LECTURE NOTES IN

GEOMETRY

Géza Csima, Ákos G.Horváth

BUTE
2019
Contents

1 Absolute geometry 4
1.1 Axiom system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.1.1 Incidence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.1.2 Order . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.1.3 Congruence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.1.4 Continuity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.1.5 Parallels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.2 Absolute theorems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.3 Orthogonality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2 Hyperbolic geometry 14
2.1 Parallel lines on the hyperbolic plane . . . . . . . . . . . . . . . . . . . . . 14
2.1.1 Perpendicular transverse of ultraparallel lines . . . . . . . . . . . . 15
2.2 Line pencils and cycles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3 Models of the hyperbolic geometry . . . . . . . . . . . . . . . . . . . . . . 17
2.3.1 Orthogonality in the Cayley-Klein model . . . . . . . . . . . . . . . 17
2.3.2 Stereographic projection . . . . . . . . . . . . . . . . . . . . . . . . 18
2.3.3 Inversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.4 Distance and angle of space elements . . . . . . . . . . . . . . . . . . . . . 21
3
2.4.1 Mutual position of space elements (E ) . . . . . . . . . . . . . . . . 21
2.4.2 Mutual position of space elements (H3 ) . . . . . . . . . . . . . . . . 21
2.4.3 Perpendicular transverse of non-intersecting hyperbolic space elements 22
2.4.4 Cross-ratio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.4.5 Distances in the hyperbolic space . . . . . . . . . . . . . . . . . . . 23
2.4.6 Angle and distance in the Cayley-Klein model . . . . . . . . . . . . 24
2.4.7 Hyperboloid model . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.4.8 Pole-polar relation . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.5 Models of the Euclidean plane . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.6 Area on the hyperbolic plane . . . . . . . . . . . . . . . . . . . . . . . . . . 27

2
3 Classification of isometries 29
3.1 Isometries of the Euclidean plane . . . . . . . . . . . . . . . . . . . . . . . 29
3.2 Isometries of the Euclidean space . . . . . . . . . . . . . . . . . . . . . . . 32

4 Analytic geometry 35
4.1 Vectors in the Euclidean space . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.2 Transformation of E3 with one fix point (origin) . . . . . . . . . . . . . . . 37
4.3 Spherical geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.3.1 Spherical Order axioms . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.3.2 Distance, angle and area . . . . . . . . . . . . . . . . . . . . . . . . 41
4.4 n-dimensional Euclidean geometry . . . . . . . . . . . . . . . . . . . . . . . 43
4.5 Classification of quadratic surfaces . . . . . . . . . . . . . . . . . . . . . . 44
4.6 Conic sections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.6.1 Tangents from external point to conic sections . . . . . . . . . . . . 47
4.7 Polyhedrons in n-dimensional Euclidean space . . . . . . . . . . . . . . . . 48
4.7.1 3-dimensional polyhedrons . . . . . . . . . . . . . . . . . . . . . . . 49
4.8 Congruence of polyhedra . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

5 Area and volume 54

5.1 Area on the Euclidean plane . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.2 Area on the hyperbolic plane . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.3 Volume in the Euclidean space . . . . . . . . . . . . . . . . . . . . . . . . . 56

3
1 Absolute geometry

1.1 Axiom system

We will follow the axiom system of Hilbert, given in 1899. The base of it are primitive
terms and primitive relations. We will divide the axioms into five parts:

1. Incidence
2. Order
3. Congruence
4. Continuity
5. Parallels

1.1.1 Incidence
Primitive terms: point, line, plane
Primitive relation: "lies on"(containment)

Axiom I1 For every two points there exists exactly one line that contains them both.

Axiom I2 For every three points there exists exactly one plane that contains all of them.

Axiom I3 If two points of a line lie in a plane, then every point of the line lies in the
plane.

Axiom I4 If two planes have a point in common, then they have another point in com-
mon.

Axiom I5 There exists four points not lying in a plane.

Remark 1.1 The incidence structure does not imply that we have infinitely many points.
We may consider 4 points {A, B, C, D}. Then, the lines can be the two-element subsets
and the planes the three-element subsets.

4
1.1.2 Order
Primitive relation: "betweeness"
Notion: If A, B, C are points of a line then (ABC) := B means the "B is between A
and C".

Axiom Os1 If (ABC) then (CBA).

Axiom Os2 If A and B are two points of a line, there exists at least one point C on the
line AB such that (ABC).

Axiom Os3 Of any three points situated on a line, there is no more than one which lies
between the other two.

Axiom Os4 (Pasch) Let A, B, C be three points not lying on the same line and let e be
a line lying in the plane ABC and not passing through any of the points A, B, C. Then,
if the line e passes trough a point D such that (ADB), it will also pass trough a point E
such that either (BEC) or (AEC).

Remark 1.2 The first three order axioms do not implies that the line has infinitely many
points. Let a model of the line be the edge of a regular pentagon. We define the betweeness
such that (XY Z) if and only if the triangle of X, Y , and Z is an isosceles triangle with
Y as edge vertex.

Lemma 1.3 In the Pasch axiom, either (BEC) or (AEC) but not both at once.

Proof. Indirect: (AEC) ∧ (BF C) ⇒ (DEF )

D
E

B F C

Figure 1.1: Proof: Lemma 1.3

We consider the Pasch axiom for the non-collinear points D, B, F and the line of AC
(see Figure 1.1). Therefore either (BAD) or (BCF ), but we supposed in the axiom that
(ADB) and in the indirect assumption that (BF C).

Definition 1.4 Let A and B be two points. The segment AB consists of those points of
C, for which (ACB).

5
Lemma 1.5 Any segment has a point. D

C
Proof.

1. C is not on the line of AB

2. Let D be such that (ACD) (see Axiom Os2) A F B

3. Let E be such that (DBE) (see Axiom Os2)

4. We use the Pasch axiom for the points A, D, B
E
and the line CE ⇒ ∃F : (AF B)
Figure 1.2: Proof: Lemma 1.5

Corollary 1.6 We can define polygonal chain, polygon, half-plane, ray, angle and con-
vexity.

Theorem 1.7 (Jordan) Any simple, closed polygon splits the plane into two parts. One
of the parts does not contain rays t this is called the inside of the polygon.

1.1.3 Congruence
Primitive relation: "congruence of segments and angles"

Axiom C1 The congruence of segments and angles are equivalence relation.

Axiom C2 Every segment can be laid off upon a given side of a given point of a given
line.

Axiom C3 Let AB and BC be two segments of a line which have no points in common
aside from B and let A0 B 0 and B 0 C 0 be two segments of the same or of another line,
likwise no point other than B 0 in common. Then, if AB ∼
= A0 B 0 and BC ∼
= B 0 C 0 , we have
AC ∼= A0 C 0 .

Axiom C4 An angle congruent to a given angle can be added to a given ray of a given
line of a given plane in both half-planes determined by the given line.

Axiom C5 If, in two triangles ABC and A0 B 0 C 0 the congruences AB ∼ = A0 B 0 , AC ∼

= A0 C 0
and BAC∠ ∼
= B 0 A0 C 0 ∠ hold, then the congruence BC ∼ = B 0 C 0 also holds.

Remark 1.8 With congruence we can

– compare segments and angles.

– introduce "length" like an additive non-negative, real function.
– introduce right angle like an angle divides the plane into four congruent parts.
– use the Descartes coordinate system with analytic geometry.

6
1.1.4 Continuity
Axiom 1.9 (Archimedes) If AB and CD are any segments then there exists an integer
number n such that n segments Cn Dn constructed contiguously from A, along the ray from
A through B, will pass beyond the point B.

Axiom 1.10 (Cantor) If Pi Qi i = 1 . . . ∞ are closed segments of a line such that Pi Qi ⊇

T∞
Pj Qj if j > i, then i=1 Pi Qi 6= ∅.

Axiom 1.11 (Dedekind) For every partition of all the points on a line into two nonempty
sets such that no point of either lies between two points of the other, there is a point of
one set which lies between every other point of that set and every point of the other set.

Theorem 1.12 The Axioms 1.9 and 1.10 are equivalent with the Axiom 1.11.

Proof. Only (D) ⇒ (C)

Let A := {S ∈ e|∃i(SPi Qi )} and B := {T ∈ e|∃i(Pi Qi T )}. Using the axiom, we get a
point C such that ∀S ∈ A and ∀T ∈ B : (SCT ) ⇒ (∀i : C ∈ Pi Qi ).

Remark 1.13 Rational number line satisfies (A) but not (C).
Question: Exists such model of the line that satisfies
incidence, order and congruence but not (A)?
Answer: YES (Veronese)
We introduce ordering as follows:
(x1 , y1 ) > (x2 , y2 ) ⇐⇒ y1 > y2 ∨ ((y1 = y2 ) ∧ x1 > x2 ) K= (0,1)
0 0
A point (x , y ) is between (x1 , y1 ) and (x2 , y2 ) if and O= (0,0) (1,0)=L
0 0
only if either (x1 , y1 ) < (x , y ) < (x2 , y2 ) or (x1 , y1 ) >
(x0 , y 0 ) > (x2 , y2 ).

Remark 1.14 This model do not satisfies (C), e.g.

Pi := (n, 0), Qi := (−n, 1). Figure 1.3: Non-Archimedean line
model

1.1.5 Parallels
Axiom 1.15 (EUC) Let a be any line and A a point not on it. Then, there is exactly
one line in the plane, determined by a and A, that passes through A and does not intersect
a.

Axiom 1.16 (HYP) Let a be any line and A a point not on it. Then, there are at least
two lines in the plane, determined by a and A, that pass through A and do not intersect a.

7
1.2 Absolute theorems

Lemma 1.17 If we connect the points A and B which

lie on different rays of an angle, then inside rays em- B
anating from its vertex O intersect the segment AB.

Proof. Let C be such that (AOC) /Axiom Os2/. We C O A

apply the Pasch axiom for the points A, B, C and the
Figure 1.4: Proof: Lemma 1.17
ray.
Lemma 1.18 (External angle) Any external angle C
M
of a triangle is larger than its any internal angle not
beside it.
Proof. Indirect CAB∠ > CBD∠ where D is such
that (ABD).
We lay off the CBD∠ upon the CAB∠ ⇒ M by A B
using 1.17 lemma. Let M 0 be such that (CBM 0 )
and BM ∼ = M 0 B. Then ABM 0 ∠ = M BD∠ =
M AB∠ ⇒ ABM 0 4 ∼ = BAM 4. Therefore ,
M AB∠ ∼
0
= ABM ∠ ⇒ M , A, M 0 are collinear and
M
A=B. Figure 1.5: Proof: Lemma 1.18
Remark 1.19 If two lines intersect another line at the same angle, then they cannot
intersect each other. ⇒ Exist non-intersecting lines in absolute geometry.
Lemma 1.20 If, in two triangles ABC4 and
,
A0 B 0 C 0 4 the congruences AB ∼
= A0 B 0 , ACB∠ ∼ = C
C*
0 0 0 ∼ 0 0 0
A C B ∠ and either ABC∠ = A B C ∠ or BAC∠ = ∼
B 0 A0 C 0 ∠ hold, then the triangles are congruent.

Proof. Let e.g. ABC∠ ∼ = A0 B 0 C 0 ∠ be true. Let C ∗

, ,
be on B 0 C 0 such that BC ∗ ∼
= B 0 C 0 . Then ABC ∗ 4 ∼ = A B
A0 B 0 C 0 4 /Axiom C5/, therefore AC ∗ B∠ ∼ = A0 C 0 B 0 ∠ Figure 1.6: Proof: Lemma 1.20
and the previous remark implies that C 0 = C ∗ .
Lemma 1.21 In any triangle, the greater angle is A
subtended by the greater side and the greater side sub-
tends the greater angle. b
Proof. Suppose that AC < BC. Let A0 be such that b
∼ 0 0 0
AC = A C and (CA B). C lies on the bisector of AA . C a A, B
Therefore CAA ∠ ∼
0 0
= CA A∠. Using the external angle Figure 1.7: Proof: Lemma 1.22
lemma, we obtain that CA0 A∠ > CBA∠.

8
 D
Lemma 1.22 For any triangle, the sum of the length
of any two sides must be greater than the length of the
b
remaining side.
C
Proof. Let D be such that (BCD) and CD ∼ = AC. a b
∼
Then ACD4 is isosceles, therefore CDA∠ = CAD∠.
But BAD∠ > CAD∠ and using the previous lemma, B c A
we obtain that |BD| > |BA|.
Figure 1.8: Proof: Lemma 1.22

Corollary 1.23 (Polygonal chain inequality) Let A1 , A2 , . . . , An be a polygonal chain.

Pn−1
Then i=1 |Ai Ai+1 | ≥ |A1 An |.

Lemma 1.24 (Arm-lemma) If, in two triangle

ABC4 and A B C 4 the congruences AB ∼
0 0 0
= A0 B 0 ,
BC ∼
= B 0 C 0 hold and ABC∠ < A0 B 0 C 0 ∠, then |AC| <
|A0 C 0 |.

Proof. We may assume that A = A0 and B = B 0 Due

to the lemma 1.17, the line BC intersects AC 0 and we
obtain D.
1. case: (BDC): AC 0 C∠ < BC 0 C∠ ∼
= BCC 0 ∠ <
ACC 0 ∠. Therefore using the triangle inequality for
the triangle ACC 0 4 we obtain that |AC| < |A0 C 0 |.
2. case: (BCD): In any isosceles triangle, the base
angle is smaller then π2 , since the sum of the two base
angle is smaller then the straight angle /External angle
lemma/. In our case BCC 0 ∠ < π
2
⇒ C 0 CD∠ > π
2
⇒
|C 0 D| > |CD|. Using the triangle inequality and, we
obtain that: |AC| < |AD| + |DC| < |AD| + |DC 0 | =
|AC 0 |.
Figure 1.9: Proof: Lemma 1.24

Theorem 1.25 (Legendre I.) For any triangle, the sum of its internal angles must be
less than or equal to π.

Proof. Indirect The internal angle sum in ABC4 is greater than π.

Let |Ai Ai+1 | = |AB| be contiguous segments on the line of AB and Ci be such that
Ai Ci Ai+1 4 ∼
= ACB4. Due to our indirect assumption ∀i = 0, 1, . . . , n − 1 : Ai Ci Ai+1 ∠ >
Ci Ai+1 Ci+1 ∠. Applying the Arm-lemma: |A0 A1 | = |Ai Ai+1 | > |Ci Ci+1 | = |C0 C1 |. Now,
ordering the polygonal chain inequality |A0 C0 | + |C0 C1 | + . . . |Cn−1 An | > n|A0 A1 |, we

9
 C=C0 C1 C n-1

A=A0 B=A1 A2 A n-1 An

Figure 1.10: Proof: Theorem 1.25

obtain that n(|A0 A1 | − |C0 C1 |) < |A0 C0 | + |Cn−1 An | which contradicts the Archimedean
axiom.

Definition 1.26 The defect of a triangle is the difference between π and the internal angle
sum of the given triangle (additive, non-negative).
C
Theorem 1.27 (External angle theorem) Any
external angle of a triangle is larger than or equal to
the sum of the other two internal angles.

Proof. Let D be such that (ABD). In the triangle

A B D
ABC4: π ≥ CAB∠ + ABC∠ + ACB∠. Therefore,
DBC∠ = π − ABC∠ ≥ CAB∠ + ACB∠. Figure 1.11: Proof: Theorem 1.27
Theorem 1.28 (Legendre II.) If, the defect of a triangle is 0, then the defect of any
triangle is 0.

Proof. Assuming, that the defect of ABC4 is 0, we can divide it into two right angled

A B ,
C

, ,
A B

Figure 1.12: Proof: Theorem 1.28

triangle with defect 0. Joining two, we get a rectangle with defect 0. For any triangle
A0 B 0 C 0 4, there exists a rectangle, which contains the triangle /Archimedean axiom/. We
may divide the rectangle into five triangles. Since the defect is additive and non-negative,
the defect of A0 B 0 C 0 4 must be 0.

Remark 1.29 For any non self intersecting polygon, we can define the defect, by dividing
it into n − 2 triangles.

10
Theorem 1.30 The defect of a triangle is 0 ⇐⇒ P K

(EU C) is true.

Proof. Let a be a line and P be a point not on it. f

Let P T be perpendicular to a (T is on a) and P K

A T a B D
be perpendicular to P T . Then, for the Remark 1.19, P

P K ∩ a = ∅.
T D1 D2 D3
⇒ Indirect: Let f be such a ray in T P K∠ that
PD1 =D1 D2 , PD 2 =D2 D3
F ∩ a = ∅. ....

We will find a point D, such that DP K∠ < f P K∠.

If, we define the points Di such that D1 = B and
|P Di | = |Di Di+1 |, then we get for the angle DDi T ∠ =
1
2i−1
DD1 T ∠. Now, f is a ray in BP D4 triangle,
L C=P K
therefore Lemma 1.17 guarantees the intersection.
⇐ Let a be the line of AB and C = P . Furthermore,
−→
let KCB∠ = CBA∠ and LCA∠ = CAB∠. Then CL a
−−→ A B
and CK are such rays, that they do not intersect a
⇒ they belong to the same line and the defect of the Figure 1.13: Proof: Theorem 1.30
triangle is 0.

1.3 Orthogonality
Definition 1.31 Two lines are orthogonal to each other if they divide the plane into four
congruent parts.

Definition 1.32 A line intersecting a plane is orthogonal to the plane, if it is orthogonal

to all the lines, passing through the intersection point.

Theorem 1.33 A line intersecting a plane is orthogonal to the plane, if it is orthogonal

to two lines, passing through the intersection point.

Proof. Let a and b be two intersecting lines on a plane and D be their intersection point.
Let n be orthogonal to both a and b such that D lies on n. Furthermore, let c be a line in
the plane, passing through the point D. There exist always points A ∈ a and B ∈ b such
that c intersects the segment AB ⇒ C. Let P 0 be a point on n such that |P D| = |P 0 D|.
Then P DA4 ∼ = P 0 DA4 and P DB4 ∼= P 0 DB4, since |P D| = |P 0 D|, they have a side
in common and the angles between them is π2 .

11
Therefore |P A| = |P 0 A| and |P B| = |P 0 B|. But then P
P AB4 ∼= P 0 AB4 ⇒ P AC∠ ∼ = P 0 AC∠ ⇒ P AC4 ∼ =
P 0 AC4 ⇒ |P C| = |P 0 C| ⇒ P DC4 ∼
= P 0 DC4 ⇒
n
P DC∠ ∼= P 0 DC∠ = π . 2
D b
Corollary 1.34 In the space, the equidistant surface a B
of two point is the orthogonal plane at the midpoint of A C
them.

Corollary 1.35 The locus of lines, orthogonal to a ,

P
given line at a given point of it is a plane, orthogonal
Figure 1.14: Proof: Theorem 1.33
to the line at that point.
Theorem 1.36 Through a given point A exactly one line can be drawn perpendicularly to
a given plane α.
Proof. 1. case: A does not lie on α
Let a be an arbitrary line on α and T be a point on
a such that AT ⊥a. First, we draw an orthogonal line
to a through T in α, then an orthogonal line to this
line through A with foot point D. Let A0 be a point A

on the line of AD such that |AD| = |A0 D| and S

a
be an arbitrary point on a. A0 T ⊥a, since a is per- D
T

pendicular to the plane of ADT . Then ADT 4 ∼ = A

, S

A0 DT 4 ⇒ |AT | = |A0 T | ⇒ AT S4 ∼
= A0 T S4 ⇒
Figure 1.15: Proof: Theorem 1.36
|AS| = |A0 S| ⇒ ADS4 ∼= A0 DS4 ⇒ AA0 ⊥DS
2. case: A lies on α
Our only solution will be the intersection line of the
planes, perpendicular to two arbitrary lines, passing
through the point A.
Theorem 1.37 (of three perpendiculars) Let a be a line on a given plane and A be
a point not lying in the plane. The the foot points T and D of the perpendiculars from A
to a and α respectively form a perpendicular line to a.
A

Proof. Let S and R be equidistant points on a

from T . Then T AS4 ∼ = T AR4, therefore |AS| = D
|AR|. But this implies that ADS4 ∼= ADR4 ⇒ S a
T
|DS| = |DR|. Now DSR4 is an isosceles triangle R
and DT ⊥SR.
Figure 1.16: Proof: Theorem 1.37

12
Definition 1.38 The orthogonal projection of a point of the space for a given plane is the
common point of the plane and the line, perpendicular to the plane through the point. The
orthogonal projection of a point set is the set of the projected points.

Definition 1.39 A mapping is a collineation if the image of collinear points are collinear
themselves.

Theorem 1.40 The orthogonal projection is a collineation and it preserves incidence.

Proof. Let A and B be two points on a line and their projection be A0 and B 0 respectively.
Then A, A0 , B and B 0 are coplanar.
Indirect: Let the line e be in the plane, perpendicular to A0 B 0 through A0 and F be
the midpoint of the segment A0 B 0 . Reflecting the lines AA0 , A0 B 0 and e respected to F ,
we obtain the lines B 0 C, A0 B 0 and f respectively. Since the orthogonal line to the plane
through a given point is unique, B is on the line B 0 C.
Now, let A, B and C be collinear points. Then the points A, A0 , B, B 0 and A, A0 , C, C 0
are coplanar ⇒ the two plane are the same ⇒ A0 , B 0 and C 0 are collinear points.

A C
B C B
A
, ,
A , C
B , ,
e A B
f
Figure 1.17: Proof: Theorem 1.40

Definition 1.41 Two planes are orthogonal to each other if one of them contains a line,
perpendicular to the other.

Theorem 1.42 For every plane and line, there always exists a plane, which is orthogonal
to the given plane and contains the given line.

Proof. If the orthogonal projection of the line is a single point, then it is orthogonal to
the plane, and any plane which contains the line will be appropriate.
If the orthogonal projection of the line is a line does not coincide with the original line,
then they determine this orthogonal plane.
If the orthogonal projection of the line is a line coincides with the original line, then
construct a perpendicular line to the given line in the given plane at any point if it. The
orthogonal plane to the constructed line through this point will satisfy the conditions.

13
2 Hyperbolic geometry

2.1 Parallel lines on the hyperbolic plane

Let T A be perpendicular to a such that T lies on
−−→ L A K
a. Furthermore let AK be perpendicular to AT
⇒ AK ∩ a = ∅. Consider the rays emanating from A
in T AK∠. They intersect the segment T K for Lemma
T
1.17. Assign a ray to the set A if it intersects a and
to B if not. Then, apply the Dedekind axiom to these Figure 2.1: Limiting parallels
sets ⇒ limiting parallels.
−→ −→ −→ −→ −−→
Definition 2.1 AC k T R, if AC is the limiting parallel to T R. If D ∈ A, then AD called
−→ −→ −→
intersecting line of T R. If E ∈ B, then AE called ultraparallel line to T R.
−→ −−→ −→ −−→
Lemma 2.2 If AB k CD, then for any point C ∗ on the line of CD: AB k C ∗ D.

Proof. One can distinguish 2 cases, both can be proved by the Pasch axiom.
Theorem 2.3 The parallelism of rays is symmetric and transitive.
−−→ −→
Proof. Symmetry: Let AK k BL be such that B TB
L
L and K are points "far enough". Draw the angle O
bisector of BAK∠ and ABL∠, they intersect each
−−→
other in O. Let TA and TB be points on AK and A TA M K
−→
BL respectively, such that OTA ⊥AK and OTB ⊥BL.
−−→ −−→
Then TA K k TB L for the previous lemma. But then
−−→ −−→
TB L k TA K, since it is symmetric to the angle bisector
of TB OTA ∠.
Transitivity (Only in the plane and only for one
−−→ −→ −→ −−→ ? −−→ −−→
case): AK k BL ∧ BL k CM ⇒ AK k CM
−−→ −−→ −−→ −→
1) If CM ∩ AK 6= ∅ |{z}
⇒ CM ∩ BL 6= ∅.
Pasch
2) If f is a ray, emanating from C in BCM ∠ ⇒
−→ −−→ −−→
f ∩ BL 6= ∅ ⇒ B 0 and B 0 L k AK ⇒ f also inter-
−−→ Figure 2.2: Proof: Theorem 2.3
sects AK
Definition 2.4 Two line is parallel to each other if they have parallel rays.

14
2.1.1 Perpendicular transverse of ultraparallel lines
Theorem 2.5 (Bolyai) Every pair of ultraparallel lines has a unique line, perpendicular
to both lines (perpendicular transverse).
Proof. We use without proof, that the angle, formed
by a given line and the intersecting rays, emanating
B S L
from a point, not on the line, changes continuously.
−−→ −→
Let the AK and BL be two rays on two ultraparallel F
π π
lines, such that BAK∠ = and ABL∠ <
2
Then 2
.
−−→
exists a point M on the ray AK such that AM B∠ ∼
= A T M K

M BL∠. Let F be the midpoint of the segment BM . Figure 2.3: Proof: Theorem 2.5
Let b be a perpendicular line to AK such that F lies
on b. Then T F M 4 ∼= SF B4 ⇒ BSF ∠ = π . 2
, ,
Definition 2.6 A quadrilateral, with two equal sides A B
G
perpendicular to the same (base) side called Saccheri
quadrilateral.

Lemma 2.7 1. AA0 B 0 ∠ ∼

= BB 0 A0 ∠ < π
2

2. A0 G ∼
= GB 0 ∧AF ∼
= F B ⇒ F G⊥AB∧F G⊥A0 B 0 A F B

Figure 2.4: Proof: Lemma 2.7

3. |A0 B 0 | > |AB|
Proof. BAB 0 4 ∼ = ABA0 4, because AB is common, AA0 ∼ = BB 0 and A0 AB∠ ∼= B 0 BA∠ =
π
2
⇒ AB 0 ∼= A0 B ∧ B 0 AB∠ ∼ = B 0 BA0 ∠ ⇒ AA0 B 0 4 ∼
= A0 BA∠ ⇒ A0 AB 0 ∠ ∼ = BB 0 A0 4 ⇒
1.
Similarly, AA0 F 4 ∼
= BB 0 F 4 and therefore A0 F G4 ∼
= B 0 F G4 ⇒ A0 GF ∠ ∼
= F GB 0 ∠ =
π
2
⇒ 2.
Consider now the triangles A0 AB4 and B 0 BA0 4. Since B 0 BA0 ∠ + A0 BA∠ = π
2
and
0 0 π 0 0
AA B∠ + A BA∠ < 2
⇒ AA B∠ < B BA∠. Using the Arm-lemma, we obtain 3.
M
,
K ,
A
K E C A ,,
A

,
L F D B B

Figure 2.5: Hilbert construction

15
Construction. (Hilbert): Suppose that AB⊥r, KAB∠ < π
2
, (KAA0 ), A0 B 0 ⊥r and
|A00 B| = |AB|. Then (A0 A00 B 0 ), because ABB 0 A00 is a Saccheri quadrilateral ⇒ BAA00 ∠ <
π
2
< BAA0 ∠. Let s0 be a line such that KAB∠ ∼ = K 0 A00 B 0 ∠. Then s ∩ s0 6= ∅, since
−→ −−→ −−→ −−→ −−0−→0 −−00−→0
BL ∩ AK = ∅ ⇒ BK k AK and B K k A K , furthermore LBK∠ ∼ = LB 0 K 0 ∠. Then
−−→ −−−→
BK and B 0 K 0 are ultraparallel rays ⇒ s ∩ B 0 K 0 6= ∅ ⇒ P . Using the Pasch axiom in
BAA0 4 with B 0 K 0 we obtain that B 0 K 0 intersects either AA0 or AB. If it is AA0 , then we
are done, otherwise in KBA4 we use the Pasch axiom again, but B 0 K 0 cannot intersect
BK ⇒ it must intersect AK. Let C be s ∩ s0 and D ∈ r b such that CD⊥r. Construct a
point E on s so that AE ∼
= A00 C. Then let F ∈ r be such that EF ⊥r. Then EF CD is a
Sacchery quadrilateral ⇒ the symmetry axis is good.

2.2 Line pencils and cycles

Definition 2.8 The lines pencils are set of lines:

– passing through a given (finite) point

– parallel to a given line
– perpendicular to a given line.

Definition 2.9 A cycle is the orbit of a point, reflecting it to a given line pencil:

– cycle (finite point)

– horocycle/paracycle (parallel)
– hypercycle/equidistant line (perpendicular).

Theorem 2.10 If, a cycle has three collinear points, then it is a line.

Proof. Let P 0 and P 00 be the reflections of the point

P respected to two elements of the line pencil ⇒ they
are the perpendicular bisector of the the segments P P 0
and P P 00 . We have the perpendicular transverse of
these lines ⇒ ultraparallel lines ⇒ the points lie on a
hypercycle ⇒ the point lie on the base line. Figure 2.6: Proof: Theorem 2.10
Definition 2.11 Let A and B be two points on two parallel rays. We say that A and B
are corresponding points if the segment AB forms equal angles with the rays.

Lemma 2.12 Circle: Locus of point, that are equidistant from a given point.
Paracycle: Locus of points, that are corresponding to a given point, respected to a given
parallel line pencil.
Hypercycle: Locus of points, that are equidistant from a given point.

16
2.3 Models of the hyperbolic geometry
1. Cayley-Klein disk model
Points: Interior of the unit disk
Lines: Chords
Axioms: I-IV trivial, since it is part of the Euclidean plane
Parallels: Chords, sharing an endpoint (boundary point)
Perpendiculars: f ⊥g if and only if f goes through the intersection point of the
tangents of g.

2. Poincaré disk model /conformal disk model/

Points: Interior of the unit disk
Lines: Diameters and circular arcs, perpendicular to the model circle
Axioms: I-IV trivial, since it is part of the Euclidean plane
Parallels: Circular arcs and diameters, sharing an endpoint (boundary point)

3. Poincaré half-plane model

Points: Upper half-plane
Lines: Rays and circular arcs, perpendicular to the model circle
Axioms: I-IV trivial, since it is part of the Euclidean plane
Parallels: Circular arcs and rays, sharing an endpoint (boundary point)

Remark 2.13 Both the Poincaré disk and half-plane model are conformal models: The
angle of lines seems real size in these models.

2.3.1 Orthogonality in the Cayley-Klein model

Lemma 2.14 If, the length of the tangents, drawn K
from an external point to two intersecting circles, are
equal, then the point lies on the common secant of the T
circles.

Proof. Let K be the external point and |KT | = A

|KR|. The ray, emanating from K through one of R
the intersection point A intersects the circles in points B
C
B and C. Using the intersecting secant theorem:
|KA||KB| = |KT |2 = |KR|2 = |KA||KC| ⇒ |KB| =
Figure 2.7: Proof: Lemma 2.14
|KC| ⇒ B = C.

17
Definition 2.15 Two chords of a said to be conjugate to each other if the intersection
point of the tangent, drawn to the circle at the endpoints of one of the chords lies on the
line of the other chord.

Lemma 2.16 Two lines are orthogonal to each other

in the Cayley-Klein model if and only if the represent-
L
ing chords are conjugate to each other.

Proof. Let KL and K 0 L0 be orthogonal lines in the K'

Poincaré disk model, intersecting each other at the

point P . Then |OP | = |OK 0 | = |OL0 |, therefore O L'
K
must lie on the common secant of the circles, deter-
mined by the points KK 0 LL0 and KP L, for the pre-
O
vious lemma.
Figure 2.8: Proof: Lemma 2.16

2.3.2 Stereographic projection

1. Defined as the projection of the sphere from the North/South pole onto the equatorial
plane.

2. This is a bijective mapping between the unit sphere and the plane (usualy S 2 →
R2 /C).

3. It is equivalent with the projection from the North pole onto the tangent plane of
the South pole.

Theorem 2.17 The stereographic projection preserves circles and it is conformal.

Proof. Let P and Q be two points on the surface of the sphere, and their projections are
P 0 and Q0 . Let the intersection of the line P Q and P 0 Q0 be N , and S be the intersection
of the line P Q and the tangent plane of the center of the projection O. First, we prove the
P QQ0 P 0 is a cyclic quadrilateral. OS is the tangent line of the triangle OQP 4. Because
of the inscribed angle theorem, the supplementary angle of QOS∠ is equal to QP O∠.
Since OS k Q0 N , the supplementary angle of QOS∠ is also equal to OQ0 P 0 ∠. Therefore
|N P ||N Q| = |N P 0 ||N Q0 |.
Now let |P Q| be a chord of a circle on the sphere. Then the product |N P ||N Q| is constant
for every |P Q| chord and equal to the product |N P 0 ||N Q0 |. This is true for every point
N , which lies on the intersection line of the projection plane and the plane, that contains
the circle. The intersection of the equatorial plane with the elliptical cone of the circle
and O is also a circle.

18
Conformal: Let P be a vertex of an angle with tangent lines t1 and t2 . Let ki be a
spherical circle in the plane of ti and O such that P and O lie on ki . Then the tangent of
these circles si meet at the same angle and they lie a parallel plane to the tangent plane
of O. Therefore t01 and t02 also meet at the same angle.

Q
R
e P
,
R* R, e Q, ,
P
.
.
. N
k
O

P
a k

t1 t2
O

s1 s2

Figure 2.9: Proof: Theorem 2.17

Theorem 2.18 Consider the unit disk with the Poincaré structure. Then, the composition
of the inverse stereographic projection with orthogonal projection back to the plane of the
circle is a bijection on the disk, the ideal points are fix points and it results in the Cayley-
Klein model.

Proof. Only the lines are different in the two models. Since a line in the Poincaré disk
model is a circle, orthogonal to the great circle, the inverse stereographic projected image
is also a circle, orthogonal to the main circle. The plane of it is perpendicular to the base
plane ⇒ the orthogonal projection of the circular arc is a line segment, connecting the
endpoints of the Poincaré line (circle).

19
2.3.3 Inversion
Definition 2.19 Let O be the center of a circle/sphere with radius r. The inverse image of
−→
a point P (6= O) is the point P 0 if P 0 lies on the ray OP and |OP ||OP 0 | = r2 . The mapping,
that assigns the inverse image to every point of the plane/space, is called inversion.

Theorem 2.20 The image of a circle or a line by inversion is either a circle or a line.
Inversion is a conformal mapping.

Proof.

– The base circle of the inversin is fixed point by point.

– Every line, through the center of the inversion is also invariant (inside ←→ outside).
P Q
– The image of a line, which does not contain O,
is a circle, through O:
Let P be the foot point of the orthogonal line ,
P
to the given line, through O and Q be an ar- Q
,

bitrary point on the given line. Let P 0 and Q0 O

be the inverse image of P and Q respectively.
|OP | |OQ0 |
Then |OP ||OP 0 | = |OQ||OQ0 | ⇒ |OQ|
= |OP 0 |
⇒ P
Q*
Q
OP 0 Q0 4 ∼ OQP 4 ⇒ P 0 Q0 O∠ = π2 ⇒ Q lies0 O K
K*
l

on the Thales circle above the segment OP 0 as l*

diameter.
– The image of a circle, which contains O is a line, , t2
M
which does not contain O.
– The image of a circle, which does not contain O, t1

is also a circle, which does not contain O: f1 f2

O1 O2
Let P Q be a secant line through O. Then the F1
E2
product of |OP ||OQ| = p is a constant indepen- M
e2
dent from the secant. Applying a scaling from O e1
F2
O
∗
by ratio 1 : p, we obtain the l circle with center E1
∗ ∗
K . The secant intersects l and the image of Q
|OQ|
is Q0 . Then |OQ∗ ||OP | = p
|OP | =1
Figure 2.10: Proof: Theorem 2.20
Conformal: Let e1 and e2 be the tangents to the curves, their intersection be M
and the image of M be M 0 . Let fi be perpendicular lines to ei through O, and Oi be
the intersection points of fi by the perpendicular bisector of OM 0 . Then the images
of ei by the inversion are circles with center Oi through O. Then ti ⊥Oi M 0 , therefore
t1 t2 ∠ ∼
= O1 M 0 O2 ∠ ∼
= O1 OO2 ∠ ∼
= e1 e2 ∠.

20
Remark 2.21 Applying the inversion to the Poincaré disk/sphere model at a boundary
point, we obtain the Poincaré half-plane/half-space model. We also have the advantage,
that this point is arbitrary, therefore we can choose the representative of one line/plane in
the model to be a ray/half plane. Because the Poincaré disk/sphere model is a conformal
model and inversion is a conformal mapping, the Poincaré half-plane/half-space model is
also a conformal model.

2.4 Distance and angle of space elements

2.4.1 Mutual position of space elements (E3 )

The parallelism of lines is an equivalence relation (parallel line pencil). We assign an ideal
point for every equivalence class. Two line intersect each other at this ideal point, if they
are parallel in the Euclidean space. The union of ideal points forms an ideal plane.

Definition 2.22 The projective space PR3 is the union of the Euclidean space and the
ideal elements.

1. Line-Line
– intersecting: if their intersection is not ideal
– parallel: if they are not intersecting in E3 but in PR3
– skew: if they are not intersecting in PR3
2. Line-Plane/Plane-Plane
– intersecting
– parallel

2.4.2 Mutual position of space elements (H3 )

1. Line-Line
– intersecting
– parallel
– ultraparallel
– skew (not intersecting in PR3 )
2. Line-Plane
– intersecting
– parallel
– skew (they intersect each other either outside of the model or only in PR3 )

21
 3. Plane-Plane
– intersecting
– parallel
– ultraparallel

2.4.3 Perpendicular transverse of non-intersecting hyperbolic space

elements
Theorem 2.23 In the hyperbolic space, there exits a unique perpendicular transverse line
for two skew lines, for two ultraparallel lines/planes.

Proof. We use the Poincaré half-plane/half-space model.

1. Ultraparallel lines (planar case)
a
2. Ultraparallel line-plane
n
Perpendicular line in the boundary plane to the b

line through the center of the plane (sphere).

3. Ultraparallel planes
We draw a perpendicular line to the intersection
of the plane and the boundary plane through the
center of the sphere. We apply the previous case n
a
for the sphere and the line, perpendicular to the S1 b

boundary plane in the given plane through the S2

O
footpoint of the previous line.

4. Skew lines
Let k be a half-sphere, perpendicular to the
boundary plane, through A, B, C ⇒ b ∈ k. F D
is the perpendicular bisector of BC ⇒ S = a

AD ∩ BC. Let N be a point on b uch that M tn

∗ n
N = S. Then n will be a circle perpendicu- tb
N
b
lar both a and b with center A and radius |AN |.
B
The perpendicularity to a is obvious. Let tb and A S
D
F
tn be the tangents of b and n in N . tb is in the C

plane of BN C ⇒ tb ⊥F D, N F ⇒ tb ⊥N D. In
the triangle AN D4 AD is the diameter of the Figure 2.11: Proof: Theorem 2.23
π
circumscribed circle ⇒ AN D∠ = 2
⇒ tn =
N D ⇒ tn ⊥tb ⇒ n⊥b.

22
2.4.4 Cross-ratio
Definition 2.24 Let z1 , z2 , z3 , z4 be complex numbers. Then the cross-ratio:
z3 − z1 z4 − z1
(z1 , z2 , z3 , z4 ) := : .
z2 − z3 z2 − z4
Theorem 2.25 The cross-ratio of 4 distinct points is real, if and only if they all lie either
on a circle or on a line.

Proof. We use the polar decomposition: z = reiΘ

z3 − z1 z4 − z1 z1 − z3 z4 − z2
: = · =
z2 − z3 z2 − z4 z2 − z3 z4 − z1
z1 − z3 iΘ3 z1 − z4 iΘ4
= e e = Rei(Θ3 +Θ4 )
z2 − z3 z2 − z4
where R is a real number. ei(Θ3 +Θ4 ) ∈ R ⇔ Θ3 + Θ4 =
kπ(k ∈ Z). If Θ3 + Θ4 = 2kπ then they lie on a circle,
otherwise they lie on a line. Figure 2.12: Proof: Theorem 2.25

2.4.5 Distances in the hyperbolic space

Definition 2.26 The hyperbolic distance in the Poincaré disk model: d(X, Y ) = log(X, Y, U, V ),
where U and V are the endpoints of the line, determined by X and Y such that X, Y, U, V
is the cyclic order of the points on the representing circle.

– point-point: definition
– point-line: distance of the given point and the footpoint of the orthogonal line to
the given line through the given point
– point-plane: distance of the given point and the footpoint of the orthogonal line to
the given plane through the given point
– line-line: length of the perpendicular transverse line segment
– line-plane: length of the perpendicular transverse line segment
– plane-plane: length of the perpendicular transverse line segment

23
2.4.6 Angle and distance in the Cayley-Klein model
– Angle: We follow the mapping from the Poincaré
structure to the Cayley-Klein structure:
α = (u, v)∠ = (u0 , v 0 )∠ = V P 0 U ∠
|U V |2 = |P 0 U |2 + |P 0 V |2 − 2|P 0 U ||P 0 V | cos(α)
|P 0 U |2 = r12 = u21 + u22 − 1, |P 0 V |2 = r22 = v12 + v22 − 1
|U V |2 = (u1 − v1 )2 + (u2 − v2)2 = u21 + u22 + v12 + v22 −
2u1 v1 − 2u2 v2

r12 + r22 − |U V |2 −2 + 2u1 v1 + 2u2 v2

= q ⇒
2r1 r2 2 (−1 + u21 + u22 )(−1 + v12 + v22 )
−1 + u1 v1 + u2 v2
cos(α) = q (C-K∠)
(−1 + u21 + u22 )(−1 + v12 + v22 )
Figure 2.13: Angle in the Cayley-
– Distance:
Klein model
−1 + x1 y1 + x2 y2
cosh(d(x, y)) = q
(−1 + x21 + x22 )(−1 + y12 + y22 )

2.4.7 Hyperboloid model

We use the V3 real vector space with the standard {e1 , e2 , e3 } base. We introduce the
symmetric bilinear form: he1 , e1 i = 1, he2 , e2 i = 1, he3 , e3 i = −1, hei , ej i = 0, (i 6= j).
∈ V3 , then hx, yi = x1 y1 + x2 y2 − x3 y3 , and
If x, y 
< 0 time − like





hx, xi = 0 light − like ⇒



 e3
> 0 space − like


 e2
e1



 two − sheeted hyperboloid r ∈ C


hx, xi = r2 cone r=0
 

one − sheeted

hyperboloid r ∈ R
We define an equivalence relation: x ∼ y ⇔ ∃c ∈
R \ {0} : y = cx ⇒ representing elements: x0 ∼
(x1 , x2 , 1). Figure 2.14: Hyperboloids
We assign the time-like vectors to the points of the Cayley-Klein model of the hyper-
bolic geometry by this equivalence relation.
1
We define the distance of two points by the d(X, Y ) = 2
log(X, Y, U1 , U2 ), where U1
and U2 are the boundary points on the line of X and Y such that X, Y , U1 and U2 are in
cyclic order. Then
β1 β2
e2d(X,Y ) = (X, Y, U1 , U2 ) = : where u1 = α1 x + β1 y, and u2 = α2 x + β2 y.
α1 α2

24
 x3
hui , ui i = αi2 hx, xi + 2αi βi hx, xi + βi2 hy, yi = 0 ⇒
βi 2
 
hx, xi + 2 αβii hx, xi + αi
hy, yi = 0
q y u1
4 hx, yi2 − 4 hx, xi hy, yi u2 x
!
βi −2 hx, yi ±
= ⇒
αi 1,2
2 hy, yi O x2
q x1
hx, yi − hx, yi2 − hx, xi hy, yi
e2d = (x, y, u1 , u2 ) = q
hx, yi + hx, yi2 − hx, xi hy, yi Figure 2.15: Distance in the hyper-
| {z
a
}
boloid model
√
√1
s
ed + e−d a+
s
a √ 1 √ 1 1
cosh(d(x, y)) = = , but a+ √ = ( a + √ )2 = a+ +2
2 2 a a a
q q
1 hx, yi − hx, yi2 − hx, xi hy, yi hx, yi + hx, yi2 − hx, xi hy, yi
a+ +2= q + q +2=
a hx, yi + hx, yi2 − hx, xi hy, yi hx, yi − hx, yi2 − hx, xi hy, yi
 q 2  q 2
2 2
hx, yi − hx, yi − hx, xi hy, yi + hx, yi + hx, yi − hx, xi hy, yi
= q 2 +2=
2 2
hx, yi − hx, yi − hx, xi hy, yi
q 2
2 hx, yi2 + 2 hx, yi2 − hx, xi hy, yi 4 hx, yi2 − 2 hx, xi hy, yi
= +2= +2=
hx, xi hy, yi hx, xi hy, yi
2
4 hx, yi 1 − hx, yi
 
= ⇒ cosh log(X, Y, U1 , U2 ) = cosh(d(X, Y )) = q .
hx, xi hy, yi 2 hx, xi hy, yi

2.4.8 Pole-polar relation

To represent a line u in the hyperboloid model, we consider a plane, orthogonal to
u(u1 , u2 , 1) through the origin: xu1 + yu2 + z = 0. Intersecting it with the plane z = 1,
we get a line with the equation xu1 + yu2 + 1 = 0.
If u21 + u22 > 1, then it can be assigned as a proper
line in the Cayley-Klein model, and X(x) lies on it, if
xu1 + yu2 + 1 = 0. The pole V (v1 , v2 , 1) of this line
u is the intersection of the tangents to the boundary
point of u. This is also the center of the circle, with
determines u in the Poincaré model. To get the radius
r of this circle, we apply the Pythagorean theorem:
r2 = a2 − 1 = v12 + v22 − 1. Figure 2.16: Pole-polar

Lemma 2.27 A line u(u1 , u2 , 1) has the pole (−u1 , −u2 , 1) and the point P (x, y, 1) lies
on it, if and only if p · u = 0.

25
Proof. The equation of the two circles are the following:
(x − v1 )2 + (y − v2 )2 = v12 + v22 − 1
x2 + y 2 = 1

Expanding the first equation, we can simplify it:

x2 − 2xv1 + S
vS12 + y 2 − 2yv2 + S
vS22 = S
vS12 + S
vS22 − 1
x2 + y 2 = 1

Finally, we get the equation of the radical line of the circles: −2xv1 − 2yv2 = 2 ⇒ −xv1 −
yv2 = 1. But the radical line is our original u line with the equation xu1 + yu2 + 1 = 0,
therefore v1 = −u1 and v2 = −u2 .

Remark 2.28 According to (C-K∠), the angle of two lines u(u1 , u2 , 1) and v(v1 , v2 , 1)
can be computed by their poles (−u1 , −u2 , 1) and (−v1 , −v2 , 1). With the defined bilinear
form:
−1 + (−u1 )(−v1 ) + (−u2 )(−v2 ) −1 + u1 v1 + u2 v2
q =q ⇒
(−1 + (−u1 )2 + (−u2 )2 )(−1 + (−v1 )2 + (−v2 )2 ) (−1 + u21 + u22 )(−1 + v12 + v22 )
hu, vi
cos(α) = q .
hu, ui hv, vi

2.5 Models of the Euclidean plane

– Horosphere: We consider the rays, orthogonal to the base plane in the Poincaré half-
space model. This is a parallel line pencil. The corresponding cycles are horospheres,
represented as planes, parallel to the base plane. Because this is a conformal model, the
interior angel sum of any triangle on this plane is π. The geometry on this horosphere is
Euclidean.

Theorem 2.29 Any two horosphere are congruent to each other.

– Hypercycle: Let O be a point on the Euclidean plane and K be a center of a unit

ball, tangent to the plane at O. We consider a unit disk, such that its plane is parallel
to the base plane, through K, as the Cayley-Klein
model of the hyperbolic plane. The points of the
P
model are the points of the unit disk, and the lines are K

the lines, through K and the corresponding hypercy-

O
cles, represented as ellipses. We project this elliptic T (P)
arc orthogonally onto the sphere (we get a great arc).
Figure 2.17: Hypercycle model
Then we project it back to the base plane through K.
With the d(A, B) := |T (A)T (B)| metric, the congruence axioms will be true, and the Eu-
clidean axiom of parallelism is obviously ture.

26
2.6 Area on the hyperbolic plane
Definition 2.30 A triangle is called asymptotic/doubly asymptotic/triply asymptotic, if
one/two/three vertex/vertices is/are boundary point(s).

Theorem 2.31 All the triply asymptotic triangles are congruent to each other.

Definition 2.32 Area is an isometry-invariante, non-negative, additive set function for

simply polygons and T (= π) is assigned to the triply asymptotic triangle.

Theorem 2.33 Any asymptotic triangle can be cur off into a pentagon.

Proof. (Poincaré disk-model): Let ABC4 be such

that C is the ideal point and A is the center of the
disk. Let D be an ideal point such that (ABD), and
M be the footpoint of the perpendicular line to CD
A
through A. Let A1 be the reflection of B in the line M1
12 A1
AM and the intersection of BC and A1 D be M1 . Let B
B1 4 34 A2
the footpoints of the perpendicular line to DC through
Q P M2
B and A1 be Q and P respectively. If the reflection of M
BC in A1 P is DA2 , then the triangle M2 A1 A2 4/4/ is
D C
congruent to A1 M1 M2 4/3/ and M1 M2 B4. Continu-
ing this process, we get ABQP A1 pentagon. Figure 2.18: Lindberg method

Theorem 2.34 If, the vertex angle of a doubly asymptotic triangle is α, then the area of
it is c(π − α) /c ∈ R+ /.

Proof. Let f (φ) be the area, if φ = π − α is the supplementary angle. The union of the
two corresponding doubly asymptotic triangle is a triply asymptotic triangle (see Figure
2.19), therefore: f (φ) + f (π − φ) = T .
N
A f

f(f) f(f)
f(p- f)

N M L f p- f M
A
Figure 2.19: f (φ) is additive

27
Figure 2.20 shows us, that T = f (φ) + f (ϕ) + f (π − φ − ϕ) ⇒ f (φ) + f (ϕ) = f (φ + ϕ)

N
A

p-y p-f C

p+y B
L M

Figure 2.20: Area and defect

Our only solution is f (x) = λx, since f is monotonously increasing and f (1) = λ ⇒
k k+1 k
f (n) = nλ. Now, if n
≤x≤ n
⇒ k ≤ nx ≤ k + 1 ⇒ kλ ≤ nf (x) ≤ (k + 1)λ ⇒ n
≤
f (x) k+1 f (x) 1
λ
≤ n
⇒ x− λ
≤ n
∀n ⇒ f (x) = λx.

Remark 2.35 Because the triply asymptotic triangle is T ⇒ λ = Tπ . Therefore, we choose

the value of T be π.

Theorem 2.36 The area of any hyperbolic triangle is its defect.

Proof. We make up the ABC4 to a triply asymptotic triangle by expanding the sides
cyclically (see Figure 2.20). The area of the three extra doubly asymptotic triangles are
α, β and γ respectively. Therefore, the area can be expressed from the formula:

π = A(ABC) + α + β + γ ⇒ A(ABC) = π − (α + β + γ) = δ.

28
3 Classification of isometries

3.1 Isometries of the Euclidean plane

Definition 3.1 Isometry is a distance-preserving bijective mapping of the space to itself.

Claim 3.2 The isometries of the space form a non-commutative group.

Theorem 3.3 Every isometry is a collineation.

Proof. Let A, B and C be three collinear points, and their image be A0 , B 0 and C 0 .
= B 0 C 0 and AB ∼
= A0 C 0 , BC ∼
Assume, that A0 B 0 C 0 form a triangle. But AC ∼ = A0 B 0 ⇒
|AC| + |BC| = |AB| = |A0 B 0 | < |A0 C 0 | + |B 0 C 0 |. ⇒⇐

Theorem 3.4 If, an isometry has two fix points, then the determined line is also fixed
point-by-point (axis).

Proof. Let A = A0 and B = B 0 be true and P be a point on this line. Then P 0 must be
also on the line and AP ∼
= AP 0 ∧ BP ∼
= BP 0 ⇒ P = P 0 .

Theorem 3.5 If, an isometry has three fix points, then the determined plane is also fixed
point-by-point (plane).
B
Definition 3.6 The reflection of the plane in a line
t assign the P 0 point to P if P P 0 ⊥t and P T ∼
= P 0T , A
0
where T = P P ∩ t.

Theorem 3.7 Reflection is an isometry.

C D
Proof. Let C and D be the footpoints of the per-
pendiculars to t from A and B respectively. Then ,
A
BCD4 ∼ = B 0 CD4 ⇒ BC ∼ = B 0 C∧BCD∠B 0 CD∠ ⇒
,
ACB∠ ∼
= A0 CB 0 ∠ ⇒ ABC4 ∼
= A0 B 0 C4 ⇒ AB ∼
= B
A0 B 0 Figure 3.1: Hypercycle model
Claim 3.8 The composition of a reflection to itself is the identical mapping.

29
Theorem 3.9 If, a non-identical isometry has exactly two fix points, then it is a reflection
in the determined line.

Proof. Let A and B be the fix points and P be an arbitrary point. Then AP ∼
= AP 0 ∧
BP ∼
= BP 0 ⇒ P = P 0 , or AB is the perpendicular bisector of P P 0 ⇒ d(P, AB) =
d(P 0 , AB) and P P 0 ⊥AB.

Theorem 3.10 If, a non-identical R isometry has exactly one fix point, then it is the
composition of two reflections, through this fix point.

Proof. Let P be an arbitrary point and A be the fix point. Then AP ∼

= AP 0 , therefore
A lies on the perpendicular bisector (t1 ) of P P 0 . The composition of the original isometry
and the reflection in this line has two fix points, therefore it is a reflection in a line t2 .
Then R ◦ t1 = t2 ⇒ R = t2 ◦ t−1
1 = t2 ◦ t1 .

Theorem 3.11 If, a non-identical isometry T has no fix point, then it is the composition
of either two or three reflections.

Proof. Let P be an arbitrary point and P 0 be its image. Then T ◦ t1 has either one, or
two fix points, where t1 is the reflection to the perpendicular bisector of P P 0 .

1. T ◦ t1 = R = t2 ◦ t3 ⇒ T = t2 ◦ t3 ◦ t−1
1 = t2 ◦ t3 ◦ t1

2. T ◦ t1 = t2 ⇒ T = t2 ◦ t−1
1 = t2 ◦ t1

Theorem 3.12 Any isometry of the plane is the composition of at most three reflections
in a line.

Proof. It follows from the above theorems trivially, since any isometry has either 0, 1, or
2 fix points or all the points on the plane is fixed point-by-point.

Lemma 3.13 The composition of two reflections can be considered as either a rotation
(intersecting axis) or a translation (parallel axis).

Proof.
– Intersecting axis: Let A be the intersection of the two lines t1 and t2 . Then P At1 ∠ ∼ =
P 0 At1 ∠ and P 0 At2 ∠ ∼
= P 00 At2 ∠, therefore P AP 00 ∠ = 2t1 t2 ∠ = 2α ⇒ Rotation around A
by 2α.
– Parallel axis: Now, we can say that d(P, t1 ) = d(P 0 , t1 ) and d(P 0 , t2 ) = d(P 00 , t2 )
and P, P 0 , P 00 are collinear points, therefore d(P, P 00 ) = d(t1 , t2 ) = 2d ⇒ Translation,
perpendicular to ti with 2d.

Remark 3.14 In the case of the rotation, only the angle of the lines matters, in the case
of the translation, only the distance and the direction matters.

30
Theorem 3.15 Every isometry of the Euclidean plane belongs to one of the following 5
groups:

1. Identity (0) 3. Rotation (2a) 5. Glide reflection (3)

2. Reflection (1) 4. Translation (2b)

Proof. We have a full discussion for 2 reflections.

1. case t1 ∩ t2 6= ∅ t2
t1 t3 t 1 t'2 t 3'

(a) t1 ∩ t2 = t2 ∩ t3
Then we can consider the t2 ◦ t3 as a rotation
around this point, and we can rotate the lines
around this center to get t1 = t02 . Then t1 ◦ t2 ◦
t3 = t1 ◦ t02 ◦ t03 = t1 ◦ t−1 0 0
1 ◦ t3 = t3 . Figure 3.2: Case 1. (a)
t3 ,
t2
(b) t1 ∩ t2 6= t2 ∩ t3 t2
t3 t1
First, we rotate t1 and t2 around their intersec- K ,
t1
tion O until t02 ⊥t3 . Then we rotate t02 and t3 O
O
around their intersection K until t002 k t01 . Then
,,
t3
t01 ◦ t002 is a translation and t01 ◦ t002 ◦ t03 is a glide
reflection.
,,
K t2
,
2. case t1 ∩ t2 = ∅ ⇒ t1 k t2 t1

(a) t1 k t2 k t3 Figure 3.3: Case 1. (b)

Then we can translate t2 and t3 along their per-
pendicular line until t1 = t02 then their compo- ,
t3
sition will be the identity, therefore only t3 re- t3
t2
,
t1
mains, so this is a translation.
t1 t2

(b) t1 k t2 ∦ t3
,,
t2
First, we rotate t2 and t3 around their intersec-
tion until t02 ⊥t1 . Then we rotate t1 and t02 around ,
t3
their intersection until t01 k t03 . Finally, we rotate ,,
t1
t002
and t03 around their intersection, until t01 k t000
2.
Then t01 ◦ t000 00
2 ◦ t3 is a glide reflection. Figure 3.4: Case 2. (b)

Remark 3.16 Two reflections are commutative to each other if their axis are orthogonal
to each other.

31
3.2 Isometries of the Euclidean space
Theorem 3.17 Any isometry of the space is the composition of at most four reflections
in a plane.

Theorem 3.18 Every isometry of the Euclidean space belongs to one of the following 7
groups:

1. Identity (0) 4. Translation (2b) 7. Screw displacement (4)

2. Reflection (1) 5. Glide reflection (3a)

3. Rotation (2a) 6. Improper rotation (3b)

where
5. Glide reflection: composition of a translation and a reflection

6. Improper rotation: composition of a rotation and a reflection

7. Screw displacement: composition of a rotation and a translation

Definition 3.19 The reflection in a line in the space means a rotation around this line
by 180◦ . This can be represented as a reflection in two orthogonal planes through the line.

Lemma 3.20 The composition of two reflections in a line is either rotation or translation
or screw displacement, if the two lines are intersecting, parallel or skew lines.

,,,,,
Proof. P
,
Assume that e = t1 ◦ t2 , f = t3 ◦ t4 . P P
,,,,

,,
1. case: The lines e and f determine a plan P
e f
We choose the positions of ti such that, t2 and t3 coin- e
f
cide the determined plane and t1 and t4 are orthogonal
,,,
to it. Then t2 ◦ t3 is identity and if e k f , then t1 k t4 , P

otherwise t1 ∦ t4 . Figure 3.5: 1. case

2. case: The lines e and f are skew lines
Let n be the perpendicular transverse of e and f . n
t1 := (e, n) ⇒ t2 ⊥n, t3 := (f, n) ⇒ t4 ⊥n. Since t2 k
t4 ∧ t3 ⊥t4 ⇒ t3 ⊥t2 ⇒ t1 ◦ t2 ◦ t3 ◦ t4 = (t1 ◦ t3 ) ◦ (t2 ◦ t4 ).
But n lies both on t1 and t3 therefore they determine e

a rotation, and t2 k t4 then they determine a trans- f

lation. The composition of these isometries is a skew
displacement.
Figure 3.6: 2. case

32
Lemma 3.21 The composition of two translations is also translation.

Proof. Let t1 k t2 and t3 k t4 be two translation. If, t2 k t3 , then it is trivial. Otherwise

t2 ∩ t3 6= ∅ ⇒ t2 ∩ t3 = e. If, we rotate t2 and t3 around e by 90◦ , then t02 ⊥t1 and t03 ⊥t4 .
Then = t1 , t02 and t03 , t4 determines two reflections in a line with parallel axis, therefore it
is a translation.

Lemma 3.22 The composition of two rotations with orthogonal axis is a screw displace-
ment.

Proof. Let t0 and t00 be the axis of the rotations and n

n
be their perpendicular transverse line. Then t0 ⊥(t00 , n) e
t
00 0
,
and t ⊥(t , n). Let t1 and t2 be reflections such that
t1 ∩ t2 = t0 and t2 = (t0 , n), furthermore t3 and t4 be t
,,

reflections such that t3 ∩ t4 = t00 and t3 = (t00 , n). Since

t2 ⊥t3 ⇒ t1 ◦ t2 ◦ t3 ◦ t4 = (t1 ◦ t3 ) ◦ (t2 ◦ t4 ). Let e and
f be the intersections of t1 , t3 and t2 , t4 respectively. f
But t1 ⊥t3 and t2 ⊥t4 , therefore this isometry is can
be represented as the composition of two reflections
in a line, which axis are in skew position ⇒ screw
displacement. Figure 3.7: Proof of Lemma 3.22
Proof of Theorem 3.18 We have a full discussion for 2 reflections. Composition of at
least three reflections:
1. case: The three plane have a common perpen-
dicular plane: P
,,,
P

The orbit of any point lies on the plane, parallel P

,

,,
to this orthogonal plane ⇒ using the planar case, P

it is the composition of 3 reflections in line ⇒ ei-

ther reflection or glide reflection ⇒ the spatial
transformation is also either reflection or glide
reflection. Figure 3.8: 1. case
2. case: If, plane 1 and plane 2 intersect each other 1'
m
in the line m and plane 3 does not intersect m,
M
then the plane orthogonal to m will be orthog- 2
2"
1 3 M
onal to plane 3 as well ⇒ 1. case. We may
3"
assume that plane 3 intersects m in the point
m
M . First we rotate plane 1 and 2 around m
M 2'
such that 20 ⊥3. Then we rotate 20 and 30 around 1'
3
their intersection such that 10 ⊥30 . Finally we get
an improper rotation. Figure 3.9: 2. case

33
3. case: If, plane 1 and 2 are parallel to each other, then plane 3 is either parallel
them both ⇒ reflection, or plane 3 intersects them in parallel lines. Then the plane,
orthogonal to the lines will be orthogonal to all the planes ⇒ 1. case
4. We can assume, that the composition of the first
3 reflections is either glide reflection or improper
1 2 1
rotation. 1 2 2
If, plane 4 is parallel to the third plane of an im-
3
proper rotation, then it is a screw displacement 3 3
by definition.
4
If, m is the intersection of plane 3 and 4, then 4 4

this isometry is the composition of two rotation

and by lemma, it is a screw displacement. 2
1
Now, we can assume, that the composition of
the first 3 reflections is a glide reflection. 1
2' m
If, plane 4 is parallel to plane 3, then this is the
composition of two translations and by lemma, 3

this is another translation. 4 n

3' 4
If, plane 4 intersects plane 3, then we rotate
plane 2 and 3 around their intersection line such Figure 3.10: 4. case
that
30 ⊥4. Then plane 3’ is orthogonal to both plane 2’ and plane 4 ⇒ plane 3’ is
orthogonal to their intersection line m. Let n be the intersection line of plane 1 and
plane 3’. Then n⊥m and 1 ◦ 2 ◦ 3 ◦ 4 = 1 ◦ 20 ◦ 30 ◦ 4, but 20 ⊥30 ⇒ 1 ◦ 20 ◦ 30 ◦ 4 =
(1 ◦ 30 ) ◦ (20 ◦ 4). Then this isometry is the composition of two rotation around
orthogonal axis, therefore this is a screw displacement.

34
4 Analytic geometry

4.1 Vectors in the Euclidean space

Definition 4.1 The representing point pairs of a translation called vector.

Definition 4.2 The sum of two vectors is a vector, that we get by the parallelogram rule.
−−→ −−→ −−→
Definition 4.3 Let t > 0, t ∈ R, then the multiplication of P P 0 by t is P P 00 if P P 00 =
−−→ −−→ −−→
t P P 0 and (P 0 P P 00 ) is not true. If t = 0, then P P 00 = 0. If t < 0, then P P 00 is such that
−−→00 −−→
P P = −t P P 0 and (P 0 P P 00 ) is ture.

Lemma 4.4 The vectors of the space with the defined addition and scalar multiplication
form a vector space.

Definition 4.5 Let a and b be vectors in the space, then ha, bi = |a||b| cos(γ) called the
dot product.

Remark 4.6 ha, bi = 0 ⇔ a⊥b, ha, bi > 0 ⇔ γ < π2 , ha, bi < 0 ⇔ γ > π2 .

Theorem 4.7 Dot product is symmetric, positive definite, bilinear function.

Lemma 4.8 Let a and v be vectors. Then v = v ⊥ + v k , where a k v k , a⊥v ⊥ and v k =

ha, vi
a, v ⊥ = v − v k .
ha, ai

|v k |
Proof. Let α be the angle between a and v. Then = cos α ⇒ |v k | = |v| cos α, but
|v|
ha, vi
ha, vi = |a||v| cos α, therefore |v k | = . Now, we need a unit vector in direction of a
|a|
a ha, vi ha, vi
to obtain v k : au = . So that v k = |v k |au = a= a.
|a| |a|2 ha, ai
Definition 4.9 Let a and b be vectors, then a × b is the cross product of a and b, where
|a × b| = |a||b| sin γ, a × b is orthogonal both a and b and the direction of a × b is given by
the right-hand rule.

35
Theorem 4.10 Cross product is antisymmetric bilinear function.

Remark 4.11 a × b = 0 ⇔ a k b, and |a × b| is equal to the area of the spanned parallel-

ogram.

Definition 4.12 Let a, b and c be vectors in the space. Then the triple product of them
is a · b · c = ha × b, ci .

Theorem 4.13 Triple product is invariant under a circular shift: a · b · c = ha × b, ci =

hb × c, ai = hc × a, bi .

Proof. This is true because of the geometric meaning. a · b · c is the (signed) volume
of the spanned parallelepiped. The area, spanned by a and b is exactly |a × b| and
h
= cos(a × b, c) ⇒ h = |c| cos γ. Then the volume of the spanned parallelepiped is
|c|
V = |a × b||c| cos γ = ha × b, ci = a · b · c.

Theorem 4.14 (Lagrange’s formula) (a × b) × c = ha, ci b − hb, ci a

Proof. (a × b) × c⊥a × b ⇒ (a × b) × c is in the plane of a and b ⇒ (a × b) × c = αa + βb

Now, we consider the dot product of both sides with b × c:
Left side: h(a × b) × c, b × ci = hc × (b × c), a × bi . Now, if v := c × (b × c), then |v| =
|c||b × c|, since c⊥b × c. But than |v| = |c|2 |b| sin γ, where γ is the angle of b and c. If
we decompose b into parallel and orthogonal elements respected to c, then |b⊥ | = |b| sin γ,
|v|
therefore = |b⊥|. It can be seen, that v k b⊥ ⇒ v = |c|2 b⊥ = |c|2 (b − bk ) =
|c| !
2

hb, ci
|c|2 b − c = |c|2 b − hb, ci c. Finally, hc × (b × c), a × bi = h|c|2 b − hb, ci c, a × bi =
|c|2
0 − hb, ci hc, a × bi = − hb, ci (a · b · c).
Right side: α ha, b × ci = α(a · b · c) and b⊥b × c ⇒ β hb, b × ci = 0
Finally, we get that α = − hb, ci. Similarly, β = ha, ci.

Theorem 4.15 (Jacobi’s formula) (a × b) × c + (b × c) × a + (c × a) × b = 0

Proof. (b × c) × a = −a × (b × c) and (c × a) × b = −(a × c) × b due to the antisymmetry.

(a × b) × c − a × (b × c) = ha, ci b − hb, ci a + ha, bi c − ha, ci b = ha, bi c − hb, ci a = (a × c) × b

Lemma 4.16 ha × b, a × bi = ha, ai hb, bi − ha, bi2

Proof. ha × b, a × bi = |a × b|2 = |a|2 |b|2 sin2 γ = |a|2 |b|2 (1 − cos2 γ) = |a|2 |b|2 −
|a|2 |b|2 cos2 γ = ha, ai hb, bi − ha, bi2

36
Theorem 4.17

i j k a1 a2 a3
T
ha, bi = a b, a × b = a1 a2 a3 , a · b · c = b1 b2 b3 .
b1 b2 b3 c1 c2 c3

Point: (x, y, z)
Line: Let P0 be a point on the line l and v its direction. Then P ∈ l ⇔ ∃t ∈ R : P = P0 +tv
x − x0 y − y0 z − z0
⇒ x = x0 + tv1 , y = y0 + tv2 , z = z0 + tv3 or = = if vi 6= 0.
v1 v2 v3
Plane: Let P0 be a point on the plane α and n(A, B, C) be orthogonal to the plane.
−−→ D−−→ E
Then P ∈ α ⇔ P0 P ⊥n ⇔ P0 P , n = 0 ⇔ A(x − x0 ) + B(y − y0 ) + C(z − z0 ) =
Ax + By + Cz + D = 0, where D = −(Ax0 + By0 + Cz0 ).

4.2 Transformation of E3 with one fix point (origin)

1. Reflection in a plane through the originD E
−−→ −→ p, n
p0 = p+P P 0 = p+2P T = p−2pk = p−2 n,
hn, ni
where n = (A, B, C)T such that |n| = 1.
     
x0 x A
    h i 
 0
y  = y  − 2 Ax By Cz B =
   
     
z0 z C
    
1 − 2A2 −2AB −2AC x x

2
    Figure 4.1: Reflection in plane
−2AB 1 − 2B −2BC  y  = Rα y 
    

    
−2AC −2BC 1 − 2C 2 z z

2. Projection to a plane through the origin

p0 = p − p k ⇒
      
x0 1 − A2 −AB −AC x x
      
 0 2
y  = 
 −AB 1−B −BC  y 
 
= Pα 
y  ,

      
z0 −AC −BC 1 − C2 z z

where n = (A, B, C)T such that |n| = 1. Figure 4.2: Projection in plane

3. Reflection in a line through the origin

First, we reflect P in the plane, orthogonal to the line l to obtain P ∗ , then we reflect
P ∗ in the origin to obtain P 0 . Let v l = n = (A, B, C)T , such that |v e | = 1.

37
     
x0 1 − 2A2 −2AB −2AC x
    
 0 2
y  = −I· −2AB 1 − 2B −2BC  y  =
  
    
z0 −2AC −2BC 1 − 2C 2 z
    
2A2 − 1 2AB 2AC x x
    
= 2AB 2B 2 − 1 2BC  y  = Rl y 
    

    
2
2AC 2BC 2C − 1 z z

4. Projection to a line through the origin

−→ −−→ −−→ −−→
OP = p = OP 0 + OP ∗ ⇒ p0 = p − OP ∗
      
x0 x 1 − A2 −AB −AC x
 
 0
     Figure 4.3: Reflection in line
y  = I y − −AB 1 − B2 −BC  y  =
    
      
z0 z −AC −BC 1 − C2 z
      
A2 AB AC x x x
      
= AB

B 2
BC  y 
 
= Pe  y  = v e v Te y  .
   
      
AC BC C2 z z z

5. Left cross product with a fix vector

 
i j k Bz − Cy
 
p0 = n × p = A B C = Cx − Az  =
 
 
x y z Ay − Bx Figure 4.4: Projection in line
    
0 −C B x x
    
= C 0 −A y  = Cn  y 
    

    
−B A 0 z z

6. Rotation around a line, through the origin

p0 = Pl (p) + cos φPα (p) + sin φCn (p) ⇒
   
A2 AB AC 0 −C B
   
AB B2 BC  + sin φ  C 0 −A +
   
   
AC BC C2 −B A 0
 
1 − A2 −AB −AC
 
+ cos φ  −AB 1 − B2 −BC  = Rnφ Figure 4.5: Rotation around a line
 
 
−AC −BC 1 − C2

38
Definition 4.18 The homogeneous coordinates of a spatial point P is the equivalence class
of the assigned point quartet.
     
  x 
 x 

x   

   


y   y 
      


P (p) = y 
 
,→ pe =   ∼ α ,α ∈ R \ {0} .
z  z
       
   
z   
   


 

1 
1 

Q is an ideal point, if Q ∼ [x, y, z, 0]T , but @R ∼ [0, 0, 0, 0]T .

7. Translation with t
  
1 0 0 t1 x
      
0 1 0 t2  y  I t
    
p0 = p + t =    = Tt ∼ α ·  T  , α ∈ R \ {0}
0 0 1 t3  z  0 1
  
  
0 0 0 1 1

8. Linear transformations
    
x x0 A 0
x0 = A x ⇒   =  T   
1 0 1 1
    
x x0 A b
x0 = A x + b ⇒   =  T   
1 0 1 1

9. Point on a line
Let X(x) and Y (y) be two points on a line l. Then U (u) ∈ l ⇔ ∃αu = y +α(x−y) =
αx + (1 − α)y). Now let α and β be real numbers, then:
       
α β
x y αx + βy x + y
α +β  =  =  α+β α+β 
1 1 α+β 1

10. Rotation around a line

 
I −t
(a) Translation of a point of the line to the origin:  T 
0 1
 
Rαn 0
(b) Rotation around a line through the origin:  
0T 1
 
I t
(c) Inverse translation:  T 
0 1
           
I t Rαn 0 I −t I t Rαn −Rαn t Rαn t − Rαn t
 · ·  = T ·  = 
0T 1 0T 1 0T 1 0 1 0T 1 0T 1

39
Definition 4.19 If, X, Y and Z are collinear points such that ze = αx e + βy e , then
β
(X, Y, Z) := is called division ratio. If, X, Y , Z and W are collinear points such
α
β1 β2 (X, Y, Z)
that ze = α1 x
e + β1 y e = α2 x
e and w e + β2 y
e then (X, Y, Z, W ) := : = is called
α1 α2 (X, Y, W )
cross ratio.
 
A t
Lemma 4.20 The linear transformation Ae =  T  preserves cross ration and pre-
u 1
serves division ratio if and only if u = 0.
     
y A t x A t
Proof. ze = α1 x e⇒A
e + β1 y gze = α1 Aex
e + β1 A
eye = α1     + β1    =
T T
u 1 1 u 1 1
   
Ax + t Ay + tα1 ] β1 ]
α1  T  + β1  =
T
A(xe) +
T
A(ye) ⇒
u x+1 uT y + 1 u x+1 u y+1
β1
uT y+1 β1
]
(A( ]
e ), A(
x ]
ye), A(ze)) = α1 = ⇔ uT x = uT y ⇔ u = 0 and
uT x+1
α1
β1 β2
] ] ] ] uT y+1 uT y+1 β1 β2
(A(xe ), A(ye), A(ze), A( e =
w)) α1 : α2 = : = (X, Y, Z, W ).
uT x+1 uT x+1
α1 α2
Definition 4.21 A collineation is a spatial transformation, which is a bijection respected
to points and lines, and preserves incidence.

Theorem 4.22 Any collineation can be represented (in homogeneous 

coordinates)

by the
A t
equivalent class of a regular linear transformation given in Ae =  T  form.
u 1
Definition 4.23 A collineation is called affinity if u = 0. Similarity is an affinity, which
preserves the ratio of segments. Isometry is a similarity, which preserves the length of the
segments.

Lemma 4.24 A collineation is similarity, if u = 0 and A is orthogonal.

D E D E D E D E
Proof. λ2 x − y, x − y = x0 − y 0 , x0 − y 0 = A(x − y), A(x − y) = (x − y), AT A(x − y)
D E D E
But λ2 x − y, x − y = (x − y), λ2 E(x − y) ⇒ AT A = λ2 E.

Remark 4.25 If λ = 1 then we have an isometry.

4.3 Spherical geometry

Points: surface of the sphere
Lines: shortest path between two points on the surface of the sphere ⇒ great arcs ⇒
great circles
Problem: Betweeness is not appropriate

40
4.3.1 Spherical Order axioms
Axiom Os5 If (ACBD) then (BCAD) and (CBDA).

Axiom Os6 If A, B and D are collinear points then there exists at least one point C
such that (ABCAD).

Axiom Os7 If four points are situated on a line, there is no more than one 2-2 partition,
such that they separate each other.

Axiom Os8 (Pasch) Let A, B and C be three points not lying on the same line and e
and f be lines, not passing through any of the points A, B, C. Then there exist the points
E ∈ e and F ∈ f such that either (AECF ) or (BECF ).

Remark 4.26 If f is the ideal line, we get the Euclidean geometry.

4.3.2 Distance, angle and area

Elliptic geometry: Antipodal points are united: A = A0 .

Spherical geometry: Every point on the surface of the
sphere. A 6= A0 .
Distance of points: Length of the arc: r · α = α if r = 1.
If x = (x, y, z)T , then x ∼ x̃ = (x0 , y 0 , 1)T if z 6= 0.
D E
ã, b̃
cos α = r D E
hã, ãi b̃, b̃
Figure 4.6: Spherical geometry
Lines: Great (half-)circles of the sphere.
Angel of lines: Angles of the orthogonal vectors of the
determined planes:

hn˜1 , n˜2 i
cos α = q
hn˜1 , n˜1 i hn˜2 , n˜2 i

Definition 4.27 Two lines on the sphere divide the

surface into four spherical lune.

Area of spherical lune:

Figure 4.7: Angel of lines, spherical
Asphere = 4π ⇒ Alune = 2α lune

41
Area of the spherical triangle:
c a b
We cover of the sphere with lunes: b g
a
2(2α + 2β + 2γ) = 4π + 4Atriangle ⇒ O

Atriangle = α + β + γ − π

Now, let a, b and c be vectors to A, B and C such that

they form a right-hand system. Then (c × a, a × b)∠ =
π − α, (a × b, b × c)∠ = π − β, (b × c, c × a)∠ = π − γ.
Figure 4.8: Spherical triangle

Theorem 4.28 (Spherical sine theorem) Let a, b and c be the sides opposite to, and
sin a
α, β and γ be the angles at the vertices A, B and C of a spherical triangle. Then =
sin α
sin b sin c
= .
sin β sin γ
Proof. According to the Lagrange formula and the definition of the triple product:

(a × b) × (b × c) = ha, b × c)i b − hb, (b × c)i a = (a · b · c)b,

since b⊥(b × c). This implies that |(a × b) × (b × c)| = |a · b · c|, but |(a × b) × (b × c)| =
|a × b| · |b × c| sin(π − β) = sin c · sin a · sin β. Similarly, |(c × a) × (c × b)| = |c · a · b| = |a · b · c|
sin a sin b
and |(c × a) × (c × b)| = sin b · sin c · sin α. Finally, we obtain that = . The other
sin α sin β
equation can be proved similarly.

Theorem 4.29 (Spherical side cosine theorem) Let a, b and c be the sides opposite
to, and α, β and γ be the angles at the vertices A, B and C of a spherical triangle. Then
cos c = cos a · cos b + sin a · sin b · cos γ.

Proof. On the one hand, h(a × b), (b × c)i = h(a × b) × b, ci = hha, bi b − hb, bi a, ci =
ha, bi hb, ci − ha, ci = cos c · cos a − cos b. On the other hand, h(a × b), (b × c)i = |a × b| ·
|b × c| cos(π − β) = − sin c · sin a · cos β ⇒ cos b = cos a · cos c + sin a · sin c · cos β. Similarly,
cos c = cos a · cos b + sin a · sin b · cos γ and cos a = cos b · cos c + sin b · sin c · cos α.

Definition 4.30 Let ABC4 be a spherical triangle. Then the A0 := b × c, B 0 := c × a

and C 0 := a × b form the polar triangle of ABC4 (The distances of the corresponding
points are less then π2 ).

Theorem 4.31 Let ABC4 be a spherical triangle and A0 B 0 C 0 4 be its polar triangle.
Then the polar triangle of A0 B 0 C 0 4 is ABC4.

42
Proof. We know, that the length of the arcs AB 0 , AC 0 , BA0 , BC 0 , CA0 and CB 0 are equal
to π
2
and the length of the arcs AA0 , BB 0 and CC 0 is less then π2 , therefore the points A,
B and C satisfy the conditions for A00 , B 00 and C 00 .

Theorem 4.32 Let ABC4 be a spherical triangle with angles α, β and γ, and A0 B 0 C 0 4
be its polar triangle with sides a0 , b0 and c0 . Then a0 + α = b0 + β = c0 + γ = π, if they are
the corresponding sides and angles.

Proof. a0 = (c × a, a × b)∠ = π − α

Theorem 4.33 (Spherical angle cosine theorem) Let a, b and c be the sides opposite
to, and α, β and γ be the angles at the vertices A, B and C of a spherical triangle. Then
cos γ = − cos α · cos β + sin α · sin β · cos c.

Proof. We apply the spherical side cosine theorem for the polar triangle A0 B 0 C 0 4: cos c0 =
cos a0 ·cos b0 +sin a0 ·sin b0 ·cos γ 0 ⇒ cos(π−γ) = cos(π−α)·cos(π−β)+sin(π−α)·sin(π−β)·
cos(π − c). Now applying that sin(π − φ) = sin φ and cos(π − φ) = − cos φ we obtain that
− cos γ = cos α·cos β −sin α·sin β ·cos c. Similarly, cos α = − cos β ·cos γ +sin β ·sin γ ·cos a
and cos β = − cos α · cos γ + sin α · sin γ · cos b.

4.4 n-dimensional Euclidean geometry

Definition 4.34 Let (V, h·, ·i) be a vector space with the usual dot product and En be
a set. If the ordered point pairs of En can be assigned bijectively to the elements of the
vectors space satisfying the following properties, then En is the n-dimensional (analytic)
Euclidean space:
−→ −→
– ∀P, Q ∈ En ∃!v ∈ V : P Q = v – ∀P ∈ En : P P = 0
−→ −→ −→ −→
– ∀P ∈ En , v ∈ V∃!Q ∈ En : P Q = v – ∀P, Q, R ∈ En : P R + RQ + QP = 0

Definition 4.35 If Vk is a k-dimensional subspace of V and x0 ∈ En , then the X =

x0 + Vk set is called a k-dimensional affine subspace of En . If k = 1, then it is called line,
if k = n − 1, then it is called hyperplane.

x1 −x01 x2 −x02 xn −x0n

– k = 1 : x = x0 + t · v ⇒ v1
= v2
= ··· = vn
, and xi = x0i if vi = 0
Pn−1
– k = n − 1 : x = x0 + i=1 αi v i ⇒ {x − x0 , v 1 , . . . , v n−1 } not linearly independent ⇒
∃!n 6= 0 : hn, x − x0 i = 0 ⇒ hx, ni = c

43
4.5 Classification of quadratic surfaces
(Q) : xT A x + 2bT x + c = 0

is a quadratic form, where A ∈ Rn×n , b, x ∈ Rn , c ∈ R and A = AT . By homogeneous

 
x
coordinates X =  :
1
 
A b
(Q∗ ) : XT  T  X = 0
b c
Now, we change the coordinate system by translating the B orthogonal system by t. Then
 
T BT A B B T (A t + b)
(Q) : X  T X = 0
(t A + bT )B tt A t + bT t + tT b + c

B can be chosen such that B T A B is diagonal and λi are the diagonal elements i = 1 . . . n.
1. case ∀i : λi 6= 0 A is invertible and t = −A−1 b.
 
Λ 0
(Q) : XT  T  X = 0,
0 −bT A−1 b + c

where Λ = diag{λ1 , λ2 , . . . , λn }.
D E
2. case ∃i : λi = 0 Let si be the ith eigenvector such that si , sj = δij . Then B =
h i
s1 s2 . . . sn , t = t1 s1 + t2 s2 + · · · + tn sn and b = b1 s1 + b2 s2 + · · · + bn sn . Then
bi
A t = ni=1 ti A si ⇒ (B T (A t + b))i = ti λi + bi . If λi 6= 0, then ti := − , otherwise ti is
P
λi
arbitrary.

n = 2, D = c − bT A−1 b D>0 D=0 D<0

λ1 > 0 λ2 > 0 ∅ x=y=0 ellipse

λ1 > 0 λ2 = 0 parabola parabola, x2 = 0 parabola, λ1 x2 = −D

λ1 > 0 λ2 < 0 hyperbola λ1 x2 + λ2 y 2 = 0 hyperbola

λ1 = 0 λ2 = 0 ∅ x, y ∈ R ∅

44
 n = 3, D = c − bT A−1 b D>0 D=0 D<0

λ1 > 0 λ2 > 0 λ3 > 0 imaginary origin ellipsoid

ellipsoid

λ1 > 0 λ2 > 0 λ3 = 0 imaginary imaginary elliptic

elliptic intersecting cylinder
cylinder, planes,
elliptic elliptic elliptic
paraboloid paraboloid paraboloid

λ1 > 0 λ2 = 0 λ3 = 0 imaginary double parallel

parallel plane, planes,
planes,
parabolic parabolic parabolic
cylinder cylinder cylinder

λ1 > 0 λ2 < 0 λ3 = 0 hyperbolic intersecting hyperbolic

paraboloid planes, paraboloid,
planes,
hyperbolic hyperbolic hyperbolic
cylinder cylinder cylinder

λ1 > 0 λ2 > 0 λ3 < 0 two-sheeted cone one-sheeted

hyperboloid hyperboloid

λ1 = 0 λ2 = 0 λ3 = 0 ∅ whole space ∅

45
4.6 Conic sections
Definition 4.36 A double circular cone is formed by a set of lines connecting a common
point (apex) to all the points of a circle, where the orthogonal line through the center of
the circle to its plane contains the apex. Intersecting a double circular cone by a plane, we
obtain conic section.
One can realize that a sphere can be inscribed by in-
creasing its radius from K and the common points of
it with the cone form a circle. Let the plane of this cir- K
cle be π and our original intersecting plane be α. Let
d be the intersection of π and α, the tangent point of M d
DT N
the sphere to α be F . Let P K be an arbitrary gener-
ating line, where P ∈ α and let D be the intersection
F
of P K and π. Finally, let M and N be the orthogonal
projection of P to π and d respectively, and T be the P
orthogonal projection of K to π.
Since KT D4 ∼ P M D4 ⇒ M P D∠ = φ, if φ is half of the aperture. |P D| = |P F |, be-
cause they are tangent segments to the inscribed sphere ⇒ |P M | = |P D| cos φ. If ψ is the
|P F | sin ψ
angle of π and α, then |P M | = sin ψ|P N | = sin ψ|P d| ⇒ = = c.
|P d| cos φ
Definition 4.37 We say, that a conic section is ellipse, parabola, hyperbola if this ratio
c is smaller than, equal to, greater than 1 respectively.
Ellipse: Now, we assume, that α has a common
K
point with every director line (c < 1). Then we D
1
d1
have two inscribed spheres. Since |P D1 | = |P F1 | and
T N1
|P D2 | = |P F2 |, therefore |P F1 | + |P F2 | = |P D1 | + E1
P F1
|P D2 | = |D1 D2 |, but |D1 D2 | is a segment on a di-
rector line ⇒ constant ⇒ 2a := |D1 D2 |. For Ei , we E2 F2
d2
can claim, that |Ei F1 | + |Ei F2 | = 2a (tangent seg- N2 D2
ments). 2c := |F1 F2 | ⇒ 2a = |E1 F1 | + |E1 F2 | =
2|E1 F1 | + |F1 F2 | ⇒ |E1 F1 | = a − c. Similarly, |E2 F2 | = Figure 4.9: Ellipse: c < 1
a − c ⇒ |E1 E2 | = 2a.
Definition 4.38 Let F1 and F2 be two points on a plane such that d(F1 , F2 ) = 2c and
a be a real number such that a > c. Then the locus of points for which the sum of the
distances to F1 and F2 is 2a is called ellipse.

Parabola: Let α be parallel to exactly one director line (c = 1). Then there is no more
Dandelin sphere and |P F | = |P d|.

46
Definition 4.39 Let F be a point and d be a line, not lying on the point. Then the locus
of points for which the distance to the point is equal to the distance to the line is equal is
called parabola.

Hyperbola: Let α be parallel to exactly one direc-

tor line (c > 1). Then we have two inscribed spheres, F2
one in each part of the double cone. Similarly to
D2 E2
the case of the ellipse, |D1 D2 | = |E1 E2 | =: 2a and
d2
|F1 F2 | =: 2c, but now c > a. It is easy to prove, that
|P F1 | − |P F2 | = 2a
d1
Definition 4.40 Let F1 and F2 be two points on a
D1 E1
plane such that d(F1 , F2 ) = 2c and a be a real number
F1
such that a < c. Then the locus of points for which the
difference of the distances to F1 and F2 is 2a is called P
hyperbola.
Figure 4.10: Hyperbola: c > 1

4.6.1 Tangents from external point to conic sections

√
Ellipse: Let F1 , F2 be the focis, a be the semi-mayor axis, b := a2 − c2 be the semi-
minor axis and K be an external point.

Lemma 4.41 If we draw a circle around F2 with radius 2a then the E1 reflection of F1
in any tangent line of the ellipse lies on this circle.

Proof. Let T be the point of tangency, then |F1 T | = |E1 T | and the tangent t is the
perpendicular bisector of E1 F1 . Suppose, that T is not on the line of E1 F2 . Then because
of the triangle inequality 2a = |T F1 |+|T F2 | = |T E1 |+|T F2 | ≥ |E1 F2 |. But then E1 F2 ∩t 6=
∅ ⇒ Q := E1 F2 ∩ t ⇒ 2a ≥ |QE1 | + |QF2 | = |QF1 | + |QF2 | ⇒ Q is inside the ellipse and
t = T Q is not a tangent, therefore T = Q and 2a = |T E1 | + |T F2 | = |E1 F2 | ⇒ E1 lies on
the circle.
Construction. Let t1 and t2 be the tangent of
the ellipse through K, and the reflection of F1 in
them be E1 and E2 respectively. Then |KE1 | =
|KF1 | = |KE2 |, since ti is the perpendicular bi-
sector of F1 Ei . Then E1 and E2 lies on the cir-
cle around K with radius |KF1 | and on the circle
around F2 with radius 2a. ⇒ ti can be constructed.
Figure 4.11: Tangents to an ellipse

47
Parabola: Let F be the foci, d be the directrix and K be an external point.

Lemma 4.42 The E reflection of F in a tangent line lies on the directrix.

Proof. Let T be the point of tangency, then the

directrix is tangential to the circle around T with
radius |T F | ⇒ d⊥T Q, where Q is the common
point of the circle and the directrix. Then T lies
on the perpendicular bisector b of F Q. Assum-
ing, that b has another common point S with the
parabola, we get, that |QS| = |F S| = dist(S, d),
but Q lies on d. Let U be the footpoint of the
perpendicular line to d through S, then U QS4 is
an isosceles triangle with two right angle ⇒ b has
Figure 4.12: Tangents to a parabola
only one common point T with the parabola ⇒ b
is the tangent to T ⇒ b = t and E = Q.
Construction. Let t1 and t2 be the tangent of
the parabola through K, and the reflection of F
in them be E1 and E2 respectively. Then |KE1 | =
|KF | = |KE2 |, since ti is the perpendicular bisec-
tor of F Ei and Ei lies on the circle around K with
radius |KF1 | and on the directrix. ⇒ ti can be
constructed.
Hyperbola: The construction process is similar
to the case of the ellipse.
Figure 4.13: Tangents to a hyperbola

4.7 Polyhedrons in n-dimensional Euclidean space

Definition 4.43 X = x0 +Vk is a k-dimensional affine subspace, if Vk is a real k-dimensio-
nal real subspace and x0 ∈ En . If k = n − 1, X is called hyperplane and can be represented
⊥
as hx, ni = c, where n ∈ Vn−1 . Every hyperplane divides the space into three parts:
H := {x| hx, ni = c}, H − := {x| hx, ni > c}, H + := {x| hx, ni < c}.

Definition 4.44 Let H be a hyperplane, then H ∪ H + is a closed hyperplane.

Definition 4.45 C is an n-dimensional convex convex polyhedron, if it is bounded and

can be obtained by the intersection of finitely many closed hyperplanes such that it contains
an n-dimensional sphere.

48
Definition 4.46 Hi ∪Hi+ is a crucial hyperplane of C, if C ∩Hi 6= ∅ and ∃Q ∈ C ∩Hi ∀j 6=
i:Q∈
/ Hj .

Definition 4.47 Let C be an n-dimensional convex polyhedron, then the (n−1)-dimensional

facets of C are the intersections of C with its crucial hyperplanes.

Lemma 4.48 The (n − 1)-dimensional facets of an n-dimensional convex polyhedron are

(n − 1)-dimensional convex polyhedrons.

Proof. Let Q be a point in C such that Hi is the closest hyperplane to Q and d :=

min{dist(Q, Hj )}. Let G be an n-dimensional sphere around Q with radius d, then G ∩ Hi
i6=j
is an (n − 1)-dimensional sphere inside C ∩ Hi and C ∩ Hi = ∩ (Hi ∩ (Hj ∪ Hj+ )) ⇒ C ∩ Hi
i6=j
is an (n − 1)-dimensional polyhedron .

Remark 4.49 We can construct a chain of polyhedrons. The facets of any k-dimensional
polyhedron are (k − 1)-dimensional polyhedrons.

Definition 4.50 The 1-dimensional polyhedrons are edges and the 0-dimensional polyhe-
drons are vertices.

Theorem 4.51 Let C be an n-dimensional convex polyhedron with {V1 , V2 , . . . Vm } ver-

tices. Then C is the convex hull of Vi : C = { αi v i |αi ≥ 0 ∧
P P
αi = 1}.

Proof. Induction For k = 0, it is true and for k = 1, we get edges x = αv i + (1 − α)v j .

Now let P be a point on a (k + 1)-dimensional hyperface. Let vi be a vertex of this
−−→
hyperface and Vi P ray intersects it in Q. If Q is a vertex, then P lies on an edge and we
Pl
are done, otherwise Q lies on a k-dimensional hyperplane ⇒ q = i=1 αi v i , where αi ≥ 0
αi = 1. Since P ∈ QVi : p = αv i + βq = αv i + β
P P P
and αj v j = (αj β)v j + (α + βαi )v i ,
i6=j
where α + β = 1 ⇒ α + β(
P
αj ) = 1.

4.7.1 3-dimensional polyhedrons

Definition 4.52 S is a star-shaped polyhedron, if any edge belongs to exactly two facets,
the facets and the whole surface is simply connected, edge connected, face connected and
∃P ∈ S∀X ∈ S : P X ⊂ S.

Theorem 4.53 (Euler) Let v, e and f be the number of vertices, edges and facets in a
star-shaped polyhedron. Then v + f = e + 2.

49
Proof. If we project the polyhedron onto a sphere around an appropriate point, then the
structure will not change. Now, we chose an inside point of a spherical facet and we apply
a stereographic projection to get a planar graph. If there is a circle in it, then erasing an
edge of it will result in a planar graph with −1 domain and −1 edge. When there is not
any circle in the graph, then we get a tree with v vertices and only 1 domain. This tree
has v − 1 edges ⇒ (v − 1) + 2 = v + 1. Then e + 2 = v + f was true at the beginning.

(−1)i fi = 1 + (−1)n−1 , where fi is the

Pn
Remark 4.54 If we are in n-dimension, then i=0
number of the i-dimensional facets.

Definition 4.55 The union of regularly connected simple polygons is called polyhedral
surface if any edge belongs to at most two polygons. The boundary of a polyhedral surface
is the union of all the edges which belong to exactly one polygon. A polyhedral surface is
closed, if the boundary of it is an empty set.

Definition 4.56 A cycle on a polyhedral surface is a closed series of edges, such that
every vertex appears exactly twice.

Lemma 4.57 A closed polyhedral surface is simply connected, if for every circle, there
/ F ∀p = e0 , e1 , . . . , en = p0 edge series ∃i : fi
exists an F set of facets, such that ∀p ∈ F, p ∈
belongs to the cycle.

Remark 4.58 The Euler theorem is true for any face connected, simply connected closed
polyhedral surface.

Definition 4.59 Let C be a convex polyhedron and V be a vertex of it. Then the vertex
figure of V is formed by the points, lying on the surface of the unit ball around V and on
the edge rays, emanating from V .

Definition 4.60 Any C polyhedron is regular, if and only if all of its facets are congruent
regular polygons and all of its vertex figures are congruent regular polygons.

Theorem 4.61 There exists exactly five regular polyhedron in 3D.

Proof. Let the facets be regular n-gons and the vertex figures be regular m-gons. Then
2e 2e
2e = f · n = v · m ⇒ f = and v = , but according to the Euler theorem:
n m
2e 2e 1 1 1 1 1
f +v = + = 2+e ⇒ + = + > and n, m ≥ 3 ⇒ (m, n) ∈
n m n m e 2 2
{(3, 3), (3, 4), (4, 3), (3, 5), (5, 3)} . Now, we consider the dual of a regular polyhedron as
follows. The vertices of the dual polyhedron is the center of the facets, and we connect to
vertex if and only if the facets are neighboring.

50
(3, 3) ⇒ tetrahedron
(4, 3) ⇒ cube ⇒ its dual is the octahedron √ ! √ !
2 1+ 5 2 1+ 5
(5, 3) ⇒ dodecahedron: vertices: (±1, ±1, ±1), 0, ± √ ,± , ± √ ,± ,0
√ 1 + 5 2 1 + 5 2
!
1+ 5 2
and ± , 0, ± √ ⇒ its dual is the icosahedron.
2 1+ 5

a b
a
b

a
b a b

a b a

c b

b c

5 -1
a:b:c= 2 :1:
2

Figure 4.14: Regular polyhedrons

51
4.8 Congruence of polyhedra
Definition 4.62 The face lattice of a polyhedron is a partially ordered set, which consists
of the vertices, edges and facets, such that order is provided by the set theory containment.
Two polyhedra are combinatorically equivalent if their face lattice are isomorphic, i.e, there
exists a bijection between vertices, edges and facets such that it preserves order.

Theorem 4.63 (Cauchy’s rigidnes theorem) If, two convex polyhedra are combina-
torically equivalent, and each pair of facets are congruent to each other, then the polyhe-
drons are congruent.

Lemma 4.64 (Spherical Arm-lemma) Let Ai and A0i be the vertices of two spheri-
cal polygon such that ∀i = 1, 2, . . . , n − 1 : Ai Ai+1 ∼
= A0i A0i+1 and ∀i = 2, . . . , n − 1 :
Ai−1 Ai Ai+1 ∠ ≤ A0i−1 A0i A0i+1 ∠. Then A1 An ≤ A01 A0n holds.
Proof. Induction: For n = 3, it is the spherical Ai
triangle inequality ⇒ true. A
i+1
Case 1: ∃i : Ai−1 Ai Ai+1 ∠ = A0i−1 A0i A0i+1 ∠
Case 2: ∀i : Ai−1 Ai Ai+1 ∠ < A0i−1 A0i A0i+1 ∠, but
An−2 An−1 An ∠ can be increased to A0n−2 A0n−2 A0n ∠, A i-1
such that A1 , A2 , . . . , Aen remains convex.
A
Case 3: ∀i : Ai−1 Ai Ai+1 ∠ < A0i−1 A0i A0i+1 ∠, but n-1

An−2 An−1 An ∠ cannot be increased to A0n−2 A0n−2 A0n ∠, A1

An ~
because A2 A1 An are collinear points, An−2 An−1 Aen ∠ = An
A0n−2 A0n−2 A0n ∠, but A1 , A2 , . . . , Ãn is non-convex. Figure 4.15: Case 2
Case 1-2: One vertex can be deleted, either Ai or An−1 . Ai
Case 3: We compare A2 , . . . , An−1 , An (n − 1)-gon
A i+1
to A02 , . . . , A0n (n − 1)-gon. Then A2 An ≤ A02 A0n , by
induction, but A2 A1 + A1 An = A2 An ≤ A02 A0n ≤
A02 A01 + A01 A0n and A02 A01 = A2 A1 ⇒ A1 An ≤ A01 A0n . A2

Assume, that A1 , . . . , An and A01 , . . . , A0n are spher- An-1

ical polygons, such that Ai Ai+1 = A0i A0i+1 . If
An
A1
Ai−1 Ai Ai+1 ∠ > A0i−1 A0i A0i+1 ∠, then we put + to Ai , if
~
Ai−1 Ai Ai+1 ∠ < A0i−1 A0i A0i+1 ∠, then we put − to Ai , An
otherwise we put 0 to Ai .
There is a sign change, if Ai = ± and Ai+j = ∓ if
∀k = 1, . . . , j − 1 : Ai+k = 0. Figure 4.16: Case 3
Lemma 4.65 (Sign lemma) If, there is a sign on a polygon, then there are at least 4
sign changes.

52
 Ak
Proof. The number of sign changes is even. If, there
+ Al
is a sign, then there ought to be another different + -
sign, otherwise we have a contradiction by our pre- > <
vious lemma. Two sign changes can only be happen
if + and − signs are in two blocks. In the two sub- A 1 +
polygons, we apply the spherical arm lemma. We get -
Aj
that the diagonal is smaller and greater in A01 , . . . , A0n
than in A1 , . . . , An at the same time. Figure 4.17: Sign lemma
Proof. (Cauchy-theorem)
Case 1: If all the dihedral angles are the same, then we build up the polyhedrons face-by-
face.
Case 2: If all the dihedral angles change, then we write + or − if the dihedral angle is
greater or smaller in P 0 than in P . We can apply the sign lemma on the surface of a
small sphere around every vertex. Sign change: Two edges, sharing a vertex on a face
" # signs ⇒ 4v ≤ δ, where δ is the number of sign changes. But on the faces
have different
k
δ≤ αk ≤ 2(k − 2)αk = 2 k · αk − 4 αk = 4e − 4f ⇒ 4v ≤ δ ≤ 4v − 4f , but
P P P P
2
k≥3 2
4v + 4f = 4e + 8, so this is a contradiction.
Case 3: If some of the dihedral angels change, then a vertex is ’real’, if it has an edge with
a sign. We delete the ghost edges and the new facets are topologically connected surfaces.
Then v + f ≤ e. We add the ghost edges back one-by one, if one of its endpoint is in the
graph. With every possibility, the v + f ≤ e inequality remains true ⇒ contradiction.

53
5 Area and volume

5.1 Area on the Euclidean plane

Definition 5.1 Area is an isometry invariant, additive, non-negative set function for
simple polygons and 1 is assigned to the unit square.

Lemma 5.2 The area of the rectangle is the product of the length of its neighboring sides.

Lemma 5.3 The area of the triangle is the half of the product of the length of one of its
side and the corresponding altitude.

Proof. We cut the triangle int two right angled triangle by one of the altitudes. Another
copies of these triangles form two rectangles, where the sum of the appropriate sides is the
side of the triangle and the other is the corresponding altitude.

Lemma 5.4 Any simple polygon can be divided into triangles by non-intersecting diago-
nals.

Proof. Induction: For n = 3 it is trivial.

Consider the leftmost point v of the polygon and its neighbors u and v. If uv is a diagonal,
then we are done. Otherwise, there is at least on vertex of the polygon inside uvw4.
Consider the furthermost of them to uv ⇒ z. Then vz is a diagonal.

Remark 5.5 We get exactly n − 2 triangles.

Definition 5.6 The area of any simple poygon is the sum of the areas of the triangles, of
which the polygon consists.

Definition 5.7 A set H has area, if the infimum of the area of the circumscribed polygons
is equal to the supremum of the area of the inscribed polygons.

Lemma 5.8 Area is a monotonic function, if A ⊆ B and both have area, then A(A) ≤
A(B).

Theorem 5.9 Every convex bounded set has area.

54
5.2 Area on the hyperbolic plane
Definition 5.10 A triangle is called asymptotic, if one of its vertex is a boundary point. A
triangle is called doubly/triply asymptotic, if two/three of its vertices are boundary points.

Theorem 5.11 All the triply asymptotic triangles are congruent to each other.

Definition 5.12 Hyperbolic area is an isometry invariant, additive, non-negative set func-
tion for simple polygons and π is assigned to the triply asymptotic triangle.

Theorem 5.13 Any asymptotic triangle can be cut off into a pentagon.
Proof. (Poincaré disk model): Let ABC4 be such
that C is the ideal point and A is the center of the
model. Let D be an ideal point such that (ABD)
and M be the footpoint of the perpendicular line to
CD through A. Let A1 be the reflection of B in the A M1
line AM and the intersection of BC and A1 D be M1 . 12
B A1
Let the footpoints of the perpendicular line to DC B1 4 34 A2
through B and A1 be Q and P respectively. If M2 Q P M2
M
is the intersection of A1 P and BC and the reflection
of BC in A1 P is DA2 , then M2 A1 A2 4 is congruent D C
to A1 M1 M2 4 and M1 M2 B4. Continuing this proce-
Figure 5.1: Lendin construction
dure, we get the ABQP A1 pentagon.
Theorem 5.14 If, the angle of a doubly asymptotic triangle at the proper vertex is α,
then the area of this triangle is (π − α).

Proof. Let f (φ) be the area, if φ = π − α. is the supplementary angle. The union of
the two corresponding doubly asymptotic triangle is a triply asymptotic triangle: π =
f (φ) + f (π − φ) A triply asymptotic triangle can be cut off into three doubly asymptotic
triangle (see Figure 5.2): π = f (φ) + f (ψ) + f (π − φ − ψ). Using the previous result for
φ + ψ we obtain that f (φ) + f (ψ) = f (φ + ψ). Our only solution is f (x) = λx, since f is
k k+1
monotonously increasing and if f (1) = λ ⇒ f (n) = n · λ. Now, if n
≤x≤ n
⇒k≤
k f (x)
nx ≤ k + 1 ⇒ f (k) ≤ f (nx) ≤ f (k + 1) ⇒ λk ≤ nf (x) ≤ λ(k + 1) ⇒ n
≤ λ
≤ k+1
n
⇒
f (x) 1
∀n : x − λ
≤ n
⇒ f (x) = λx. Finally π = f (0) + f (π) = f (π) ⇒ λ = 1.

Theorem 5.15 The area of any hyperbolic triangle is its defect.

Proof. Let D, E and F be boundary points, such that (ABD), (BCE) and (CAF ).
Then the area of DEF 4 = π but

π = area(F DE4) = area(ABC)+area(ADF )+area(BDE)+area(CEF ) = area(ABC)+α+β+γ

Therefore area(ABC) = π − (α + β + γ).

55
 N
A f

f(f) f(f)
f(p- f)

N M L f p- f M
A

N
A

p-y p-f C

p+y B
L M

Figure 5.2: Hyperbolic area

5.3 Volume in the Euclidean space

Area: We only used the principum that two polygons have the same area if and only if
the first can be cut into finitely many polygonal pieces that can be reassembled to yield
the second.
Question: Can we extend this to 3D? (NO!)

Definition 5.16 Two polyhedrons are scissors-congruent if the first can be cut into finitely
many polyhedral pieces that can be reassembled to yield the second.

Dehn-invariant: We assign a value to every polyhedron such that two scissors-congruent

Pn
polyhedrons have the same value and D(P) = i=1 D(Pi ) if P has been cut into P1 , P2 , . . . , Pn .
Let f be an additive function such that f (0) = f (π) = 0, l(e) be the length of the edge
e and Θe be the dihedral angle between the two facets meeting at e. Then D(P) :=
P
e∈edges f (Θe )l(e)

• If e ∈ Pk and e is inside P, then the sum of the dihedral angles around e is 2π ⇒

f (2π)l(e) = 2f (π)l(e) = 0.

• If e ∈ Pk and e belongs to a facet of P , then the dihedral angles around e is

π ⇒ f (π)l(e) = 0.

• If e ∈ Pk and e belongs to the e0 edge of P, then f (Θe0 )l(e)

56
Theorem 5.17 The regular tetrahedron T and the cube C are not scissors-congruent.
 
1
Proof. Let l(e) := 1 for every e edge in T . Then D(T ) = 6f (Θ), where Θ = arccos 3
is
the dihedral angle of the regular tetrahedron. It is known, that neither Θ nor π is rational
 
π
∃f additive function such that f (Θ) = 1 and f 2
= 0. Since Θ and π are independent
vectors in R1 over Q ⇒ ∃ base with Θ and π. Let f (Θ) be 1 and f (b) be 0 for all b ∈base
(f is not linear over R). Then f satisfies all conditions and D(T ) = 6 but D(C) = 0.

Cavalieri principle: The set function V satisfies the Cavalieri principle if the following
property is true for any H1 and H2 sets in the domain of V : If every element of a parallel
plane pencil intersects both H1 and H2 in cross-sections of equal are, then V (H1 ) = V (H2 ).

Definition 5.18 Volume is an isometry invariant, non-negative, additive set function for
simple polyhedrons, which satisfies the Cavalieri principle and 1 is assigned to the unit
cube.

Theorem 5.19 V (T ) = 13 area(ABC4 ) · m, where m = dist(D, ABC).

Proof.

f b
g f a c
d e f d e g
c e cg d
c b e
b a
a
Figure 5.3: Volume of the tetrahedron

57