Hyperbolic Geometry

David Aguilera and Steven Guild June 7, 2013

Introduction

Hyperbolic geometry is a branch of a type of geometry known as Euclidean geometry. Non-Eucledian geometry is the logical counterpart of Euclidean geometry, which was developed by Euclid of Alexandria, a Greek mathematician, around the year 300BC. His text Elements, which has become the classical framework for studying geometry, is based upon a set of ve postulates. The ve postulates of Euclidean geometry are: 1. A straight line segment can be drawn joining any two points; 2. Any straight line segment can be extended indenately as a straight line; 3. Given any straight line segment, a cicrcle can be drawn having the segment as radius and one endpoint as center; 4. All the right angles are congruent (meaning any isometry will take them into right angles) 5. If two lines are drawn which intersect a third in such a way that the sum of the inner angles on one side is less tha two right angles, then the two lines inevitably must intersect on that side, if extended far enough. The fth postulate seems slightly less articial if worded in the following equivalent manner: Given any straight line l, and a point p not on it, there exists one and only one straight line which passes through p, and never intersects l, regardless of how far these lines are extended. This line is dened as the line parallel to l that goes through p. 1

This last statement seemed far too unnatural to most mathematicians (including Euclid himself), while the rst four seemed a lot more reasonable to assume. Because of this, generations of mathematicians attempted to prove Euclids fth postulate by using the previous four, but they never succeeded; in fact, many incorrect proofs were published on this subject. It wasnt until 1868 (more than 2000 years later!) that the Italian mathematician Eugenio Beltrami published his work on a new eld of mathematics, Non-Euclidean Geometry, that the reason for this diculty became clear. This work was further developed by other mathematicians, but the fundamentalworks were made by Beltrami, Nikolai Lobachevsky, J anos Bolyai, Felix Klein and Henri Poincar e. Non-Euclidean Geometry arises by negating the fth postulate. Since there are two logical ways of negating it, there are two branches of Non-Euclidean Geometry: Hyperbolic and Projective Geometry. Projective Geometry is obtained by stating that there are no parallel lines in the Euclidean sense, i.e., that every line that goes through a point p in projective space, not contained in l, will intersect l inevitably. Hyperbolic Geometry is obtained by relaxing the uniqueness condition, i.e., allowing for more parallel lines to exist. We will explore the latter case in more detail. In this context, the word line refers to a geodesic. We will introduce the notion of Hyperbolic Spaces, and we will talk about a few of such spaces, observing that these properties will be satised on them. Hyperbolic spaces are those in which the Gaussian curvature K < 0 always. Many properties are natural in these spaces, including the rst four Euclidean postulates, and the Hyperbolic version of the fth, as we shall see. The rst of such spaces described in this sense was the pseudosphere. This space can be obtained by assuming that the Gaussian curvature K of a surface is constant and negative, say K = 1, and that the surface is a surface of revolution. By doing this, and recalling that the Gaussian curvature of a surface of revolution is K = ff , we may obtain the following parametrization of a pseudosphere: (u, v ) = (eu cos v, eu sin v, 1 e2 u cosh 1 (e u )), which has a rst fundamental form ds2 = du2 + e2 dv 2 . This is the rst parametrization we have of a hyperbolic space, but it is far from being simple. This can be simplied by applying the transforma1 1 tion w = eu , v = v , which leads to (w, v ) = ( w cos v , w sin v ,

1 w 2

cosh1 w ). This reparametrization of the surface has a rst fundamental form dv 2 +dw2 2 ds = w2 . This reparametrization is convenient because it is conformal to the plane. This, however, is dened for v (, ) and w (1, ), so to simplify the domain of the surface, and to relate this half plane to the complex plane in a simple manner, we simply consider the translation w = w 1, which preserves the rst fundamental form. This parametrization is known as the Upper Half Plane Model, and its usefulness comes from the fact that it is conformal to the plane, but, as we will see, this model has many other interesting and useful features. First of all, the geodesics, which were quite complicated in the pseudosphere model, get mapped into the arcs of circles that intersect the v axis perpendicularly and the straight lines that intersect the v axis perpendicularly. Thus, we dene the hyperbolic space H by: H = {(v, w) R|w > 0} Equipped with rst fundamental form: dv 2 + dw2 , w2 It is convenient to identify H with the upper half of the complex plane in the following manner: ds2 = H = {z C|Im(z ) > 0}. Now, lets see how lines behave in this space. Recalling that our hyperbolic lines come from all the half circles and straight lines perpendicular (in the euclidean way) to the imaginary axis, its interesting to note that, for a line l in H, and a point p not on it, there may be innitely many lines passing through it, and that do not intersect l, and there are actually two lines that intersect l at the w axis. These two lines are said to be parallel to l at p, and the rest of them are said to be ultraparallel. This proves that the fth postulate of Euclidean geometry breaks down in this space. Since it is embedded in a half plane, then all the angles are actually consistent with Euclidean geometry, but parallel lines are not unique. It must be noted that, in the terminology of the second postulate, these lines can be extended indenately, in the sense that one can always nd a point on a line that is arbitrarily farther away from any other point in that

line, if one considers the distance between two points arises from the rst fundamental form given for this space. The distance between two points a and b is given by: dH (a, b) = 2 tan1 There are four types of isometries in H: 1. Translations parallel to the real axis 2. Reections in lines parallel to the imaginary axis 3. Dilations of a factor a > 0 4. Inversions in circles with centers on the real axis Another way to study the properties of the hyperbolic space, is to transform it once again, but to now conne it to the unit disk by means of the map: zi . P (z ) = z+i This conformal map will take the whole upper plane onto the unit disk, and the resultant model is known as the Poincar e Disk Model. The inverse map 1 P is given by: z+1 P 1 (z ) = . i(z 1) Lets see that for z = u + iw, P (v + iw) =
v 2 +w2 +12w 1 2 (v 2 +w 2 +1+2w ) , v +i(w1) , v +i(w+1)

|b a| |b a|

so P (v + iw) =

so P (v + iw) < 1 if w > 0, P (v + iw) = 1 if w = 0, and P (v + iw) > 1 if w < 0, so this clearly takes our upper half plane into the unit circle, taking the real axis into the boundry of the unit disk. The Poincar e Disk Model DP of H can be then dened as the unit disk, equipped with a rst fundamental form such that the dieomorphism that takes H into DP is an isometry. The rst fundamental form will then become: ds2 DP = 4(dv 2 + dw2 ) . 1 v 2 w2

Proof: Let P 1 (v + iw) = v + iw . If P is an isometry, then dv 2 + dw 2 = dv 2 + 2w v 2 w 2 2 1 dw . By the formula given for P before, v = (v1)2 +w2 and w = (1 , v 1)2 +w2 so 1 dv 2 = 4(v 1)wdv 2((v 1)2 w2 )dw ((v 1)2 + w2 )2 4

and w 2 = 1 2((v 1)2 w2 )dv 2((v 1)2 w2 )dw , ((v 1)2 + w2 )2

so, combining these equations we get dv 2 + dw 2 4(dv 2 + dw2 ) = = ds2 DP . w 2 1 v 2 w2 Since the Poincar e and the Half Plane models are isometric, then the isometries of DP are the compositions of P and the isometries of H, and the distance formula can be obtained likewise, and it turns out to be ds2 H = dDP (a, b) = 2 tanh1 |b a| . |1 ab|

Its easier to visualize lines and paralelism in this model. Since the Poincar e model is conformal to the Upper Half Plane Model, right angles are preserved after applying P , and thus the lines in this model are those circles that intersect the unit circle perpendicularly at both ends, and straight lines through the origin. If two lines meet only at the boundry once, theyre said to be parallel, because, just like parallel lines in the Euclidean sense, they dont intersect until innity. If two lines dont intersect even at innity (i.e. the edge of the circle), theyre said to be ultraparallel. Another model for H that permits an easier visualization of lines, and relates it to projective geometry, is the Beltrami-Klein Model. This is achieved by making a composition of the inverse of the stereographic projection and the vertical projection P r. The inverse of the stereographic projection of DP takes the unit disk onto the southern hemisphere of the units sphere, while the vertical projection of the inverse of the inverse of the stereographic projection takes the southern hemisphere of the sphere back onto the unit disk, so this coimposition map is bijective if the domain is taken to be the z disk. Let K = P r , then, if we make z = x + iy , then K (z ) = |z|2 2 +1 . This is clearly not conformal to H, since ds2 K = (1 w2 )dv 2 + 2vwdvdw + (1 v 2 )dw2 , (1 v 2 w2 )2

but it has a critical advantage over the other two models: hyperbolic lines from DP get mapped into straight lines in the unit disk. Since we are embedded in projective space, some notions of projective geometry become important and well introduce a few that come into play with important properties 5

of the Beltrami-Klein Model. First of all, if a, b, c and d are four points in ac)(bd) . Let a, b, c and d C, then their cross ratio is given by (a, b; c, d) = ( (ad)(bc) lie on the same line in C, then (a, b; c, d) > 0. In fact, if we make p a point not in the same line as the others, and then take the lines that go through p and each of the other points, and intersect those lines with another set of points a , b , c and d , then (a, b; c, d) = (a , b ; c , d ), which means that cross ratio is a projective invariant. Now, to nd the distance between two points in the Beltrami-Klein Model, consider the points a and b in DK . Let l be the line through a and b, and let c and d be the points at which l intersects the unit disk. Since isometries preserve distances, we can nd isometries such that l is taken to the real axis, and c and d are taken to 1 and 1 respectively, and dDK (a, b) = dDK (a , b ). Now, since we know the distance between the inverse images of a and b on DP , and K : DP DK is an isometry, then dDK (a , b ) = dDP (K 1 (a ), K 1 (b )). Using the rst fundamental form of the Poincar e model to compute the distance, we get:
K 1 (b )

dDK (a , b ) =

K 1 (a )

(1 + K 1 (a ))(1 K 1 (b )) 2dv = log . 1 v2 (1 K 1 (a ))(1 + K 1 (b )) 1 2 ),

1 Now, from the denition of K (z ), we know that K 1 () = (1 so: 1+ 1 + K 1 () = 1 1 K () 1

dDK (a , b ) = log

1 (1 + b )(1 a ) 1 (1 + K 1 (a ))(1 K 1 (b )) = log = log(a , b ; 1, 1). 1 1 (1 K (a ))(1 + K (b )) 2 (1 b )(1 + a ) 2

Finally, since we had isometries take a, b, c and d into a , b , 1 and 1, and (a, b; c, d) = (a , b ; c d ), then: dDK (a , b ) = dDK (a, b) = 1 log |(a, b; c, d)|. 2