nLab Euclidean geometry

Idea

Euclidean geometry (Euclid 300BC) studies the geometry of Euclidean spaces. These spaces come equipped with a canonical metric in the sense of analysis, in fact a Riemannian metric, and Euclidean geometry may be regarded as the local model for Riemannian geometry, in the sense of Cartan geometry.

In distinction from non-Euclidean geometry (hyperbolic geometry or elliptic geometry), Euclidean geometry is Riemannian geometry of Riemannian manifolds with zero sectional curvature.

Properties

• Pythagorean theorem

• pentagon decagon hexagon identity

• conic section

• Hadwiger's theorem

• Non-Euclidean geometry

Axiomatizations

Euclidean geometry in the sense of the Idea section makes# reference to the extrinsic notion of real number (to delineate the notion of Euclidean space) and whatever surrounding structure is used to found that notion, but it has also received more intrinsic axiomatic treatments in the spirit of synthetic geometry. Notable among these are Hilbert's axioms, as set out in his Grundlagen der Geometrie, and Tarski‘s axioms, which form a classical first-order axiomatic theory.

Hilbert’s axioms

Here we will just give an extremely sketchy description (for now, see Wikipedia for a fuller account). It should be said right away that two of the Hilbert axioms are not first-order.

Hilbert’s theory has

• Three sorts: points, line, planes (or if you like: tables, chairs, and beer mugs¹);

• Six relations: betweenness (a ternary relation on points), three incidence relations (one for points and lines, one for points and planes, one for lines and planes), and two congruence relations (a relation $L(a, b, c, d)$ on points whose intuitive meaning is that the the line segment $a b$ is congruent to the line segment $c d$, and a relation $A(a, b, c, d, e, f)$ on points whose intuitive meaning is that the angle $a b c$ is congruent to the angle $d e f$);

• A bunch of axioms (20 or 21 depending on which specific edition or translation or exposition one is working with; Wikipedia says 20 but actually has, confusingly, 21 – see the Talk page for more information). They are organized as

  • Axioms of incidence (8),
  • Axioms of betweenness (4),
  • Axioms of congruence (6),
  • Axiom of parallels (the parallel postulate) (1),
  • Axioms of completeness (2), mirroring the archimedean property and the completeness properties of the real numbers. These are not first-order.

I (Todd Trimble) assume Hilbert’s theory is, or is at least intended to be, categorical in the model-theoretic sense, at least considered against a fixed set-theoretic background. In other words I assume the real numbers as complete ordered field would be a definable structure and the axioms are uniquely modeled by $\mathbb{R}^3$ with its Euclidean structure.

Tarski’s axioms

Tarski’s theory EPG (Elementary Plane [Euclidean] Geometry) is a one-sorted theory in classical first-order logic with equality, with variables representing points. There are no constants or operations in the signature, and just two relations:

• A ternary relation $B$ (betweenness), with $B(x, y, z)$ meaning $y$ is between $x$ and $z$ ($y$ is on the line segment between $x$ and $z$) – we will write instead $B x y z$ to conserve space;

• A 4-ary relation $C$ (congruence), with $C(x, y, z, w)$ meaning that a line segment $x y$ is congruent (of the same length) as $z w$. For readability, we write this as $x y \equiv z w$.

The theory can be presented (in some sense very parsimoniously) by 10 axioms and one first-order axiom scheme. Axioms 8 and 9 constrain the dimension to $n = 2$ (Euclidean plane geometry), but there are analogues of 8 and 9 that constrain the dimension to any chosen number. Axiom 10 is an extremely disguised version of the Parallel Postulate, with a complicated history behind it. The axiom scheme 11 plays the role of a Dedekind cut axiom expressed in first-order terms.

1. $\vdash \; a b \equiv b a$,

2. $a b \equiv p q, a b \equiv r s \;\; \vdash \;\; p q \equiv r s$,

3. $a b \equiv c c \vdash a = b$,

4. $\vdash \; (\exists x)\; B q a x \wedge a x \equiv b c$ (segment construction: given a line segment $b c$ and a line segment from $q$ to $a$, one can extend that line segment in that direction by a segment $a x$ congruent to $b c$),

5. $a \neq b, B a b c, B a' b' c', a b \equiv a' b', b c \equiv b' c', a d \equiv a' d', b d \equiv b' d' \;\; \vdash \;\; c d \equiv c' d'$ (5 segment axiom),

6. $B a b a \; \vdash \; a = b$,

7. $B a p c, B b q c \;\; \vdash \;\; (\exists x)\; B p x b \wedge B q x a$ (Pasch axiom: roughly, if a line goes through one side of a triangle and not through a vertex, then it must go through another side),

8. $\vdash \; (\exists a, b, c)\; \neg (B a b c \vee B b c a \vee B c a b)$ (there exist three non-collinear points: the dimension is at least 2),

9. $p \neq q, a p \equiv a q, b p \equiv b q, c p \equiv c q \;\; \vdash \;\; B a b c \vee B b c a \vee B c a b$ (if three points are each equidistant from two given distinct points $p$ and $q$, then those three points are collinear: the dimension is at most 2),

10. $B a d t, B b d c, a \neq d \;\; \vdash \;\; (\exists x, y)\; B a b x \wedge B a c y \wedge B y t x$ (Parallel Postulate (!)),

11. (Cut schema) Let $\phi(x)$ be a formula in which variables $a, b, y$ do not occur freely, and let $\psi(y)$ be a formula in which $a, b, x$ do not occur freely. Then

    $$(\forall x, y)\; (\phi(x) \wedge \psi(y) \Rightarrow B a x y) \;\; \vdash \;\; (\exists b)(\forall x, y)\; (\phi(x) \wedge \psi(y) \Rightarrow B x b y)$$

    (if every element $y$ in the set defined by $\psi$ lies beyond every element $x$ in the set defined by $\phi$ as seen from the perspective of $a$, then there is some $b$ between every such $x$ and every such $y$).

See Tarski-Givant for a thorough discussion of this and related axiom systems, and an extensive bibliography.

One important aspect of the meta-theory lies in the relation of models of this theory to real closed fields. If $R$ is a real closed field, then there is a corresponding model of EPG whose underlying set is $R \times R$ and where the relations $B$ and $C$ are interpreted by expected conditions:

$$C(x, y, z, w) \;\;\; iff \;\;\; (x_1 - y_1)^2 + (x_2 - y_2)^2 = (z_1 - w_1)^2 + (z_2 - w_2)^2$$

$$\,$$

$$B(x, y, z) \;\;\; iff \;\;\; (x_1 - y_1)(y_2 - z_2) = (x_2 - y_2)(y_1 - z_1) \; \wedge \; 0 \leq (x_1 - y_1)(y_1 - z_1) \; \wedge \; 0 \leq (x_2 - y_2)(y_2 - z_2).$$

It is mostly straightforward to verify that this structure on $R \times R$ satisfies the EPG axioms (all but the cut schema hold for any ordered field). More importantly,

Theorem

For every model $M$ of EPG there is a real closed field $R$ such that $M \cong R \times R$ as models.

Of course there is some arbitrariness involved in the choice of isomorphism; for example, any line segment $x y$ can be used to stipulate a unit length and a coordinate axis. The proof is lightly sketched in Tarski; again only the first ten axioms are needed to produce an ordered field structure on $R$, with the cut schema being used to verify the real closure.

The significance of this result is that EPG is a complete and decidable theory, just as the theory of real closed fields is complete and decidable. Hence there exists an algorithm which, on being fed a sentence in the language of Elementary Plane Geometry, will produce a proof of the sentence or a proof of its negation in the theory.

Mac Lane’s axioms

To be written, see Mac Lane (1959) for details.

Related concepts

• trigonometry

Basic notions and facts in ( differential) Euclidean geometry (see also at: differential geometry of curves and surfaces)

• parallel postulate

• length, area, volume

• curve

  • velocity

  • areal velocity

  • acceleration

  • jerk

• circle

  • circumference of a circle

  • area enclosed by a circle

• regular polygon

  • center of a regular polygon

• space form

Other flavors of geometry

synthetic geometry
Euclidean geometry
hyperbolic geometry
elliptic geometry

Riemannian geometry, Cartan geometry:

geometric context	gauge group	stabilizer subgroup	local model space	local geometry	global geometry	differential cohomology	first order formulation of gravity
differential geometry	Lie group/algebraic group $G$	subgroup (monomorphism) $H \hookrightarrow G$	quotient (“coset space”) $G/H$	Klein geometry	Cartan geometry	Cartan connection
examples	Euclidean group $Iso(d)$	rotation group $O(d)$	Cartesian space $\mathbb{R}^d$	Euclidean geometry	Riemannian geometry	affine connection	Euclidean gravity
	Poincaré group $Iso(d-1,1)$	Lorentz group $O(d-1,1)$	Minkowski spacetime $\mathbb{R}^{d-1,1}$	Lorentzian geometry	pseudo-Riemannian geometry	spin connection	Einstein gravity
	anti de Sitter group $O(d-1,2)$	$O(d-1,1)$	anti de Sitter spacetime $AdS^d$				AdS gravity
	de Sitter group $O(d,1)$	$O(d-1,1)$	de Sitter spacetime $dS^d$				deSitter gravity
	linear algebraic group	parabolic subgroup/Borel subgroup	flag variety	parabolic geometry
	conformal group $O(d,t+1)$	conformal parabolic subgroup	Möbius space $S^{d,t}$		conformal geometry	conformal connection	conformal gravity
supergeometry	super Lie group $G$	subgroup (monomorphism) $H \hookrightarrow G$	quotient (“coset space”) $G/H$	super Klein geometry	super Cartan geometry	Cartan superconnection
examples	super Poincaré group	spin group	super Minkowski spacetime $\mathbb{R}^{d-1,1\vert N}$	Lorentzian supergeometry	supergeometry	superconnection	supergravity
	super anti de Sitter group		super anti de Sitter spacetime
higher differential geometry	smooth 2-group $G$	2-monomorphism $H \to G$	homotopy quotient $G//H$	Klein 2-geometry	Cartan 2-geometry
	cohesive ∞-group	∞-monomorphism (i.e. any homomorphism) $H \to G$	homotopy quotient $G//H$ of ∞-action	higher Klein geometry	higher Cartan geometry	higher Cartan connection
examples			extended super Minkowski spacetime		extended supergeometry		higher supergravity: type II, heterotic, 11d

References

• Euclid, Elements, 300BC

• Alfred Tarski, What is elementary geometry?, in The axiomatic method. With special reference to geometry and physics Proceedings of an International Symposium held at the Univ. of Calif., Berkeley, Dec. 26, 1957-Jan. 4, 1958 (ed. L. Henkin, P. Suppes, and A. Tarski), Studies in Logic and the Foundations of Mathematics, Amsterdam: North-Holland (1959), pp. 16–29. (pdf)

• Alfred Tarski and Steven Givant, Tarski’s system of geometry, Bull. Symb. Logic, Vol. 5 No. 2 (1999), 175–214. (web)

• Saunders Mac Lane, Metric postulates for plane geometry, American Mathematical Monthly, Vol. 66 No. 7 (1959), 543–555. (doi:10.2307/2309851, web)

Textbook account:

• Elmer Rees, Notes on Geometry, Springer 2005

Expository lecture notes:

• Drew Armstrong, Introduction to Geometry, 2017 (pdf)

Textbook accounts of the axiomatization of Euclidean geometry include these:

• Wolfram Schwabhäuser, Wanda Szmielew, Alfred Tarski, Mathematische Methoden in der Geometrie, Springer 1983

• Anton Petrunin, Euclidean Plane and its Relatives; a minimalistic introduction, (arXiv:1302.1630)

Full formalization of this book in Coq (as synthetic geometry but following Tarski’s work) is discussed at

• Gabriel Braun, Pierre Boutry, Julien Narboux, GeoCoq, La géométrie de Tarski en Coq

A formalisation of Euclid using diagrammatic reasoning is in

• J. Avigad, E. Dean, J. Mumma, A formal system for Euclid’s Elements , arXiv:0810.4315 (2009).

Another formalization of Euclidean geometry in Coq:

• Evgeny V. Ivashkevich, On Constructive-Deductive Method For Plane Euclidean Geometry (arXiv:1903.05175)

---

1. After a famous saying attributed to Hilbert which in some sense heralds the axiomatic spirit: ‘Man muß jederzeit an Stelle von “Punkte, Geraden, Ebenen” “Tische, Stühle, Bierseidel” sagen können’. The 1891 anecdote is reported by O. Blumenthal in Hilbert, Gesammelte Abhandlungen III. Band , Springer Berlin 1935, p.403. The three classes became ‘Liebe, Gesetz, Schornsteinfeger’ (love, law, chimney-sweep) in Hilbert’s letter to Frege (Dec. 29th 1899). The question raised in Frege’s rejoinder (Jan. 6th 1900) whether his pocket watch is a point is still an open problem. ↩