You are on page 1of 26

Copyrighted material

Prinld inmeUnited Sw.e::; of America

12 u1009 OS 109, 8 7 6 5 4 3 2 I
l"ibrary or Congress Cataloging-to-MUcatlOcD Data
Hitehman, Michael P.
GeometrY with !lj)introducrion to cosmic topology I M.icltael P. Hitehma.n.
ISBN 918-0-7637-5457-0 (casebaund/herd cover) 1.. OW1l1etry.2. Cosmic magnetic fields. 3. Topology. 1. Title.
QAM5.H48 2008
PrtKlocnon C-eits
Aequisitioas EdilOr: Tinl0lby Anderson
Editorial Assistant: M,eL issaP'(}OOt
Production. Director: Am.y Rose
Associate Production Editor: Melissa Elmore
Senior Marketing ~1allnger: .AndreaDefronzo
VP., Manufacturing Il.ndInvenlory Cuntrol: Therese Connell
Composition: NorrhetL St Q:)nrpo. jtOf:S~lAc.
Cover Design: Brian Moore
Cover hllage: ESA,(Image by C. Cerreaa)
Printing and Binding: MAlloy~Inc.
COVe'f Printing: Mulloy. Inc.
AU rights reserved. No part' of we m.'llcritd prQtectcd by this c.opytigbl rnay be reproduced or utilized in any form, elec-
tronic or mechanical, including phOtooop,ying," at by any infOtrrultion storage and retTieval system. without
written permission from the ropyri,ght owner,
Copyright @2009 by J ones and Bartlett Publisners, .L L C
Substantial discounts on bulk quantities of J ones and Bartlett' publications are available to corporations. profes-
sional associations, and other qualified org:Wlizations. For details and ~cifie diSCOUtll tnforlt1atioo. contacr the
speCial sales depzu1ment at J ones and BartL el via me above eoatact information or send an email to special.
J ones and Bartrett's hooks and products are available through mosr bookstores and online booksellers. To eonracs J oaes
and i3MtJ eu, Publishers directly, callSOOS32-0034. fax 978-443~8OC)(), or visit [our Wt..hsitcwww.jbpul).c-om.
J ones and Bartlett Publishers
Barb Honse, Barb Mews
L ondon W6 7PA
United Kingdom
J ones and Bartlett Publishers
'6339 Ormindale 'Vay
~fissjsaaQga.Ontario LSV 1J 2
J one end Bartlett Publishen;
40Tall Pille Drive
Sudbury, MA 01776

. . . . . . . . . . . . . . . . . ' . .
3. 5 Mobius Transformations: A Closer Look .

. . . . or . . . . . . . . . .

3. 4 Mobius Transformations .
51 3. 3 The Extended Plane
. . . . . . . . . . . . . . . .

. . . . . . . . . .

3. 2 Inversion. . . . . . . .

3. 1 Basic 'Iransformations of C
29 3 Transformations
24 . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. 4 Complex Expressions .
21 . . . . . . . . . . . . . . . . . . . . . . . 2. 3 Division of COIn lex Numbers . . . . . . . . . . . . .
18 . . . . . . . . . . . . . . . . . . . . . . . . 2. 2 Polar FOTln of a Complex Number .

15 2 TIlePlane and COlnplex Numbers
8 . . . ' . .
. . . . . . . . . . . . . . . . 1. 3 Geometr on Surfaces: A First Look .
3 . . . . . . . . . . . . . . . . . . . . . . 1. 2 A Brief History of Goometrv
1. 1 Introductjon. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1 All Illvitation to Geometl'Y 1
COPYrighted material
. . . . . . . . .. 192 t .. I! ... of
~ .' " l ~ ~ .. ;0 4 186 ..... ~.. " 7.6 Oeornctry of Surfaces , . .
7.7 Quociont Spaces ,.....
. . . .. 170
".. .. . 7.4 Ob: erving Curvature in a Universe (optional)
7.5 Surfa ',.\ . .
o," o,....
. 157 7.2 Elliptic Geometry with Curvatur J ... >0 . . . . .
7.3 Hyperbolic Geometry with Curvature k : <0 ..
. . . " 153 .. ,.. .. .. .
' . .. ... .. .. Curvature . . . . . . 7.1
~ .. .. iI .. .. .. .. .. ..., 150 ... .. .. ...' "

.. .. .. .. .. .. .. . . . . .. 133
. . . . .. 141 ., .. ..
... ..." Oi
6.2 Elliptic Ceometry . . . .. . ...
6.3 Measurement iII Elliptic G-eometry .
6.4 Revisitiug Euclid's Pcstuletee . . . .
7 Geometry on Surfaces

.._ .... .!'

. ...... .....

. .. 107 .. .... < to ..

........ _ -. ... 6- . ~ 5.3 Measurement irl Hyperbolic Geometry
5.4 Area and Triangle Trigonometry . . .
5.5 The Upper Half-Plane Model (optional) .
6 Elliptic Geometry
6.1 Antipodal Points
.. .. .. .. .. .. .. ..
.. ........ ..
5 Hyperbolic Geom.etry
5.1 T,ll,CHyperb ;,licTransforrnation Group . . . .. .
5.2 Figuros of Hyperbolic Geometry ,... ..
.. . ~........ " .... ...... ..
.. .. ... "" .. .. ..... ' .
4 Geometry
4.1 Tllc Basics .... . .
4.2 Mobius Geometry . . . .. .
xii Contents
COPYrightedrna rial
Bibliography 233
I ndex 235
8 Cosmic Topology 203
8.1 Three-Dimensional Geometry and 3-n1aI1ifolds.
8.2 Cosmic Crystallography
8.3 Circles inthe Sky.
8 . 4 Final Remarks
Contents xiii
The infinite plane model of the two-dimensional universe works well enough for most pur-
poses, but cosmologists and mathematicians, whonotice that everything within the universe
is finite, consider the possibility that the universe itself is finite. Would a finite UIliVCf!)C
have a boundary? Can it have an edge, a. point beyond which one cannot travel? This
possibility is unappealing because a boundary point would be phy ically different fromthe
rest of space. But how can afinite universe have no boundary?
In a stroke as bold as it was simple, a two-dimensional mathematician suggested that the
universe might look likea rectangular region with opposite edges identified.
Consider a fiat, two-dimensional rectangle. Infact, visualize a perfectly rectangular flat-
screen computer monitor. Nowimagine that youareplaying the video gameshowninFigure
_ .
_. c

Imagine you are atwo-dimensional being living in atwo-dimensional universe. Mathemati-
cians in this universe often represent its shape as a.ninfinite plane, exactly like the x-y
plane you've used as the canvas for your calculus courses.
Your two-dimensional self has been taught in geometry that triangles have angles that
add up to 1800You may have even constructed some triangles to check. Builders use the
Pythagorean theorem to check whether two walls meet at right angles, and houses are
1.1 Introdudion
A n I n v i t a t i o n t o G e o m e t r y
COPYrightedrna rial
Of course your two-dimensional self would not be able to see this torus surface in 3-space,
but you could understand the space perfectly well in its rectangle-with-edges-identified
form. It is clearly a finite universe without any edge.
A sphere, like the surface of a beach ball, is another finite two-dimensional surface without
an)' edge. Consideration of a finite-area universe leads to questions about the type of geom-
etry that applies to the universe. Consider thesphere. On asmall scale, Euclidean geometry
Figure 1.2 Our videoscreen:isequivalent to a torus.

Thus, the left edge of the rectangle has been identified, point by point, with the right edge.
Inthree dimensions one can physically achieve this identification, or gluing of the edges.
In particular, one can bend the rectangle to produce a. cylinder being careful to glueonly
the left and right edges together, and not to glue any other points together. The top and
bottom edges of the rectangle have now become the top and bottom circles of the cylinder,
which themselves get identified, point by point. Bend the cylinder to achieve this second
gluing, and one obtains a donut, also called. atorus.
Figure 1.1 A finite two-dimensional world with no boundary.
1.1. As you shoot the boulders and move your ship arouncl the screen you find that if you
go off the top of the screen your ship reappears on the bottom' and ifyou go off the screen
to the left you reappear on the right. InFigure 1.1there are just five boulders. One has
partially moved off the top of the screen and reappeared below, while a second is halfway
off the right-hand "edge" and is reappearing to the left.
2 CHAPTER 1 An IJ lvitation to Geometry
COPYrighted material
Geometry is one of the oldest branches of mathematics, and most important among texts
is Euclid's Elements. The Elements begins with 23 definitions, 5 postulates, and 5 common
notions. From there Euclid starts proving results about geometry using a rigorous logical
method, and many of us have been asked to do the same in high school.
The Elements served as the text on geometry for over 2000years and ithas been admired as
a brilliant work in logical reasoning. But one of Euclid's fivepostulates was also the center
of ahot debate. It was this debate that ultimately led to the non-Euclidean geometries that
can be applied to different surfaces.
We won't explain this equation here, but wewill point out that geometry is on the left side
of the equation, and topology is on the right. So if a two-dimensional being can deduce
what sort of global geometry holds in her world, she can greatly reduce the possible shapes
for her universe. OUf immediate task in the text is to study the other, non-Euclidean types
of geometry that may apply on surfaces.
kA =2r.X.
Infact, there is a wonderful relarlonship between the topology (shape) of a surface and the
type of geometry that it inherits. A primary goal of this book is to arrive at this relationship,
given by the pristine Gauss-Bonnet equation
_ . . . - - - - - - - - - ,~. - _ -
works \ \ 1011; small triangles have an angle SlJID essentially equal to 1800I which is a defining
feature of Euclidean geometry. But on a larger scale) things go awry. A very large triangle
drawn on the surface of the sphere 11M an angle sumfar exceeding 1800Note here that a
triangle consists of three points and three segments drawn to represent the shortest path
between the points (we'll discuss this more carefully later). Consider the triangle formed
by the north pole and two points on the equator. The angle at each point on the equator
is 900l so the total angle sum exceeds l80 by the amount of the angle at the north pole.
\ "Aleconclude that a non-Euclidean geometry applies to the sphere on aglobal scale.
1.2 A Brief History of Geometry 3
Copyriphted matenal
The parallel postulate debate carne to a head in the early 19t1l century. Farkas Bolyai
(1775-1856) of HIIOgt\l1..spent much of his lifeon the problem of trying to provetile parallel
postulate fromthe other four, He failed, and he fretted when his son J anos (1802-1860)
Reformulation Si of the parallel postulate is called Playfair's Axiom after J ohn Playfair
(1748--1819). This version of tIle fifth postulate will bethe one wealter inorder to produce
nOD-Euclidean geometry.
.Intryingto make senseof theparallel postulate) many equivalent statements emerged. Tbe
t\VO equivalent statements most relevant to ow' study are these;
5'. Given a line and a point not on. the line, there is exactly
one line through the point t11~lt does not intersect tile given line.
5 / 1, The sum of the angles of an}' triangle is 180.
Indeed, the parallel postulate immediately gavephilosophers and other thinkers .fits, and
many tried to prove that the fifth postulate followed from the first four}to no avail. Euclid
himself may have been bothered at some level b)' the parallel poo~ul&,tesince he avoids
using it unti] the proof .of the 29th proposition in The 111ement,s.
Does one postulate not look like the others? The first four postulates are short, simple,
and intuitive. Well, the second might seem a bit odd, but all Euclid is saying here is that
}~OUcan produce a line segment to any length you want, However, the fifth one called the
parallel postulate; is Iong-winded, unclear, and sounds more BItesomethingyou would
try to prove than something you would take as a given,
1. One can draw astraight line froman,y point to any point.
2. One can produce a finite straight linecontinuously ina straight line.
3 " One can describe a circle with any center and radius
4. All right angles equal one another.
5. If astraight line falling Oll two straight lines makes the interior angles
011 the same aide less than ewe right angles, the two straight lines; if
produced indefinitely) meet on that side 011 which the angles are less
than two right angles.
Here areEuclid's fivepostulates!
4 CHAPTER 1 An Invita.tiou to G-eometry
Copynghted material
Inelliptic gOO}J :lctry, the SUtl1of the angles of any tria;ngleisgre8.ter tban 1800,afact proven
inChapter 6.
TIlearrival of non-Euclidean geometry soon caused astir incircles outside themathematics
eommunity, Fyodor Dostoevsky thought non-Euelidean geometry was interesting enough
to include in The Brothers Ksrasnaso, first published in 1880. Early in the novel two of
the brothers, 1"8.11axld Alyoaha~get reacquainted at atavern..Ivan disccurages hisyounger
lSee, for Inssaaoe, lvlartin Gardner's The Colcssal Book e J Matn.etrto:tics (W.'~.Norton &; COlnpallY,
2001), 176.
5E. Givenaline and a point not on the line, there are no lines through the
POUlttllat do not intersect the given line.
In hyperbolic geometry, the sumof the angles of any triangle isless than 1800, afact proven
inChapter 5.
The second type of non-Euclidean geometry in this text is called.elliptic geometry. which
models the goometr)7 of the sphere. In tilis geometry, EtlClidls fifth postulate is replaced IlY
5H. Givena lineand a point not on the line, there are m,any lines through
the point that do not intersect the givenline.
But. J anos continued to work on the problem, as did a Russian matheIl:latician, Nikolai
Lobachevsky (1792-1856). They independently discovered that a well-defined geometry is
possible in which the first four postulates hold but the fifth doesn't. In p.a;rtic,war, they
demonetreted that the ,fifthpostulate is not a necessary conse que nce of tile first four.
In this text, wewill study two types of non-Euclidean geometry. 'The first type is called
hyperbolic geometry, and is the geometry that. Bolyai and Lobachevsky disecvered. (The
great Carl Friedrich Gauss (1777-1855) had also discovered this geometry; however1 he
didn't publish his work because he feared it would be too controversial for the eetabliah-
ment.) In hyperbolic geometry, Euclid's fifth postulate is replaced l)y this:
For God's sake, I beseech you, giveit up. Fear it no less than the sensual passions
because it 'too may take all your tinl and deprive you of your health, peace of
mind and happiness in life.!
started followingthe same tormented path. In an oft-quoted letter, the father beggedthe
son to end the obsession:
1.2 A Brief History of Geometry 5
Copyriphted matenal
2&13, fur instance, Fyodcr Do.sto0vsl{j', The Brotllers. Kammazov, (6. newtranslation by Richard Pevear
and Larissa 'Vowkhousky) (NorchPoint Press, 1990), 2.3.5,
;}Relativity: The Special and Ge1u:ral Theory (Crown Publications Inc., 1961). 14.
brother from thinking about whether God exists, arguing that ifone cannot fathom non-
Euclid.ean geometry, then one has no hope of uaderstandlng questicns about God.2
One of the first challenges of non-Euclidean geometry was to determine its logical consis-
tency. By changing Euclid's parallel postulate, was a. system created that led to contra-
dietary theorems? III 1868the Italian mathernaticiaJ l Enrico Beltrami (183&-1900) Sl10WL~
that the new non-Euclidean geometry could be constructed within the Euclidean plane
so t}mt.)as long as Euclidean geometry was consistent, non-Euclidean geometry would be
consistent as well. Non-Euclideen geometry W8.1lJ thns placed on solid ground.
This text does not develop geOflletl')! as Euclid and Bolyai and Lobaehevsky did. Instead, it
approaches the subject t'4'S theGerman mathetnati(;ian FelixKlein (1849-1925) did, \Vhereas
Euclid's approach to geometry was additive (he started with basic definitions and. axioms
and proceeded to build a. sequence of resules depending onprevious ones}, Klein's approach
was subtractive. Hestarted with a space and agroup of allowable trmworma,tions of tllat
space. Hethen threw out all concepts th~t did not remain unchanged under these transfor-
mations. Geometry, to Klein, is the study of objects and fnnctkms that remain unchanged
under allowable transforntations.
Klein's approach to geometry, called the Erlangen Program after the university at. which
beworked a,~the tiD1:e~has Iiliebenefit that all threegeo!rletries (ElIClideaIl, hyperbolic, and
elliptic) emerge as special cases froma general space and a general set of transformations.
The next three chapters will be devoted to IDa-king sense of, and working through, the
preceding two paragraphs,
Like so much of ma.thelua;tic!s, the development of non-Euclidean geometry anticipated
applications. Albert Einstein s theory of special relativity illustrates the power of Klein's
appro.acll to geometry. Speeial relativity, says Ein teln, is derived fromthe notion tha,t the
laws of nature are invariant with respect to Lorentz t.ransformations.3
Even witll non-Euclldeen geometry in band, Euclidean geometry remains central to mod-
ern mathematics because itis an excellent model for our local geometry. TIle angles of a
triangle drawn on this paper do add up to 180. Even 'galactic" triangles determined by
the positions of three nearby stars 118>Vean angle sum indistinguisha,ble from. 1800.
However) on a larger scale) things might be different.
Maybe weliveinauniverse that looksfiat (Le., Euclidean) on smallish scaleabut is curved
globally. This is not so hard to; believe. A bug living on a particularly fiat patch on the
6 CHAPTER 1 An. Invitationto G~om,etry
COPYrightedrna rial
F i g u r e 1.3 Two consequences of the parallel postulate.
.A C
(a) (b)
1.2.2 Use the parallel postulate and the previous problem to prove that the sum of the angles of
any triangle is 1800Youmay findFigure 1.3(b) helpful, where segment CDis parallel tosegment
1.2.1 Usethe parallel postulate to provethe alternate interior angles theorem. That is inFigure 1.3
(a), assume the linethrough BD isparallel to the linethrough AG. Prove that LBAC =LABD.
surface of Earth might reasonably conclude that he is living on an infinite plane. The
bug cannot sense the fact that his fiat, visible world is just a small patch of a curved
surface (Earth) livingin three-dimensional space. Likewise,our apparently Euclidean three-
dimensional universemight becurving in someunseen fourth dimension, meaning that the
global geometry of the universemight benon-Euclidean.
Under reasonableassumptions about space; hyperbolic, elliptic, andEuclidean geometry are
theonlythree possibilities fortheglobal geometry of our universe. Researcherscurrently are
gathering data in hopes of decidingwhichgeometry is ours. Deducing the geometry of the
universecan tell us much about the shape of the universeand perhaps whether it isfinite.
Ifthe universe is elliptic, then it must befinite in volume, If it is Euclidean or hyperbolic,
then it could be either finite or infinite. Moreover, each geometry type corresponds to a
class of possible shapes. And, ifthat isn't exciting enough, the overall geometry of the
universe is fundamentally connected to the fate of the universe. Clearly there is no more
grand application of geometry than to the fate of the universe.
1.2 A Brief History of Geometry 7
Copynghtedma nal
On a sphere, geodesics follow great circles.A great circle is a circle drawn on the surface
of the sphere whose center corresponds to the center of the sphere. Put another way, agreat
circle is a circle of maximum diameter drawn on the sphere. As shown in the accompanying
It is an important matter to decide what we mean, exactly, by a triangle on a surface. A
triangle consists of three points and three edge.')connecting these points. An edge connecting
point A to point B is drawn to represent the path of shortest distance between A and B.
SUCll a path is called a geodesic; that is, for the two-dimensional bug, a "straight line
from A to B is simply the shortest path from A to B.
F i g u r e 1.4 Thetorussurface isnothomogeneous.
The surface of a donut inthree-dimensional space (see Figure 1.4) is not homogeneous, and
a two-dimensional bug living on the surface of the donut could tell the difference between
various points. One approach to discovering differences in geometry involves triangles.
If Y01I draw atriangle on this page, the angles of thetriangle will add up to 1800. In fact, any
triangle drawn anywher'e on the page has this property. Euclidean geometry on this fiat page
(aportion of the plane) ishomogeneous: the local geometry of the plane isthe same at all
points. Our three-dimensional space appeal's to be homogeneous as well. This is nice, for it
means that if we buy a 5 ft3freezer at the appliance store, it will not shrink to 0.5 ft3when
weget it home. A sphere is another example of a homogeneous surface. A two-dimensional
bug living on the surface of asphere could not tell the difference (geometrically) between.
any t\VOpoints.
Think for a minute about the space we live in. Think about objects that live in ow' space.
Do the features of objects change when the)' move around inour space? If I pick up this
paper and move it across the rOOID, will it shrink? Will it become a broom?
1 . 3 G e o m e t r y o n S u r f a c e s : A F i r s t L o o k
8 CHAPTER 1 An Invitation to Geometry
COPYrighted material
Getting back to the donut, a two-dimensional bug could use triangles to tell the difference
between a "convex" point on an outer wall and it "saddle-shaped point on an inner wall
(seeFigure 1.4). A bug could draw atriangle around the convexpoint, determine the angle
Slim, and then move around the surface to a saddle-shaped point and determine the angle
sumof anew triangle (WllOS legs are the same length as the first triangle). The bug would
scratch his head at the different anglesums beforerealizing he'd stumbled onsomething big.
He'd go home and write up his results, emphasizing the fact that a triangle in the convex
region will have an angle S11m greater than 1800, while a triangle inthe saddle-shaped
region will have an angle sum less than 1800. This happy bug will conclude that his donut
surface isnot homogeneous. Hewill then sit 'back and watch the accolades pour in. Perhaps
even aNobel Prize. Thus, small triangles and their angles can help a two-dimensional bug
distinguish points on the surface on which it lives.
The donut surface is not homogeneous, so let's build one that is.
Figure 1.5 A triangle on a surface is formed by connecting three points with geodesics (paths of
shortest distance). Depending on the shape triangles can havean anglesumgreater than 1800:less
than 1800, or exactly equal to1800
!\ f \
On the Euclidean plane, geodesics arc Euclidean lines.
One way to physically determine ageodesic OD asurface is to pin somestring at A and B
and then draw the string tight on the surface. TIle taut string will followthe geodesic. In
Figure 1.5wehave drawn geodesic triangles on three different surfaces,

__ .. .J __ ...... _.
_ . . - . -.

illustration, the equator is a great circle (seecircle a), as are all the longitudes (likecircle
b), but the other latitudes (likecircle c) are not great circles.
1.3 Geometry on Surfaces:A First Look 9
1.3.2 Saddleland. Repeat the previous exercise but with circle wedges having 0 >27T '. Identifying
the radial edges in this case produces a saddle-shaped surface.
b. Draw tile segments connecting the three points. You should get a triangle with the tip of tile
cone in its interior. (T his triangle should actually look like a triangle if you re-form the cone.) If
you don't get the tip of the cone on the inside of the triangle, adjust the points accordingly.
c. With your protractor, carefully measure the angle 0 subtended bythe circular sector. T oemphasize
O 's role inthe shape of the cone, we lct S (O ) denote the cone surface determined by O .
d. With your protractor, carefully measure the three angles of your triangle. T he angle at point C
is the SIUD of the angles formed by the triangle legs and the radial egments. Let A denote the sum
of these three angles.
e. State a conjecture about the relationship between the angle 0 and A, the Slimof the angles of tile
triangle. Your conjecture can be in the formof an equation. T hen prove your conjecture. Hint: if
you draw a segment connecting the two copies of point C, what is the angle sumof the quadrilateral
1.3.1 Coneland. Here webuild cones fromfiat wedge , and measure the angles of some triangles.
a. Begin with a circular disk with a wedge removed, likea
pizza missing a slice or two, J oining the two radial edge
produces 1\ cone. Try it with a. cone of your O"'Ll to make
sure it works. Now. with tbe cone fiat again, pick three
points, labeled A, B, and C, such that C is on the radial
edge. T his means that in this flattenedversionof thecone)
point C actually appears twice: once on each radial edge.
T hese two representatives for C should get identified when
you match the radial edges.
Example1.3.1 T heFlatT orus:Ahomogeneous surface
Con ider again the world of Figure 1.1. T his world is called a flat torus. At eveIJ ' spot ill
this world, the pilot of the ship would report flat urroundings (triangle angles adding up
to 10). Unlike the donut surface living in three dimensions, the flat torus is homogeneous.
Locally, geometry is the same at every point, and thanks to a triangle check, this geometry
is Euclidean. But the world as a whole is much different than the Euclidean plane. For
instance, if the pilot of the ship has a powerful enough telescope, he'd be able to see the
back of his ship. Of course, if the ship had windows just so, he'd be able to see the bade of
his head. T he flat torus isafinite, Euclidean, two-dimensional worldwithout any boundary.
10 CHAPT ER 1 AnInvitation to Geometry
COPYrightd rna rial
HoweverI the angle of each corner is 1200, and gluing two of these COrnerstogether will
create acone point. This will be the case for the other two pairs of corner points as well.
As inExercise 1.3.1) apilot can distinguish acorner point froman interior point here. She
Suppose the pilot of a ship wants to By around one of the corners of the hexagon. If she
begins flying around the upper right corner labeled 1in Figure 1.6, she would fly off the
screen to the right and reappear just above the lower left corner labeled 1. Continuing
around she would complete her journey after circling this comer. Similarly, she would find
that the other corners meet in groups of two.
Ex ampl e 1 . 3 . 2 A no n- Eudi de an ho mo g e ne o us s urf ac e
Consider the surface inFigure 1.6,obtained by identifying theedges of the hexagon as indi-
cated. Inparticular) the edges arematched according to their labels and arrow orientation.
Therefore, if a ship fliesoff the hexagonal screen at a spot on the lower edge marked 'a',
sa)' then it will reappear at the matching spot on the other edge marked 'a'.
F i g u r e 1.6 A hexagonal video screen.
Remember, ahomogeneous surface is aspace that has the same local geometry at every
point. Our fiat torus is homogeneous, having Euclidean geometry at every point. However,
our cones S(O) in the previous exercises are not homogeneous (unless 0 happens to be 27r).
Ifa triangle in 8(8) does not contain the tip of the cone in its interior, then the angles
of the triangle will add to 7 T ' radians; ifthe triangle does contain the tip of the cone in
its interior, then the angle sum will not be 1r radians. A two-dimensional bug, then, could
conclude that S(8) is not homogeneous.
1.3 Geometry on Surfaces: A First Look 11
Copynghtedma nal
It turns out that e'uerlj surface can be given 011eof three t)'PCS of homogeneous geometry:
Euclidean, hyperbolic, or elliptic. \\fO will return to the geometry of surfaces (and of our
universe) after wedevelop hyperbolic and elliptic geometry. Ifit doesn't make a ,vholelot of
sen eright now, don't sweat it, but pleaseusc these facts as motivation for learning about
these non-Euclidean geometries.
F i g u r e 1.7 A surface with homogeneous elliptic geometry.
,.,.,..- - ..~..- - . - -~.._ ..- ~
.--~. - _
But how can weincrease the corner angles? By putting the hexagon on the sphere. Imagine
stretching the hexagon onto the northern hemisphere of a sphere (sec Figure 1.7). In this
case we can think of the six points of our hexagon as lying on the equator. Then each
corner angle is 1800, and when weglue the edges together, each pair of corner angles adds
IIp to exactly 360; therefore, the surface is homogeneous, The homogeneous geometry of
this surface is the geometry of the sphere (elliptic geometry), not the geometry of the plane
(Euclidean geometry).
So the surface is not homogeneous if it is drawn in: the plane. However, the surface does
admit a homogeneous geometry. 'Ve can get rid of the cone points if we can increase each
corner angle to 1800By doing SOl two corners would come together to form a perfect 3600
patch around the point.
call look at triangles: a triangle containing one of the cone points will have an angle sum
less than 1800; any other triangle will have an angle sum equal to 1800
12 CHAPTER 1 An Invitation to Geometry
Copyrighted material
1.S.S C.ircumfer'ence VS,RadilJ.8. IIIaddition to triangles, ill ewe-dimensional bug can use circles
to screen for 'difereut geometries. Inparticular 1a. bug can study the relationship between the radius
andthe circumference of acircle. To make sure we are thinking like the bug, here is how we define
acircleon asurface: Gi\'en apolnt P on the surfaceandareal number r >0, the circleeentered
at P with radius r is the set of all points are 11distance of r awa)' from P, where the distance
between P and apoint is 'the length of the shortest [lath connecting them (tl10geodesic),
a. Pick yow: favorite circle inthe plane. '\.Vllat is the relationship between the circle's radius andits
circumference? Isyour answer true for arty circle inthe plane?
b. Consider the Coneland surface of E..xereise1.3.1, Construct acircle centered at thetip of the cone
and derive at relaeionahip between its circumference and its radius, Is ,C== 21fThere? Ifnot which
is true: C>21i1' or C <21ir?
c. Oonsider the Saddleland surface of Exercise 1.3.2.,Construct a, circle centered at the tip of the
saddle and derive a. relationship between its circumference andits radius. Is 0= 21fT' here? If not,
which is true: C >21f'ror C <21r1 7
1.3 Geometry onSW"faees:AFirst Look 13
k (a, b) =(ka, kb).
and ifk is a real number, wedefine scalar multiplication by
(a b) +(c, d) =(a +c, b +d),
Addition in C is componentwise,
Given the complex number z =(a, b), a is called the real part of z, denoted Re(z), and b
is called the imaginary part of z denoted Im(z). Tile set of real numbers is a subset of
C under the identificatiou a +-+ (a,O) for any real number a.
c={C a , b) I a, b are real numbers].
Algebraically, the set of complex numbers is obtained by adjoining the number ito the set
of real numbers, where iisdefined by tile property that i2=-1. Wewill take ageometric
approach and define a complex number tobe an ordered pair (a, b) of real numbers, "'Te
let C denote the set of a ll complex numbers,
2.1 Basic Notions
Tostudy geometry usingKlein's Erlangen Program, weneed todefineaspaceand agroup
of transformations on the space. Our space will be the complex plane.
T h e P l a n e a n d C o m p l e x N u m b e r s
Copynghted material
E x a m p l e 2.1.1
Suppose z =3- 4i and w =2 +Ti.
Note that Izi givesthe Euclidean distance of z to the point (0,0).
One final bit of terminology: ITz =a +bi, the conjugate of z, denoted z ; is a - bi.
The modulus of z: denoted I zl isgiven by
(a +bi) . (c+di) =ac +bci +adi +bdi2
= (ae - bd) +(be +ad)i.
We define complexmultiplication using the fact that i2 =-1.
The expression (1. +in iscalled the Cartesian formof the complex number. This form can
be helpful when doing arithmetic of complex numbers, but it can also be a bit gangly. We
often let a single letter such as z or w represent a complex number. So z =a +bi me8J :1S
that the complexnumber we'recallingz corresponds to the point (a, b) inthe plane.
It issometimes helpful to viewacomplex number as avector, and complex addition cor-
responds to vector addition in the plane. The same holds for scalar multiplication. The
complex number z - w can be represented by the vector from w to z in the plane (see
Figaro 2.1).
( a b) =( a, O ) +(0, b)
=(a,O) +b( O , 1 )
= a +bi.
Within this framework i=(0 1), meaning that any complex number ( a, b) can beexpressed
as a +bi, as suggested here:
Figure2.1 Complex numbers as vectors inthe plane.
- - _ .-
_ .-
_ - -

z +l1>'
16 CHAPTER 2 The Plane and ComplexNumbers
Copyrighted material
d. z +Z =2Re(z). (Hence, z +z is a real number.]
e. z - z =2Im(z)i.
f. I z i =1 1 1
c. z+w =z+w.
2.1. 3 Suppose z ::;; a +b -i end 'tV =C +di are twocomplex numbers. Prove thefollowingproperties
of the conjugate end the modulus.
a. I w , z l =I wl' [z ],
. w . . c . 'W"
:2.1.2 Showthat z . z =tZ 1 2, where z is the conjugate of z,
2.1~1 In each case, determine z +fl }, SZ. [z],and z . ~w.
a. Z = 5+2i , s = -41W :=-1+2i
h. z =3i , 8= 1/2 w = -3 +2i .
c. z =1+i , 8=0.6, w =1- i.
4z =12-16i
I z i =-/32 +(_4)2 =5
z u: .=34- l3i.
A few other computations:
z - w =(3- 4i)(2 +7i)
=6+28- 8i +21i
Then z+w =5+3i;l and
:a,l Basic' Notions 11
Copynghtedma nal
(all-star equation)
e77r +1=O .
For instance, et1</2 =cos(Tr/2) +isin(rr/2) =0+'i. 1=i.
Similarly, eiO =cos(O ) +isin(O ) =1, and it's a quick check to see that ei1r =-I, which
leads to asimpleequation involving the most famous numbers in mathematics (besides 8),
truly an all-star equation:
Definition 2.2.1 For ally real number B 1 wedefine
eiO =cos(B) +i sin(D).
Using these relationships, we Call rewrite
a +V i =r cos(O ) +r sin(B)i
=r(cos(8) +'isiu(B)).
(0. b)
a =rcoo(8)
b =rsin(B).
A point (a, b) in the plane can be represented in polar form (r\ B ) according to the
relationshi ps
2 . 2 P o l a r F o r m o f a C 1 m p l e x N u m b e r
a. Prove that a2 +b2 =c2.
h. Find the complex number z =x +yi that generates the famous triple (3,4,5).
c. Find the complex number that generates the triple (5,12,13).
d. Find fiveother Pythagorean triples) generated using complex numbers of the form z =x +yi.
where x and 'y are positive integers with no common divisors.
c =ze.
2.1.4 Pythagorean triples. A Pythagorean triple consists of three integers (a, b, c) suchthat
a2 +b2 =c2. \\re can use complex numbers to generate Pythagorean triples. Suppose z =z +'Vi,
where z and y are positive integers. Let
18 CHAPTER 2 The Plane and Complex Numbers
COPYrighted material
o This completes tile proof.
(sum of angles formulas) =(1'8) cos(B +{3) +sin(B +fJ)i
=(r 8)ei(O+,B).
=(r s) cos()cos{3 - sin()sin.B +(cosB sin,8 +sin()cos,8)i
Proof. Weusc the definition of the complexexponential and sometrigonometric identities.
'reifJ . sei.{:J =.,'(cos8+i sin8) . s( cos{3 +'i sin(3)
=(r s) (cose +isinB) . (cos.B +i sin,8)
T h e o r e m 2.2.3 The product of two comple x numbe rs in polar form is give n by
reiO seif3 =(r s )ei(o+t3) .
To convert z = -3 +4i to polar form, refer to the right side of Figure 2.2. Note that
r =";9+16=5, and tan(a:) =4/3, so ()=7r - tan-1(4/3) ~2.214. Thus,
Figure2.2 The polar form of a complex number.
- 3
3 4
Ifz =a +In and (a, b ) has polar form (r, B) then z ='re iO. The non-negative scalar 11' 1 is
the modulus of z, and the angle 8 is called the argument of z, denoted arg(z}.
E x a m p l e 2.2.2 E x p l o r i n g t h e polar form
On the left side of Figure 2.2, weplot the points z =3ei1r/4
and u =4e -i1 T
2.2 Polar Form of a Complex Numbor 19
COPYrightd rna rial
2.2.4 If z =re'", prove that z =-:".
2.2.3 Modify the all-star equation to involve8. Inparticular, write an expression involving e, i, 71', 1,
and 8 that equals O.You may use no other numbers.
2.2.2 Express the followingpoints in Cartesian formand plot them: z =2eirr/3, w =_2eifr/4,
tt =4 e ~ 5 1 1 " /3, and z ' 1.
2.2.1 COJ IVert the following points to polar form and plot them: 3+i -1 - 2i, 3- 4 i, 7.002,001 ,
and -4i.
Thus, by adding 11' to the angle if necessary, wemay always assume that z =re iO 1 where r
is non-negative.
(since -1=ei1 r)
(Theorem 2.2.3)
re if J =-lrle iO
=(eirr) . Irle iO
=Irle ' i(O+1 r).
wheretheequation istaken modulo 27r. That is, depending onour choicesfor thearguments,
wemay have arg(vw) =arg(v) +arg(w) +k27r for some integer k.
E x a m p l e 2.2.4
When representing a complex number z in polar formas z =re i(J , wemay assume that r
is non-negative. Ifr <0, then
arg(vw} =arg(v) +arg(w),
Thus, the product of two complex numbers is obtained by multiplying their magnitudes
and adding their arguments, and
20 CHAPTER 2 The Plane and Complex Numbers
COPYrighted material
where equality is taken modulo 27r.
=arg(z) - arg(w),
That is,
1'ei8 r
_ i(O-{J)
- se- - - : i" '{ 3 - se .
to obtain
z 1
- =z,-
w w
If z =rei8, apply Theorem 2.2.3 to the quotient
1 1-i{J
- -= -e .
w s
Ex am pl e 2.3.3
Ifw =seif 3 , then one checks that
8- i
8 1.
=13- 131,
2 +i 3- 2i

3+2i 3- 2i
(6+2) +(-4 +3)i
2 +i
Ex am pl e 2.3.2
We convert the following quotient to Cartesian form:
For instance, ; =-i because 1=i .(-i).
Inpractice, division of complex numbers can be done by multiplying the top and bottom
of the quotient by the conjugate of the bottom expression.
De finition 2.3.1 If z and w are complex numbers (w i=0), then T v is the complex number 'U
that satisfies the equation z =w . 'U.
2 . 3 D i v i s i o n O f C o m p l e x N u m b e r s
2.3 Divisionof Complex Numbers 21