0 1 0 0
    
3 4 −2 1 1
f¯(~u) = f¯(1, 0, 1) =  0 1 0   0  =  0  = u
0 0 1 1 1
    
3 4 −2 0 0
f¯(~v ) = f¯(1, 0, 1) =  0 1 0   1  =  1  = v.
0 0 1 2 2
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

2.4 Some examples of affine transformations of a space in itself

Let (A, V, φ) be an affine space and let f : A −→ A be an affine transfor-
mation with associated linear map f¯: V −→ V . Let Mf (R) be the matrix
associated to f in the coordinate system R.

2.4.1 Translations
Given a vector v ∈ V , we define the translation of vector v as the transfor-
mation Tv : A −→ A such that f (P ) = P + v.

Proposition Every translation Tv is an affine transformation where the asso-

ciated linear map is the identity.
Proof Given P, Q ∈ A the following statement holds:
T v (P Q) = Tv (P )Tv (Q) = Tv (P )P + P Q + QTv (Q)
= −v + P Q + v = P Q.
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

2.4.2 Projections

An affine transformation f : A −→ A is a projection if f 2 = f . Therefore, if

Mf (R) is idempotent (Mf (R)2 = Mf (R)). The linear transformation asso-
ciated to a projection is also idempotent f¯2 = f¯.

2.4.3 Homotethy

An affine transformation f : A −→ A is an homothety of ratio r if f¯ = rIdV ,

where IdV is the identity map on V .

Remark
A homotethy of ratio r has only one fixed point C called center of the ho-
mothety. The image of any other point P is obtained as follows:
f (P ) = C + rCP .
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

How to calculate the center of a homothety

Let C ∈ A be the center of the homothety f . We have:

C = f (C) = f (P + P C) = f (P ) + f¯(P C) = f (P ) + rP C =⇒ f (P )C = rP C
We also have:
P C = P f (P ) + f (P )C = P f (P ) + rP C =⇒ (1 − r)P C = P f (P ).
Therefore, the fixed point C verifies
1
C=P+ P f (P ).
1−r
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

Example 1
Study whether the affine transformation f (x, y, z) = (1 + 23 x, −1 + 23 y, 2 + 23 z)
has a fixed point or an invariant subspace.
Solution.
The matrix associated to f is
 
1 0 0 0
 1 32 0 0 
 
Mf (R) = 
 −1 0 23 0 

2 0 0 23
and the matrix associated to the linear transformation f¯ is
2
Mf (B) = IR3 .
3
Therefore, f is a homothety of ratio r = 32 . The center of the homothety is:
1
C=P+ P f (P )
1 − 23
for every P ∈ A. We take P (0, 0, 0) then f (P ) = f (0, 0, 0) = (1, −1, 2) and
P f (P ) = (1, −1, 2), therefore
3
C= (1, −1, 2) = (3, −3, 6).
3−2

The invariant subspaces of f are:

- The center C since it is a fixed point
- The lines that contain the center
- The planes that contain the center

Example 2
Study whether the affine transformation f (x, y, z) = (x + 1, y + 2, z + 3) has
fixed points or invariant subspaces.
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

Solution.
The matrix associated to f is
 
1 0 0 0
1 1 0 0
 
Mf (R) = 
2 0 1 0

3 0 0 1

and the matrix associated to the linear transformation f¯ is the identity. The-
refore, f is a translation of vector v = Of (O) = (1, 2, 3) − (0, 0, 0) = (1, 2, 3).
The translations do not have fixed points.
The invariant subspaces of f are:
- The lines whose direction is the direction of the translation vector; this is,
lines r ≡ P + hvi.
- The planes whose direction contains the translation vector; this is, planes
of the form π ≡ P + hv, wi.
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

Example 3
Study whether the affine transformation f (x, y, z) = (−2 + 2x − y, −4 + 2x −
y, z) has fixed points or invariant subspaces.
Solution.
The matrix associated to f is
 
1 0 0 0
 −2 2 −1 0 
 
Mf (R) = 
 −4 2 −1 0 

0 0 0 1

and the matrix associated to the linear transformation f¯ is

 
2 −1 0
A = Mf (B) =  2 −1 0  .
0 0 1
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

The eigenvalues of A are the roots of:

 
 2 − λ −1 0 
det(A − λI) =  2 −1 − λ 0  = −λ (λ − 1)2 .

 0 0 1−λ 

The matrix A may be idempotent because its eigenvalues are λ = 0 and

1. We can check that A2 = A and therefore, A is idempotent. So f is a
projection.
The fixed points of f satisfy the following equation:
0 = (A − I) P + b
this is,       
0 1 −1 0 x −2
 0  =  2 −2 0   y  +  −4 
0 0 0 0 z 0
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

equivalently 
 0=x−y−2
0 = 2x − 2y − 4
0=0


Therefore the plane π of equation x − y − 2 = 0 is a plane of fixed points

(its associated vector space is the one of eigenvectors associated to the
eigenvalue λ = 1).
The subspace of eigenvectors associated to the eigenvalue λ = 0 is:
      
 2 −1 0 x 0 
V (0) = (x, y, z) such that  2 −1 0   y  =  0 
0 0 1 z 0
 
= {(x, y, z) such that 2x − y = 0, z = 0}
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

On the other hand

P f (P ) = f (P ) − P = (−2 + 2x − y, −4 + 2x − y, z) − (x, y, z)
= (−2 − x − y, −4 + 2x − 2y, 0) ∈ V (0)
as the components of the vector P f (P ) verify the equation of V (0). There-
fore, the lines with associated vector space V (0) = h(1, 2, 0)i, are invariant
lines of f .
The invariant subspaces of f are:
- The lines with associated vector space V (0) = h(1, 2, 0)i.
- The planes containing invariant lines.
- The plane of fixed points π ≡ x − y − 2 = 0.
- The lines contained in the plane of fixed points are lines of fixed points.
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

Example 4
Obtain the analytic expression of the affine tranformation f : A3 −→ A3 that
sends the origin to the point (3, 1, 1) and whose plane π of cartesian equa-
tion x1 + 2x2 − x3 + 1 = 0, is a plane of fixed points.
Solution.
As the plane π is a plane of fixed points, the vector plane associated with
π is a plane of eigenvectors associated with the eigenvalue λ = 1 of the
associated linear transformation f¯.
As π ≡ P + hu1, u2i with P (0, 0, 1), u1 = (1, 0, 1), u2 = (0, 1, 2) then P ∈ π (this
is, the coordinates of P are a solution of the equation of π) and the vectors
u1,u2 ∈ π (their coordinates are a solution of the associated homogeneous
equation: x1 + 2x2 − x3 = 0).
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

Therefore, we have:
f (0, 0, 0) = (3, 1, 1)
f (0, 0, 1) = (0, 0, 1)
f¯(u1) = u1 =⇒ f¯(1, 0, 1) = (1, 0, 1)
f¯(u2) = u2 =⇒ f¯(0, 1, 2) = (0, 1, 2)
From the first two conditions we obtain
f¯(OP ) = f (P ) − f (O) = (0, 0, 1) − (3, 1, 1)
= (−3, −1, 0).
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

Thus, considering the coordinate system R0 = P ; OP , u1, u2 (notice that



OP , u1, u2 are linearly independent), we obtain:

 
1 0 0 0
 3 −3 1 0 
 
0
Mf (R , R) =  .
 1 −1 0 1 
1 0 1 2

Since  
1 0 0 0
0 0 1 0
 
0
M (R , R) =  ,
0 0 0 1
1 1 1 2
 AFFINE AND PROJECTIVE GEOMETRY, E. Rosado & S.L. Rueda

we have
Mf (R, R) = Mf (R0, R) · M (R0, R)−1
  −1
1 0 0 0 1 0 0 0
 3 −3 1 0   0 0 1 0 
  
 1 −1 0 1   0 0 0 1 
  
1 0 1 2 1 1 1 2
 
1 0 0 0
 6 4 6 −3 
 
= .
 2 1 3 −1 
1 1 2 0
So the analytic expression of f is:
f (x1, x2, x3) = (6 + 4x1 + 6x2 − 3x3, 2 + x1 + 3x2 − x3, 1 + x1 + 2x2).