Discover millions of ebooks, audiobooks, and so much more with a free trial

Only $11.99/month after trial. Cancel anytime.

Differential Geometry
Differential Geometry
Differential Geometry
Ebook658 pages5 hours

Differential Geometry

Rating: 5 out of 5 stars

5/5

()

Read preview

About this ebook

This outstanding textbook by a distinguished mathematical scholar introduces the differential geometry of curves and surfaces in three-dimensional Euclidean space. The subject is presented in its simplest, most essential form, but with many explanatory details, figures and examples, and in a manner that conveys the geometric significance and theoretical and practical importance of the different concepts, methods and results involved.
The first chapters of the book focus on the basic concepts and facts of analytic geometry, the theory of space curves, and the foundations of the theory of surfaces, including problems closely related to the first and second fundamental forms. The treatment of the theory of surfaces makes full use of the tensor calculus.
The later chapters address geodesics, mappings of surfaces, special surfaces, and the absolute differential calculus and the displacement of Levi-Cività. Problems at the end of each section (with solutions at the end of the book) will help students  meaningfully review the material presented, and familiarize themselves with the manner of reasoning in differential geometry.
LanguageEnglish
Release dateApr 26, 2013
ISBN9780486318622
Differential Geometry

Related to Differential Geometry

Titles in the series (100)

View More

Related ebooks

Mathematics For You

View More

Related articles

Reviews for Differential Geometry

Rating: 5 out of 5 stars
5/5

4 ratings0 reviews

What did you think?

Tap to rate

Review must be at least 10 words

    Book preview

    Differential Geometry - Erwin Kreyszig

    INDEX

    I

    PRELIMINARIES

    1. Notation. In this section the meaning of some symbols occurring very frequently in our text will be explained and the page on which these symbols are introduced for the first time indicated. The reader will find another index of definitions, formulae, and theorems at the end of this book.

    2. Nature and purpose of differential geometry. In differential geometry properties of geometric configurations (curves, surfaces) are investigated by means of differential and integral calculus. All our considerations will take place in three-dimensional Euclidean space and will, in general, be restricted to real geometric configurations. We will, however, occasionally extend our methods to the complex domain.

    A geometric property is called local, if it does not pertain to the geometric configuration as a whole but depends only on the form of the configuration in an (arbitrary small) neighbourhood of a point under consideration. For instance, the curvature of a curve is a local property. Since differential geometry is concerned mainly with local properties, it is primarily a geometry in the small or a local geometry.

    This fact does not exclude the possibility of considering geometric configurations as a whole. This kind of investigation belongs to what we call global differential geometry or differential geometry in the large. In this book we will consider only a small number of global problems, for example, in connexion with the theorem of Gauss–Bonnet. We may say that global problems are problems in which ‘macroscopic’ properties are related to ‘microscopic’ ones. For further study in this field see, for example, W. Blaschke, Vorlesungen über Differentialgeometrie (3 vols., Berlin, 1945, 1923, and 1929).

    As is natural, concepts, methods, and results of analytic geometry will be constantly used in differential geometry. The following sections are consequently devoted to a brief review of some of the topics from analytic geometry which we will need for our further investigations. We may restrict ourselves to the analytic geometry of three-dimensional Euclidean space in which all our considerations will take place.

    3. Concept of mapping. Coordinates in Euclidean space. The concept of mapping is of basic importance in differential geometry.

    Let M and M′ be two sets of points in three-dimensional Euclidean space R3. (M or M′ may contain all points of R3 or only a subset of these points.) If a rule T is stated which associates a point P′ of M′ to every point P of M we say that a mapping or transformation (more exactly: point transformation) of the set M into the set M′ is given. P′ is called the image point of P, and P is called an inverse image point of P′. The set of the image points of all points of M is called the image of M. If every point of M′ is an image point of at least one point of M the mapping is called a mapping of M onto M′.

    A mapping T of M onto M′ is called one-to-one if the image points of any pair of different points of M are different points of M′. Then there exists the inverse mapping of T, denoted by T–1 which maps M′ onto M such that every point P′ of M′ is mapped onto that point P of M which corresponds to P′ with respect to the mapping T.

    The set of all points whose distance from a point P is smaller than a positive number η is called a neighbourhood of P. Consequently this neighbourhood consists of all points in the interior of a sphere of radius η with centre at P. There are arbitrarily many different neighbourhoods of P each of which corresponds to a certain value of η. A mapping of a set M into a set M′ is said to be continuous at a point P of M if, for every neighbourhood U′ of the image P′ of P there exists a neighbourhood U of P whose image is contained in U′. The mapping is said to be continuous if it is continuous at every point of M. A one-to-one continuous mapping whose inverse mapping is also continuous is called a topological mapping. Point sets which can be topologically mapped onto each other are said to be homeomorphic.

    A mapping is called a rigid motion if any pair of image points has the same distance as the corresponding pair of inverse image points.

    We will now discuss some basic facts of the analytic geometry of the three-dimensional Euclidean space R3 which we will need in our later investigations.

    We first introduce a right-handed system of orthogonal parallel coordinates x1, x2, x3 whose unit points on the axes, that is, the points with coordinates (1, 0, 0), (0, 1, 0), (0, 0, 1), respectively, have the distance 1 from the origin, cf. Fig. 1. Such a special right-handed system will be called a Cartesian coordinate system.

    FIG. 1. Right-handed system of orthogonal parallel coordinates

    FIG. 2. Left-handed system of orthogonal parallel coordinates

    In general, a coordinate system is called right-handed if the axes, in their natural order, assume the same sort of orientation as the thumb, index finger, and middle finger of the right hand. A system is said to be left-handed if the axes, in their natural order, assume the same sort of orientation as the thumb, index finger, and middle finger of the left hand, cf. Fig. 2.

    The notation x1, x2, x3 for the coordinates is more convenient than the familiar x, y, z, for it enables us to use the abbreviated form (xi) for the coordinates x1, x2, x3 of a point.

    is related to the given one by a special linear transformation of the form

    whose coefficients satisfy the conditions

    and

    The quantity δkl is called the Kronecker symbol.

    The transition from one Cartesian coordinate system to another can be effected by a certain rigid motion of the axes of the original system. Such a motion is composed of a suitable translation and a suitable rotation. A rigid motion which carries a Cartesian coordinate system into another Cartesian coordinate system is called a direct congruent transformation (or displacement). We will now investigate (3.1) in somewhat greater detail.

    Let m and n be natural numbers. A system of m · n quantities arranged in a rectangular array of m horizontal rows and n vertical columns is called a matrix. The quantities are called elements of that matrix. If m equals n the matrix is said to be square, and the number n is called the order of the matrix.

    The coefficients aik figuring in (3.1) form a quadratic matrix

    The corresponding determinant (3.1c) will be denoted by

    If in particular A equals the ‘unit matrix

    then (3.1a) is of the form

    This transformation corresponds to a translation of the coordinate system. If, moreover, bi = 0 (i = 1, 2, 3), we obtain from (3.2)

    i.e. the transformed coordinates are the same as the original ones. Such a special transformation is called the identical transformation.

    If bi = 0 (i = 1, 2, 3), and the coefficients aik are arbitrary but such that the conditions (3.1b) and (3.1c) are satisfied, then (3.1a) corresponds to a rotation of the coordinate system with the origin as centre. A rotation is also called a direct orthogonal transformation.

    We note that a transformation of the form

    can be geometrically interpreted as a motion composed of a rotation about the origin and a reflection in a plane. A transformation of this type is called an opposite orthogonal transformation. It transforms a right-handed coordinate system into a left-handed one and vice versa. An example of a reflection in a plane (in this particular case, in the x2 x. Direct and opposite orthogonal transformations are called orthogonal transformations, and the corresponding matrices are referred to as orthogonal matrices. A transformation which is composed of translations, rotations, and an odd number of reflections is called an opposite congruent transformation. We should note that every translation or rotation can be composed of two suitable reflections.

    We will now point out that (3.1) can be interpreted in two different ways:

    (Alias.) Formerly we interpreted (3.1) as a coordinate transformation; (xii) are then the coordinates of one and the same point with respect to two different Cartesian coordinate systems.

    (Alibi.) The relation (3.1) can also be interpreted as a mapping or point transformation. Then (xii) represent the coordinates of two different points with respect to one and the same Cartesian coordinate system; that is, the coordinate system remains fixed and the location of the points is changed.

    Both interpretations are closely related to each other. For, in consequence of the above remarks, the transition from one interpretation to the other can be effected in the following manner. Instead of imposing a direct congruent transformation on the given Cartesian coordinate system, one can just as well move the geometric configuration, that is, change its location in space, while the coordinate system is kept fixed. Geometric configurations which differ only by their location in space are said to be congruent to each other. Two congruent geometric configurations can always be made to coincide by means of a displacement. (Remark: Since we do not include reflections the concept of congruence is used here in a narrower sense than is usual.)

    If we wish to investigate geometric facts analytically we have to introduce a coordinate system. Consequently, a coordinate system is a useful tool but no more. A property of a geometric configuration consisting of points of R3 will be called geometric if it is independent of the special choice of the Cartesian coordinates. In other words, a property of a configuration is called a geometric property if it is an invariant with respect to direct congruent transformations of the coordinate system, or, what is the same, with respect to direct congruent transformations of the configuration under consideration.

    In order to consider this fact more precisely and from a general point of view it is advantageous to use the concept of a group and to mention an important connexion between geometry and group theory.

    We assume the mappings

    used in the following to be one-to-one mappings which are defined in the whole space R3, i.e. they are one-to-one mappings of the space R3 onto itself. If several mappings of this kind are given, for example,

    we can conceive these mappings to be effected successively; in this manner we obtain what is called a composite mapping or product of these mappings. In the case of the foregoing example we have

    A set G of mappings (3.4) is called a group of mappings or a transformation group if G has the following properties:

    1. The identical mapping i = xi(i = 1, 2, 3), is contained in G.

    2. The inverse mapping T –1 of any mapping T contained in G is also an element of G.

    3. For any arbitrary pair of (not necessarily distinct) mappings T1, T2 contained in G the product mapping of these mappings is also an element of G.

    The direct congruent transformations form a group.

    By means of every one of these groups we may classify the geometric configurations in R3 into equivalence classes: Two configurations are called equivalent with respect to a certain group (that is, they are contained in one and the same equivalence class) if that group contains a mapping which maps one of the configurations onto the other.

    In the case of the group of direct congruent transformations this concept of equivalence is identical with the concept of congruence as defined above.

    The three conditions involved in the definition of a group, if considered from the standpoint of classification, are nothing but the axioms of equivalence:

    1. Every geometric configuration is equivalent to itself.

    2. If a configuration A is equivalent to a configuration B, then B is also equivalent to A.

    3. If a configuration A is equivalent to a configuration B and B is equivalent to a configuration C, then A is also equivalent to C.

    Let now a certain group G of mappings be given and consider geometric configurations each of which, for the sake of simplicity, is an ordered system of a certain finite number of points of R3. Every property of each of these objects can be analytically described by a certain expression which is a function of the coordinates of the points of the configuration under consideration. We denote by I such an expression, corresponding to a certain property of an object. This property will be called geometric with respect to a group G if and only if the numerical value of I remains unaffected by any mapping contained in G. Then I is said to be an invariant of the group G.

    For example, the distance

    of two points (xi) and (yi) of R3 is an invariant with respect to the group of the displacements; such a motion will, in general, cause a change of the values of the coordinates, but d will retain its value.

    In consequence of the foregoing reasoning one may consider a geometry to be identical with the theory of invariants of a certain transformation group. The investigation of geometry from that ‘group theoretical’ point of view has been initiated by F. Klein [1], [2, vol. i, p. 460]† in his so-called Erlangen Program (1872). Klein’s idea created a certain order among the different types of geometry and also had a considerable influence on the development of geometry.

    For example, affine geometry corresponds to the group of affine transformations,

    projective geometry to the group of projective transformations,

    (where the determinant det(bik) of the coefficients is of order 4), etc.

    Problem 3.1. Do the following mappings form a group or not? (a) Translations, (b) rotations about a fixed axis, the angle of every rotation being always an integral multiple of 120°, (c) direct congruent transformations, (d) direct orthogonal transformations, (e) opposite orthogonal transformations.

    4. Vectors in Euclidean space. Vectors are essential tools which we will constantly need in our investigations. The concept of a vector was first used by W. Snellius (1581–1626) and L. Euler (1707–83).

    A line segment S in Euclidean space R3 is uniquely determined by its end points P and Q. The segment is said to be directed if we designate one of the two points as the initial point and the other as the terminal point. Obviously, there are two possibilities in ‘directing’ a segment or giving it a sense.

    A directed segment is called a vector. The length of the segment is called the length or magnitude of the vector.

    We define: Two vectors are equal if and only if they are parallel, have the same length, and have the same sense.

    Hence a vector may be arbitrarily displaced parallel to itself, or, what amounts to the same, its initial point may be chosen arbitrarily.

    A vector of length 1 is called a unit vector.

    Vectors in Euclidean space R3 will be denoted by bold-face letters, e.g. a, b, u, v, w, z, etc. The length of a vector a will be denoted by |a|.

    Let us denote by z the vector which we obtain by directing the segment S with end points P and Q such that P becomes the initial point and Q the terminal point, cf. Fig. 3. We assume that in R3 a certain Cartesian coordinate system has been introduced with respect to which P has the coordinates (xi) and Q has the coordinates (yi). As follows from (3.2), the values of the coordinate differences

    remain unchanged if z undergoes a translation. These quantities z1, z2, z3 are called the components of the vector z with respect to that coordinate system. We write z = (z1, z2, z3), or z = (zi). Similarly, the components of a vector a will be denoted by a1, a2, a3, etc.

    If we introduce the null vector 0, that is, a vector with components 0, 0, 0, then every ordered triple of numbers z1, z2, z3, also the triple 0, 0,0 can be chosen as the components of a certain vector. The correspondence between the ordered triples of numbers and the vectors is one-to-one. Moreover, if the null vector is introduced, addition of vectors (see below) can be done without exceptional cases.

    FIG. 3

    Since vector components are invariant with respect to translations, the transformation (3.1) corresponds to a transformation of the components of a vector which is of the form

    The transformation property (4.2) is basic for vectors in Euclidean space: A vector in that space can also be defined as an ordered system of three numbers, called components, which behave according to (4.2) if a transformation (3.1) is imposed. From this definition it follows that the differences of the corresponding coordinates of two points of R3 are the components of a vector. Hence the present definition of a vector is equivalent to the one given previously. While the first definition may be easier to grasp, the second has the advantage that it can be extended to more general spaces, cf. Section 29.

    Vectors are useful for the following two reasons: Many geometrical (and physical) concepts are vectors. Rules of vector calculation can be defined which in several respects are similar to the rules governing the system of real numbers and which enable us to conduct many investigations without regard to the special choice of the coordinate system. We will discuss the basic rules of vector calculus in the following section.

    As was shown at the beginning of this section, a vector may be arbitrarily translated; its initial point may be chosen arbitrarily. It is sometimes advantageous, however, to choose a certain fixed point P as the initial point of a vector. Then the vector is said to be bound at P. A vector whose initial point is left undetermined is sometimes called a free vector.

    A vector having the origin of the chosen coordinate system as initial point and a point Q as terminal point is called the position vector of Q with respect to that particular coordinate system, cf. Fig. 4. The components of the position vector have the same numerical value as the coordinates of the point Q. Given a fixed coordinate system, any point in space can be uniquely determined by a certain position vector. It should be stressed that the origin of the coordinate system will in general change if the coordinates are transformed according to (3.1); hence the components of the position vector with respect to the transformed coordinate system will be obtained by using (3.1) and not (4.2).

    FIG. 4. Position vector

    5. Basic rules of vector calculus in Euclidean space. The product ka. of a vector a by a real number k is, by definition, a vector having the components kai. Obviously, the vectors ka and a are parallel to each other; they have the same sense if k is positive, and opposite sense if k is negative. Instead of (—1)a we write simply —a. The vectors —a and a are parallel to each other and are of the same length but of opposite sense.

    The sum

    of two vectors a and b is, by definition, a vector c with components

    Instead of a + (—b) we write simply ab. Vector addition is associative,

    Furthermore, we have

    where k, k1, and k2 are real numbers.

    The expression

    is called the scalar product of the vectors c = (ci) and d = (di), where ci and di are the components of these vectors with respect to a Cartesian coordinate system. Scalar products were first used by H. Grassmann (1809–77). As can be seen from (5.3), the scalar product is a number. Instead of c·d the notation cd, without dot, can also be found in the literature.

    Scalar multiplication is commutative,

    and distributive with respect to addition,

    The scalar product is invariant with respect to coordinate transformations of the form (3.1). Using (4.2) we obtain from (5.3)

    The scalar product can be geometrically interpreted in the following manner: We choose a point with coordinates (xi) to be the initial point of a vector z = (zi) and denote by (yi) the coordinates of the corresponding terminal point of z. Then, by the Theorem of Pythagoras, z has the length

    Hence the scalar product of a vector by itself is equal to the square of its length. In general, if c and d are two arbitrary vectors,

    where α(0 ≤ α ≤ π), denotes the angle between c and d. In order to prove (5.5) we transform the Cartesian coordinates in such a way that c lies in the x1-axis and d lies in the x1 x2-plane. Then c = (c1, 0,0) and d = (d1, d2, 0).

    The value of the scalar product c·d is invariant under the coordinate transformation imposed. As follows from (5.3), we have in the transformed coordinate system c·d = c1 d1 = |c||d| cos α, cf. Fig. 5.

    Relation (5.5) enables us to represent angles between vectors in terms of scalar products,

    As can be seen from (5.3) or (5.5) the scalar product of a vector by the null vector is zero; moreover we have from (5.5)

    FIG. 5

    Theorem 5.1. Two non-null vectors are orthogonal to each other if and only if their scalar product vanishes.

    The vector product v = a × b

    of two vectors a and b is defined by the following expression:

    where ei denotes a unit vector having the (positive) direction of the ith coordinate axis of the Cartesian coordinate system in space R3.

    The vector product v = a × b can be interpreted as a vector orthogonal to a and b in such a way that the vectors a, b, v, in this order, have a right-handed sense, i.e. the same sense as the positive rays of the coordinate axes, taken in their natural order; the absolute value |v| of v equals the area of a parallelogram with sides a and b. We have

    FIG. 6

    where α(0 ≤ α ≤ π), denotes the angle between a and b, cf. Fig. 6. Developing the determinant (5.7) by the first column we obtain the following explicit expressions for the components of v:

    Since the determinant in (5.7) changes sign when the last two columns are interchanged, we have

    Such a multiplication is consequently called anticommutative. From the definition of vector multiplication we easily derive the following rules:

    where k is a real number. The relation

    is called the Identity of Lagrange. From (5.11) follows

    Applications of vector products are numerous in differential geometry. Other applications can be found in physics as is well known.

    FIG. 7

    In our later investigations we shall often meet scalar products of a vector by a vector product. If we form the scalar product of a vector a and a vector product b × c then, in consequence of the definitions (5.3) and (5.7) we obtain

    We write

    and call this expression the mixed product, scalar triple product, or determinant of the vectors a, b, and c.

    As can be found by a simple computation, the mixed product is invariant under direct congruent transformations of the coordinate system and changes its sign under a reflection. The absolute value of the mixed product |abc| can be geometrically interpreted as the volume of a parallelepiped having the edge vectors a, b, and c, cf. Fig. 7. In consequence of the rules governing determinants we find

    where k denotes a scalar. From (5.12) we obtain

    We assume that a is a vector whose components ai are differentiable functions of a variable t. Then the derivative a′(t) of the vector a = a(t) with respect to t is defined as the vector with components a′i(t), where the prime denotes the derivative with respect to t. By applying the usual rules of calculus we immediately find

    where all the vectors occurring are assumed to have components which are differentiable functions of a variable t.

    We finally consider the concept of linear dependence of vectors, p vectors a(1), a(2),…, a(p) are said to be linearly dependent if and only if there are p real numbers k1, k2,…, kp, not all zero, such that

    If no such p numbers exist the vectors a(1), a(2),…, a(p) are said to be linearly independent.

    If the vectors a(i) are linearly independent then (5.17) can hold only if k1 = k2 =…= kp = 0. Conversely, if (5.17) holds only if every ki(i = 1, 2,…, p), vanishes then the vectors a(i) are linearly independent. If at least one of the vectors a(i) is a null vector these vectors are always linearly dependent.

    The geometric interpretation of the linear dependence of two non-null vectors is that the vectors are parallel to each other. From (5.8) we obtain

    Theorem 5.2. Two vectors are linearly dependent if and only if their vector product is the null vector.

    From (5.13) we obtain

    Theorem 5.3. Three vectors are linearly dependent if and only if the determinant of these vectors vanishes.

    The linear dependence of three vectors means that these vectors, if bound at the same point, lie in one and the same plane. Four or more vectors of the three-dimensional Euclidean space R3 are always linearly dependent, cf. Problem 5.3.

    Problem 5.1. Give a representation of a plane using a scalar product.

    Problem 5.2. Prove the invariance of the mixed product under direct congruent transformations of the coordinate system.

    Problem 5.3. Prove: Four vectors of the three-dimensional Euclidean space are always linearly dependent.


    † Cf. the bibliography at the end of this book.

    II

    THEORY OF CURVES

    6. The concept of a curve in differential geometry. We will now investigate curves in three-dimensional Euclidean space R3. We restrict ourselves to real curves. We assume that a Cartesian coordinate system has been introduced in R3. Then every point in space can be uniquely determined by its position vector

    In order to introduce the concept of a curve we start from a real vector function

    In (6.1), the components

    of the vector X are (single-valued) functions of a real variable t defined in the interval I: a t b. By means of (6.1), to every value of t a point of R3 is associated whose position vector is X(t); the vector function (6.1) determines a point set M in R3. (6.1) is called a parametric representation of the set M, and the variable t is called the parameter of this representation.

    Since we will apply calculus to geometric problems we have to require that the vector function (6.1) possess a certain number of derivatives. Moreover, we wish to exclude the trivial case where M consists of one point only. Consequently we make the following

    Assumptions:

    (1) The functions xi(t) (i = 1, 2, 3) are r (≥ 1) times continuously differentiable in I (where the value of r will depend on the problem under consideration).

    (2) For every value of t in I, at least one of the three functions

    is different from zero.

    A representation of the form (6.1) satisfying these conditions will be called an allowable parametric representation.

    As follows from Assumption

    Enjoying the preview?
    Page 1 of 1