Hilbert Eng
Hilbert Eng
Hilbert Eng
A SHORT INTRODUCTION
BJØRN JAHREN
Euclid’s “Elements” introduced the axiomatic method in geometry, and for more
than 2000 years this was the main textbook for students of geometry. But the 19th
century brought about a revolution both in the understanding of geometry and of
logic and axiomatic method, and it became more and more clear that Euclid’s sys-
tem was incomplete and could not stand up to the modern standards of rigor. The
most famous attempt to rectify this was by the great German mathematician David
Hilbert, who published a new system of axioms in his book “Grundlagen der Geome-
trie” in 1898. Here we will give a short presentation of Hilbert’s axioms with some
examples and comments, but with no proofs. For more details, we refer to the rich
literature in this field — e. g. the books ”Euclidean and non-Euclidean geometries”
by M. J. Greenberg and ”Geometry: Euclid and beyond” by R. Hartshorne.
Hilbert also treats geometry in 3-space, but we will only consider the 2-
dimensional case. We begin by agreeing that the basic ingredients in our
study are points and lines in a plane. At the outset the plane is just a set
S where the elements P are called points. The lines are, or can at least
be naturally identified with certain subsets l of S, and the fundamental
relation is the incidence relation P ∈ l, which may or may not be satisfied
by a point P and a line l. But we also introduce two additional relations:
betweenness, enabling us to talk about points lying between two given points,
and congruence, which is needed when we want to compare configurations in
different parts of the plane. Hilbert formulated three sets of axioms for these
relations: incidence axioms, betweenness axioms and congruence axioms. In
addition to these we also need an axiom of continuity to make sure that lines
and circles have “enough” points to intersect as they should, and of course
the axiom of parallels. As we introduce Hilbert’s axioms, we will gradually
put more and more restrictions on these ingredients, and in the end they
will essentially determine Euclidean plane geometry uniquely. The axioms
are also independent, in the sense that for each axiom A there is a model
satisfying all the rest of the axioms, but not A.
Note that although circles also are important objects of study in classical
plane geometry, we do not have to postulate them, since, as we shall see,
they can be defined in terms of the other notions.
Before we start, maybe a short remark about language is in order: An
axiom system is a formal matter, but the following discussion will not be
very formalistic. After all, the goal is to give a firm foundation for matters
that we all have a clear picture of in our minds, and as soon as we have
introduced the various formal notions, we will feel free to discuss them in
more common language. For example, although the relation P ∈ l should,
strictly speaking, be read: “P and l are incident”, we shall use “l contains
P ”, “P lies on l” or any obviously equivalent such expression.
2 BJØRN JAHREN
We are now ready for the first group of axioms, the incidence axioms:
I1: For every pair of distinct points A and B there is a unique line l
containing A and B.
I2: Every line contains at least two points.
I3: There are at least three points that do not lie on the same line.
We let AB denote the unique line containing A and B.
These three axioms already give rise to much interesting geometry, so-
called “incidence geometry”. Given three points A, B, C, for example,
any two of them span a unique line, and it makes sense to talk about the
triangle ABC. Similarly we can study more complicated configurations.
The Cartesian model R2 of the Euclidean plane, where the lines are the
sets of solutions of nontrivial linear equations ax + by = c, is an obvious
example, as are the subsets obtained if we restrict a, b, c, x, y to be rational
numbers (Q2 ), the integers (Z2 ), or in fact any fixed subring of R. However,
spherical geometry, where S is a sphere and the lines are great circles, is not
an example, since any pair of antipodal points lies on infinitely many great
circles — hence the uniqueness in I1 does not hold. This can be corrected by
identifying every pair of antipodal points on the sphere. Then we obtain an
incidence geometry called the (real) projective plane P2 . One way to think
about the points of P2 is as lines through the origin in R3 . If the sphere
has center at the origin, such a line determines and is determined by the
antipodal pair of points of intersection between the line and the sphere. A
“line” in P2 can then be thought of as a plane throuh the origin in R3 , since
such a plane intersects the sphere precisely in a great circle. Notice that in
this interpretation the incidence relation P ∈ l corresponds to the relation
“the line l is contained in the plane P ”.
There are also finite incidence geometries — the smallest has exactly three
points where the lines are the three subsets of two elements.
The next group of axioms deals with the relation “B lies between A and
C”. In Euclidean geometry this is meaningful for three points A, B and C
lying on the same straight line. The finite geometries show that it is not
possible to make sense of such a relation on every incidence geometry, so
this is a new piece of structure, and we have to declare the properties we
need. We will use the notation A ∗ B ∗ C for “B lies between A and C”.
Hilbert’s axioms of betweenness are then:
B1: If A ∗ B ∗ C, then A, B and C are distinct points on a line, and
C ∗ B ∗ A also holds.
B2: Given two distinct points A and B, there exists a point C such that
A ∗ B ∗ C.
B3: If A, B and C are distinct points on a line, then one and only one
of the relations A ∗ B ∗ C, B ∗ C ∗ A and C ∗ A ∗ B is satisfied.
B4: Let A, B and C be points not on the same line and let l be a line
which contains none of them. If D ∈ l and A ∗ D ∗ B, there exists
an E on l such that B ∗ E ∗ C, or an F on l such that A ∗ F ∗ C, but
not both.
HILBERT’S AXIOM SYSTEM 3
C4 and C5 are the obvious analogues of C1 and C2, but note that C4
says that we can construct an arbitrary angle on both sides of a given ray.
C6 says that a triangle is determined up to congruence by any angle and
HILBERT’S AXIOM SYSTEM 5
its adjacent sides. This statement is often referred to as the “SAS” (side–
angle–side) congruence criterion.
In the Euclidean plane R2 we define congruence as equivalence under
actions of the Euclidean group of transformations of R2 . This is generated
by rotations and translations, and can also be characterized as the set of
transformation of R2 which preserve all distances. It is quite instructive to
prove that the congruence axioms hold with this definition.
These three groups contain the most basic axioms, and they are sufficient
to prove a large number of propositions in book I of the “Elements”. How-
ever, when we begin to study circles and “constructions with ruler and com-
pass”, we need criteria saying that circles intersect (have common points)
with other circles or lines when our intuition tells us that they should. The
next axiom provides such a criterion.
First a couple of definitions:
Definition: Let Γ be a circle with center O and radius OA. We say that
a point B is inside Γ if OB < OA and outside if OA < OB.
We say that a line or another circle is tangent to Γ if they have exactly
one point in common with Γ.
We can now formulate Hilbert’s axiom E :
E: Given two circles Γ and ∆ such that ∆ contains points both inside
and outside Γ. Then Γ and ∆ have common points. (They “inter-
sect”.)
(It follows from the other axioms that they will then intersect in exactly
two points.) This is an example of what we call a continuity axiom. The
following variation is actually a consequence of axiom E:
E’: If a line l contains points both inside and outside the circle Γ, then
l and Γ will intersect. (Again in exactly two points.)
Hilbert gives the Axiom of parallels the following formulation — often
called “Playfair’s axiom” (after John Playfair i 1795, although it goes back
to Proclus in the fifth century):
P: (Playfair’s axiom) Given a line l and a point P not on the line. Then
there is at most one line m through P which does not intersect l.
If the lines m and l do not intersect, we say that they are parallel, and
we write m k l. The existence of a line m through P parallel to l can be
shown to follow from the other axioms, so the real content of the axiom is
the uniqueness.
With these axioms we are able to prove all the results in Euclid’s “Ele-
ments” I–IV, but they do not yet determine the Euclidean plane uniquely.
The standard plane (R2 with the structure defined so far) is an example,
and it is an instructive exercise to prove this in detail, but we obtain other
examples by replacing the real numbers by another ordered field where ev-
ery element has a square root! For uniqueness we need a stronger continuity
axiom, as for instance Dedekind’s axiom:
6 BJØRN JAHREN
Exercises.
1. Find all incidence geometries with four or five points.
2. Let V be a vector space of dimension at least 2 over a field F .
Show that V satisfies I1-3, if we define lines to be sets of the form
{A + tB|t ∈ F }, where A, B ∈ V , B 6= 0.
3. Show that Q2 satisfies axioms B1–4, but Z2 does not.
4. Prove that ’being on the same side of the line `’ is an equivalence
relation on the complement of `, with exactly two equivalence classes.
5. Show that Q2 does not satisfy C1. Try to determine conditions on
an algebraic extension F of Q such that F 2 will satisfy C1.
6. Show that the center of a circle is uniquely determined.
7. Discuss which axioms are needed in order to bisect a given segment.
2
8. Which axioms are satisfied by Q , where Q is the algebraic closure
of Q?
9. Show that the axiom of Archimedes can be used to define a length
function on segments.
10. Suppose given a geometry with incidence, betweenness and congru-
ence, and let r be a ray with vertex O. r determines a unique line `
containing it.
Show that we can give ` the structure of an ordered abelian group,
with O as neutral element and such that P > O if and only if P ∈ r.
Show that two different rays give rise to isomorphic ordered groups.