V .

7 : Non – Euclidean geometry in modern mathematics

Hyperbolic geometry, which was considered a

th
dormant subject ... [in the middle of the 20
century], has turned out to have extraordinary
applications to other branches of mathematics.
Greenberg, p. 382
It seems appropriate to conclude this unit on non – Euclidean geometry with a brief
discussion of the role it plays in present day mathematics. Questions of this sort arise
naturally, and In particular one might ask whether objects like the hyperbolic plane
are basically formal curiosities or if they are important for reasons beyond just
showing the logical independence of the Fifth Postulate. In fact, hyperbolic
geometry turns out to play significant roles in several contexts of independent interest.
Some of these date back to the 19th century, and others were discovered during the last
few decades of the 20th century.

Additional models for hyperbolic geometry

Most of the ties between hyperbolic geometry and other topics in mathematics involve
mathematical models for the hyperbolic plane (and spaces of higher dimensions) which
are different from the Beltrami – Klein models described in the preceding section. There
are three particularly important examples. One model (the Lorent zian model) is
discussed at length in Chapter 7 of Ryan, and two other basic models are named after
H. Poincaré. We shall only consider a few of properties of the Poincaré models in these
notes. Further information can be found at the following online sites:
http://www.geom.uiuc.edu/docs/forum/hype/model.html
http://www.mi.sanu.ac.yu/vismath/sazdanovic/hyperbolicgeometry/hypge.htm
http://math.fullerton.edu/mathews/c2003/poincaredisk/PoincareDiskBib/Links/PoincareDiskBib_lnk_1.html
http://mathworld.wolfram.com/PoincareHyperbolicDisk.html
http://www.geom.uiuc.edu/~crobles/hyperbolic/hypr/modl/
Probably the most important and widely used model for hyperbolic geometry is the
Poincaré disk model. In the 2 – dimensional case, one starts with the points which lie
in the interior of a circle (i.e., in an open disk) as in the Beltrami – Klein model, but the
definitions of lines, distances and angle measures are different. The lines in this model
are given by two types of subsets.
(1) Open “diameter” segments with endpoints on the boundary circle.
(2) Open circular arcs whose endpoints lie on the boundary circle and meet
the boundary circle orthogonally (i.e., at each endpoint, the tangent to the
boundary circle is perpendicular to the tangent for the circle containing the arc).
An illustration of the second type of “line” is given below.

276
The drawing below illustrates several lines in the Poincaré disk model.

(Source: http://www.geom.uiuc.edu/~crobles/hyperbolic/hypr/modl/pncr/ )
The Poincaré disk model distance between two points is given by a formula which
resembles the comparable identity for the Beltrami – Klein model, and it is given in the
first online reference in the list of online sites at the beginning of this section. On the
other hand, one fundamentally important feature of the Poincaré disk model is that its
angle measurement is exactly the same as the Euclidean angle between two
intersecting curves (i. e., given by the usual angle between their tangents); such angle
measure preserving models are said to be conformal. In contrast, both the distance
and the angle measurement in the Beltrami – Klein model are different from their
Euclidean counterparts.
The second Poincaré model in two dimensions is the Poincaré half – plane model, and
2
its points are given by the points in the upper half plane of R ; in other words, the points
are all ordered pairs (x, y) such that y > 0. The lines in this model are once again
given by two types of subsets.
(1) Vertical open rays whose endpoints lie on the x – axis.
(2) Open semicircular arcs whose endpoints lie on the x – axis.
The drawing below illustrates several lines in the Poincaré half – plane model.

277
 (Source: http://www.geom.uiuc.edu/~crobles/hyperbolic/hypr/modl/uhp/ )
The Poincaré half – plane model distance between two points is given by a formula in
the first online reference in the list of online sites at the beginning of this section. As in
the preceding case, one fundamentally important feature of the Poincaré half – plane
model is that its angle measurement is exactly the same as the Euclidean angle
between two intersecting curves (i.e., given by the usual angle between their tangents).

Euclidean models of the hyperbolic plane. In all the preceding models, it was
necessary to introduce a special definition of distance in order to make everything work
right. It would be very satisfying if we could give a nice model for the hyperbolic plane
in Euclidean 3 – space for which the distance is something more familiar (i.e., the
hyperbolic distance between two points is the length of the shortest curve in the model
joining these points), but unfortunately this is not possible. The first result to show that
no reasonably nice and simple model can exist was obtained by D. Hilbert (1862 – 1943)
in 1901, and it was sharpened by N. V. Efimov (1910 – 1982) in the 1950s. One
reference for Hilbert’s Theorem is Section 5 – 11 in the following book:
M. Do Carmo, Differential Geometry of Curves and Surfaces, Prentice – Hall,
Upper Saddle River, NJ, 1976. ISBN: 0–132–12589–7.
In contrast, during 1950s N. H. Kuiper (1920 – 1994) proved a general result which
shows that the hyperbolic plane can be realized in Euclidean 3 – space with the “right”
distance, but the proof is more of a pure existence result than a method for finding an
explicit example, and in any case the results of Hilbert and Efimov show that any such
example could not be described very simply. Kuiper’s result elaborates upon some
fundamental results of J. Nash (1928 – ); another an extremely important general result
of Nash implies that the hyperbolic plane can be realized nicely in Euclidean n – space if
n is sufficiently large; it is known that one can take n = 6, but apparently there are
open questions about the existence of such realizations if n = 5 or 4. Here are
references for the realizability of the hyperbolic plane in Euclidean 6 – space; the first is
the original paper on the subject, and the second contains a fairly explicit construction of
a nice model near the end of the file.
D. Blanuša, Über die Einbettung hyperbolischer Räume in euklidische
Räume. Monatshefte für Mathematik 59 (1955), 217 – 229.
http://www.math.niu.edu/~rusin/known-math/99/embed_hyper
In yet another — and more elementary — direction, it is not difficult to represent small
pieces of the hyperbolic plane nicely in Euclidean 3 – space. In particular, this can be

278
done using a special surface of revolution known as a pseudosphere. Further
information on this surface can be found in many differential geometry books and notes,
including pages 96 – 97 of the following online reference:
http://math.ucr.edu/~res/math138A/dgnotes2006.pdf

Footnote. (This is basically nonmathematical information.) The extraordinary life of

John Nash received widespread public attention in the biography, A Beautiful Mind, by
S. Nasar, and the semifictional interpretation of her book in an Academy Award winning
film of the same name. During the 1950s Nash proved several monumental results in
geometry, but in nonmathematical circles he is better known for his earlier work on game
theory, for which he shared the 1994 Nobel Prize in Economics with J. Harsányi (1920 –
2000) and R. Selten (1930 – ); an ironic apsect of this is noted in the footnote at the
bottom of page 565 in Greenberg.

Hyperbolic geometry models and differential geometry

We have already mentioned Riemann’s approach to the classical non – Euclidean

geometries, which views the latter as special types of objects now called Riemannian
geometry. Numerous properties of hyperbolic n – spaces play fundamental roles in
many aspects of that subject, including some that have seen a great deal of progress
over the past three decades. Three books covering many of these advances are
discussed in a relatively recent book review by B. Kleiner [ Bull. Amer. Math. Soc. (2) 39
(2002), 273 – 279. ]

The Poincaré disk model and functions of one complex variable

The investigation of the symmetries of a given

mathematical structure has always yielded the
most powerful results.
E. Artin (1898 – 1962)
There also is an important connection between the models described above and the
subject of complex variables. In the latter subject, one considers complex valued
functions that are defined in a region of the complex plane and defines the concept of
differentiability in complete analogy with the real case; specifically, given a function f, the
complex derivative at a point c is the limit of

f ( z ) − f (c )
z−c
as z approaches c, provided the limit exists. Functions which have complex derivatives
at all points are said to be complex analytic. Many results and examples involving
differentiable functions from ordinary calculus have analogs for complex analytic
functions; eventually two subjects become quite distinct, but a discussion of such
matters is beyond the scope of these notes. Our objective here is to state the following
important relationship between the Poincaré disk model and analytic function theory.
Theorem 1. Given the Poincaré disk model of the hyperbolic plane, let W be the
underlying set of points viewed as a region in the plane. Then a 1 – 1 correspondence

279
ϕ from W to itself is a hyperbolic isometry if and only if either (1) ϕ and its inverse
are complex analytic or (2) the complex conjugates of ϕ and its inverse are
complex analytic.
The proof of this result involves methods and results from the theory of complex
variables. A detailed treatment appears in Section VII.2 of the following textbook:
th rd
S. Lang, Complex Analysis (4 Ed., corrected 3 printing). Springer –
Verlag, New York, 2003. ISBN: 0–387–98592–1.

Theorem 1 reflects the importance of hyperbolic geometry in a branch of complex

analysis called the theory of Riemann surfaces. The latter play important roles in
many mathematical subjects. For example, the proof of Fermat’s Last Theorem, mainly
by A. Wiles (1953 – ), uses the hyperbolic geometrical structures that exist on most
Riemann surfaces (Note: This fact led to some misunderstanding of Wiles’ work in some
nontechnical books on the subject; these are summarized and accurately analyzed in
the online article http://en.wikipedia.org/wiki/Marilyn_vos_Savant. ).
There are also important connections between the theory of hyperbolic symmetries and
a subject called inverse geometry; the latter studies a class of transformations called
inversions, and some of these inversions correspond to reflections in the hyperbolic
plane or hyperbolic 3 – space. Further information on this subject can be found in the
previously cited book by Wallace and West (mainly Section 5.5), Chapters 24 and 25 of
the previously cited book by Moïse, and also in the following online references:
http://en.wikipedia.org/wiki/Inversive_geometry
http://www.maths.gla.ac.uk/~wws/cabripages/inversive/inversive0.html
http://www.maths.gla.ac.uk/~wws/cabripages/klein/models.html
Some basic facts about inverse geometry are also included in the exercises for this unit.

Regular tessellations

For some minutes Alice stood without speaking,

looking out in all directions over the country ... “I
declare it’s marked out just like a large chessboard
... all over the world — if this is the world at all.”
Lewis Carroll (C. L. Dodgson, 1832 – 1898),
Through the Looking Glass

Although a precise and comprehensive description of hyperbolic geometry’s place in

modern mathematics is beyond the scope of these notes, we shall describe one
geometric manifestation of its role. However, before doing so we shall summarize the
corresponding results for Euclidean and spherical geometry.
The planar case. A regular tessellation (or tiling) of the Euclidean plane is a
decomposition of the plane into closed regions (see Section I I I.7) bounded by regular
convex polygons such that the following hold:
1. All the bounding polygons have the same number of sides.
2. If the intersection of two distinct regions is nonempty, then it is a common
side or vertex of the bounding polygons.

280
There are three obvious ways to construct such regular tilings of the Euclidean plane. If
the regular polygons are squares, then one example corresponds to covering a flat
surface by square tiles that do not overlap each other, and if the regular polygons are
hexagons, then another example corresponds to the familiar honeycomb configuration of
hexagons. A third example this type is the covering of a flat surface by tiles that are
equilateral triangles. All of these are illustrated below.

Greek mathematicians (probably as early as the Pythagoreans) realized that the

preceding examples were the only ones.
The spherical case. On the surface of the sphere, the regular tessellations
correspond to regular polyhedra whose vertices lie on the sphere. More precisely, the
vertices of the regular tessellation for the sphere are just the vertices of the regular
polyhedron, its edges are the great circles joining two vertices (one for each pair of
vertices which lie on a common edge of the polyhedron), and its faces are the regions
bounded by the spherical polygons bounded by appropriate edges.
In view of the preceding discussion, the description of regular tessellations for the
sphere reduces to the description of the possible types of regular polyhedra. The two
simplest examples of regular polyhedra are a cube and a triangular pyramid such that
each face is an equilateral triangle. Each of these illustrates the fundamental properties
that all regular solids should have.
1. Every 2 – dimensional face should be a regular n – gon for
some fixed value of n ≥ 3.
2. Every 1 – dimensional edge should lie on exactly two faces.
3. Every vertex should lie on r distinct faces for some fixed
value of r ≥ 3.
4. No three vertices are collinear.
5. Given a face F, all vertices that are not on F lie on the same
side of the plane containing F.
Regular polygons beyond triangles and squares were known in prehistoric times, and in
fact archaeologists have also discovered early examples of stones carved and marked
to represent several (in fact, most and maybe all) 3 – dimensional regular polygons.
One major achievement of Greek mathematics (which appears at the end of Euclid’s
Elements) was the proof that there are exactly five distinct types of regular
polyhedra, and they are illustrated below:

281
 (Source: http://www.goldenmeangauge.co.uk/platonic.htm )

Examples in the hyperbolic plane

The situation in the hyperbolic plane is entirely different. One important reason is given
by the following result, which is also mentioned on page 176 of Ryan.

Theorem 2. Let n be an integer greater than 2, and let θ be a positive number less
than 180(n – 2)/n. Then there is a regular hyperbolic n – gon such that all the sides
have equal length and the measures of all the vertex angles are equal to θ.
In particular, if n is greater than 4 and θ = 90, then one might expect that we can form
a regular tessellation of the hyperbolic plane with regular n – gons such that four meet at
each vertex. In fact, this is possible. This is a special case of the following general
result:

Theorem 3. Suppose that m, n ≥ 3 are integers such that

1 1 1
+ < .
m n 2
Then there is a regular tessellation of the hyperbolic plane into solid regular n – gons
with m distinct polygons meeting at each vertex. Conversely, if there is a regular
tessellation of the hyperbolic plane into solid regular n – gons with m distinct polygons
meeting at each vertex, then the displayed inequality holds.

282
There are several ways to prove this theorem. In particular, algebraic results of W. F.
von Dyck (1856 – 1934) give an approach which is related to the viewpoint of Ryan’s
book.
Since there are infinitely many pairs of positive integers m and n satisfying these
conditions, it follows that there are infinitely many distinct regular tessellations of
the hyperbolic plane. The type of such a tessellation is generally denoted by the
ordered pair (n, m); note that the first coordinate gives the number of sides. We
shall give a few illustrations below; they are taken from the following sources:
http://aleph0.clarku.edu/~djoyce/poincare/poincare.html
http://www.hadron.org/~hatch/HyperbolicTesselations/
http://www.google.com/search?hl=en&q=hyperbolic+tessellations&btnG=Google
+Search
http://www.btinternet.com/~connectionsinspace/Patterns_and_Space_Filling/Hyp
erbolic_Geometry/body_hyperbolic_geometry.html
http://www.d.umn.edu/~ddunham/dunham04.pdf#search=%22hyperbolic%20tess
ellations%22
http://en.wikipedia.org/wiki/Tilings_of_regular_polygons

(5, 4)

283
(3, 12)

(4, 8)

284
 (4, 6)
Regular tessellations of the hyperbolic plane also appear in some of the artwork created
by M. Escher (1898 – 1972). For example, the angels and devils in the picture Circle
Limit IV fit together to form a tessellation by regular hexagons with right angles at
every vertex (type (6, 4) in our notation). This can be seen from the illustrations below:

(Source: http://www.d.umn.edu/~ddunham/mam/essay1.html )

285
 (Source: http://www.allposters.com/-sp/Circle-Limit-IV-Posters_i96944_.htm )

The following book contains further information on the interaction between art and
progress in geometry during the past two centuries:
L. D. Henderson, The Fourth Dimension and Non – Euclidean Geometry in
Modern Art. Princeton Univ. Press, Princeton, 1983. ISBN: 0–691–04008–7.
A recent summary of this book (which is currently out of print) is available at the
following online site:
http://www.space.com/scienceastronomy/generalscience/astrobizarre_fourD_010619.html

Final remarks

There is an interesting relationship between Theorem 3 and the results for regular
tessellations of the sphere and the Euclidean plane. In the Euclidean plane, there is a
regular tessellation into solid regular n – gons with m distinct polygons meeting at each
vertex if and only if

1 1 1
+ = .
m n 2
because this equation holds if and only if (n, m) is equal to one of the three ordered
pairs (3, 6), (4, 4) or (6, 3). Similarly, on the sphere there is a regular tessellation of
the hyperbolic plane into solid regular spherical n – gons with m distinct polygons
meeting at each vertex if and only if

286
 1 1 1
+ >
m n 2
because this equation holds if and only if (n, m) is equal to one of the five ordered pairs
(3, 3), (3, 4), (3, 3), (3, 5) or (5, 3). If we combine these observations with Theorem
3, we obtain the following unified conclusion:

For each ordered pair (n, m) such that m, n ≥ 3, there is a

regular tessellation of either the Euclidean plane, the hyperbolic
plane or the sphere into solid regular n – gons with m distinct
polygons meeting at each vertex, and the specific type of plane
supporting such a configuration is given by comparing ½ to the
previously described sum of reciprocals:

1 1
+
m n
In particular, the relevant geometry will be spherical if this sum is
greater than ½, it will be Euclidean if this sum is equal to ½,
and it will be hyperbolic if this sum is less than ½.

Honeycomb patterns in higher dimensions. asdfasf

V .8 : Summarizing the impact of non – Euclidean geometry

The last few sections of Unit V discuss the impact of non – Euclidean geometry on
mathematics and related subjects from several different viewpoints. The following list
summarizes the discussion:
1. In contrast to the laws of physics and chemistry, the laws for whole number
arithmetic seem to be logically inevitable. However, the development of non –
Euclidean geometry during the late 18th and most of the 19th centuries showed
very conclusively that the classical setting for (Euclidean) geometry was not
equally logically inevitable.

2. In many respects, the classical non – Euclidean geometry — which is often

called hyperbolic geometry — behaves like a spherical geometry in which the
radius of the sphere is an imaginary number.

3. Non – Euclidean was one of several important factors leading to the creation of a
more rigorous logical foundation for mathematics during the late 19th and early
20th centuries. Mathematical models for both Euclidean and non – Euclidean
geometry can be constructed within these frameworks for mathematics.

4. The development of non – Euclidean geometry eventually led to many alternative

mathematical theories for describing the physical universe, including the currently

287
 standard models of Relativity Theory. There are some relationships between the
latter and non – Euclidean geometry, but neither can really be described part of
the other.
5. For many (maybe most) practical purposes, Euclidean geometry is an extremely
good first order approximation for studying small regions in any reasonable
theory of space.

Non – Euclidean geometry has turned out to be more than just a logical curiosity, and
many of its basic features continue to play important roles in several branches of
mathematics and its applications. Two simply stated examples involve the proof of
Fermat’s Last Theorem and several of Maurits C. Escher’s artistic creations.
Finally, it seems appropriate to end with the following quotation from page 105 of
Greenberg:
Let us not forget that no serious work toward constructing new axioms
for Euclidean geometry had been done until the discovery of non –
Euclidean geometry shocked mathematicians into reexamining the
fourndations of the former. We have the paradox of non – Euclidean
geometry helping us to better understand Euclidean geometry!

288