ANALYSIS MTHA5001Y (2015–16)
David Asper´o University of East Anglia
Contents
1 The Complex Plane 3 1.1 Polar coordinates ...... 8
2 Metric spaces and continuity 9 2.1 Continuity ...... 11
3 Open sets, topologies, uniform continuity, and convergence 13 3.1 Small interlude on linear orders ...... 21
4 Sequences 22 4.1 Cauchy sequences ...... 26
5 Series 27 5.1 Absoluteconvergence...... 29
6 Sequences and series of functions 30 6.1 Seriesoffunctions...... 32 6.2 Power series ...... 33
1 7 Holomorphic functions 37 7.1 The Cauchy–Riemann equations ...... 41 7.2 Harmonicfunctions...... 45 7.3 Somepopularfunctions...... 47
8 Functions as Transformations 52
9 The reciprocal function 53
10 Fractional linear transformations 55 10.1 The extended plane ...... 56 10.2 M¨obiustransformations ...... 56
11 Conformal mappings 59
12 Curves, paths, and connectedness 63
OVERVIEW
1. AUTUMN SEMESTER. We will lay the basics of complex analysis. The main focus will be the notions of series of complex functions and of holomorphic function. The following are the main topics to be covered.
The complex plane; the modulus as a measure of distance. • Basic topological notions. • Continuity. • Sequences of functions. Uniform and point–wise convergence for sequences of • functions. Power series representing real and complex functions. Radius of convergence. • Di↵erentiability of complex functions. Holomorphic functions. • Cauchy–Riemann equations. • Elementary functions. • Fractional linear transformations and conformal functions. • 2. SPRING SEMESTER. We will focus on the theory of complex integration along paths. This theory will contain the main highlights of the module. The following is a list of topics likely to be covered then (changes may occur).
2 More topology of the complex plane. • Review of di↵erentiability of complex functions, holomorphic functions, the • Cauchy–Riemann equations and elementary functions defined by power series. Paths, contours, connectedness and simple-connectedness. • Integration along a path, the Estimation Theorem, integration of power series, • the Fundamental Theorem. Cauchy’s Theorem, Cauchy’s Integral Formulae, Taylor’s Theorem, Liouville’s • Theorem, the Identity Theorem, Laurent expansions. Singularities, residues and Cauchy’s residue theorem, techniques for finding • residues, summation of series and other applications.
1 The Complex Plane
This material is familiar but we shall take this opportunity to review it and set the basis for the rest of the course. Our starting point will be the problem of finding solutions of polynomial equations with real coe cients, i.e., solutions of an expression of the form
n anx + ...+ a1x + a0 =0 where n is some integer and ai R for all i. 2 Linear equations (when n =1)ofcourseareeasyandhavealwaysexactlyonesolution when a =0. 1 6 Quadratic equations are a bit trickier: The solutions of
x2 + px + q =0
2 2 are given by x = p + p q and x = p p q,providedtheseexpressions 0 2 4 1 2 4 p2 make sense. In other words,q these solutions will existq (in R)ifandonlyif q 0. 4 The simplest quadratic equation without a real solution is
x2 +1=0
Is this the end of the story? It won’t be for us. For various reasons mathematicians have found it desirable to work with numbers which, “if they existed,” would satisfy equations like x2 +1=0.1
1It would be perhaps more accurate to say that mathematicians were naturally driven to work with fields containing all real numbers and square roots of negative real numbers.
3 We will stipulate the existence of a new number, i,theimaginaryunit,2 which satisfies
i2 = 1 Of course, i is not in R.Wewouldliketobeabletousethisnumberincombination with the usual real numbers. In other words, we would like to have a number system C which
(1) contains all the real numbers together with this new number i.
This number system should come endowed with new arithmetical operations of ad- dition, ,andmultiplication, , such that (2) when restricted to the real numbers, and are exactly the usual operations, let us call them + and ,onR. · Moreover,
(3) and should have all the usual properties that we are used to from working with the real numbers (they should be associative, commutative, and so on).
Putting (1), (2) and (3) together we have, in particular, that if a, b are reals, then a (i b) should be a number in our system, and that for all reals a1, a2, b1, b2 the following should hold:
(a1 (i b1)) (a2 (i b2)) = ((a1 + a2) (i b1)) (i b2) =(a + a ) ( i (b + b )) 1 2 1 2 (a1 (i b1)) (a2 (i b2)) = a1a2 (((i i) b1b2) (i (a1b2 + a2b1))) =( a a b b ) (i (a b + a b )) 1 2 1 2 1 2 2 1 where the last equation should hold since i i = 1. By now we have quite a lot of information on how our intended number system should behave. Of course, from now on we will write + to represent both of + and ,and to represent both of and ,evenifthesearecompletelydi↵erentoperations.Wewill · also adopt the customary· abbreviations in use when doing arithmetic with the reals; in particular, we will remove parentheses in certain cases (as for example when we write a + b c, which doesn’t really make sense, in order to denote a +(b c)). We will also tend to· suppress the symbol (and thus will tend to write ab instead· of a b), and will write a2 to denote a a,etc.Thisway,wemaywritetheaboveequationsasfollows:· · · (a1 + ib1)+(a2 + ib2)= (a1 + a2)+ib1 + ib2 =(a1 + a2)+i(b1 + b2)
2This name is very misleading. i is not any more ‘imaginary’ than, say, ⇡, p2 or, for that matter, 1.
4 2 (a1 + ib1) (a2 + ib2)=a1a2 +(i b1b2 + i(a1b2 + a2b1)) · =(a a b b )+i(a b + a b ) 1 2 1 2 1 2 2 1 Our next move is to stipulate that our new numbers, which we are going to call ‘complex numbers’, are exactly the numbers of the above form (in other words, that they are expressions of the form a + ib,wherea and b are reals and i is some new object), and that the operations satisfy exactly the above requirements. Noting that we can unambiguously identify an expression like a + ib,wherea, b R,withtheordered 2 pair (a, b) R2 prompts us to make the following definition. 2
Definition 1.1 The complex plane, C, is R2 endowed with the following operations + and : ·
1. (a1,b1)+(a2,b2)=(a1 + a2,b1 + b2) 2. (a ,b ) (a ,b )=(a a b b ,a b + a b ) 1 1 · 2 2 1 2 1 2 1 2 2 1 where all expressions ai + aj, a1a2 b1b2, aibj refer of course to the usual addition, subtraction and multiplication in the reals.
We will of course identify a complex number (a, b)inC (which is nothing but the real plane R2)withtheexpressiona + ib. Addition and multiplication of complex numbers then satisfy the following.
(a1 + ib1)+(a2 + ib2)=(a1 + a2)+i(b1 + b2)(1)
(a + ib )(a + ib )=a a b b + i(a b + a b )(2) 1 1 2 2 1 2 1 2 1 2 2 1 where a1, a2, b1 and b2 are reals, and where the expressions a1 + a2, a1a2 and so on of course refer to the usual operations on the reals. Throughout the notes, if I write something like a + ib Iwillusuallyunderstand implicitly that a and b are reals. On the other hand, in an expression like u + iv, u and v will be typically complex numbers but not necessarily reals. If necessary, I will make this precise in the relevant context, though. I will also often use letters like z, w to denote complex numbers. We identify x R with (x, 0) (or, what amounts to the same thing, with x + i0), so 2 R C. With this identification, R is also called the real axis. Also, the line ia : a R is called✓ the imaginary axis. You can naturally visualize the real axis and the{ imaginary2 } axis as being orthogonal to each other and meeting at exactly 0 (0 is also called the origin).
5 Notation 1.2 Given z = a + ib C with a and b reals, we call a the real part of z, and denote it by Re(z), and call b the2 imaginary part of z, and denote it by Im(z).
For example, Re(i)=Re( i)=0,Im(i)=1,andIm( i)= 1. It is easy to check – which we are not doing here – that (C, +, )isacommutative ring with identity,whichmeansthatitisanicelybehavedalgebraicstructure,inthe· sense that the usual arithmetical computations that we are used to from working with the reals can be performed in it. In particular, given any number a + ib there is another number, namely a ib,suchthat (a + ib)+( a ib)=0 (in other words, a ib is the additive inverse of a + ib). Note that a ib is obtained by reflecting a + ib with respect to the origin. The existence of additive inverse means that C comes also with a natural operation of subtraction that behaves exactly in the way you would expect. Also, given a + ib =0,3 we have that 6 1 (a + ib) (a ib)=1 (3) · a2 + b2 (in other words, 1 (a ib), which of course we may write also as a2+b2 a ib , a2 + b2 is the multiplicative inverse of a + ib). The existence of these multiplicative inverses means that (C, +, )is,notonlyacommutativeringwithidentity,butinfactafield.In · other words, (C, +, )comesalsoequippedwithdivision.4 ·
Notation 1.3 Let z = a + ib C, with a and b reals. 2 1. The complex number a ib is called the conjugate of z and is denoted by z. 2. The real number pa2 + b2 is called the modulus of z and is denoted by z . | | Note that the conjugate of z is obtained by reflecting z it along the real axis. Also, the modulus of z is a measure of “how far” z is from the origin (we will make this more precise in a while).
3And of course a + ib = 0 if and only if at least one of a, b is not 0. 6 4However, unlike the reals, (C, +, ) is not an ordered field, i.e., there is no order relation on C · compatible with the operations. In fact, we can naturally picture C as a plane but not as a line. We will see a proof of this later on.
6 From equation 3 it follows that if z = 0 is a complex number, then its multiplicative 6 1 inverse, which of course we can denote z ,isgivenby 1 z = (4) z z 2 | | 1 Thus, z is obtained by reflecting z along the real axis, and then multiplying the 1 resulting complex number by the real number z 2 . In particular it follows that if z is 1 | | 1 very far from 0, then z is very close to 0 and, vice versa, if z is very close to 0, then z is 1 very far from 0. Later on we will take a closer look at the function sending z to z . In general, dividing complex numbers needs a trick, which relies on the fact that (a + ib)(a ib)=a2 iab + iba i2b2 = a2 + b2.Fromthiswegetthat (a + ib) (a + ib)(c id) (ac + bd)+i(bc ad) = = . (c + id) (c + id)(c id) c2 + d2 It is remarkable that the addition of i to R and closure under multiplication and addition lets us not only solve the equation x2 +1=0(whosesolutionsarei and i), but in fact every polynomial equation: Theorem 1.4 (Fundamental theorem of algebra) Any equation n anx + ...+ a1x + a0 =0, where n is some integer and ai C for all i, has some solution in C. 2 Aconsequenceoftheabovetheoremisthateverypolynomialp(x)withcomplex coe cients can be factored as p(x)=(x ↵ )n1 ... (x ↵ )nm 1 · · m for ↵i C and integers ni. 2 Theorem 1.4, despite its pompous name, is not that fundamental in algebra, but it does say something remarkable; that (C, +, )isanalgebraically closed field.Afieldis · algebraically closed if every polynomial equation has some solution in it. Q and R are certainly not algebraically closed. C is by no means the only algebraically closed field. In fact the algebraic closure Q⇤ of Q is an algebraically closed field contained in C and much smaller than C.5 5 5 3 p2, i, p7+p11 + i, and so on, belong all to Q⇤. On the other hand, many interesting numbers, such as ⇡ or e, do not belong to Q⇤.ThenumbersinQ⇤ are called algebraic numbers, and the numbers in C which are not in Q⇤ are called transcendental numbers. Surprisingly, it is not known whether or not e + ⇡ is algebraic and whether or not e⇡ is algebraic (although it is expected that none of them is). It is not even known whether or not e + ⇡ is rational, and whether or not e⇡ is rational (!). On the other hand it is easy to see that at least one of e + ⇡, e⇡ is transcendental. The proof is as follows: Consider the polynomial p(x)=(x e)(x ⇡)=x2 2(e + ⇡)x + e⇡.Ife + ⇡ is algebraic, then so is 2(e + ⇡)sinceQ⇤ is a field and therefore closed under the operations of arithmetic. Hence, if e⇡ were also algebraic, then p(x) would have all its coe cients in Q⇤.ButQ⇤ is algebraically closed. Therefore, all of its roots, namely e and ⇡, would be in Q⇤. But it is known, by deep theorems in analysis, that e and ⇡ are in fact both transcendental.
7 There are proofs of of Theorem 1.4 which should be accessible to you in a little while, and it would be a good idea to look at them later this semester.6 In any case a proof will be given in the second semester.
Exercise 1.5 Prove that the following is true for all z, w C. 2 1 z¯ Re z =(z +¯z)/2, Im z =(z z¯)/(2i), z 2 = zz,¯ = , | | z z 2 | | z w =¯z w,¯ zw =¯zw,¯ z/w =¯z/w,¯ z + w z + w . ± ± | || | | |
1.1 Polar coordinates
We can represent the number z = a + ib on the plane using its coordinates (a, b), but we can also use polar coordinates. Namely we can represent z uniquely as a pair z , ✓ where z is the distance from z to the origin, and ✓ is the angle between the positive| | part of| the| x-axis and the line connecting z with the origin. In other words, and using elementary trigonometry, the number z just described is the number z (cos(✓)+i sin(✓)). | |
Notation:Iwilltemporarilyusethenotationcis(✓)todenotethenumbercos(✓)+ i sin(✓).7
Using the above notation, the number z Iamtalkingaboutis z cis(✓). | | The angle ✓ is certainly not unique; in fact, the map ✓ (cos ✓, sin ✓)fromR into 7! the unit circle in R2 is 2⇡-periodic. On the other hand, if z =0,thenthereisaunique ✓ ( ⇡,⇡]suchthatz = z cis(✓). This ✓ is called the principal6 argument of z,butwe will2 sometimes refer to it simply| | as the argument of z.8 We will denote this number also by arg(z). Recall now the formulas relating the cosine and sine of an angle ↵ + to those of ↵ and
cos(↵ + )=cos↵ cos sin ↵ sin (5)
sin(↵ + ) = sin ↵ cos +sin cos ↵ (6)
The following proposition is an immediate consequence of the above equations.
6See for example “The fundamental theorem of algebra: an elementary and direct proof”, by Oswaldo Oliveira. 7Later on we will prove that there is a certain ‘exponentiation function’ ez, defined for all z C, and 2 such that ei✓ =cis(✓) for every ✓ R. Here e denotes the familiar Euler’s constant from real analysis. 8Of course, for z = 0, any ✓ is2 such that z = z cis(✓). | |
8 Proposition 1.6 Given complex numbers z1 = r1 cis(✓1) and z2 = r2 cis(✓2),
z z = r r cis(✓ + ✓ ) 1 · 2 1 2 1 2 The following are some immediate consequences of the above proposition, some of them quite useful or illuminating.
For all z , z ,arg(z z )=arg(z )+arg(z ). • 1 2 1 2 1 2 For every z C, z z¯ = z 2. • 2 · | | Given any number z, iz is the result of rotating z anti-clockwise around the origin • ⇡ by 2 . For all reals r 0, ✓,andeveryintegern 1, (r cis(✓))n = rn cis(n✓). In particular, • (cos(✓)+i sin(✓))n =cos(n✓)+i sin(n✓)(7)
which is the so–called de Moivre’s formula for the case ✓ R. 2 Given any z =0andanyintegern 1, z has exactly nn-th roots. If z = r cis(✓), • these roots are6 ✓ +2⇡k z = pn r cis( ) k n for k =0, 1,...n 1(notethattheyarealldi↵erentandthatz can have at most nn-th roots). In other words, the polynomial xn z can be written xn z = (x z ) k k