A Before starting a new topic, let us recall what we did so far. Can you briefly describe this in your own words, sparing all details?

Similar documents
6. Define log(z) so that π < I log(z) π. Discuss the identities e log(z) = z and log(e w ) = w.

Partial Fractions. Combining fractions over a common denominator is a familiar operation from algebra:

MATH 52: MATLAB HOMEWORK 2

Math 4310 Handout - Quotient Vector Spaces

5.3 Improper Integrals Involving Rational and Exponential Functions

3 Contour integrals and Cauchy s Theorem

The Method of Partial Fractions Math 121 Calculus II Spring 2015

Zero: If P is a polynomial and if c is a number such that P (c) = 0 then c is a zero of P.

TOPIC 4: DERIVATIVES

Partial Fractions. p(x) q(x)

1 if 1 x 0 1 if 0 x 1

Lies My Calculator and Computer Told Me

Solving Rational Equations

Real Roots of Univariate Polynomials with Real Coefficients

Continued Fractions and the Euclidean Algorithm

Representation of functions as power series

3. Mathematical Induction

Solutions to Homework 10

Mechanics 1: Conservation of Energy and Momentum

Vieta s Formulas and the Identity Theorem

36 CHAPTER 1. LIMITS AND CONTINUITY. Figure 1.17: At which points is f not continuous?

1.7 Graphs of Functions

Activity 1: Using base ten blocks to model operations on decimals

Notes on metric spaces

MATH10212 Linear Algebra. Systems of Linear Equations. Definition. An n-dimensional vector is a row or a column of n numbers (or letters): a 1.

Math Workshop October 2010 Fractions and Repeating Decimals

Quotient Rings and Field Extensions

Differentiation and Integration

Complex Function Theory. Second Edition. Donald Sarason >AMS AMERICAN MATHEMATICAL SOCIETY

1. Prove that the empty set is a subset of every set.

DIFFERENTIABILITY OF COMPLEX FUNCTIONS. Contents

MOP 2007 Black Group Integer Polynomials Yufei Zhao. Integer Polynomials. June 29, 2007 Yufei Zhao

3.3 Real Zeros of Polynomials

Zeros of Polynomial Functions

3.1. RATIONAL EXPRESSIONS

INTRODUCTION TO ALGEBRAIC GEOMETRY, CLASS 24

1 Lecture: Integration of rational functions by decomposition

FEGYVERNEKI SÁNDOR, PROBABILITY THEORY AND MATHEmATICAL

1.6 The Order of Operations

Chapter 17. Orthogonal Matrices and Symmetries of Space

Continued Fractions. Darren C. Collins

Stanford Math Circle: Sunday, May 9, 2010 Square-Triangular Numbers, Pell s Equation, and Continued Fractions

0.8 Rational Expressions and Equations

Algebraic and Transcendental Numbers

Domain of a Composition

Section 1.1 Linear Equations: Slope and Equations of Lines

Integrals of Rational Functions

Limits and Continuity

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.436J/15.085J Fall 2008 Lecture 5 9/17/2008 RANDOM VARIABLES

Homework 2 Solutions

BANACH AND HILBERT SPACE REVIEW

March 29, S4.4 Theorems about Zeros of Polynomial Functions

If A is divided by B the result is 2/3. If B is divided by C the result is 4/7. What is the result if A is divided by C?

(Refer Slide Time: 2:03)

Undergraduate Notes in Mathematics. Arkansas Tech University Department of Mathematics

a 11 x 1 + a 12 x a 1n x n = b 1 a 21 x 1 + a 22 x a 2n x n = b 2.

RAJALAKSHMI ENGINEERING COLLEGE MA 2161 UNIT I - ORDINARY DIFFERENTIAL EQUATIONS PART A

PYTHAGOREAN TRIPLES KEITH CONRAD

x 2 + y 2 = 1 y 1 = x 2 + 2x y = x 2 + 2x + 1

Increasing for all. Convex for all. ( ) Increasing for all (remember that the log function is only defined for ). ( ) Concave for all.

U.C. Berkeley CS276: Cryptography Handout 0.1 Luca Trevisan January, Notes on Algebra

Metric Spaces. Chapter Metrics

Microeconomic Theory: Basic Math Concepts

Introduction to Fractions

Math 319 Problem Set #3 Solution 21 February 2002

The last three chapters introduced three major proof techniques: direct,

Separable First Order Differential Equations

Chapter 7 - Roots, Radicals, and Complex Numbers

Polynomial Invariants

THE DIMENSION OF A VECTOR SPACE

Mathematics 31 Pre-calculus and Limits

by the matrix A results in a vector which is a reflection of the given

THE FUNDAMENTAL THEOREM OF ALGEBRA VIA PROPER MAPS

3. Reaction Diffusion Equations Consider the following ODE model for population growth

Alex, I will take congruent numbers for one million dollars please

LEARNING OBJECTIVES FOR THIS CHAPTER

4. Complex integration: Cauchy integral theorem and Cauchy integral formulas. Definite integral of a complex-valued function of a real variable

Numerical Analysis Lecture Notes

Part 1 Expressions, Equations, and Inequalities: Simplifying and Solving

Student Outcomes. Lesson Notes. Classwork. Discussion (10 minutes)

Sample Induction Proofs

Linear Algebra Notes for Marsden and Tromba Vector Calculus

Georg Cantor ( ):

THE BANACH CONTRACTION PRINCIPLE. Contents

a 1 x + a 0 =0. (3) ax 2 + bx + c =0. (4)

WHAT ARE MATHEMATICAL PROOFS AND WHY THEY ARE IMPORTANT?

OSTROWSKI FOR NUMBER FIELDS

LIES MY CALCULATOR AND COMPUTER TOLD ME

Availability of a system with gamma life and exponential repair time under a perfect repair policy

This unit will lay the groundwork for later units where the students will extend this knowledge to quadratic and exponential functions.

Lecture Notes on Polynomials

Week 13 Trigonometric Form of Complex Numbers

Equations, Inequalities & Partial Fractions

The General Cauchy Theorem

Graphing Rational Functions

MATH 4330/5330, Fourier Analysis Section 11, The Discrete Fourier Transform

MATH 60 NOTEBOOK CERTIFICATIONS

3 1. Note that all cubes solve it; therefore, there are no more

To discuss this topic fully, let us define some terms used in this and the following sets of supplemental notes.

Multiplication and Division with Rational Numbers

Transcription:

COMPLEX ANALYSIS PART 4: MEROMORPHIC FUNCTIONS AND RESIDUES Q What are we going to discuss today? A Before starting a new topic, let us recall what we did so far. Can you briefly describe this in your own words, sparing all details? Q Why do we have to review at this moment? I am eager to learn new things. A In math, new things are often developed from old ones. Other subjects such as physics are all alike in this respect. To understand Einstein s relativity theory, you have to know Newton s mechanics. Besides, learning a math subject is like taking a long journey. From time to time we have to recall the places we have passed in order to find out where we are, to avoid being completely lost. Q I m afraid that I can t remember everything correctly. A Give a try. You may consult your notes if necessary. But don t look up the notes without giving a try. In doing this, conduct a self-evaluation to see how well you can visualize the big picture. Q OK. let me go through this briefly. We begin with basics of complex numbers. Then we move to complex functions and we define a function to be analytic if it is complex differentiable at each point of its domain. This leads to CR-equations, which can be simply put as f/ z = 0. Then we introduce contour integrals and we use Green s formula to prove Cauchy s theorem, which says for an analytic function f Cauchy s integral formula, which says f(z) dz = 0 in a suitable set-up. This theorem is used to establish f(z 0 ) = 2πi C w z 0 dw (z 0 D) ()

for an analytic function in an appropriate set-up. Then we use this formula to deduce several important properties of analytic functions, such as the existence of power series expansion at each point. A You are doing pretty well. Can you elaborate power series expansions a bit? Q It says that if f is an analytic function with an open set U as its domain, and if z 0 is a point in U, then we can write f(z) = a 0 + a (z z 0 ) + a 2 (z z 0 ) 2 +, (2) where the series on the right hand side converges uniformly to f(z) on any closed disk D(z 0 ; R) contained in U centered at z 0. Here the coefficients a n can be described in two ways: a n = f (n) (z 0 ), and a n = n! 2πi w z 0 =R dw. (3) (w z 0 ) n+ Here I simply write z z 0 = R for the circle C z0 ;R (z 0 ; R) oriented in the counterclockwise manner. A What happens if all derivatives f (n) (z 0 ) at z 0, imcluding n = 0, vanish? Q In that case the above power series vanishes and hence f D(z 0 ; R). is zero on the entire disk A This is rather amazing: what happens at a point z 0 can affect the entire neighborhood D(z 0 ; R). Actually, more is true: it can affect the whole domain U as long as U is connected. Let me make this precise: if f is an analytic function on a connected open set U and if f (n) (z 0 ) = 0 (n = 0,, 2,... ) at some point z 0 U, then f is a zero function: f 0. Q How do you prove this? A Take any point z in U. We have to show f(z ) = 0. Because U is connected, we can link up z 0 and z by a curve C. Due to a topological consideration, we know that there is a positive number R > 0 such that z w > R for all z on C and all w not in U. Another topological consideration allows us to find a finite sequence of open disks D(c ; R), D(c 2 ; R),..., D(c N ; R) 2

of radius R such that c = z 0, c N = z and c k c k+ < R ( k N ), that is, each disk contains the center of the next disk. We have seen that f vanishes on the first disk D(c ; R) because c = z 0 and f (n) (z 0 ) = 0 (n 0) by our assumption. Since c 2 belongs to D(c ; R) and since f vanishes on D(c ; R), we have f (n) (c 2 ) = 0 (n 0) and consequently f vanishes on D(c 2 ; R). This argument shows in general that the property f vanishes on D(c k ; R) passes on to the next disk. Now you can see why f(z ) f(c N ) = 0. Q From now on let us assume that f is not identically zero. Thus, in the power series expansion, there is at least one n such that a n 0. Now, what can we say about the local behavior of f at z 0? A Let N be the smallest integer such that a N 0. Thus we have f(z) = a N (z z 0 ) N + a N+ (z z 0 ) N+ + = (z z 0 ) N F (z) where F (z) = a N + a N+ (z z 0 ) + with F (z 0 ) = a N 0. When N, we call z 0 a zero of f and N the order of this zero. If f is a polynomial, then of course z 0 in this case is a root of f and N is the multiplicity of this root. From the expression (z z 0 ) N F (z) we know that the zero z 0 a neighborhood N of z 0 such that z 0 is the only zero of f in N. Q Let me try to prove this. Since F (z 0 ) = a N 0, by the continuity of F that there is a neighborhood N is isolated: there is we know on which F never vanishes. WE are going to show that z 0 is the only zero of f in N as follows. Suppose that z N is a zero of f in N. Then (z z 0 ) N F (z ) = f(z ) = 0 but F (z ) 0. Hence (z z 0 ) N = 0, which gives z = z 0. A Good. Next we consider the reciprocal /f. What is its local behavior at z = z 0? Let us write f(z) = (z z 0 ) N F (z) again, assuming N and F at any point in a neighborhood N does not vanish of z 0. (We may assume that N is an open disk, say N = D(z 0 ; r).) Now /F is an analytic near z 0 and hence we have a power series expansion of /F at z z 0, say /F (z) = b k(z z 0 ) k. Thus f(z) = (z z 0 ) N F (z) = 3 b k (z z 0 ) k N = n N c n (z z 0 ) n,

where c n = b k in case k N = n, or c n = b n+n. Q What are you going to do next? A Next, I split the sum n N c n(z z 0 ) n partial fractions: into two parts. The first part is a sum of n= N c n(z z 0 ) n c N (z z 0 ) N + + c 2 (z z 0 ) 2 + c z z 0. (4) It is called the principal part. The second part is h(z) n=0 c n(z z 0 ) n, which represents an analytic function defined on a neighborhood of z 0. (Our discussion here can be repeated in the future for a meromorphic function with z 0 as a pole.) Q Suppose that, instead of /f, we have g/f, where g is another analytic function defined on U. What happens? A We write down f(z) = (z z 0 ) N F (z) again. Similarly, we write g(z) = (z z 0 ) M G(z), where G(z 0 ) 0. Then we have g(z) f(z) = (z z M N G(z) 0) F (z). When M N, g/f is analytic at z 0. When M < N, we can repeat the above discussion... Q I know how to handle the rest. A Now let us entertain a general situation. Assume that f is a function analytic on U, except at some isolated points called singularities. Let z 0 for f. That z 0 be a point of singularity is isolated means that we can take a disk D(z 0 ; R) centered at this point so that f is analytic at each point of this disk except its center z 0. In a neighborhood of this singularity, we can express f by a Laurent series: f(z) = <n< a n(z z 0 ) n + a 2 (z z 0 ) 2 + a (z z 0 ) + a 0 + a (z z 0 ) + a 2 (z z 0 ) 2 + (5) which converges uniformly in an annulus {z : δ z R}, where δ is any positive number less than R. Q How do you prove (5)? 4

A Laurent series expansions generalize power series expansions. So you expect that the proof is similar. Alas, this is not quite the case. The first step is to guess what a n in (5) are. For power series expansion at a regular point (a point is regular if it is not a singularity) we have two ways to put a n : see (2) and (3) above. (2) is no good for the present situation because z 0 is a singularity of f and hence f (n) (z 0 ) does not make sense. So we guess a n is given by (3). This is a correct guess. From this we can start the proof. But we are not planning to do here because it is a bit technical, and not suitable for our conversation. Q But can you give a rough outline of this? A OK, a very rough outline. I let you fill in the detail. Define a n = 2πi w z 0 =r dw (w z 0 ) n+ where r is any number satisfying 0 < r < R. First we check that a n of the choice of is independent r > 0. This can be achieved by applying Cauchy s theorem to the function f(z)/(z z 0 ) n+ on the annulus {z : r z z 0 r 2 } for arbitrary r, r 2 with 0 < r < r 2 < R. Next, fix a point z in the open disk D(z 0 ; R) with z z 0. Take any r, r 2 > 0 such that 0 < r < z z 0 < r 2 < R and let D be the annulus {w : r < w z 0 < r 2 }. By Cauchy s formula, f(z) = 2πi ( dw = w z 0 2πi w z 0 =r 2 If w z 0 = r, then w z 0 < z z 0 and hence w z 0 =r ) dw w z 0. = w z 0 (w z 0 ) (z z 0 ) = z z 0 (w z 0 )/(z z 0 ) = (w z 0 ) k (z z 0 ) k+. Consequently 2πi = n= w z 0 =r ( 2πi ( ) dw = (w z 0 ) k dw w z 0 2πi w z 0 =r ) (w z 0 ) n dw (z z 0 ) n = a n (z z 0 ) n. w z 0 =r n= (z z 0 ) k+ In the same way we get w z 0 =r 2 = n=0 a n(z z 0 ) n. Done. 5

Q How do you classify singularities? A Back to (5), consider two cases: the case that there exits an integer n 0 such that a n = 0 for all n < n 0 and the case that such n 0 does not exist. In the latter z 0 is called an essential singularity of f. For example, the Laurent expansion of e /z at z = 0 is given by e /z = + z + + n!z n + and hence e /z has an essential singularity at z = 0. In the first case we can pick n 0 such that a n0 0 and a n = 0 for n < n 0. Then, as before, we may write f(z) = (z z 0 ) n 0 F (z), where F is an analytic function with F (z 0 ) 0. Here n 0 negative; otherwise z 0 would not be a singular point. In the first case, z 0 is called a pole of f. Q I ve heard that meromorphic functions are those for which all singularities are poles. Why do we bother to give a name for them? A They are important for several reasons. First, complex functions encountered in practice are mostly meromorphic functions, such as tan z. Second, all meromorphic functions defined on something called Riemann surface S form a field M S : if f, g are meromorphic, then so are f + g, fg and f/g (assuming g 0). This gives a link between field extensions and branched coverings of Riemann surfaces, one from the area of algebra and the other from topology. It relates field automorphisms to covering transformations. But we are not going to touch upon this interesting but advanced aspect of meromorphic functions: we keep our discussion elementary. For our purpose here, meromorphic functions are important because of the residue theorem. The residue calculus allows us to manipulate many intriguing integrals, most of which are inaccessible by elementary calculus. So it is considered to be a major application of complex analysis. Q How should we think of meromorphic functions? A One may think that analytic functions generalize polynomials and meromorphic functions generalize rational functions. We have seen that the quotient of two analytic functions is meromorphic. The converse holds at a local level. Indeed, if z 0 is a pole of a meromorphic function f, we can write f(z) = (z z 0 ) n 0 F (z), where n 0 is a negative integer and F is analytic on a neighborhood N of z 0, which is a 6 is

quotient of f and (z z 0 ) N, where N = n 0 > 0. This indicates that the relation of meromorphic functions to analytic functions is like that of rational functions to polynomials. There is another way to think of meromorphic functions: at a pole z 0 a meromorphic function is an analytic function plus a sum of partial fractions, say P (z; z 0 ), which is an expression of the form given in (4) above, called the principal part of f at z 0. Suppose that f has only finitely many poles, say z 0, z,..., z N. Let R(z) = N P (z; z k), the sum of all principal parts of f. Let F (z) = f(z) R(z). Then g has no poles and hence is analytic. rewrite F = f R as f = R + F, we see that the meromorphic function f with finitely many poles is the sum of a rational function R and an analytic function F. Q In the first year calculus you use partial fractions to compute integrals of rational functions. But why one should put partial fractions in a particular pattern is rarely explained. Now I can tell that we are getting close to a good explanation. A Oh, yes. Suppose that f complex plane C. Write f = R + F is rational. Clearly f has only finitely many poles in the as before. Then F = f R is also a rational function. But F has no poles in C. So F must be a polynomial. Thus f = F + R gives us the partial fraction decomposition of f used in the first year calculus. Q You have mentioned the residue theorem. I am eager to know this important thing. What is it about? A It is about a contour integral f(z) dz, of a meromorphic function f with finitely many poles in D and no poles on. Again write z 0, z,..., z N for these poles and let f = R + F be the decomposition as before. Since F is analytic, we have F (z) dz = 0 in view of Cauchy s theorem. So it remains to consider R(z) dz. Here R is a sum of principle parts. Let us take one of them, say P (z; z 0 ) = N k= a k(z z 0 ) k. Thus it remains to consider integrals of the form (z z 0) k dz. When k =, the answer is 2πi, obtained by applying Cauchy s formula () to f. When k, we can write (z z 0 ) k as the derivative of (z z 0 ) k+ /( k + ) and hence the integral is zero. (If C is a path running from z 0 to z, then C g (z) dz = g(z ) g(z 0 ) for any function g analytic along C. In particular, C g (z) dz = 0 if C is closed.) So we have C P (z; z 0) dz = a C (z z 0) dz = 2πi a. The number a is called the residue of f at z 0 and will be denoted by Res(f; z 0 ). We have shown that P (z; z 0) dz = 2πi Res(f; z 0 ). 7

Now we put things together: f(z) dz = R(z) dz = N P (z; z 0 ) dz = 2πi N Res(f; z k). Thus we obtain the following identity, which concludes the residue theorem: f(z) dz = N 2πi Res(f; z 0). That is, the integral (/2πi) f(z) dz of the meromorphic function f is equal to the sum of all its residues in D. Q After Cauchy s formula the residue theorem is the next important theorem, I suppose. A Quite so. But Cauchy s theorem can be regarded as a special case of the residue theorem. Indeed, the integrand g(z) f(z)/(z z 0 ) in () is a meromorphic function with z 0 as its only pole. We can compute its residue as follows. Write down the power series expansion of f at z 0 : f(z) = n=0 a n(z z 0 ) n. Then we have the Laurent expansion at z = z 0 : g(z) = f(z) z z 0 = a 0 z z 0 + a + a 2 (z z 0 ) + a 3 (z z 0 ) 2 + So Res(g; z 0 ) = a 0 = f(z 0 ). Thus, applying the residue theorem to g, we get Cauchy s formula. By the way, the method of figuring out the residue of g at z 0 here is applicable in many situations. Q I think now I understand the residue theorem well. Tell me its application to evaluation of integrals. A When we try to use the residue theorem to evaluate an integral, we have to take three major steps: find a suitable meromorphic function, choose an appropriate contour, and calculate residues. The second step is often more tricky. In applying to some improper integrals, often we have to move parts of contour to infinity or to shrink some semicircle to a point. Many examples can be found in standard textbooks. Q OK. Let us stop now. 8