Mathematical Analysis T. M.
Apostol
Chapter 3
Mathematical Analysis by Tom M. Apostol Chapter 3: Elements of Point Set Theory: Notes
Let E1 denote the set of all real numbers (the real line). Let E2 denote the set of all complex numbers (the complex plane). An open interval (a, b) is defined as (a, b) = {x  a < x < b}. Closed intervals, halfopen intervals and infinite intervals can be similarly defined. The real line is sometimes referred to as the open interval (∞, ∞). A single point is sometimes considered as a “degenerate” closed interval. Let h > 0, and let x be a given point. The open interval (xh, x+h) is called a neighbourhood with x as centre and of radius h. We denote the neighbourhood by N(x; h), or simply by N(x) if the radius is unimportant. Let S be a set in E1 and assume that x ∈ S. Then x is called an interior point of S if there is some neighbourhood N(x) all of whose points belong to S. Let S be a set in E1. S is called an open set if every point of S is an interior point of S. Thus an open set is such that each of its points can be enclosed in a neighbourhood which is completely contained in the set. The simplest kind of open set is an open interval. The empty set is also open, as is the real line E1. The union of any collection of open sets is an open set. The intersection of a finite collection of open sets is open. Note that arbitrary intersections will not always lead to open sets. The union of a countable collection of disjoint open intervals is an open set and, remarkably enough, every open set on the real line can be obtained in this way. The structure of open sets in E1. A set of points in E1 is said to be bounded if it is a subset of some finite interval. Let S be an open set in E1 and let (a, b) be an open interval which is contained in S, but whose endpoints are not in S. Then (a, b) is called a component interval of S. If S is a bounded open set, then (i) each point of S belongs to a uniquely determined component interval of S; and (ii) the component intervals of S form a countable collection of disjoint sets whose union is S. From this, it follows that an open interval in E1 cannot be expressed as the union of two disjoint open sets when neither set is empty. Every open set in E1 is the union of a countable collection of disjoint open intervals. Accumulation points and the BolzanoWeierstrass theorem in E1. Let S be a set in E1 and x a point in E1, x not necessarily in S. Then x is called an accumulation point of S, provided every neighbourhood of x contains at least one point of S distinct from x. If x is an accumulation point of S, then every neighbourhood N(x) contains infinitely many points of S.
. 3. x+y ≤ x + y. The set of x such that xx0 ≤ r form a closed sphere. The set of all ndimensional points is called the ndimensional Euclidean space and is denoted by En. the ndimensional open interval (a... xn)  ak < xk < bk. .. If S is open. or length... An ordered set of n real numbers (x1. and x = 0 iff x = 0.... Let a = (a1.1. If ak < bk for every k. Thus a closed interval is a closed set. .. an) and b = (b1. . if S is closed. x = (x1.. (These last two sets are also open!) A set which is not closed need not be open. n = 1.. is not a closed set because it does not contain its endpoints.2.... both of which are accumulation points of the set. xn)... however. A set which has no accumulation points is automatically closed and. b] = {(x1. It follows from the above that the union of a finite collection of closed sets is closed and that the intersection of an arbitrary collection of closed sets is closed... .. such that ak ≤ bk for each k = 1. has 0 as an accumulation point. The empty set is also closed. δk. Extensions to higher dimensions. which is neither open nor closed. b).. The usual vector rules apply. 2.. for example. n 1 x > 0.. (2) The set of rational numbers has every real number as an accumulation point. By an open sphere of radius r > 0 and centre at the point x0 in En.g. 2.k = 1.. defined by δk. the boundary of the sphere being the set of x with xx0 = r. xx0 < h... The distance between x and y is given by xy = S i=1 (x i − y i ) 2. Let n > 0 be an integer. If S is open. n}. n}.
. 2. where xk is the kth coordinate of the point or vector x.. Closed sets in E1.. Points or vectors will be denoted by single boldfaced letters. Conversely.j is the “Kroneker delta”.. . as is the whole real line E1. The absolute value. we mean the set of all x in En such that xx0 < r. bn) be two distinct points in En. r) if r is the radius.. or norm of x n is given by x = x 2 + . n). b) is the set (a.. still makes sense if x and x0 belong to higher dimensional spaces. .n) (k = 1.. An open interval. then the complement E1—S is closed. e.. every finite set is closed. x2. where δk. An open sphere with centre at x0 is called a neighbourhood of x0 and is denoted by N(x0) — or by N(x0. .. The ndimensional closed interval [a.. . the halfopen interval (a. 2. as. Note: Unit coordinate vectors: uk = (δk. b] is an accumulation point of the set of numbers in the open interval (a. k = 1. Examples: (1) The set of numbers of the form 1/n.Theorem 313 (BolzanoWeierstrass): If a bounded set S in E1 contains infinitely many points. in particular. . yn) be in En. n.. . (3) Every point of the closed interval [a. .. δk.. b) = {(x1.. The definition of a neighbourhood.j = 0 if k ≠j. xn) and y = (y1. Generalisation: If S is closed. x2. and δk.. k = 1.. . xn) is called an ndimensional point or a vector with n components.. A set is called closed if it contains all its accumulation points. b] is defined to be the set [a. + x 2 . The corresponding closed sphere is denoted by N 0). then there is at least one point in E1 which is an accumulation point of S. then the complement of S (relative to any open set containing S) is open. b].. . The open sphere with its centre removed is (x called a deleted neighbourhood of x0 and is denoted by N’(x0). 2.. xy = yx. then the complement of S (relative to any closed set containing S) is closed. xn)  ak ≤ xk ≤ bk. then E1—S is open. Let x = (x1.
(a. Assume that S ⊂ A ⊂ En. for some r > 0. If A is open and if S is closed. we have S ⊂ N(0. b2) × . Any ndimensional open sphere and any ndimensional open interval is an open set in En. The union of a finite collection of closed sets in En is closed and the intersection of an arbitrary collection of closed sets in En is closed. that is. A collection F of sets is said to be a covering of a given set S if S ⊂ ∪A∈F A. In fact. Assume that S ⊂ En. bk). If A is closed and if S is open. bn) of the n onedimensional open intervals (ak. the first quadrant.. The set S is said to be open if each of its points is an interior point. for example. then every neighbourhood N(x) contains infinitely many points of S. then A—S is closed. if N’(x) ∩ S is not empty. it makes no difference whether a neighbourhood is taken to be a “sphere” or an “interval” (we will use spheres). Examples of open sets in the plane are: the interior of a disk. Caution: an open interval (a. × (an. The HeineBorel covering theorem. and assume that x ∈ S. If x is an accumulation point of S.. Theorem 329 (BolzanoWeierstrass): If a bounded set S in En contains infinitely many points. The union of an arbitrary collection of open sets in En is an open set in En. and in what follows. that is. An open interval in En could then be called a neighbourhood of any of its points. The interior of S is the collection of its interior points. A set which is closed in E1 is also closed in En for n > 1. r). b) = (a1. Then x is called an accumulation point of S if every neighbourhood N(x) contains at least one point of S distinct from x. b1) × (a2. x ∈ En. A set S in En is said to be bounded if S lies entirely within some sphere. no subset of E1 (except the empty set) can be open in E2. An ndimensional closed interval is a closed set. We begin by defining a covering of a set. The collection F is also said to cover S. then there exists at least one point in En which is an accumulation point of S. b) can be considered as the “cartesian product” (a. If F is a collection of open sets. Let S be a set of points in En. A set S in En is closed iff En—S is open. the whole space. A set S in En is said to be closed if it contains all its accumulation points. because such a set can contain no twodimensional neighbourhoods. then F is called an open covering of S. Then x is called an interior point of S if there exists a neighbourhood N(x) ⊂ S.Thus.
. Examples: A closed sphere in En is a closed set. then A—S is open. and the intersection of a finite collection of open sets in En is an open set in En. b) in E1 is no longer an open set when it is considered as a subset of the plane.
Assume that x ∈ En. n+2). (2) The real axis E1 is covered by the collection of all open intervals (a. Every open set in E2 of the form {z z > r > 0} is called a neighbourhood of ∞.) is an open covering of the interval 0 < x < 1.
. By the extended complex number system E2* we shall mean the union of the complex plane E2 with a symbol ∞ which satisfies the following properties: (1) If z ∈ E2. Then the intersection ∩k=1∞ Qk is closed and non empty. +∞) is called a neighbourhood of +∞. a) is called a neighbourhood of ∞.} denote the countable collection of neighbourhoods in En having rational radii and centres at points with rational coordinates. 4... namely all intervals of the form (n. (ii) Each set Qk is closed and Q1 is bounded. and it is +∞ if the set is not bounded above. Let G = {A1. The converse result is as follows: Let S be a compact set in En. We have (i) every infinite subset of S has an accumulation point in S. but z ≠ 0. z/∞ = 0. 3.. E1* = [∞. where n runs through the integers. Every open interval (a. ∞]. The HeineBorel theorem states that every closed and bounded set in En is compact. Theorem 338 (HeineBorel) Let F be an open covering of a closed and bounded set A in En. b). then we have x(+∞) = +∞ and x(∞) = ∞. That is. x+(∞) = ∞. then we have x+(+∞) = +∞. . However. Then there is a countable sub collection of F which also covers A. we have x ∈ Ak ⊂ S for some Ak in G.. and every open interval (∞. (3) If x < 0. and z/0 = ∞.. E1 = (∞.. Then a finite sub collection of F also covers A. A2. and x/(+∞) = x/(∞) = 0. . x(∞) = +∞. (2) If z ∈ E2.Examples: (1) The collection of all intervals of the form 1/n < x < 2/n (n = 2. then we have ∞ < x < +∞.. and let F be an open covering of A. (∞)+(∞) = (+∞)(∞) = ∞. then we have x(+∞) = ∞ and x(∞) = +∞. (4) (+∞)+(+∞) = (+∞)(+∞) = (∞)(∞) = +∞.). By the extended real number system E1* we shall mean the union of the set of real numbers E1 with two symbols +∞ and ∞ which satisfy the following properties: (1) If x ∈ E1. Then at least one of the neighbourhoods in G contains x and is contained in S. Let {Q1. . then z(∞) = ∞. Note that ∞ is not needed here because no ordering is involved with complex numbers. 2. (3) ∞+∞ = (∞)(∞) = ∞. it contains a countable covering of E1. and let S be an open set in En which contains x. Note that E1* does not satisfy all the axioms for the real number system. (5) If x ∈ E1. then z+∞ = z∞ = ∞. This is an example of a countable covering. 3.} be a countable collection of non empty sets in En such that (1) Qk+1 ⊂ Qk (k = 1. (2) If x > 0. (ii) the set S is closed and bounded. Assume that A ⊂ En. Q2. x(+∞) = ∞. Every set in E1* has a sup — it is finite if the set is bounded above. A set S in En is said to be compact iff every open covering of S contains a finite sub collection which also covers S. This covering is not countable. ∞). ..
b). x+h). x ∈ (a. so that xh < y < x+h. x+h).Chapter 3: Selected Exercises
31. What we have to show is that one of these neighbourhoods is entirely contained within the interval (a. Definition 2: Let S be a set in E1 and assume that x ∈ S. S is an open set if every point of S is an interior point of S. To prove that an open interval in E1 is an open set. Definition 1: An open interval (a. we can do the following manipulation: h ≤ ½(xa) ⇒ xh ≥ x½(xa) ⇒ xh ≥ ½(x+a) ⇒ xh > ½(a+a) (because a < x) ⇒ xh > a proving part (1) From (ii) above. we need to prove that every point in an open interval is an interior point. where h > 0. b) is defined by (a. so ( that h = ½min(xa. b) for every x and for some h > 0. x+h) ⊂ (a. From (i) above. and that (2) x+h < b. b). which is what we want to prove. b).
xh x x+h ( )
) b
Let us now analyse what we know about the point y. x+h). we also have y ∈ (a. and that (ii) h ≤ ½(xb). we can do the following manipulation: h ≤ ½(xb) ⇒ x+h ≤ x+½(xb) ⇒ x+h ≤ ½(3xb) ⇒ x+h < ½(3bb) (because x < b) ⇒ x+h < b proving part (2)
. we can rewrite h as h = ½min(xa. From this definition. Consider an arbitrary open interval (a. b) = {x  a < x < b}. Let x be any point in this interval. all of whose points belong to the open interval. we have to show that for every y ∈ (xh. xb). we must prove that every point in an open interval has an associated neighbourhood. we know that y (by definition) sits in the interval (xh. A neighbourhood of x is given by the interval (xh. Definition 3: If S is a set in E1. we see that (i) h ≤ ½(xa). all we have to show is that (1) a < xh. Prove that an open interval in E1 is an open set and that a closed interval is a closed set. Let us choose h to be given by half the value of the minimum distance from x to each of the two endpoints. we have to show that (xh. To do this. To do this. First of all. Answer: Let us first look at some definitions.e. Because we know that x > a a and that x < b. i. Then x is called an interior point of S if there is some neighbourhood N(x) all of whose points belong to S. In order to show that a < y < b. b). bx).
Because S contains all its accumulation points (of which there are none!).). a is an accumulation point of S so that the set of accumulation points of S is given by the set {x  a ≤ x ≤ b}. Answer: Let us first remind ourselves of the definition of an accumulation point: Let S be a set in E1 and x a point in E1. . . (h) All numbers of the form (1)n/[1+(1/n)]. as the complement of a closed interval is an open set. then we have proved that an open interval is an open set. (g) All numbers of the form (1/n) + (1/m). consider an arbitrary neighbourhood N(x) of x. . so that E1—[a. (b) Let S = (a. x not necessarily in S. Determine all the accumulation points of the following sets in E1 and decide whether the sets are open or closed (or neither). 2. Consider an arbitrary point x ∈ S.. . b]. n = 1. x+h) ⇒ y ∈ (a. ∞). 32.). with h > 0.5). 2. b]. 2. the union of two open intervals.). a)∪(b. Taking an arbitrary neighbourhood N(x) of x.. n = 1. S contains no accumulation points because every point x has a neighbourhood (x½.. (c) All numbers of the form 1/n. x+h). (a) All integers. (d) All rational numbers. b). Justification: to show that each point x = 2n is an accumulation point of S (n = 1. 2.). . then the point y = 2n+5l will be in N(x) and in S. If we now choose an integer l so that 5l ≤ h/2 (and there will always be such an l. then a closed interval must therefore be a closed set. QED. Further. (m.... To show that a closed interval is a closed set. b] = (∞. S is not open because b is not an interior point (in the neighbourhood (bh. m = 1.. Let us consider an arbitrary closed interval [a. We therefore conclude that all the points in S are accumulation points. b] is an open set.). So. Further.. Further. Claim: all numbers of the form x = 2n (n = 1. (a) Let S = {x  x ∈ Z}. (m.. the points in (b. 2. (m. then it is clear that the point max{xa/2.. with h > 0. the union of two open sets is also an open set (Theorem 3. 2. b+h) do not belong to S — but do belong to the neighbourhood.) are accumulation points of S. 3. It follows that b is not an interior point). x+½) which contains no other integers.. and hence not an open set (because of exercise 35).
. 2. (e) All numbers of the form 2n + 5m. . we use the following trick: the complement of an open set is a closed set. x+h). N(x) = (xh. 2. it follows that S is a closed set. . .).. (e) Let S = 2n+5m (n. 2. n = 1.. (n = 1. with any h > 0. The complement of this interval is given by E1—[a. S is not closed because it doesn’t contain the accumulation point a.. Then x is called an accumulation point of S provided every neighbourhood of x contains at least one point of S distinct from x. (b) The interval (a. (f) All numbers of the form (1)n + (1/m).Because we have shown that y ∈ (xh. b+h)... N(x) = (xh. xh/2} will always be in S and in N(x) so that every neighbourhood of a point in S will always contain at least one other point from S.).. which we know are open sets by the above. b+h) ⊂ (bh. (n = 1. so that x is not an accumulation point.) and of the form y = 5m (m = 1. . given by l = ceiling(log(n/2)/log(5))). b]..
we see that S is an alternating series. To start with.) are accumulation points of S.. b]. no neighbourhood of the largest element of S. 5/6. which we shall denote by T. 3/4. and so every neighbourhood of this point (i. Take the point in S given by setting n = 1 and m = 1. namely the point 7/10.. and the other half tending to 1. 2 /3. . we want to show that C can be written as C = ∩n=1∞ On.
From this series representation.e. 2/3. and it can be expressed as the intersection of a countable collection of open sets as follows: S = ∩n=1∞ (x1/n. There is one other accumulation point of S: zero (every neighbourhood of zero will contain a point from S if n and m are large enough).) are accumulation points of S. where every On is an open set. This closed set can be expressed as the intersection of a countable collection of open sets as follows: T = ∩n=1∞ (a1/n. Answer: Let us first consider some examples of closed sets and the intersection of open sets that is equivalent to the closed set in question. S = {1/2. 7/10+h) for some h > 0) will contain elements which are larger than 7/10. Show that every closed set in E1 is the intersection of a countable collection of open sets. . . x+1/n). . which we have shown in a previous exercise to be a closed set. To prove that it is not. Because S does not contain its two accumulation points (1 ∉ S and 1 ∉ S).. and because 0 ∉ S. it is sufficient to find a single point in S which does not have a neighbourhood all of whose points belong to S. Now consider a closed interval [a. the interval (7/10h. 36. and so the point 7/10 ∈ S is not an interior point — so that S is not an open set. 2. Writing S out as a sequence. By the same argument as in part (e). We therefore have some evidence that the statement that we are trying to prove is correct... 2. This set is closed because it has no accumulation points.}.e. There are no other accumulation points in the set. so that 2/3 is not an interior point of S.. 2. consider the set S consisting of a single point x ∈ E1. because the set has elements which are arbitrarily close to 1 and 1. 4/5.It therefore follows that all points of the form x = 2n (n = 1.). and so S is not an open set. (h) Let S =
(−1) n 1+ 1 n
(n = 1. then S is not a closed set.. 1 and 1 are accumulation points of the set S. elements which cannot be expressed in the form 2n+5m. Therefore. one half of the series tending to 1. b+1/n).. i. The question remains as to whether S is an open set. A similar argument can be used to show that all points of the form y = 5m (m = 1. will contain elements just from S. We now want to show that every closed set C can be expressed as the intersection of some countable collection of open sets. then S is not a closed set. Clearly this is the largest element of the set S.
.
If x ∈ C. When we take the intersection of all the Ej’s. then we refer to the proof of claim 2 below to show that x ∈ F. x+1/j). We have already shown that F is the intersection of an arbitrary collection of open sets. End of Proof. being that every interval Ej will contain the point x. Claim 1: F contains all of C’s accumulation points. then we will also show that F = C because for a closed set C. Proof of Claim: Consider an arbitrary accumulation point of the set C. Conclusion: we have expressed an arbitrary closed set C as the intersection of a countable collection of open sets. It follows that y belongs to each Ej. every ( neighbourhood N(x) of x will contain a point y ∈ C. But for every E 1 such y. then z belongs to all of the sets Ej. D is the ‘1/j interval’ of a point from C. If the accumulation point x is not in C. j = (y1/j. j is an open set because it consists of an open interval. we “close E3 in” on the point x (see the diagram on the right). and j ∈ R>0. show that F = C. End of Proof. To complete the proof. then there must be a set Dx. i. If z belongs to all of the sets Ej. QED. then we reach the required conclusion. C ⊂ F. To do this. so that F ⊂ C∪C’. there exists a corresponding interval Dy. then there is a set Dy. which we know (from exercise 31) to be an open set. It follows that F is the intersection of a countable collection of open sets. Each Ej is an open set because it is an arbitrary union of open sets. It follows (by the definition of Di. If z ∈ F. as C= C∪C’ (see exercise 312 for this definition). namely the set F. if we can . Proof of Claim. the union over all points of C of ‘1/j intervals’ of all those points. Because y ∈ C ⇒ y ∈ F. j. and so it is in the right form to be considered as the solution to this exercise. then C∪C’ ⊂ F.e. We will in fact show that F = C i. Consider a point y ∈ C. If y ∈ C. then we first need to show that F contains all of C’s accumulation points and all of the points from C. which (by theorem 35) is an open set. j. and so z ∈ F ⇒ z ∈ C∪C’. j = (x1/j. QED. and so y belongs to F = ∩j=1∞ Ej. the crucial feature etc. Finally. so that y ∈ F. If we can do this. We see immediately that Di. j. In other words. say the point x. we have (looking at part (f) of exercise 312) C = C Therefore. The question now arises as to whether we have the conclusion F = C. Now let us define Ej to be the following expression: Ej = ∪i Di. where j ∈ R>0. Therefore. we need to show that F ⊂ C∪C’. let us define F to be the set given by F = ∩j=1∞ Ej.To start with.
x Following the definition of an accumulation point. j and by the proof of claims 1 and 2 above) that z ∈ C∪C’.
. j = (di1/j.
Claim 2: F contains all of the points from C. Conclusion: Because we have shown that C ⊂ F and that C’ ⊂ F. show . where di ∈ C. let us define D to be given by Di.j which in turn is also E2 in every Ej. that F is the closure of the set C. F will also contain the point x. di+1/j). then we must take a little more care.e. where Ej = ∪i Di. y+1/j). where j ∈ R>0. then C ⊂ F. whether they are in C or not. where Ej = ∪i Di.
which implies that either (1) x ∈ S’ ⊂ S’∪T’.
. then either x ∈ S’ ⊂ (S∪T)’. S must contain all its accumulation points. so that every neighbourhood of y contains a point z from T. (b) If S ⊂ T. and so S’ ⊂ T’. (d) (S)’ = S’.39. then it follows that x must be an accumulation point of the set S. which implies that either (1) x is an accumulation point of S. Answer: Let I be the set of all interior points from S. (S’)’ ⊂ S’. i. so that x ∈ (S’)’ ⇒ x ∈ S’. But S’ ⊂ S as S = (S∪S’).) Show that (a) S’ is a closed set. let S’ denote the set of accumulation points of S. Answer: (a) If x ∈ En and x ∈ (S’)’. (i) If x ∈ (S∪T)’. so that x ∈ S’∪T’ ⇒ x ∈ (S∪T)’. In all cases. QED. all the points in A must be interior points. (e) To be a closed set. QED. then either y ∈ S. then x ∈ S ⇒ x ∈ T. It follows that all neighbourhoods N(y) of y will contain a point z ∈ S. QED. If S is a set in En. From (d). another accumulation point of S. we must prove that (i) (S∪T)’ ⊂ S’∪T’. which is also (because S ⊂ T) a point from T. then x is an accumulation point of the set S’. or (3) x ∈ S’∩T’ ⊂ S’∪T’. then S’ ⊂ T’. and therefore the interior of a set S in En is an open set. Because all the points in our set I are interior points of S. Note that in order to do this. If x is an accumulation point of the set S’. Show that the interior of a set S in En is an open set. QED. (b) If S ⊂ T. which implies that y ∈ T. we must have Ε < max(dist(x. we know that the set of the accumulation points of S is the set S’.e. εdist(x. or y ∉ S. then y is an accumulation point of the set S. and so (S’)’ ⊂ S’ as required. and that (ii) S’∪T’ ⊂ (S∪T)’. (c) (S∪T)’ = S’∪T’. or (3) an accumulation point of both S and T. (The set S’ is called the derived set of S and S is called the closure of S. But if y is an accumulation point of S. Bringing the two cases together. S = S. (e) S is a closed set. x ∈ (S∪T)’ ⇒ x ∈ S’∪T’. we conclude that Sis a closed set. it follows that I is an open set. QED. then y ∈ T’.y)). and so y is also an accumulation point of the set T. and so y is an accumulation point of the set T. So as S contains all its accumulation points. that is. (c) To prove that (S∪T)’ = S’∪T’.y). then x is an accumulation point of the set S∪T. or (2) an accumulation point of T.e. (ii) If x ∈ S’∪T’. then every neighbourhood Nε(x) of x contains at least one other point y ∈ S’ (y ≠ x). Pick a N Ε(y) y neighbourhood NΕ(y) such that NΕ(y) ⊂ Nε(x) and x ∉ NΕ(y). or (2) x ∈ T’ ⊂ S’∪T’. so that (S∪T)’ ⊂ S’∪T’. and so S’∪T’ ⊂ (S∪T)’. For a set A to be an open set. we see that if y ∈ S’. but every neighbourhood of y contains a point from z from S. (by definition) I is the interior of S. (d) (S)’ = (S∪S’)’ = (from (c)) = S’∪(S’)’ = S’∪T (where T ⊂ S’) = S’. 312. (f) S is closed if. and only if. or x ∈ T’ ⊂ (S∪T)’. and let S = S∪S’. i. QED. Now if y ⊂ S’. So as x N ε(x) every neighbourhood N(x) of x contains a point z ∈ S (with z ≠ x).
QED. every neighbourhood of the point z we converge onto will contain one of the regions which we know to be uncountable (i. 1). . where n is a finite number. Therefore. 3. 1/ng]. i.. (n = 2. where α > ng. S’ will consist of elements from S. and so S is closed by definition. then at least one of these ‘regions’ will have to be uncountable as well. .) 319. But G (by construction) has no elements of this type. and repeat the above with this set (i. 2/3). If S is closed.. < ng.. say g elements. 1). QED. so that the set G covers no part of the interval (0. A point x in En is said to be a condensation point of S if every neighbourhood N(x) has the property that N(x)∩S is not countable.e. the set G must contain elements of the form (1/α. Therefore. 2/n1). n. i. We must now show that G does not cover the interval 0 < x < 1. then there exists a point x in S such that x is a condensation point of S.. If the set S is not countable.e. Answer: Let us first consider some elements from F. i = 1. QED. (1/n2. Only If.e.. n = 2: we have the interval (½. 315. 1/ng] ⊂ (0.
S S1 S2 S3 Sn
Answer: Divide the set S up into n (possibly partially overlapping) sets Si. n = 3: we have the interval (1/3. and so S = (S∪S’) = (using S’ ⊂ S) = S in this case. 1) but 1/ng ∉ ∪G. 2/n2). To cover at least part of this interval. z is a condensation point. and is again a closed set by definition. so that S’ ⊂ S. is an open covering of the interval 0 < x < 1. (1/ng. We therefore conclude that the set G does not fully cover the interval (0. Justification: with every subdivision. Show that if S is not countable.. Let us denote by G any finite subcollection of F. S contains no accumulation points. 4.
. 1). or S’ consists of points from S. 2/ng)}. S’ ⊂ S.(f) If. Because G is a finite subcollection..e. To do this. Conclusion: given an uncountable set S. and ∪i Si = S. it is sufficient to show that G does not cover the interval (0. If we continue to do this.e. i. If we order the elements in G with respect to the number n. (Note: a quicker proof is to show that 1/ng ∈ (0.. then it follows that we can write G in the following form: G = {(1/n1. then either S’ = φ. 2. 2/α). The collection F of intervals of the form 1/n < x < 2/n. it will have a finite number of elements... 1/ng]. then this process will converge onto a point z ∈ S which will be the condensation point we are looking for. and ng ∈ R (ng < ∞). then it contains all its accumulation points. In the limit of this process. every neighbourhood of this point z has the property that N(z)∩S is not countable. etc.). divide it up into a finite collection of smaller sets. some region Sj ⊂ N(z)). where n1 < n2 < . Assume that S ⊂ En. at least one of which will be uncountable).
Choose one of the sets Si which is uncountable. so that again S contains all its accumulation points. where (0. we can always find a condensation point z ∈ S. If S = S. we find a smaller and smaller “region” which is uncountable. or else we will have a finite collection of countable sets and thus S will be countable — and we will have a contradiction. Show (without using Theorem 340) that no finite subcollection of F covers the interval 0 < x < 1. .
(2) We know that S is not countable. N1(z) cannot z N 1(z) possibly be countable. and so T has no accumulation points that are not in T. Because z is not a condensation point. it follows that no point in T is an isolated point. But if T is the set of condensation points of S. it follows that not all neighbourhoods N(z) are uncountable. It follows (by contradiction) that S—T is countable. Answer: (a) If S—T was not countable. Pick one such neighbourhood. we use three pieces of information: (1) From set theory. It follows that all the neighbourhoods of the point z must be uncountable. QED (b) The set S∩T is the set of condensation points which are in S. part (a)). Pick a neighbourhood of z1 which is contained in N1(z). QED. We know that every point x ∈ T is a condensation point. so that every point x ∈ T is an accumulation point. Show that (a) S—T is countable. (d) T contains no isolated points. (c) By definition. If S—T is countable. say N1(z). QED. N(x)∩S is not countable). not an accumulation point. say the neighbourhood N2(z1) (We can always do this — see the answer to N 2(z 1) z1 exercise 312. Assume that z is an accumulation point of the set T — with z ∉ T. Let T denote the set of condensation points of S. we know that S∩T = S—Tc.e.
Note that Exercise 319 is a very special case of (b).
. Assume that S ⊂ En and assume that S is not countable. then S—Tc must be non countable. QED (d) Because we concluded in (c) that every point in T is an accumulation point. But the neighbourhoods of z1 are uncountable (by the definition of a condensation point). By the definition of an accumulation point. then S—T contains no condensation points. then by 319 it would contain a condensation point. N1(z) contains an element z1 ∈ T. This implies (by Exercise 319) that the neighbourhood contains another condensation point y ∈ T (N(x)∩S uncountable ⇒ ∃ a condensation point y ∈ N(x)∩S). contradicting our assumption that N1(z) was countable. so that the point x is an accumulation point of the set T. T is closed if it contains all its accumulation points. (c) T is a closed set. This applies for all condensation points. To do this. We must therefore prove that the set of condensation points in S is not countable. (b) S∩T is not countable. so that every neighbourhood N(x) of a point x ∈ T is not countable (more precisely. Using (1). It remains to show that T has no accumulation points that are not in T. Because N2(z1) is uncountable. and so we must have z ∈ T. i.320. and if S is not countable. it follows that S∩T must also be non countable. and (3) We know that S—T is countable.
321.e. [Hint: Use Exercise 320. we know that S—T is countable. Answer: Let T be the set of condensation points of S.e.]. i. Also from 320. where A is a perfect set and B is a countable set (CantorBendixon theorem). S is the union of all the condensation points of S together with the points of S which are not condensation points. A set S is said to be perfect if S = S’. and B = S—T is countable. we know that all the points in T∩S are accumulation points. Show that if F is an uncountable closed set in En. S = (T∩S)∪(S—T) is the required form. (Note: S’ is the set of accumulation points of S). QED. From 320. we have T∩S = (T∩S)’. It follows that T∩S is a perfect set.
. We can therefore write S = (T∩S)∪(S—T). if S is a closed set which contains no isolated points. that is. then F can be expressed in the form F = A∪B. It follows that T∩S is the set of condensation points in S. Therefore. i. and that S—T is the set of points in S which are not condensation points. where A = T∩S is a perfect set.
then this process will converge onto a point z ∈ S which will be the condensation point we are looking for. If we continue to do this. Adjusted answer: Answer: Divide the set S up into a countable collection of (possibly partially overlapping) sets Si. Justification: with every subdivision. every neighbourhood of the point z we converge onto will contain one of the regions which we know to be uncountable (i. at least one of which will be uncountable). some region Sj ⊂ N(z)). Choose one of the sets Si which is uncountable. or else we will have a countable collection of countable sets and thus S will be countable — and we will have a contradiction. z is a condensation point. QED.e.
. then at least one of these ‘regions’ will have to be uncountable as well. i. we find a smaller and smaller “region” which is uncountable.e. every neighbourhood of this point z has the property that N(z)∩S is not countable. Conclusion: given an uncountable set S. we can always find a condensation point z ∈ S. and repeat the above with this set (i. If the set S is not countable. Therefore.Possible Further Work / Evaluation
Exercise 319: I could make a slight adjustment in the answer due to the comment in Draft 2 — to divide the set S up into a countable collection of sets instead of into a finite collection of sets. In the limit of this process. so that ∪i Si = S.e. divide it up into a countable collection of smaller sets.
Chapter 4
.
defined by the equation f(x) = (f1(x).. there is an integer N such that whenever n > N. Note 2: If limx→a f(x) exists (finite or infinite). Strictly speaking. If we have several realvalued functions.. there is another number δ > 0 such that whenever 0 < xx0 < δ. The parts of the definitions concerned with ε can also be rephrased in neighbourhood terminology.. This will usually be clear from the context but.
. then the symbolism xda f(x) = b is defined to mean the following: For every neighbourhood N(b) ⊂ Ek. . For example. then f(x) will be as near to A as desired. the inequality f(x)A < ε states that f(x) ∈ N(A. we see that both above definitions involve the same principle. and b is called the righthand limit at a. If f is a realvalued function defined at a point x = (x1. it is extremely convenient to introduce a vectorvalued function f.
If we use neighbourhood terminology. To say that 0 < xx0 < δ means that x is in a neighbourhood of x0. Also. ε). we will write lim f(x) = b to emphasise the fact more explicitly. we require that a be an accumulation point of S in order to make certain that the intersection N’(a)∩S will never be the empty set... Also. similarly. but x ≠ x0. An xda
x`S
important special case of this occurs when S is an interval in E1 having a as its left endpoint. x is in a deleted neighbourhood N’(x0). its value is uniquely determined. there exists a neighbourhood N(a) ⊂ Em such that x ∈ N’(a)∩S implies f(x) ∈ N(b). ε) and. The statements “x sufficiently near to x0” and “f(x) will be as near to A as desired” are made mathematically precise by the following type of definition: the symbolism lim 0 f(x) = Ameans that for every xdx number ε > 0. then f(x)A < ε. xm) in Em. (lefthand limits xda x`S are similarly defined). xnA < ε means that xn ∈ N(A. . we should somehow indicate the fact that the limit just defined depends on the set S through which x is allowed to range.. If a is an accumulation point of S and if b lim ∈ Ek. . The basic principle involved in both definitions of limit is that for every neighbourhood N(A) there must exist a neighbourhood N(x0) such that x ∈ N’(x0) implies f(x) ∈ N(A). we use either f(x1. then xnA < ε. Note 1: we write x ∈ N’(a)∩S rather than x ∈ N’(a) in order to make certain that x is in the domain of f. fk defined on a common subset S of Em.Chapter 4: The Limit Concept and Continuity: Notes
Let us begin with the equation lim xdx0xn = A.. which is meant to convey the idea that when x is sufficiently near to x0.. xm) or f(x) to denote the value of f at that point. Our definition of limit now assumes the following form: Let f be a function defined on a set S in Em and let the range of f be a subset T of Ek.. if x ∈ S. to say that n > N is the same as saying that n is in a deleted neighbourhood of +∞.
lim We write ndº xn = A to mean that for every ε > 0... . say f1. fk(x)). lim We then write lim f(x) = xda+ f(x) = b.. That is. if necessary.
The above theorem tells us that S can have no further accumulation points. such lim that each term xn ≠ a but such that ndº xn = a. when we deal with continuous functions. There exists a point y in Ek such that limn→∞ xn = y if.. Theorem 47 (Cauchy condition for functions). If f is continuous at every point of S. (ii) limx→a f(x) = f(a). x2. there is a neighbourhood N(a) such that x and y in N’(a)∩S implies f(x)f(y) < ε. the limit symbol may be interchanged with the function symbol. and let a be an accumulation point of S. S has exactly one accumulation point. say a = limn→∞ xn. it follows at once from the definition of limit that a is an accumulation point of S. Theorem 46 (Cauchy condition for sequences). in which case the accumulation point is also the limit of the sequence. and only if. We say that f is continuous at the point a provided that (i) f is defined at a. we say f is continuous at a provided only (i) holds. (iii) xda f(x)/g(x) = A/B if B ≠ 0. such that limn→∞ xn = a. Let f be defined on a set S in Em. Let f be defined on a set S in Em with function values in Ek. (ii) xdaf(x)g(x) = AB. Therefore. and only if. without knowing its value in advance. There exists a point b in Ek such that limx→a f(x) = b if. the following condition is satisfied: For every ε > 0. . xdag(x) = B. If a is not an accumulation point of S.. Let {xn} be an infinite sequence whose terms are points in Ek. we say f is continuous on the set S. each defined on a set S in En. with function values in E1 or lim lim in E2. Observe also that continuity at a means that for every neighbourhood N(f(a)). Then we have (i) If limx→a f(x) = b. If S is infinite. Note: Suppose a sequence {xn} has a limit. then all these sequences have the same limit (call it y) and also limx→a f(x) exists and equals y. Definition 49. we can write part (ii) of the lim lim above definition as follows: xdaf(x) = f( xdax). Thus. there exists an integer N such that n > N and m > N implies xnxm < ε.Theorem 44. The following theorem gives us a condition (called the Cauchy condition) which enables us to determine. the following condition holds: For every ε > 0. Then lim lim lim we also have (i) xda [f(x)±g(x)] = A±B. (ii) Conversely. Continuity. there exists a neighbourhood N(a) such that f[N(a)∩S] ⊂ N[f(a)]. whether such a point b exists. Let a be an accumulation point of S. Let {xn} be an infinite sequence whose terms are points of S. if for each such sequence we know that limn→∞ f(xn) exists. Let f and g be two functions. the function values being in Ek.
. Let a be an accumulation point of a set S in Em. it was assumed that the limiting value b was given. a sequence {xn} whose range S is infinite has a limit if. and only if. In the definition of the statement limx→a f(x) = b. Theorem 45.} denote the range of the sequence. and let S = {x1. and let a be an accumulation point of S. Let a be an accumulation point of S and assume we have xdaf(x) = A. Let f be defined on a set S in En with function values in Em. It is convenient to note that whenever f is continuous at a. Then there exists an infinite sequence {xn} whose terms are distinct points of S. then limn→∞ f(xn) = b.
. that is to say. and in the case there is no ambiguity in the notation f1(Y). Theorem 414.Theorem 410. and let the range of f be a set T in Ek. since f maps E1 onto the open interval (π/2. and let gf be the composite function defined on S by the equation gf(x) = g[f(x)]. then Y = f(X). (ii) If Y = f(X). However. Suppose that we have (i) The point a is an accumulation point of S and f is continuous at a. Then if Y is an open subset of T. xdag[f(x)] = g[f(a)]. Let f be a function with domain S and range T. Let f and g be continuous at the point a in En and assume that f and g have function values in E1 or E2. Then if Y is a closed subset of T. and let Y be a subset of T. π/2). the inverse image f1(Y) will be an open subset of S. and f•g are each continuous at a. Theorem 411. in general. Then the composite function gf is also continuous at a. and let the range of f be a set T in Ek. A rational function g/f is continuous whenever the denominator does not vanish. Assume X ⊂ S and Y ⊂ T. Note: If f has an inverse function f1. we cannot conclude that Y = f(X) implies X = f1(Y). that is. the identity function (f(x) = x) and hence all polynomial functions are continuous. Let f be a function which is continuous on an open set S in Em. f(x) ∈ Y}. We have already seen that the image of an open set under a continuous mapping is not always open. The example in E1. Let g be defined on f(S) with values in Ek. shows that the image of a closed set under a continuous mapping need not be closed. Note that the product f•g should not be confused with the composition fg. Let f be a function defined on a set S in En and assume f(S) ⊂ Em. X ⊂ f1[f(X)]. Let f be a function which is continuous on a closed set S in Em. defined by fg(x) = f[g(x)]. are continuous whenever they are defined. By the inverse image of Y under f. This will be proved in the next theorem. we shall mean the largest subset of S which f maps onto Y. Then f+g. f1(Y) = {x  x ∈ S. defined by the equation f(x) = tan1(x). Definition 412. then X ⊂ f1(Y). the inverse image f1(Y) will be a closed subset of S. Then we have (i) If X = f1(Y). provided that g(a) ≠ 0. the inverse image of Y under f is the same as the image of Y under f1. Note that the statements in this theorem can also be expressed as follows: Y = f[f1(Y)]. Let f be a function whose domain is S and whose range is T. Theorem 415. Theorem 413. The familiar realvalued functions of elementary calculus such as exponential. In E2. Examples of continuous functions. The quotient f/g is also continuous at a. trigonometric and logarithmic functions. constant functions (f(x) = c). denoted by f1(Y). It should be observed that. fg. (ii) The point f(a) is an accumulation point of f(S) and g is continuous lim at f(a). the image of a compact set under a continuous mapping is always compact. Functions continuous on open or closed sets.
Topological mappings. sup f(S). Then f is said to have an absolute maximum on the set S if there exists a point a in S such that f(x) ≤ f(a). Let f be a realvalued function which is continuous on a compact set S in En. An immediate consequence of the above theorem is the intermediate value theorem for continuous functions (Theorem 422): If f is realvalued and continuous on a closed interval S in E1. that is. Let f be defined on a set S in En with function values in Em. Absolute minimum and relative minimum are similarly defined. Definition 423. Definition 418. assume f(a)f(b) < 0. f1 is continuous on f(S).Functions continuous on compact sets. for all x in N(a)∩S. Theorem 421 (Bolzano). and assume that f(S) ⊂ Ek. It follows from the definition that uniform continuity on a set implies continuity on the set. Let f be continuous on S and assume that f is onetoone on S so that the inverse function f1 exists. and assume that f(S) ⊂ Ek. given the accumulation point a and given ε. Then f(S) is a compact set. we have limx→a f(x) = f(a). Then f is said to be uniformly continuous on S if the following statement holds: For every ε > 0. there is a number δ > 0 (depending on a and on ε) such that 0 < xa < δ implies f(x)f(a) < ε. the function is called uniformly continuous on the set S. and suppose that f(a) and f(b) have different signs. for all x in S. In the ε and δ terminology. a global property. Definition 419.
. when it does. Uniform Continuity. then f(x)f(y) < ε. Suppose we have a function f continuous on a set S. Let f be a function which is continuous on a compact set S in Em. Note: A property of a set which remains invariant under every topological mapping is called a topological property. b) such that f(x) = 0. Let f be a function which is continuous on a compact set S in Em. and its minimum. using f(x) > f(a). Then there is at least one point x in the open interval (a. If a ∈ S and if there is a neighbourhood N(a) such that f(x) ≤ f(a). In general. so that for each accumulation point a of S. Let f be realvalued and continuous on a closed interval [a. inf f(S). this means that. however. then f is said to have a relative maximum at the point a. Then f1 is continuous on f(S). Theorem 420. Let f be a realvalued function defined on a set S in En. Then f has an absolute maximum and an absolute minimum on S. b] in E1. we cannot expect that for a fixed ε the same value of δ will serve equally well for every such point a. there exists a δ > 0 (depending only on ε) such that if x ∈ S and y ∈ S and xy < δ. in addition. Theorem 417. Suppose further that f is onetoone on S so that the inverse function f1 exists. Properties of realvalued continuous functions. then f assumes every value between its maximum. that is. Observe that uniform continuity is a property of the whole set S. Theorem 416. The converse of this statement is also true when the set is compact. Then f is called a topological mapping or a homeomorphism if. This might happen.
for every pair of points x and y in S. then f is said to be strictly increasing on S. then f is a decreasing function. then f is said to be increasing (or nondecreasing) on S. Definition 427. we write f(c+) to denote the right hand limit. regardless of the value of A. b].) We say that f is continuous from the right at x if the righthand jump is equal to 0. b). (Similar terminology applies from the left. f(x+)f(x). f(0) = A. the point x is also said to be a removable discontinuity. If c ∈ (a. d] must be a strictly monotonic function. (b) If both f(x+) and f(x) exist but have different values.Theorem 424 (Heine). and only if. the difference f(x+)f(x) is called the righthand jump of f at x. x < y implies f(x) ≤ f(y). (For the endpoints a and b. It is clear that we have f(x+) = f(x) = f(x) if. f is continuous at x. since the discontinuity can be removed by redefining f at x to have the value f(x+) = f(x). then f is sometimes said to have an infinite discontinuity at c. Definition 426. If f is continuous on S. This theorem tells us that a continuous. has a discontinuity at x = 0. Then we have: (i) The inverse function f1 exists and is strictly increasing on its domain. every topological mapping of an interval [a. Let f be a realvalued function defined on an interval [a. (ii) If f is continuous on [a. b]. Definition 425. strictly increasing function is a topological mapping. we define f(c) = limx→cf(x). only onesided jumps are considered. Monotonic Functions.) A function is called monotonic on S if it is increasing on S or decreasing on S. Let f be a realvalued function defined on a set S in E1. is called the jump at x. whereas f is discontinuous from the right if either f(x+) does not exist or the righthand jump is ≠ 0. b]. If either limx→c+ f(x) or limx→c. If f is a realvalued function defined on an open interval (a. b].
. then x can be a discontinuity of f only under one of the following conditions: (a) If either f(x+) or f(x) does not exist.) In the case that f(x+) = f(x) ≠ f(x). (c) If f(x+) = f(x) ≠ f(x). (Decreasing functions are similarly defined. then the difference f(x)f(x) is called the lefthand jump of f at x. Let f be strictly increasing on [a. f(b)]. and their sum. neither f(0+) nor f(0) exists. If f is an increasing function. In this case.f(x) is infinite (+∞ or ∞). b) and we have f(x0) ≤ f(x0) ≤ f(x0+). then f(x0+) and f(x0) both exist for each x0 in (a. Discontinuities of realvalued functions. If f is increasing on [a. Example: The function f defined by f(x) = sin(1/x) if x ≠ 0. Conversely. b) in E1 and if c is a point in [a. then f is also uniformly continuous on S. b] onto an interval [c. If f is defined at x. b]. Theorem 428. we have f(a) ≤ f(a+) and f(b) ≤ f(b). If any one of these three numbers is different from 0. If. If x < y implies f(x) < f(y). then x is called a jump discontinuity of f. whenever the limit exists. limx→c+ f(x). then f1 is continuous on [f(a). At the endpoints. If f(x+) and f(x) both exist at some interior point x. Let S be a compact set in En. Theorem 429.
A is the intersection of B with a set which is open relative to En. Theorem 431. Assume that A ⊂ B ⊂ En. and every open set in En is open relative to En. Let A and B be two sets in En. then f1(Y) is open relative to S. It is clear that every closed set in En is closed relative to En. Continuous functions can be characterised in a very elegant fashion if we introduce a slight generalisation of the notions of open and closed sets. This statement.
.Necessary and sufficient conditions for continuity. It is also clear that every set is both open and closed relative to itself. A set which is open relative to B need not. Definition 430. with A ⊂ B. which will form the basis of b B the following theorem. of course. such a set must be the intersection of a B with an open set. (b) If Y is closed relative to T. We say that A is open relative to B if the complement B—A is closed relative to B. However. Using these concepts. and let T = f(S) be a subset of Em. then f1(Y) is closed relative to S. Note that this theorem also holds with “open” replaced by “closed” throughout. Then the following three statements are equivalent: (a) The function f is continuous on S. and only if. (c) If Y is open relative to T. We say that A is closed relative to B if those accumulation points of A which lie in B are also in A. Then A is open relative to B if. we can give the following characterisation of continuous functions: Theorem 432. Let f be a function defined on a set S in En. is illustrated as shown. be an open set open set S in E 2 (relative to En). except for the endpoints a B∩S open relative to B and c) is open relative to B. The shaded c region (which includes the arc abc.
f(x) f(x)B−AB+AB−Ag(x)+AB−AB B(f(x)−A)+AB−A(g(x)−B)−AB  Bg(x)
=
B(f(x)−A)−A(g(x)−B)  Bg(x)
Now g(x) = B+g(x)B < B+ε’ < B+1.(A+B) = f(x)A + g(x)B ≤ f(x)A + g(x)B (by the triangle rule) < ε’ + ε’ = 2ε’. In this question. each defined on a set S in En. the conclusion we want is to show that f(x)/g(x) . (ii) xdaf(x)g(x) = AB. (i) Let an arbitrary ε > 0 be given and let ε’ be a second number (ε’ > 0) which will be made to depend on ε in a way to be described later.(AB) = f(x)A . Proof. Conclusion: f(x)±g(x) . Prove statements (i) and (iii) of Theorem 48. Theorem 48. Answer: Let us first remind ourselves of the statements that we are trying to prove. QED.A/B < ε whenever x ∈ N’(a)∩S.Chapter 4: Selected Exercises
41. then we have both f(x)A < ε’ and g(x)B < ε’. And f(x)g(x) . Choose a neighbourhood N(a) such that if x ∈ N’(a)∩S.A/B < ε whenever x ∈ N’(a)∩S.(g(x)B) ≤ f(x)A + g(x)B (using αβ ≤ α+β) < ε’ + ε’ = 2ε’. QED. Then we also have (i) xda[f(x) ± g(x)] = A ± B. we see that we have f(x)/g(x) . and this proves (iii). and choosing ε = ε’/2. Now  g(x) − A  =  = B Bg(x) −A(g(x)−B) B(f(x)−A) Be Â −Ae Â [  Bg(x)  +  Bg(x)  <  Bg(x)  +  Bg(x) . with function lim values in E1 or in E2. Choose a neighbourhood N(a) such that if x ∈ N’(a)∩S.(A±B) < 2ε’. Let f and g be two functions. we see that f(x)±g(x) .
. so that Be Â  −Ae Â  e Â  −Ae Â  Be Â −Ae Â eÂ  Bg(x)  +  Bg(x)  = Bg(x) + Bg(x) < B+1 + B(B+1) = B+1 +
Ae Â B(B+1)
=
e Â (B+A) B(B+1) .(A±B) < ε as required. Let a be an accumulation point of S and assume that we have xdaf(x) = A lim lim lim and xda g(x) = B. then we have both f(x)A < ε’ and g(x)B < ε’. and lim (iii) xda f(x)/g(x) = A/B if B ≠ 0. (iii) Let an arbitrary ε > 0 be given and let ε’ be a second number (ε’ > 0) which will be made to depend on ε in a way to be described later.
Choosing ε’ = (ε(B(B+1)))/(B+A). Now f(x)+g(x) .
If (i) holds. according to whether x is irrational or rational. (using the triangle inequality.43.
Adding equations (1) and (2) together. we will have m! = 1×2×. leaving us with A = απa. If m is large enough. Therefore. but hd0 f(x+h)f(x) = 01 = 1.
.
lim Answer: (a) Consider an arbitrary point x ∈ (a. Consider the two statements lim (i) hd0 f(x+h)f(x) = 0.. then m!πx = m!πa/b. 1]. Conclusion: the limit in question exists when x is rational. so that we have ndº cos 2n (m!ox)= 1 provided that lim lim m is large enough. Recall that cos(2kπ) = 1 for all k > 0. (b) Let us define a function f to be given by the following: f(x) = x² if x ≠ 0. given an arbitrary ε > 0. QED.×b×. hd0 f(x+h)f(xh) = 0 as required.e. For any m > 2.e. given an arbitrary ε > 0. and f(x) = 1 lim lim if x = 0 (with x ∈ E1). m!π will always be an even number of π’s. and so it follows that cos(m!πx) ∈ (1.×m). In other words. 45. b). 1). Give an example in which (ii) holds but (i) does not hold. A+B ≤ A+B) f(x+h)+f(xh) < ε. and it has the value 1.. A will still consist of an even number of π’s (a large m! will still have an even factor — lots of them in fact!). we have f(xh)f(x) < ε/2 ((2)). then lim f(x+h)+f(xh) < ε. so that cos(A) = 1 for large m. and that its value is 0 or 1.. where α = m!/b. then m!x cannot possibly be an even or an odd number (no irrational number is ever even or odd). Similarly. we will have an ‘infinite power’ of cos(m!πx). In other words. At x = 0. hd0 f(x+h)f(xh) = 00 = 0. when x is rational and when m is large enough. our expression consists of an lim ‘infinite power’ of 1’s which will still be 1. f(x+h)f(x)+f(xh)f(x) < ε. then if h < δ (which is equivalent to the condition h < δ). say x = a/b. lim (ii) hd0 f(x+h)f(xh) = 0. If x is any real number in [0. show that the following limit exists: lim lim 2n mdº [ ndº cos (m!ox)]. where a. b). Answer: When n → ∞ in the inner limit. Let f be defined on the interval (a. If x is rational. so that it is of the form 2kπ for some integer k > 0. then f(x+h)f(x) < ε/2 ((1)). b) and assume x ∈ (a. if we replace h by h (which holds since h = h). there exists a δ > 0 such that if h < δ. and thus the b’s cancel in the numerator and the denominator. b ∈ R. If x is irrational. there exists a δ > 0 such that if h < δ. i. mdº [ ndº cos 2n (m!ox)] = 1 when x is rational. It follows that we must have cos(m!πx) ≠ 1 in this case (for any m). then hd0 f(x+h)f(x) = 0. (a) (b) Show that (i) always implies (ii). Therefore. then b will be a factor of m! (i. we get f(x+h)f(x) + f(xh)f(x) < ε..
.. But yn < ε happens exactly when nlogy < log(ε)... 1) we have ndº cos2n(m!πx) = 0. there is a rational number z such that f(z) = 0 and so f(x) ∈ N(0). it follows that for any lim cos(m!πx) ∈ (1. then ndº yn = 0.. and it is given by zero. a3. an) in En.. 1). . b].. an) in En. so that condition (i) is satisfied for an arbitrary real number x = a1. an) for an arbitrary x ∈ E1 and fixed a2.
47. and (ii) limx→y f(x) = f(y). Conclusion: f(x) = 0 for all x ∈ [a. (ii) the function s(x) is continuous at x. . or when n > log(ε)/logy. a2... Let f be continuous at the point a = (a1. (This is sometimes stated as follows: “A continuous function of n variables is continuous in each variable separately. b]. a2. Conclusion: the limit in question exists when x is irrational.. an fixed and define a new function g of one real variable by the equation g(x) = f(x. . it follows that (i) f is defined at y.. a3. Therefore. but what is this limit? It must be zero. (i) We know that any real number is an accumulation point of the real line E1. a2. . . a2. Looking at Theorem 411. then f is continuous at every point in [a. Let f be continuous on [a. then we can say that g is continuous at the point x = a.. then for any ε > 0 we will always have yn < ε for any y ∈ (1. Keep a2. We know that f(x) = 0 when x is rational.. We now know that f(y) = limx→y f(x) for any irrational point y. we need to show that (i) the arbitrary point x ∈ E1 is an accumulation point of E1.. so by definition 49.. QED. because for any y ∈ (1. 1) we have ndºyn = 0. But every irrational number y is an accumulation point of the set of rational numbers. We need to find an N such that if n > N. Show that f(x) = 0 for every x in [a. then for an arbitrary ε > 0 and a particular y ∈ (1.lim Claim: if y ∈ (1.. Show that g is continuous at the point x = a1. b] and let f(x) = 0 when x is rational. (iii) The point s(x) is an accumulation point of En. Proof of Claim.. we see that in order to prove that g(x) is continuous at the point a = (a1.. b]. 0) and any odd n. 48. 1) we have yn < ε (We need the modulus signs in because if we don’t put them in. because y is an accumulation point.. as in every neighbourhood N(y). an ∈ E1. and by knowing that f is continuous at the point y. and (iv) f is continuous at the point s(x).
. This will hold for any m and so we have lim lim 2n mdº [ ndº cos (m!ox)] = 0 when x is irrational. thus satisfying the definition for a limit. an). If we can show that the four conditions above hold. lim Therefore. if we choose N = log(ε)/logy.”) Answer: Let us define a new function s: E1 → En by the following definition: s(x) = (x. a3. Answer: If f is continuous on [a. Looking at the above definition and at the question itself. then we will always have yn < ε for all n > N. End of Proof. b]. We will therefore prove a stronger condition than the one we need). we see that we can now write g(x) as follows: g(x) = f(s(x)) for an arbitrary x ∈ E1.
In order to prove that the above definition cannot be satisfied. and can subsequently say that the function g is continuous at the point x = a1. so that if xy < δ = ε. 1] which is not a half.
Finally. If x = ½. a n ) = s(x). then f(x)f(y) = ½(1y) = y½. with y ∈ [0. then f(x)f(y) < ε. Consider that we fix x to be a number in [0. so if xy < δ = ε. 1]. xy = ½y = y½ = f(x)f(y). Now consider the case when x ≠ ½.. (b) f assumes every value between 0 and 1. and we know by the assumption made in the question that f is continuous at the point a in En. If y is a rational number in N’(½)∩[0... . then f(x)f(y) < ε. every neighbourhood of this point x will contain both rational and irrational numbers from [0. . then for every arbitrary ε > 0. it is sufficient to find a single ε1 > 0 such that there is no neighbourhood N’(x)∩[0. then s(a1) = (a1. 1]. and let f(x) = 1x if x is irrational. Therefore. then we will always have f(x)f(y) < ε. 1]. For condition (iii) to hold. But by the same argument as in condition (i).(ii) To prove that the function s(x) is continuous at an arbitrary point x ∈ E1. adx (a 2 ). But in this case. then s(x) = (x. . We can therefore say that condition (ii) holds. we must show that f is continuous at the point s(x). we need to find a δ > 0 such that if xy < δ. As stated above. let f(x) = x if x is rational.. But if we choose δ = ε again. 1] will always contain irrational and rational numbers from [0. as s(x) is a point in En. then given an arbitrary ε > 0.. we need to find a δ > 0 such that if xy < δ. If we choose δ = ε. 1].. then we want to show that if xy < δ = ε. we must show that the point s(x) is an accumulation point of the set En. satisfying condition (iii). a2. QED. the deleted neighbourhood N’(½)∩[0. a 2 . condition (iv) is satisfied. But if we let x = a1... with y ∈ [0. (In the above we used the fact that AB = BA).
lim lim lim lim lim But adx s(a) = adx (a.. 1] ≠ x. then f(x)f(y) = ½y. xy = ½y = f(x)f(y). Show that (a) f is continuous only at the point x = ½. . Answer: Let us first prove that f is continuous at the point x = ½. and as all points in En are accumulation points. Consider the case when x is rational and we therefore have f(x) = x. we need to lim show that the function is defined at x and that adx s(a) = s(x). 1] in which all the points within δ of x in [0. then we will always have f(x)f(y) < ε. a 2 . It remains to show that we have adx s(a) = s(x) . adx (a n )) = (x. then f(x)f(y) < ε. 1]. an) is an accumulation point of the set En. . 411. a n ) = ( adx (a).. a2. If f(x) is continuous for x ≠ ½. an) = a.. then f(x) is continuous.
. 1] satisfy f(x)f(y) < ε1. as before we need to show that if xy < δ = ε. then f(x)f(y) < ε... Now if y is an irrational number. But in this case.. Now for any arbitrary δ > 0.. a 3 . So we have shown that if x = ½. To show that f is continuous at this point. The first part follows immediately because the function s(x) is defined for all real numbers x ∈ E1 by its definition lim on the previous page. then x is rational and so f(x) = ½. and thus we have shown that all four conditions hold. For each x in [0.
Assume that such a δ exists.e. then we need 2x1+δ < ε2. we will have x+y1 < x+(x+δ)1 = 2x1+δ. Therefore. we will have x+y+1 < x+(x+δ)+1 = δ+1. Therefore. Now if xy < δ. i. QED. 1] and x ≠ ½. It follows that f(x)f(y) = x(1y) = x+y1.Consider also that we define a number ε1 = 2x1 > 0. Conclusion: f(x) is not continuous when x is a rational number in [0.
. i. Conclusion: f(x) is not continuous when x is an irrational number in [0. y < x+δ. with xy < δ. δ+1 < 1. then we will always have f(x)f(y) < ε1. we have f(x) = 1x and f(y) = y. if we pick ε1 = 2x1 > 0. then f(x)f(y) < ε1 for ALL irrational y in this neighbourhood. 1] ≠ x. 1]: x = ½. and x ≠ ½. 1] which is an irrational number. and let us pick out an element y from the neighbourhood N’(x)∩[0. Consider now that for this fixed irrational number x. we need to find a δ > 0 such that if xy < δ for y ∈ [0.e. we have f(x) = x and f(y) = 1y. and let us pick out an element y from the neighbourhood N’(x)∩[0. But if f is continuous at x. 1]. we need to find a δ > 0 such that if xy < δ for y ∈ [0. The other case to consider is when x is a fixed irrational number in [0. In this case. so that f(x) is not continuous. 1]. Therefore. so that f(x) is continuous. In this case. then y is again bounded from above by x+δ. i. Given this ε1 (which will be greater than zero if x is not a half). then we will always have f(x)f(y) < ε2. 1] which is a rational number. Given this ε2. But if f is continuous at x. 2x1+δ < 2x1. then we need δ+1 < ε2. then there is no δ > 0 that will suffice so that if xy < δ. i. We have now considered all cases for x in [0. Therefore. Assume that such a δ exists. we conclude that when x is irrational.e. There is no possible δ > 0 that we can pick that will satisfy the above inequation. y < x+δ. Now if xy < δ. 1].e. we define a number ε2 = 1. if we pick ε2 = 1. then there is no δ > 0 that will suffice so that if xy < δ. we conclude that when x is rational and not equal to a half. with xy < δ. But there is no possible δ > 0 that we can pick that will satisfy the above inequation. It follows that f(x)f(y) = 1x+y = x+y+1. then f(x)f(y) < ε1 for ALL rational y in this neighbourhood. then y is bounded from above by x+δ.
lim Does the limit hd0+ Ω f(N(x. let us look at what exactly is sup{f(x1)f(x2)  x1 ∈ (N(x.
Claim: If h2 > h1. then f(x1) must be the highest value of f(x) in the neighbourhood N(x. h2)∩S). h1)∩S) ≤ Ω f(N(x. then Ω f(N(x. f is continuous at x. 1]. If this is true. we have sup{f(x1)f(x2)  x1 ∈ (N(x. h)∩S (x.e. Therefore. and so we can write
A=
lim hd0+
Ω f(N(x. Our only problem occurs in showing that f assumes the value of every irrational number in the interval [0. h)∩S. as f(x) = x for all rational numbers in [0. QED. 1] such that f(z) = y.e. then Ω f(N(x. then sup{f(x)  x ∈ N(x. all we need do is to show that if h2 > h1. then 1y will also be an arbitrary irrational number in this interval. h1)∩S) ≤ Ω f(N(x. x2 ∈ ((x. h)∩S)} . f(x1)f(x2)) will also be finite. h)∩S.(b) It is a trivial matter to show that f assumes the value of every rational number in the interval [0. x2 ∈ ((x. say. h)∩S). The proof centres on the fact that if h2 > h1. and f(x2) must be the smallest value of f(x) in the neighbourhood N(x. h2)∩S). the entry where h = 1). 414. Because f(S) is bounded (i. h)∩S) will exist and is a finite value. and so their difference (i. Proof of Claim. h)∩S. 1]. h)∩S. b]. i.) If T is a subset of S. the number Ω f(T) = sup{f(x)f(y)  x ∈ T. h)∩S)} = sup{f(x1)  x1 ∈ ((x. h)∩S). To start with. assume that f(S) is a bounded set. 1]. h1)∩S} > inf{f(x)  x ∈ N(x. If x ∈ S. The value ωf(x) is thus determined by two points (say x1 and x2) in N(x. (That is. i. then we need to find a number z ∈ [0. and thus z = 1y is the number we are looking for so that f(z) = y for all irrational numbers y ∈ [0. h1)∩S} ≤ sup{f(x)  x ∈ N(x. in the neighbourhood of x in S. Ω f(N(x. it is the maximum value of f(x1)f(x2) in the neighbourhood N(x. Conclusion: f assumes every value between 0 and 1. Ω f(T) is the biggest value of f(x)f(y) in T (x.e. h)∩S). 1].e. say. Answer: By definition. h)∩S). It follows that ωf(x) is the biggest value of f(x)f(y) in N(x. h)∩S) exist? To show that it does exist. Thus f(1y) = 1(1y) = y (because 1y is irrational). y ∈ T). Let f be defined and bounded on the closed interval S = [a.inf{f(x2)  x2 ∈ ((x. 1)∩S). and only if. where 0 ≤ A ≤ Ω f(N(x. h)∩S}. h)∩S). h)∩S)}? Well. y ∈ T} is called the oscillation of f on T. 1]. and inf{f(x)  x ∈ N(x. h)∩S) given by decreasing h will be a bounded monotonic sequence of finite positive numbers (bounded above by. h2)∩S}
. and thus we can apply Theorem 126 to say that this sequence converges to a value. 1]. for a particular h. the oscillation of f at x is defined to be the number lim ωf(x) = hd0+ Ω f(N(x. But if this is so. If y is our arbitrary irrational number in the interval [0. then the sequence of numbers Ω f(N(x. then the two points in question will have finite values. h2)∩S}. Show that this limit always exists and that ωf(x) = 0 if. say A. if we have an arbitrary irrational number y ∈ [0. Therefore. y ∈ N(x. f(S) is a subset of some finite interval).
h)∩S. h)∩S)} = 0 as required. h)∩S)} → 0 when h → 0+. we note that limy→x f(y) = f(x) can be equivalently written as the two lim lim following equations: hd0+ (x + h) = f(x) and hd0+ (x − h) = f(x).I will not prove these statements (intuitively. h)∩S. h2)∩S}. h)∩S. From our previous analysis of sup{f(x1)f(x2)  x1 ∈ (N(x. x2 ∈ ((x. so that f is defined for all x ∈ [a. lim hd0+ sup{f(x1)f(x2)  x1 ∈ (N(x. we know that f(x1) must be the highest value of f(x) in the neighbourhood N(x. h)∩S). h)∩S). h)∩S. x2 ∈ ((x. h1)∩S)} ≤ sup{f(x1)f(x2)  x1 ∈ (N(x. h1)∩S). so that for a given ε > 0. or Ω f(N(x. b]. points (x+h) and (xh).
. we need to show that f is defined at x and that limy→x f(y) = f(x). we have sup{f(x1)  x1 ∈ ((x. h2)∩S) as required. h)∩S. h)∩S). Only If. then the distance between x1 and x2 is at most 2h. The first bit follows immediately from the fact that f is defined and bounded on the interval [a. h)∩S. For a given h. h)∩S)} if y. h)∩S). x2 ∈ ((x. If. then we need to show that f is continuous at x. then for a point x3 in the neighbourhood N(x. then lim hd0+ sup{f(x1)f(x2)  x1 ∈ (N(x. x2 ∈ ((x. x2 ∈ N(x. x2 ∈ ((x. Knowing this. h)∩S)}. QED. there are less points to consider. h1)∩S} ≤ sup{f(x1)  x1 ∈ ((x. Now if ωf(x) = 0. x1x2 < 2h ⇒ f(x1)f(x2) < 2ε. and if y > z. h)∩S)}. We now need to show that ωf(x) = 0 ⇔ f is continuous at x.inf{f(x2)  x2 ∈ ((x. If f is continuous at x. x2 ∈ ((x. h1)∩S)} . or sup{f(x1)f(x2)  x1 ∈ (N(x. Note also that we must have 0 ≤ f(y)f(z) ≤ sup{f(x1)f(x2)  x1 ∈ (N(x. h2)∩S)}. h)∩S)} = 0. and that f(x2) must be the smallest value of f(x) in the neighbourhood N(x.inf{f(x2)  x2 ∈ ((x. If ωf(x) = 0. in other words. h2)∩S). there exists a h > 0 such that if xy < h. x2 ∈ ((x. lim lim or hd0+ (x + h) − f(x) = 0 and hd0+ f(x) − (x − h) = 0. h2)∩S)} . z ∈ N(x. h)∩S). the neighbourhood given by N h)∩S will contain the point x and the (x. where y ∈ N(x. For the second bit. Assuming that these statements are correct. then f(x1)f(x2) → 0+ so that sup{f(x1)f(x2)  x1 ∈ (N(x. QED. if we are considering a smaller neighbourhood. x2 ∈ ((x. then f(x)f(y) < ε. To do this. Now if x1. b]. h1)∩S) ≤ Ω f(N(x. h)∩S). then we must have f(x3)f(x) ≤ sup{f(x1)f(x2)  x1 ∈ (N(x. or. But if h → 0+. and so the maximum value in a smaller neighbourhood will be smaller and the smallest value in a smaller neighbourhood will be bigger). then given an arbitrary ε > 0.
then Y contains all its accumulation points (of which there are none of).f(x) ≤ sup{f(x1)f(x2)  x1 ∈ (N(x. h)∩S). lim lim hd0+ f(x) . b]. x2 ∈ (N(x. T = f(S)). h)∩S)}. i. Let Y be the subset of T consisting only of the zero point.f(xh) = 0 as required (because we must have f(x)f(xh) > 0).f(x) ≤ 0. we can say that [x1. h)∩S)} = 0.
lim But again we have assumed that hd0+sup{f(x1)f(x2)  x1 ∈ (N(x. i. Further.e. Answer: Consider the interval [x1. this interval is a closed set. h)∩S). By Exercise 31. x2] as well. Technical Note: There is a slight mistake in the above in that f(x ± h) ∉ N(x. h)∩S). QED. then it will be continuous on the subinterval [x1. 419. Let f be continuous on a closed interval [a.f(x) = 0 as required (because we must have f(x+h)f(x) > 0). then A = {φ}. Answer: Assume that the range of f is a set T in E1 (i. the elements of S which map onto 0.f(xh) ≤ hd0+ sup{f(x1)f(x2)  x1 ∈ (N(x. i. we have hd0+ f(x) . the interval [x1. b].f(x) ≤ hd0+ sup{f(x1)f(x2)  x1 ∈ (N(x. x2 ∈ (N(x. h)∩S)}. h)∩S). x2 ∈ ((x. but I have assumed that in taking the limit this is of no consequence to the proof.e. and is a closed set by definition.f(xh) ≤ 0. QED. Show that there must be a third point between x1 and x2 where f has a local minimum. But f1(Y) is the set A. lim lim so that hd0+ f(x + h) . Note: in the case that 0 ∉ f(S).
417. and thus Y is a closed subset of T. the inverse image f1(Y) is a closed subset of S. Therefore. Because Y consists of just a single point. Therefore.e. and thus A must be a closed subset of S.f(xh) ≤ sup{f(x1)f(x2)  x1 ∈ (N(x. f(x) = 0}. b]. Therefore. x2]. h)∩S). x2] ⊂ [a.
. x2] is a closed interval. we have hd0+ f(x + h) . and so by the note following Definition 339 (and by knowing that [x1. h)∩S)}. x2] is a bounded set (it is a subset of [a. x2] is a compact set. then we see that we have (i) f(x + h) . Y = {0}. b]).e.If we let x3 = x ± h. x2 ∈ (N(x. lim lim h)∩S)} = 0.
lim But we have assumed that hd0+ sup{f(x1)f(x2)  x1 ∈ (N(x. h)∩S)}. x2 ∈ ((x. h)∩S. x2 ∈ ((x. so that hd0+ f(x) . Let f be defined and continuous on a closed set S in En. lim lim hd0+ f(x + h) . By Theorem 414. Suppose that f has a local maximum at x1 and a local maximum at x2.
(ii)
f(x) .e. i. Because f is continuous on the closed interval [a. we can apply Theorem 420 to say that f has an absolute maximum and an absolute minimum on [x1. Let A = {x  x ∈ S. Show that A is a closed set. h)∩S).
x2] such that f(x) > f(c) for all x ∈ [x1. where δ depends only on ε. x2]. then we need to show that f(x)f(y) < ε. But this is exactly the condition we want so that the function f is continuous on S (it doesn’t matter that δ does not depend on a). Therefore. then for every ε there exists a δ > 0 (depending only on ε) such that if x ∈ S. there exists a δ > 0 (depending only on ε) such that if x ∈ S and y ∈ S and xy < δ. x1+h).If f has an absolute minimum on [x1. or else c would have been incorrectly found. we redefine c to be given by c = x2h or by c = (x1+x2)/2 as appropriate). Note: if x1+h > x2. where h is the radius of the neighbourhood N(x1). using the above. 421. Answer: Consider that we are given two arbitrary points x and y in E1. then we must have f(x) = f(x1) for x ∈ [x1. if we find that we arrive at the conclusion c = x1 or c = x2. x2]. then there exists a point c ∈ [x1.
. f is said to be continuous on S provided that at each accumulation point a. we have f(x) ≤ f(x1). and therefore we have reached the conclusion we want. If we now take any neighbourhood around the point c such that N(c) ⊂ [x1. Show that a function which is uniformly continuous on a set S is also continuous on S. Answer: Let f be defined on a set S in En with function values in Em. But because c is the absolute minimum of f on [x1. b]. 422. then there is a neighbourhood N(x1) such that for all x ∈ N(x1)∩[a. because x1 is a local maximum. The only detail left to mop up is to make sure that c is not either x1 or x2 (we need a third point between x1 and x2). The case c = x2 goes through in much the same way as above (i. x2]. f(x1) = f(x1+h). We need to show that if the distance between x and y is less than δ. In other words. in that uniform continuity on S does imply continuity on S. x2]. then f(x)f(a) < ε. then the distance between f(x) and f(y) is less than ε. x2]. and xa < δ. then just take a point half way between x1 and x2 to represent c — in this situation f must be a constant function between x1 and x2. then we can say that f has a local minimum at the point c in question. given ε. x2]. f is said to be uniformly continuous on S if the following statement holds: For every ε > 0. if xy < δ. then f(x)f(y) < ε. Show that the function f defined by f(x) = x² is not uniformly continuous on E1. But. Let a denote any arbitrary accumulation point in S. and so we can equally well define c to be given by c = x1+h to represent our absolute/local minimum on [x1. QED. then we proceed as follows: for the case c = x1. QED.e. there is a number δ (depending on a and on ε) such that 0 < xa < δ ⇒ f(x)f(a) < ε. so that we have found a point c between x1 and x2 which is a local minimum in the interval [x1. If f is uniformly continuous. and given ε. and let ε > 0 be an arbitrary number.
If we let δ2 = ε1. i. So we have a contradiction. However. and thus cannot define δ in terms of ε only so that the function f(x) = x² in uniformly continuous on E1. and so as x+y ≤ 2.e. by (2) we can always find a δ2 > 0 (dependent only on ε2) such that f(x)f(y) < δ2 ⇒ g(f(x))g(f(y)) < ε2. and therefore cannot define δ = ε/n for a given arbitrary ε > 0 so that f(x)f(y) < ε. But if ab < δ2. We know that if xy < δ1. if x. so that ab < ε1 ⇒ ab < δ2. and so gf is uniformly continuous on S. we can always find a δ1 > 0 (dependent only on ε2) such that xy < δ1 ⇒ f(x)f(y) < ε1 = δ2 ⇒ g(f(x))g(f(y)) < ε2. In the example in the book where we considered points in the interval (0. 425. Let g be defined on f(S) with values in Ek. given an ε2 > 0. the definition of uniform continuity). δ = ε/n. Assume for the moment that δ2 = ε1. we conclude that given an ε2 > 0.e. q ∈ Em). Let f be a function defined on a set S in En and assume that f(S) ⊂ Em. then given an ε1 > 0. QED. but f(c)f(d) = (cd)(c+d) > (cd)(n+n) = 2n(δ/2) = nδ = ε. so that we cannot choose a value n so that f(x)f(y) < δn. so that if we choose δ = ε/2 for a given arbitrary ε > 0. d ∈ E1) and which are at a distance of δ/2 apart.
. then we get the required result: f(x)f(y) < ε if xy < δ. Proof by Contradiction: Suppose that there was a positive number ‘n’ so that if we defined δ = ε/n for a given arbitrary ε > 0. then f(x)f(y) < 2δ. then f(x)f(y) < δx+y. xy < δ1 ⇒ g(f(x))g(f(y)) < ε2 (x. the value x+y was bounded above by 2. then x+y is not bounded above. then given an ε2 > 0. If xy < δ. we can always find a δ1 > 0 (dependent only on ε1) such that xy < δ1 ⇒ f(x)f(y) < ε1. y ∈ En). we get ε = δn. and QED. y ∈ E1. 1]. y ∈ En). End of Proof. we can (by (1)) always find a δ1 > 0 (which now is dependent only on ε2) such that xy < δ1 ⇒ f(x)f(y) < ε1. and let gf denote the composite function defined by gf(x) = g[f(x)]. then ab < ε1. so that we have cd < δ. g(f(x))g(f(y)) < ε2. Now let a = f(x) and let b = f(y).Now f(x)f(y) = x²y² = (xy)(x+y) = (xy)(x+y). then we would get f(x)f(y) < ε for every xy < δ (i. (2) If g is uniformly continuous on f(S).e. then g(a)g(b) < ε2. i. Answer: (1) If f is uniformly continuous on S. then given that particular ε1 (which is dependent only on ε2). Putting the above all together. But we can always choose two numbers c and d which are greater in magnitude than n (because c. show that gf is uniformly continuous on S. Therefore. we can always find a δ2 > 0 (dependent only on ε2) such that pq < δ2 ⇒ g(p)g(q) < ε2. if x ∈ S. Note: with our definition of δ. (x. If f is uniformly continuous on S and if g is uniformly continuous on f(S). (p.
b). If f is not increasing on (a. 429. (b) (c) Again the function is continuous at 1 x x every point in E1 except at x = 0. f(0) = 0.427. then we must throw away this neighbourhood from our subinterval (c. d). f(0) = 0.= 1 and limx→0+ = 1. As limx→0. As f(x) has a lefthand jump at x = 0. Answer: (a) f(x) is f(x) f(x) (b) continuous at every point in E1 (a) 1 except at x = 0. f(x) is continuous at every other point in E1. (d) In this final example. f(x) is continuous everywhere except at x = 0. But we can apply the above argument with any point in the subinterval (c. d) ⊂ (a. (c) In this case. f(0) = 0. and f(0+) = ∞. where f(x) has an infinite jump discontinuity: f(0) = 0. i. then f(x) has a lefthand jump discontinuity at x = 0. d). f will be increasing in a neighbourhood of that point. then because (c. as before. f is strictly decreasing in this subinterval. and f(0) does not exist. f(x) has an infinite righthand jump discontinuity as in part (b). b). b) in which f is not increasing. Because of this. b) are interior points. b) is an open set — so therefore all points x ∈ (a. Let f be defined in the open interval (a. This time. 1/x (b) f(x) = e if x ≠ 0. d) in which we are claiming that f is strictly decreasing. and thus there is some neighbourhood of z in which f is increasing. This implies that for all points x ∈ (a. b). and thus there cannot possibly be any such subinterval (c. QED.e. Answer: We saw in exercise 31 that an open interval (a. 1/x 1 (c) f(x) = e + sin( /x) if x ≠ 0. f(0+) = 0. f(0) = 0. We therefore conclude that f(x) is discontinuous at x = 0. we have f(0) = 1. there exists a neighbourhood N(x) in which f is increasing. Let us now prove that f is increasing by contradiction. As the function has an infinite righthand jump. and f(0) = 0. We therefore conclude that f must be increasing on the interval (a. f(0) = 0. b) and assume that for each interior point x of (a. we have z ∈ (a. Show that f is an increasing function throughout (a. d). b). we can make the function continuous by f(x) f(x) (d) redefining f(0) to be f(0) = 1. then it is discontinuous at x = 0 (and cannot be made continuous at this point). If z ∈ (c. d)) of (a. Locate and classify the discontinuities of the function f defined on E1 by the following equations: (a) f(x) = (sin x)/x if x ≠ 0. (d) f(x) = 1/(1e1/x) if x ≠ 0. b) there exists a neighbourhood N(x) in which f is increasing. b). b). Consider a point z ∈ (c.
. d) in which f is strictly decreasing because for any point in such an interval. where f(x) has a x x removable discontinuity. then there is a subinterval (say (c. Apart from this.
b]. The interval [x1. x2] is a bounded set (it is a subset of (a. d] must be strictly monotonic. there exists one and only one point x in the domain such that f(x) = y. b]. By Exercise 31. enabling us to say that f(x1) ≤ f(x2) as required. and so by the note following Definition 339 (and by knowing that [x1. say a collection of n neighbourhoods. Answer: If f is onetoone on [a. with x ∈ [x1. x2] is a closed interval.
. there will be a corresponding neighbourhood N(x) in which f is increasing. Because of this. show that f must be strictly monotonic on [a. prove that every topological mapping of [a. In other words. 431. then we know that f is defined for all points in the interval [a. x2]. b) so that a < x1 < x2 < b. a finite number of these neighbourhoods will cover [x1. then the intermediate value theorem says that f assumes all values between y1 and y2. Consider the case where y1 < y2 (we will only consider this case — the other case can be derived from the following by replacing all ‘<’ by ‘>’ and viceversa). we know that f(a) will have a value: say f(a) = y1. b] onto an interval [c. then f will be increasing throughout the interval [x1. That is. Suppose that we pick two arbitrary numbers x1 and x2 from the interval (a. b]. and because this finite collection of neighbourhoods covers [x1. x2]. Consider that we define a covering of the compact set [x1. we can write [x1. i=1 Because f is increasing throughout all of these neighbourhoods. x2] in which f is increasing. x2]. Because f is onetoone. If f is continuous on [a. and thus we must have either y1 > y2 or y1 < y2.
Now because [x1. If f is onetoone and continuous on [a. b]. then for all points y in the range. b]. Now for every point x ∈ [x1. x2] given by the union of all such neighbourhoods of points in [x1.x 2 ] N(x) . b)) we can say that [x1. x2]. To show that f is an increasing function throughout (a. x2] is a compact set. x2] = 4 n N i (x) . b]. then we cannot have y1 = y2. Therefore. this interval is a closed set. QED. Because of the continuity of f on the closed interval [a. x2] =
4 all x c [x 1 . and that f(b) will have a value: say f(b) = y2. b). x2] is a compact set. x2].Here is a second proof for the above question. all we must show is that f(x1) ≤ f(x2) for our arbitrary numbers x1 and x2. we can write [x1.
we would reach the following conclusion: y1 < f(c) < f(d) < y2. Further (as seen in Diagram 1). and this cannot be the case if f is onetoone (because then we will have f(q) = f(a) = y1). b) where f(q) = y1.Consider an arbitrary point c ∈ (a. we cannot f(p) have f(c) > y2 as then (according to the IVT) there will be a point p in the interval (a.e. and thus f is strictly increasing on [a. i. we conclude that f is either a strictly monotonically increasing function or a strictly monotonically decreasing function. given any pair of points c and d in [0. where y1 > y2. QED. with c < d. b]. 1]. b). we conclude that we must have y1 < f(c) < y2. b). b]. In other words. Putting all of this together. we cannot have f(c) < y1 as then (again according to the IVT) there will be a point q in the f(a) interval (c. Similarly (as seen in Diagram 2).
Diagram 1
f(c)
f(b) f(b) f(q) f(c)
Diagram 2
If we apply the same sort of argument for the point d ∈ (c. f(c) cannot be equal to either y1 or y2 because f is a onetoone function. With the other case. Therefore. What is f(c)? First of all. c) where f(p) = y2. then we must have f(c) < f(d). and this cannot be the case if f is onetoone (because then we will have f(p) = f(b) = f(a) y2).
. we would reach the conclusion that f would be strictly decreasing on [a. f is a monotonic function.
In this case. if we consider an arbitrary irrational number in [a. or a supremum. Given this ε2. 1] which is a rational number. a finite number of these neighbourhoods will cover [x1. b]. we define a number ε2 = 1. with yi ∈ [x1.7: Let A be a set of real numbers bounded above. N(yn). then x is called an upper bound for the set A and we say that A is bounded above.Possible Further Work / Evaluation
Exercise 47: Clarification in the wording. x2] is a compact set. i=1
.. Therefore.g. I made a slip up with a negative sign. Exercise 414: I was mistaken in thinking that “sup” was the same thing as “maximum”.
Such a number x is called a least upper bound. I therefore looked at the following definitions in Chapter 1: Definition 1. and let us pick out an element y from the neighbourhood N’(x)∩[0. then x ≤ y.. with xy < δ. Assume that such a δ exists. QED. In the second part. New version: . Definition 1. . we can write [x1.The other case to consider is when x is a fixed irrational number in [0.... say a collection of n neighbourhoods N(y1).6: Let A be a set of real numbers. the result follows because of the continuity of f? Exercise 411: A ‘small gap’ left unfilled in the first part of the answer.Now because [x1. Exercise 429: Need to correct the indexing in the penultimate paragraph: . If there is a real number x such that a ∈ A implies a ≤ x. we need to find a δ > 0 such that if xy < δ for y ∈ [0. or possibly an alternative solution. e. Suppose there is a real number x satisfying the following two conditions: (i) (ii) x is an upper bound for A. Consider now that for this fixed irrational number x. 1] ≠ x. x2]. 1]. But this is what we had in the first section of the answer so the result follows by the validity of this first section.... x2] ⊂ 4 n N(y i ) . and because we know that there exists a sequence of rational numbers xn converging to y. Exercise 422: Need to prove the general case — not just for a particular formula for δ in terms of ε... of the set A. It follows that f(x)f(y) = 1xy = xy+1 = x+y1 (as AB = BA).. and if y is any upper bound for A. x2].. then we will always have f(x)f(y) < ε2. we have f(x) = 1x and f(y) = y..
Chapter 13
.
Let T denote the set of these points x in S for which the second sequence converges... n) = 0 otherwise.. If each term of the sequence {fn} has a certain property. will be referred to as the limit function of the sequence {fn} and we will say that {fn} converges pointwise on the set T. and the limit function has the value 0 at each point in [0. or. if x ∈ T. but it does not provide the complete answer to the question. The last equation can also be written as follows: xdx0 ndº f n (x) = ndº xdx0 f n (x) . n). even if it does exist. 1]. . Examples can be found in which the order of the two limits can be interchanged although the sequence is not uniformly convergent. We shall find that uniform convergence is a farreaching sufficient condition for the validity of interchanging certain limits. this will be explicitly stated). limn→∞ fn(x) exists if 0 ≤ x ≤ 1. it may not be equal to f(x0). Given a sequence {fn}. 2. it is stated that Σm=1∞ Σn=1∞ f(m. n = 1. and f(m. . Here. f(m. In chapter 12. n) is not necessarily equal to Σn=1∞ Σm=1∞ f(m. n) = 1. 2.. but Σn=1∞Σm=1∞ f(m. we will be dealing with complexvalued functions defined on certain subsets of E2 (but if results hold only for realvalued functions defined on subsets of E1. The most important of these is the notion of uniform convergence. fn has a local maximum at x = 1/(n+1). n) = 1 if m = n+1. consider the following function: f(m. .. the limit in lim the equation ( xdx0 fn(x) = fn(x0)) may not exist. An example of a sequence of realvalued functions: Let fn(x) = n²x(1x)n if x ∈ E1.. For example. we are really asking whether the equation xdx0 fn(x) = fn(x0) implies the equation xdx0 f(x) lim lim lim lim = f(x0). each term of which is a function defined on a set S. then to what extent does the function f also possess this property? In general.
. The question arises frequently as to whether we can change the order of the limit processes. for each x in S we can form another sequence {fn(x)} whose terms are the corresponding function values. The lim function f defined by f(x) = ndº fn(x). our question about continuity amounts to this: Can we change the limit symbols in the above? In general. n) = 1 if m = n1. Therefore. First of all. we need a study of “stronger” methods that do preserve these properties. but fn(1/(n+1)) → +∞ as n → ∞. 2...Chapter 13: Sequences of Functions
In this chapter. n) = 1. Then Σm=1∞Σn=1∞ f(m. we shall see that we cannot interchange the symbols. n = 1. n = 1. When we ask whether continuity at each fn at x0 implies continuity of the limit function f lim lim at x0.
For each n = 1. Definition 135: Given a sequence {fn} of functions defined on a set T. n)) exists. 2. If the same N works equally well for every point in T. Theorem 132. there is a useful geometric interpretation of uniform convergence. The inequality fn(x)f(x) < ε is then equivalent to the two inequalities f(x)ε < fn(x) < f(x)+ε. . 2. let sn(x) = Σk=1n fk(x) (n = 1. Note 1: If x0 is an accumulation point of T.
.Let {fn} be a sequence of functions which converges pointwise on a set T to a limit function f. where g(m) = limn→∞ f(m. If there exists a function f such that sn → f uniformly on T. n) also exists and has the same value. When each term of the sequence {fn} is realvalued. then it is easy to prove that {fn} is uniformly bounded on T. the conclusion implies that lim lim lim lim xdx 0 ndº f n (x) = ndº xdx 0 f n (x) . there exists an N (depending only on ε) such that n > N implies fn(x)f(x) < ε. Let f be a double sequence and let P denote the set of positive integers. then the limit function f is also continuous at x0. If the iterated limit limm→∞ (limn→∞ f(m. the set {(x. for every x in T. The Cauchy condition for uniform convergence. then the double limit limm. A sequence {fn} is said to be uniformly bounded on T if there exists a constant M > 0 such that fn(x) ≤ M for all x in T and for all n = 1. Uniform Convergence and Continuity. That is. We denote this symbolically by writing fn → f uniformly on T. the convergence is said to be uniform on T.. If each fn is continuous at a point x0 of T... the following condition (called the Cauchy condition) is satisfied: For every ε > 0 there exists an N such that m > N and n > N implies fm(x)fn(x) < ε. for every x in T. n).n→∞ f(m. For each x in T. if m ∈ P. Assume that gn → g uniformly on P. Note 2: Uniform convergence of {fn} is sufficient but not necessary to transmit continuity from the individual terms to the limit function. this means that the entire graph of fn (that is. n). y)  y = fn(x).. x ∈ T}) lies within a “band” of height 2ε situated symmetrically about the graph y = f(x)ε of f (see the diagram on the right). 2. Uniform Convergence on infinite series... . we have the following definition: A sequence of functions {fn} is said to converge uniformly to f on a set T if. we say that the series Σfn(x) converges uniformly on T and we write Σn=1∞ fn(x) = f(x) (uniformly on T). this means that for each point x in T and for each ε > 0. The number M is called a uniform bound for {fn}. Going back to the basic definition of limit. there exists an N (depending on at most both x and ε) such that n > N implies fn(x)f(x) < ε. This observation often enables us to conclude that a sequence is not uniformly convergent. define a function gn on P as follows: gn(m) = f(m.). . There exists a function f such that fn → f uniformly on T if. If this is to hold for y = f(x)+ε y = f(x) y = fn(x) all n > N and for all x in T. If each individual function is bounded and if fn → f uniformly on T.. Theorem 133: Assume that fn → f uniformly on T. and only if. Theorem 134: Let {fn} be a sequence of functions defined on a set T. for every ε > 0.
Theorem 137: (Weierstrass’ Mtest). if 2/3 ≤ t ≤ 4/3. In fact.. Let α = (α1. Therefore. Next. 2. if 4/3 ≤ t ≤ 5/3. We will show that Γ ‘fills’ the unit square. that Γ = [0.. But then we get 2/3 ≤ dk ≤ 1. for n = 1. where dk = Σn=1∞ cn+k/2n. and this will give us α1(c) = a. Extend the definition of φ to all of E1 by the equation φ(t+2) = φ(t). for each k = 0. 2.. and every x in T. 2. n=1 n n=1 n º c where each an and each bn is either 0 or 1.. 1]×[0. Clearly. Hence. it is clear that 0 ≤ α1(t) ≤ 1 and that 0 ≤ α2(t) ≤ 1 for each t. then f is also continuous at x0. That is. If we can prove that φ(3kc) = ck+1. We can apply the above theorem to construct an example of what is known as a spacefilling curve.e. 1]. for every ε > 0 there is an N such that n > N implies n+p  S k=n+1 fk(x) < ε.. α2(c) = b. Now let c = 2 S n=1 3n . Consider the following example: Let φ be defined on the interval [0.. 2n
2
1 0
1 2
3 4
Both series converge absolutely for each real t and they converge uniformly on E1.. 1] under α. we write a and b in the binary system. . Γ is a subset of the unit square. The series Σfn(x) converges uniformly on T if.. φ(3kc) = ck+1 in this case. Therefore. This is a continuous curve in E2 that passes through every point of the unit square [0. 1]. . i. φ(t) = 3t1. 2n
α2(t) = S º n=1
v(3 2n−1 t) . 1]. and φ(t) = 3t+5. 1. we write 3kc = 2 S k 3 n−k + 2 S º n=1 n=k+1 3 n−k = (an even integer) + dk.Theorem 136: (Cauchy condition for uniform convergence of series). n = 1. Note: If x0 is an accumulation point of T. n=1 n=1 A spacefilling curve. Since φ has period 2. and only if. then we have 0 ≤ dk ≤ 2Σn=2∞ 3n = 1/3. and hence φ(dk) = 0. or if 5/3 ≤ t ≤ 2. α2). If each fn is continuous at a point x0 of T. and for every x in T. For this purpose. for each p = 1. ((1)). and φ(32n1c) = c2n = bn.. 0 ≤ c ≤ 1 since 2Σn=1 3 = 1.
c
c
. . then we will have φ(32n2c) = c2n1 = an. where c2n1 = an and c2n = n ∞ n bn. since Σn=1∞ 2n = 1. and let Γ denote the image of the unit interval [0. 2] by the following formulas: φ(t) = 0. φ(3kc) = ck+1 in all cases and this proves that α1(c) = a. b) ∈ [0. We will show that α1(c) = a and that α2(c) = b. Theorem 138 tells us that α1 and α2 are also continuous on E1. b) ∈ Γ whenever (a. . The only other case to consider is the case ck+1 = 1. the Weierstrass Mtest is applicable with Mn = 2n. we write a = S º 2n . b = S º 2n . φ(t) = 1. Then Σfn(x) converges uniformly on T if ΣMn converges. it follows that φ(3kc) = φ(dk). since φ(t) ≤ 1 for all t. If ck+1 = 0.. This makes φ periodic with period 2 (as shown below). if 1/3 ≤ t ≤ 2/3. α2(c) = b. Since f is continuous on E1.
a b
n n To prove equation (1). First. 1]×[0. Let {Mn} be a sequence of nonnegative numbers such that 0 ≤ fn(x) ≤ Mn. Hence. Now define two functions α1 and α2 by the following equations: α1(t) = S º n=1
v(3 2n−2 t) . and hence φ(dk) = 1. Γ fills the unit square. Theorem 138: Assume that Σfn(x) = f(x) (uniformly on T). this lim lim theorem permits us to write xdx0 S º fn(x) = S º xdx0fn(x). if 0 ≤ t ≤ 1/3. we must show that (a. 1]×[0. 2.
An application to repeated series. As a second application of Theorem 138, we derive the following generalisation of Theorem 1243 (which went as follows: Theorem 1243: Suppose f(m, n) > 0 for all m, n. Assume that Σn=1∞ f(m, n) converges for each fixed m = 1, 2, ..., and that Σm=1∞Σn=1∞ f(m, n) converges. Then: (a) Σm=1∞ f(m, n) converges for each n = 1, 2, ...; and (b) Σn=1∞Σm=1∞ f(m, n) converges and equals Σm=1∞Σn=1∞ f(m, n).): Theorem 139. Let f be a complexvalued double sequence. Assume that Σn=1∞ f(m, n) converges absolutely for each fixed m and that Σm=1∞Σn=1∞ f(m, n) converges. Then (a) Σm=1∞ f(m, n) converges absolutely for each n, and (b) Σn=1∞Σm=1∞ f(m, n) converges absolutely and equals Σm=1∞Σn=1∞ f(m, n). Uniform convergence and differentiation. Theorem 1313: Assume that each term of {fn} is a realvalued function having a finite derivative at each point of an open interval (a, b). Assume that for at least one point x0 in (a, b) the sequence {fn(x0)} converges. Assume further that there exists a function g such that f’n → g uniformly on (a, b). Then (a) There exists a function f such that fn → f uniformly on (a, b); (b) For each x in (a, b), the derivative f’(x) exists and equals g(x). When we reformulate the above theorem in terms of series, we obtain the following: Theorem 1314: Assume that each fn is a realvalued function defined on (a, b) such that the derivative f’n(x) exists for each x in (a, b). Assume that, for at least one point x0 in (a, b), the series Σfn(x0) converges. Assume further that there exists a function g such that Σf’n(x) = g(x) (uniformly on (a, b)). Then (a) There exists a function f such that Σfn(x) = f(x) (uniformly on (a, b)); (b) If x ∈ (a, b), the derivative f’(x) exists and equals Σf’n(x). Sufficient conditions for uniform convergence of a series. In some of the preceding theorems, the importance of uniformly convergent series has been illustrated. It seems natural to seek some simple ways of testing a series for uniform convergence without resorting to the definition in each case. We have seen one such test already (Weierstrass’ Mtest), but when the Mtest is not applicable, the following may be useful: Theorem 1315: (Dirichlet’s test for uniform convergence). Let Fn(x) denote the nth partial sum of the series Σfn(x), where each fn is a complexvalued function defined on a set T. Assume that {Fn} is uniformly bounded on T. Let {gn} be a sequence of realvalued functions such that gn+1(x) ≤ gn(x) for each x in T and for every n = 1, 2, ..., and assume that gn → 0 uniformly on T. Then the series Σfn(x)gn(x) converges uniformly on T. Example: Let Fn(x) = Σk=1n eikx. In a previous chapter, the inequality Fn(x) ≤ 1/sin(x/2) was derived, valid for every real x ≠ 2mπ (m is an integer). Therefore, if 0 < δ < π, we have the estimate Fn(x) ≤ 1/sin(δ/2) if δ ≤ x ≤ 2πδ. Hence, {Fn} is uniformly bounded on the interval [δ, 2πδ]. If {gn} satisfies the conditions of the above theorem, we can conclude that the series Σgn(x)einx converges uniformly on [δ, 2πδ]. In particular, if we take gn(x) = 1/n, this establishes the uniform convergence of the series Σn=1∞ (einx/n) on [δ, 2πδ] if 0 < δ < π. Note that the Weierstrass Mtest cannot be used to establish uniform convergence in this case, since einx = 1.
Bounded convergence. Arzelà’s theorem. Definition 1316. A sequence of functions {fn} is said to be boundedly convergent on T if {fn} is pointwise convergent and uniformly bounded on T. Theorem 1317 (Arzelà). Assume that {fn} is boundedly convergent on [a, b] and suppose each fn is Riemannintegrable on [a, b]. Assume also that the limit function f is Riemannintegrable on [a, b]. Then limn→∞ ∫ba fn(x)dx = ∫ba limn→∞ fn(x)dx = ∫ba f(x)dx. Note: it is easy to give an example of a boundedly convergent sequence {fn} of Riemannintegrable functions whose limit f is not Riemannintegrable. If {r1, r2, ...} denotes the set of rational numbers in [0, 1], define fn(x) to have the value 1 if x = rk for some k = 1, 2, ..., n, and put fn(x) = 0 otherwise. Then the integral ∫01 fn(x)dx = 0 for each n, but the limit function f is not Riemannintegrable on [0, 1]. Mean convergence. Definition 1318. Let {fn} be a sequence of Riemannintegrable functions defined on [a, b]. Assume that f ∈ R on [a, b]. The sequence {fn} is said to converge in the mean to f on [a, b], and we write l.i.m.n→∞ fn = f on [a, b], if limn→∞ ∫ba fn(x)f(x)²dx = 0. If the inequality f(x)fn(x) < ε holds for every x in [a, b], then we have ∫ba f(x)fn(x)²dx ≤ ε²(ba). Therefore, uniform convergence of {fn} to f on [a, b] implies mean convergence, provided that each fn is Riemannintegrable on [a, b]. A rather surprising fact is that convergence in the mean need not imply pointwise convergence at any point of the interval. This can be seen as follows: For each integer n > 0, subdivide the interval [0, 1] into 2n equal subintervals and let I 2 n +k denote that subinterval whose right endpoint is (k+1)/2n, where k = 1, 2, ..., 2n1. This yields a collection {I1, I2, ...} of subintervals of [0, 1] of which the first few are: I1 = [0, 1], I2 = [0, 1/2], I3 = [1/2, 1], I4 = [0, 1/4], I5 = [1/4, 1/2], I6 = [1/2, 3/4], and so forth. Define fn on [0, 1] as follows: fn(x) = 1 if x ∈ In, and fn(x) = 0 if x ∈ [0, 1]—In. Then {fn} converges in the mean to 0, since ∫10fn(x)²dx is the length of In, and this approaches 0 as n→∞. lim sup On the other hand, for each x in [0, 1], we have ndº fn(x) = 1 and lim inf fn(x) = 0. Hence, ndº {fn(x)} does not converge for any x in [0, 1]. Theorem 1319. Assume that l.i.m.n→∞ fn = f on [a, b]. If g ∈ R on [a, b], define h(x) = ∫ba f(t)g(t)dt, hn(x) = ∫xa fn(t)g(t)dt, if x ∈ [a, b]. Then hn→h uniformly on [a, b]. Theorem 1320. Assume that l.i.m.n→∞ fn = f and l.i.m.n→∞ gn = g on [a, b]. Define h(x) = ∫ f(t)g(t)dt, hn(x) = ∫xa fn(t)gn(t)dt, if x ∈ [a, b]. Then hn→h uniformly on [a, b].
x a
Power Series. An infinite series of the form a0+Σn=1∞an(zz0)n, written more briefly as Σn=0∞an(zz0)n, is called a power series in zz0. Here, z, z0 and an (n = 0, 1, 2, ...) are complex numbers. With every power series, there is associated a circle, called the circle of convergence, such that the series converges absolutely for every z interior to this circle and diverges for every z outside this circle. The centre of the circle is at z0 and its radius is called the radius of convergence of the power series. (The radius may be 0 or +∞ in extreme cases.) The next theorem establishes the existence of the circle of convergence and provides us with a way for calculating its radius.
Theorem 1321. Given a power series Σn=0∞ an(zz0)n, let λ = ndº a n , r = 1/λ, (where r = 0 if λ = +∞ and r = +∞ if λ = 0). Then the series converges absolutely if zz0 < r and diverges if zz0 > r. Furthermore, the series converges uniformly on every compact subset interior to the circle of convergence. Note: If the limit limn→∞ an/an+1 exists (or if this limit is +∞), its value is also equal to the radius of convergence of Σn=0∞an(zz0)n. Example: The two series Σn=0∞zn and Σn=1∞an/n² have the same radius of convergence, namely r = 1. On the boundary of the circle of convergence, the first converges nowhere, but the second converges everywhere. This illustrates why the theorem at the top of the page makes no assertion about the behaviour of a power series on the boundary of the circle of convergence. Theorem 1322. Assume that the power series Σn=0∞ an(zz0)n converges for each z in N(z0;r). Then the function f defined by the equation f(z) = Σn=0∞an(zz0)n, if z ∈ N(z0;r), is continuous on N(z0;r). Proof. Since each point in N(z0;r) belongs to some compact subset of N(z0;r), the conclusion follows at once from Theorem 138. Note: The series f(z) = Σn=0∞an(zz0)n, if z ∈ N(z0;r), is said to represent f in N(z0;r). It is also called a power series expansion of f about z0. Functions having power series expansions are continuous inside the circle of convergence. Much more than this is true, however. We will later prove that such functions have derivatives of every order inside the circle of convergence. The proof will make use of the following theorem: Theorem 1323. Assume that Σan(zz0)n converges if z ∈ N(z0;r). Suppose that the equation f(z) = Σn=0∞an(zz0)n is known to be valid for each z in some open subset of N(z0;r). Then, for each point z1 in S, there exists a neighbourhood N(z1;R) ⊂ S in which f has a power series expansion of the form f(z) = Σk=0∞ bk(zz1)k (equation (1)), where bk = Σn=k∞ ( n )a n (z 1 − z 0 ) n−k (k = 0, 1, 2, ...) (equation (2)). Note that the proof of this theorem asserts that k we may use any R > 0 that satisfies the condition N(z1;R) ⊂ S. Theorem 1324. Assume that Σan(zz0)n converges for each z in N(z0;r). Then the function f defined by the equation f(z) = Σn=0∞an(zz0)n, if z ∈ N(z0;r), has a derivative f’(z) for each z in N(z0;r), given by f’(z) = Σn=1∞nan(zz0)n1. Note 1: The two series above have the same radius of convergence. Note 2: By repeated application of f’(z) = Σn=1∞nan(zz0)n1, we find that for each k = 1, 2, ..., the derivative f(k)(z) exists in N(z0;r) and is given by the series f(k)(z) = Σn=k∞ n!/(nk)!an(zz0)nk. If we put z = z0 in this series, we obtain the important formula f(k)(z0) = k!ak (k = 1, 2, ...). This equation tells us that if two power series Σan(zz0)n and Σbn(zz0)n both represent the same function in a neighbourhood N(z0;r), then an = bn for every n. That is, the power series expansion of a function f about a given point z0 is uniquely determined (if it exists al all), and is given by the formula f(z) = Σn=0∞ (f(n)(z0)/n!)(zz0)n, valid for each z in the circle of convergence.
lim sup n
.r)∩N(0. q0 = 1/p0.. Theorem 1327.r). say f(z) = Σn=0∞anzn. Let Γ be a piecewise smooth curve in N(z0. 2. .). We shall restrict our considerations in the remainder of this chapter to real power series. As an application of the substitution theorem. but first we introduce further terminology and notation. if z ∈ N(0. If.r). . if z ∈ N(0... Consider that we are given two power series expansions about the origin. 1.z 1 ) Σn=0∞an(zz0)ndx = Σn=0∞ (an/n+1)(z1z0)n+1. we will show that the reciprocal of a power series in z is again a power series in z. Then we can certainly form the power f (n) (x ) series Σn=0∞ n! 0 (xx0)n. In this section. if z ∈ N(0.h). where p(0) ≠ 0. where cn(p) = m 1 +. Then ck = Σn=0∞anbk(n) for k = 0.δ) in which the reciprocal of p has a power series expansion of the form 1/p(z) = Σn=0∞qnzn.a m p . 2. 1. if z ∈ N(0.) are real numbers. we get f(z)² = Σn=0∞cnzn. all the results of the last three sections can be interpreted in an obvious way as theorems on the real line. Then there exists a neighbourhood N(0.Theorem 1325. we have f(z)p S = Σn=0∞cn(p)zn. 2. provided that the constant term is not 0. If neighbourhoods are taken to mean onedimensional neighbourhoods. Note: The series Σk=0∞ckzk is the power series which arises formally by substituting the series for g(z) in place of z in the expansion of f and then rearranging terms in increasing powers of z.R). we shall deal with the following type of problem: Suppose that we are given a realvalued function f defined in some neighbourhood of a point x0 in E1. then for this z we can write f[g(z)] = Σk=0∞ckzk. If x. Assume that we have p(z) = Σn=0∞pnzn. The symbol S m 1 +m 2 =n indicates that the summation is to be extended over all nonnegative integers m1 and m2 whose sum is n. Note: If the two series are identical.R).R). 1. and g(z) = Σn=0∞bnzn. Multiplication of power series. The substitution theorem. for any integer p > 0. Furthermore. the answer to both questions is “No”. Assume that Σan(zz0)n converges for each z in N(z0.. where cn = Σk=0nakbnk (n = 0. then the product f(z)g(z) is given by the power series f(z)g(z) = Σn=0∞cnzn. x0 and an (n = 0. for a fixed z in N(0.R). Does this series converge for any x besides x = x0? If so. the series Σan(xx0)n is called a real power series. Then °G(z 0 . we have Σn=0∞bnzn < r.
. where the coefficients ck are obtained as follows: Define the numbers bk(n) by the equation g(z)n = (Σk=0∞bkzk)n = Σk=0∞bk(n)zk. say f(z) = Σn=0∞anzn. Real Power Series.+m p =n a m 1 . and g(z) = Σn=0∞bnzn. is its sum equal to f(x)? In general.. Given two power series expansions about the origin.r) joining z0 to z1. if z ∈ N(0. if z ∈ N(0. .r).. Similarly. A necessary and sufficient condition for answering both question in the affirmative can be given by using Taylor’s formula (f(x) = f(x0) f (n) (x ) f (k) (x ) + Σk=1n1 k! 0 (xx0)k + n! 1 (xx0)n).. where cn = Σk=0nakank = S m 1 +m 2 =n a m 1 a m 2 . Theorem 1328. Theorem 1326. and suppose that f has derivatives of every order in this neighbourhood. Its circle of convergence intersects the real axis in an interval called the interval of convergence..
if x0 ≤ x < b. Assume that there is a neighbourhood N(x0) and a constant M (which might depend on x0) such that f(n)(x) ≤ M for every x in N(x0)∩[a. we have f(x) = Σk=0∞ (f(k)(x0)/k!)(xx0)k. Hence a necessary and sufficient condition for the Taylor’s series to converge to f(x) is that limn→∞(f(n)(x1)/n!)(xx0)n = 0. is valid for x in N(x0. but we have no guarantee that it represents f(x) if x < x0. so that it extends to the left of x0. the power series Σn=0∞(f(n)(x0)/n!)(xx0)n is called the Taylor’s series about x0 generated by f. To indicate that f generates this series. Note that Bernstein’s Theorem gives us a power series expansion about x0 valid in a halfopen interval of the form [x0. if x0 is an interior point of [a. The binomial series.(a−n+1) a 1. if x0 ≤ x < b. Theorem 1331 (Bernstein’s Theorem). where c > 0 and x < 1. Then f(x) = Σk=0∞ (f(k)(x0)/k!)(xx0)k.... .. b]. and suppose that f and all its derivatives are nonnegative in the halfopen interval [a. b]. this series also converges in a neighbourhood of radius bx0 about x0. b]. for every x0 in [a. we can always increase the range of validity of the series f(x) = Σk=0∞ (f(k)(x0)/k!)(xx0)k. b). where a is an arbitrary real number. As an example illustrating the use of Bernstein’s Theorem. Since (xx0)n/n! → 0 as n → ∞.R) ⊂ (a. Bernstein’s theorem is not n! directly applicable in this case. Theorem 1332. Theorem 1330. If f has derivatives of every order at each point of I. known as the binomial series: (1+x)a = Σn=0∞ ( a xn.. Let f be a realvalued function defined on an interval I in E1. we will obtain the following expansion.. b]. 2. b). Assume that f ∈ C∞ on [a. we can argue as follows: Let f(x) = (1x)c. Assume that f ∈ C∞ on an open interval of the form (aδ. where δ > 0. and ( n ) = . Let R be such that N(x0.Definition 1329.(c+n1)(1x)cn. we write f ∈ C∞ on I. Thus we can state the following sufficient condition for representing a function by a Taylor’s series.
. then. where x1 is some point between x and x0.R). and for every n. x0. we have f(x) = Σk=0n1 (f(k)(x0)/k!)(xx0)k + (f(n)(x1)/n!)(xx0)n. for each x in N(x0)∩[a. Then. However. we have f(x) = Σn=0∞ (f(n)(x0)/n!)(xx0)n. Assume that f satisfies the hypotheses of the previous theorem and suppose that x0 is an interior point of [a. b).. if x0 ≤ x < b. b). In practise it may be quite difficult to deal with this limit because of the unknown position of x1. if 1 < x < n) a(a−1). b]. for every x in [a. b). Then. b] and if x0 ∈ [a. however. provided that x < 1. b] and let x0 ∈ [a. However. The question we are interested in is this: When can we replace the symbol ~ by the symbol =? Taylor’s formula states that if f ∈ C∞ on the closed interval [a. we write f(x) ~ Σn=0∞ (f(n)(x0)/n!)(xx0)n. If f ∈ C∞ on some neighbourhood of a point x0. b). By Theorem 1321. Then f(n)(x) = c(c+1). b] and every n = 1. a suitable upper bound can be obtained for f(n)(x1) and the limit can be shown to be zero. and hence f(n)(x) > 0 for each n. and on n. limn→∞(f(n)(x1)/n!)(xx0)n = 0 will certainly hold if the sequence {f(n)} is uniformly bounded on [a. b). The point x1 depends on x. In some cases.
we have Σn=0∞cn = (Σn=0∞an)(Σn=0∞bn). we may apply Theorem 1332. if a is a positive integer. Assume that we have f(x) = Σn=0∞ anxn. However. Let f(x) = Σn=0∞anxn for 1 < x < 1.
Abel’s limit theorem. A large number of such results are now known and they are referred to as Tauberian theorems. Of course. As an application of Abel’s theorem. Then f(x) = 1/(1+x) if 1 < x < 1 and f(x) → ½ as x → 1. then ( m ) = 0 for n > m. Theorem 1333 (Abel’s limit theorem). if 1 < x < 1. n n reduces to a finite sum (the Binomial Theorem).
. say a = m. if f is given by f(x) = Σn=0∞ anxn. If 1 < x < 1. Tauber (1897) discovered that by placing further restrictions on the coefficients an. The simplest of these. then Σn=0∞an converges and has sum S. If we put x = 1 in the righthand side of log(1x) = Σn=1∞ (xn/n). Replacing c by a and x by x in the above. Can we also put x = 1 in the lefthand side? The next theorem answers this question in the affirmative. For example. also valid for 1 < x < 1. is valid for each a < 0. We find (1−x) c = Σk=0∞( −c )(1)kxk. and assume that limn→∞ nan = 0. take an = (1)n. sometimes called Tauber’s first theorem. Example: we may put x = 1 in log(1x) = Σn=1∞ (xn/n) to obtain log 2 = Σn=1∞ ((1)n+1/n). with x0 = 0. and the formula (1+x)a = Σn=0∞ ( a )xn. we can derive the following result on multiplication of series: Theorem 1334. if 1 < x < 1. If f(x) → S as x → 1. one can obtain a converse to Abel’s theorem. The converse of Abel’s limit theorem is false in general. That is. Let Σn=0∞an and Σn=0∞bn be two convergent series and let Σn=0∞cn denote their Cauchy product. namely Σ(1)n+1/n. Tauber’s theorem. then the limit f(r) may exist but yet the series Σanrn may fail to converge. and if r < x < r. then the limit limx→rf(x) exists and we have limx→rf(x) = Σn=0∞ anrn. integration of the geometric series 1/1x = Σn=0∞xn gives us the series expansion log(1x) = Σn=1∞ (xn/n). If Σn=0∞cn converges. we find that the k formula (1+x)a = Σn=0∞ ( a )xn. we obtain a convergent alternating series. Σ(1)n diverges. is the following: Theorem 1335 (Tauber). if 1 < x < 1. if r < x < r. A. a = 1 and b = 1. If the series also converges at x = r.1 Therefore. But now the validity of this n formula can be extended to all real a by successive integration.
gn(x) = 1/n if x = 0 or if x is irrational. . Let hn(x) = fn(x)gn(x). 132..
. . Show that {fn} is uniformly bounded on S. fN. Show that {hn} does not converge uniformly on any finite interval. (a) (b) Show that both {fn} and {gn} converge uniformly on every finite interval.. Choosing n = N+1 for simplicity. If fn → f uniformly on S. all function values fn(x) are within ε of f(x). then there exists a constant Mn such that fn(x) ≤ Mn for all x in S. 2. and another bound (MN+1+2ε) for fN+1.. say x = a/b. we have shown that the sequence {fn} is uniformly bounded. The number M is called a uniform bound for {fn}. n = 1. In other words. then O is an upper bound for each the fi from i = 1 to i = N. Defining P = max(O. 2. there exists an N (depending only on ε) such that n > N implies fn(x)f(x) < ε for every x in S. where N is associated with a particular ε > 0.. Knowing that fn(x) is bounded above by the value Mn. f2. But f(x) itself is bounded above as it is within ε of every function value fn(x) (using ab = ba). We now know that the sequence {fn} is uniformly bounded on S if we have n > N. and gn(x) = b+1/n is x is rational.. and so every value of fn(x) for n > N is bounded above by Mn+2ε. But what about the function values f1. Define two sequences {fn} and {gn} as follows: fn(x) = x(1+1/n). b > 0. max each of these is bounded above by a particular M. Summary: we now have a bound (O) for f1. fN? Well.. . .... If each fn is bounded on S. If fn → f uniformly on S. then for every ε > 0.. and that P is an uniform bound for {fn}. .2. we can say that fn(x) is bounded above for n > N by the value MN+1+2ε. it follows that P is an upper bound such that fn(x) ≤ P for all x in S and all n = 1.. fN+2. MN+1+2ε).. if x ∈ E1. Answer: A sequence {fn} is said to be uniformly bounded on S if there exists a constant M > 0 such that fn(x) ≤ M for all x in S and all n = 1.Chapter 13: Selected Exercises
131.. In other words. Assume that fn → f uniformly on S and that each fn is bounded on S... f2... it follows that f(x) is bounded above by Mn+ε.. . every value of fn(x) for n > N is bounded above by f(x)+ε. then given an ε > 0 and an associated (finite) N. we know that for all values of n > N. QED.. 2.N Mi. so if we define O = i=1..
Show that hn → h uniformly on S. then we have g(fn(x))g(f(x)) < ε1. x(1+1/n)x < ε. Claim: f(x) = x.e. if x ∈ S. and because fn(x) ≤ M for all x in S and all n. so that if we choose n > max{c. then given a δ1 > 0 (the same one as above). there exists an N (depending only on ε) such that n > N implies gn(x)g(x) < ε for all x ∈ S. gn(x)g(x) = 1/n = 1/n. We need to find an N such that n > N implies fn(x)f(x) < ε. we need to show that for every ε > 0.Answer: (a) Consider that our arbitrary finite interval is given by S = [c. M). Therefore. δ1 > 0 (dependent only on ε1) such that if xy < δ1. i. Claim: g(x) = 0 if x = 0 or if x is irrational. To prove this claim. consider that our arbitrary interval is again given by S = [c. For fn(x). there exists an N (the same one as above — which was dependent only on δ1. 134. then g(x)g(y) < ε1. We want to show that gn(x) → g(x) on [c. But if fn(x)f(x) < δ1. This happens exactly when n > 1/ε. This proves our initial claim. It follows that given an ε1 > 0.
. there exists an N (depending only on ε) such that n > N implies fn(x)f(x) < ε for all x ∈ S. then there exists an N such that for all n > N and all x ∈ S. we need to show that for every ε > 0. so that N = ceiling(1/ε) will do. then x/n < ε certainly holds. In other words. Consider an arbitrary ε > 0. hn → h uniformly on S. with b > 0. given an arbitrary ε > 0. d}/ε. given an ε1 > 0. d]. Let g be continuous on the closed disk N M) and define (0. For gn(x). then x/n < ε certainly holds. Therefore. we want to show that fn(x) → f(x) on [c. Answer: Because g is continuous on N(0. d] for some function g(x). To prove this claim. d]. In all cases (with our definition of g(x)). we have fn(x)f(x) < δ1. which in turn was only dependent on ε1 — so the N is only dependent on ε1) such that if n > N and if x ∈ S. there exists a (0. So we need to find an N such that n > N implies 1/n < ε. d}/ε) such that n > N implies fn(x)f(x) < ε for all x ∈ S. we have found a suitable N (N = ceiling(max{c. Assume that fn → f uniformly on S and suppose there is a constant M > 0 such that fn(x) ≤ M for all x in S and all n. d}. But we know that x ≤ max{c. we have found a suitable N (N = ceiling(1/ε)) such that n > N implies gn(x)g(x) < ε for all x ∈ S. then we have g(fn(x))g(f(x)) < ε1. Consider an arbitrary ε > 0. d] for some function f(x). and g(x) = b if x is rational. given an arbitrary ε > 0. Because we know that fn → f uniformly on S.e. then g is uniformly continuous on N M). This proves our second claim. x/n < ε for all x ∈ S. say x = a/b. x/n < ε. x/n < ε If we choose n > x/ε. We need to find an N such that n > N implies gn(x)g(x) < ε. h(x) = g[f(x)]. i. QED. hn(x) = g[fn(x)].
x0 ∈ S). where δ3 = min{δ1.136. some k in A satisfies fk(x)f(x) < δ. Now what can we say about fk(x)f(x)? If we add equations (1) and (2) together.. If we now choose any arbitrary δ > 0. 2. Assume that fn → f uniformly on S. If fn → f uniformly on S. kr}. then fk(x)fk(x0) < ε/3 (x. Assume that conditions (i) and (ii) hold.
Hint: To prove the sufficiency of (i) and (ii). show that for each x0 in S there is a neighbourhood N(x0) and an integer k (depending on x0) such that fk(x)f(x) < δ if x ∈ N(x0). In other words. for every given integer k > 0. then f(x)f(x0) < ε/3 ((1)). Therefore. and given an ε > 0. there always exists an m > 0 (dependent only on ε) such that if n > m. then f is uniformly continuous on S. say A = {k1. We need to show that fn → f uniformly on S. . Uniform convergence is an easy consequence of this fact. given an ε > 0. 2. then fk(x)fk(x0) < ε/3 ((2)). there is a δ1 > 0 (dependent only on ε) such that if we have xx0 < δ1. For every ε > 0. then for every x0 ∈ S. This proves part (i).. x0 ∈ S). there exists an N > 0 (depending only on x0 and on ε) such that n > N implies fn(x0)f(x0) < ε/3. Now because we know that {fn} converges pointwise on the set S... the functions fn are all uniformly continuous on the set S. then for a given x ∈ Nδ3(x0). then given an ε > 0.. Answer: If. and only if. . This proves part (ii). . we know that given an ε > 0. Conclusion: For every ε > 0. In other words. Show that fn → f uniformly on S if. Only If. if x ∈ Nδ1(x0). there is a δ2 > 0 such that if xx0 < δ2.. there exists an m > 0 and a δ > 0 such that n > m and fk(x)f(x) < δ implies fk+n(x)f(x) < ε for all x in S and all k = 1. δ2}.. It therefore follows by Theorem 133 that the limit function f is continuous on S. Because the limit function f is continuous on S. the following two conditions hold: (i) (ii) The limit function f is continuous on S. We know that every function fn is continuous at every point x0 in S. Similarly.. Therefore. then fn(x)f(x) < ε. Let {fn} be a sequence of continuous functions defined on a compact set S and assume that {fn} converges pointwise on S to a limit function f. then we automatically have fk+n(x)f(x) < ε precisely because fn(x)f(x) < ε. By compactness. if x ∈ Nδ2(x0). for each x in S. a finite set of integers. and because S is a compact set. then f(x)f(x0) < ε/3 (x. there exists an m > 0 and a δ > 0 such that n > m and fk(x)f(x) < δ implies fk+n(x)f(x) < ε for all x in S and all k = 1. Our task is to find an N (depending only on a given ε > 0) such that n > N implies fn(x)f(x) < ε for all x ∈ S. then if we have fk(x)f(x) < δ. we have (see over)
. has the property that.
Let us now define an open covering of S given by the union of all neighbourhoods 4 Nδ3(x).. Use the sequence in Exercise 135(a) to show that compactness of S is essential in Dini’s theorem..f(x)f(x0) + fk(x)fk(x0) < ε/3 + ε/3 f(x0)f(x) + fk(x)fk(x0) < 2ε/3 f(x0)f(x)+fk(x)fk(x0) < 2ε/3 fk(x)f(x) + (f(x0)fk(x0)) < 2ε/3.
. we have an integer ki so that for any x in a particular neighbourhood. Nδ3(x2). Therefore. Therefore.. 137.
But we already know that condition (i) holds because it is assumed in the question. so that a finite number of such neighbourhoods cover the set S. . and given an ε > 0.. we have fn(x)f(x) < ε for any n > K.. and so fk(x)f(x) < ε/3 + 2ε/3 (for all x ∈ Nδ3(x0))..
(b)
Answer: (a) Exercise 136 says the following: Let {fn} be a sequence of continuous functions defined on a compact set S and assume that {fn} converges pointwise on S to a limit function f. there is a neighbourhood Nδ3(x0) and an integer k > N (where N corresponds to a particular ε > 0 and is dependent on x0) such that fk(x)f(x) < ε for x ∈ Nδ3(x0). Therefore.. (a) Use Exercise 136 to prove the following theorem of Dini: If {fn} is a sequence of realvalued continuous functions converging pointwise to a continuous limit function f on a compact set S. 2. then fn → f uniformly on S. . Because S is a compact set.. where each xi ∈ S. we conclude that fn → f uniformly on S. Therefore. QED. this open covering of S reduces to a finite covering of S. .. all we need show is that condition (ii) holds. Let this collection of neighbourhoods be denoted as Q = {Nδ3(x1). we have defined an open covering P = xcSNδ3(x) ⊃ S. where i = ik 1. q. To show that fn → f uniformly on S. where K depends solely on ε. . For every ε > 0. then f(x0)fk(x0) < ε/3 for all k > N. fk(x)f(x) < ε (for all x ∈ Nδ3(x0)). for each x0 ∈ S. to show that fn → f uniformly on S. If we define K = max i. we have fn(x)f(x) < ε for n > ki.. there exists an m > 0 and a δ > 0 such that n > m and fk(x)f(x) < δ implies fk+n(x)f(x) < ε for all x in S and all k = 1. and if fn(x) > fn+1(x) for each x in S and every n = 1. Nδ3(xq)}. we only need to show that the following two conditions hold: (i) (ii) The limit function f is continuous on S.
(using AB = BA) (using A+B ≤ A + B)
Now as fk(x0)f(x0) < ε/3 for all k > N. then we see that for any x ∈ S.. 2. where x ∈ S. For each of the above neighbourhoods.
we see that we have fn(x) > fn+1(x) for all x ∈ (0. and therefore we conclude that fn → f uniformly on S. 2.. 1).
. then we can apply Dirichlet’s test for uniform convergence (Theorem 1315) to say that the series Σfn(x)gn(x) = Σ(1)n+1gn(x) converges uniformly on T. then given any δ > 0. 1310. and thus fn(x) is larger than fn+1(x). The function f(x) to which the functions fn(x) converge pointwise to is f(x) = 0 (as n → ∞... then we definitely have fn(x) > fn+1(x) for all x in (0. But Exercise 135(a) shows that {fn} does not converge uniformly on (0. This shows that condition (ii) above holds. and suppose that gn → 0 uniformly on T.
1
(b) Recall that the sequence in Exercise 135(a) was the following sequence: fn(x) = /(nx+1) for 0 < x < 1 and n = 1. for every ε > 0. Let Fn(x) denote the nth partial sum of the series Σfn(x). then it follows immediately that fk+n(x)f(x) < ε for any integer n because we know (from the question) that fn(x) > fn+1(x) for all x ∈ S.. we note that the set S = (0.. . Looking at the information on the previous page.. we note that all the conditions of Dini’s theorem are satisfied but for the compactness of the set S on which the functions fn and the function f are defined upon. there will always be an integer n such that fk+n(x)f(x) < ε. 3. Assume that gn+1(x) ≤ gn(x) for each x in T and each n = 1. Finally. Show that Σ(1)n+1gn(x) converges uniformly on T. knowing that {gn} is a sequence of realvalued functions such that gn+1(x) ≤ gn(x) for each x in T and for all n = 1. then again because fn(x) > fn+1(x) for all x ∈ S. we conclude that the compactness of S is essential in Dini’s theorem. (b) If ε < δ. QED. . F3(x) = (1)²+(1)³+(1)4 = 11+1 = 1. 2.. But if we are then given another number ε > 0. 1): if we fix x... there exists an m > 0 and a δ > 0 such that n > m and fk(x)f(x) < δ implies fk+n(x)f(x) < ε for all x in S and all k = 1. From the above. QED. QED. Therefore. as fn(x) > fn+1(x). It follows immediately that f(x) is a continuous function because it is a constant function. Therefore. 1). the functions fn(x) converge pointwise to a function f(x) (Exercise 135 asks you to show that this is true). 2. 2. the denominator of fn(x) tends to ∞ and thus fn(x) tends to 0). F4(x) = (1)²+(1)³+(1)4+(1)5 = 11+11 = 0. every function fn(x) is continuous because fn(x) is a rational function and the range of x does not include the single discontinuity of fn(x) which occurs at x = 1/n for each n = 1. etc. . F2(x) = (1)²+(1)³ = 11 = 0... Answer: Let us define the function fn(x) = (1)n+1 defined for all x ∈ T. we can always find an integer k such that fk(x)f(x) < δ.Because we know that the sequence of functions {fn} converges pointwise to a continuous limit function f. Let us analyse the properties of this sequence of functions. then (a) if ε > δ. Now F1(x) = (1)² = 1. Thirdly.
First of all. Therefore. 2. . Knowing this. we see that {Fn} is uniformly bounded above by the number 1. and knowing that gn → 0 uniformly on T. .. and because {fn} converges pointwise to f. then we see that the denominator of fn(x) is smaller than the denominator of fn+1(x). 1) is not compact (it is an open set). Secondly.
1] and assume that fn → f uniformly on [0. b]. then there exists a positive real number M such that gn(x) < M for all x ∈ T and for all n = 1. By the above discussion. If {gn} is uniformly bounded on T and if Σfn(x) converges uniformly on T.. Prove or disprove 1 lim 1−1/n f n (x)dx = ° 0 f(x)dx. Knowing that Σfn(x) converges uniformly on T. given an ε > 0. 1314. 2.. 1]? To be uniformly bounded. This proves that Σfn(x)gn(x) converges uniformly on T. . Assume also that the limit function f is Riemannintegrable on [a.1311... 1]. Then limn→∞ ∫ba fn(x)dx = ∫ba limn→∞ fn(x)dx = ∫ba f(x)dx. ndº ° 0 Answer: In this question. . then every function fn is uniformly continuous on [0. So. 1] (see Exercise 131). 2. there n 2 +p exists an integer N2 such that if n2 > N2. then we know that given an ε2 > 0. and for every x in T. Answer: If {gn} is uniformly bounded on T.. 1]. there exists an integer N2 (associated with ε2 and thus now with ε) such that if n > N2. we shall apply Theorem 1317: Theorem (Arzelà): Assume that {fn} is boundedly convergent on [a.
. b]. . we need to show that for every ε > 0. To prove that Σfn(x)gn(x) converges uniformly on T. 2. 1].. 2. n 2 +p n 2 +p  S k=n 2 +1 fk(x)gk(x) ≤  S k=n 2 +1 fk(x)×M < ε2×M. . Prove Abel’s test for uniform convergence: Let {gn} be a sequence of realvalued functions such that gn+1(x) ≤ gn(x) for each x in T and for every n = 1. . or ε2×M < ε. 1]? To be boundedly convergent. pick ε2 so that ε2 < ε/M. and for every x in T. then Σfn(x)gn(x) also converges uniformly on T. then we certainly know that {fn} is pointwise convergent. Now because {gn} is uniformly bounded above by M. We know that our sequence {fn} is a sequence of realvalued continuous functions defined on [0.. 1] is a compact set (every closed and bounded set such as [0. b] and suppose each fn is Riemannintegrable on [a. Is {fn} boundedly convergent on [0. then we have n+p  S k=n+1 fk(x)gk(x) < ε2×M < ε for each p = 1. Because each function fn is continuous and because [0.. 1]. there is an integer N such that n > N n+p implies that we have  S k=n+1 fk(x)gk(x) < ε for each p = 1. 1] (see Theorem 424). then we have  S k=n 2 +1 fk(x) < ε2 for each p = 1. 1] is a compact set — see the note following definition 339). Let {fn} be a sequence of realvalued continuous functions defined on [0... QED. {fn} must be pointwise convergent and uniformly bounded on [0. and for every x in T. and we know that fn → f uniformly on [0. Is {fn} uniformly bounded on [0. 2. Therefore. all we must show is that each individual function is bounded because we know that fn → f uniformly on [0. 1]. we have (for any integers n and p) n+p n+p  S k=n+1 fk(x)gk(x) ≤  S k=n+1 fk(x)×M. Because we know that fn → f uniformly on [0.
Looking at Theorem 927. and if we have n > N. 1] — so that we can apply part (i) of Theorem 136 to say that f is continuous on [0. Now that we have shown that all the conditions of Theorem 1317 have been satisfied when applied to this exercise. and knowing that each function fn is continuous on [0. and so we have proved the hypothesis asked in the question. 1]. and that fn → f uniformly on [0. because f is continuous on [0. if we choose N = ceiling(1/ε²). 1] for all x ∈ E1. 1] — see Exercise 423.e. 1] (we know that {fn} is a sequence of continuous functions defined on a compact set [0. 1]).
. Answer: Let us define fn(x) =((1)n/√(n))sin(1+(x/n)). we get f’n(x) = ((1)n/n×√(n))cos(1+(x/n)). 1/n < ε². Thus the claim that an converges to zero has been proven. we can say the following: lim 1 1 lim 1 n→∞ ∫ 0 fn(x)dx = ∫ 0 n→∞ fn(x)dx = ∫ 0 f(x)dx. Let us now define the terms an = 1/√n.. and we can prove that an → 0 as n → ∞: Claim: limn→∞ an = 0. 1] as well.Because each function fn is uniformly continuous on the bounded set [0.. then we get an0 < ε. i. To prove the above claim. then an0 < ε. 2. knowing that each function fn is bounded on [0. But as limn→∞ ∫10 fn(x)dx = limn→∞ ° 0 fn(x)dx (as limn→∞ 11/n = 1).. or an < ε. Therefore. Similarly. n > 1/ε².
If n = 1. then we can say that 1 lim 1−1/n f n (x)dx = ° 0 f(x)dx. then we can say with certainty that each fn is Riemann integrable on [0. 1]. It is obvious that an is a decreasing sequence. we need to find an N such that if n > N. then the series Σfn(0) is given by Σfn(0) = Σn=1∞ ((1)n/√(n))sin(1+(0/n)) = sin(1)Σn=1∞ ((1)n/√(n)). 1/√n < ε. If we consider the point x = 0. that an+1 ≤ an. we know that {fn} is uniformly bounded on [0. 1]. The next question to ask ascertains as to whether each function fn is Riemann integrable or not on [0. Taking the derivative. then we see that the derivative f’n(x) exists for all x ∈ E1 as cos(1+x/n) ∈ [1. Therefore. 3.. ndº ° 0
1−1/n
1317. 1]. 1]. 1]. . Show that the series Σn=1∞ ((1)n/√(n))sin(1+(x/n)) converges uniformly on E1. QED. and so {fn} is boundedly convergent on [0. 1]. then f is Riemann integrable on [0. then given an arbitrary ε > 0. then f must be bounded on the set [0. 1/√n < ε. 1].
1] for all z). QED. The red bit is also bounded above by 1 for all possible n because 1/(n×√n) = 1/n3/2 is another decreasing sequence bounded above by 1 (for n = 1. Because of the above. Our next task is to show that the series Σf’n(x) converges uniformly on E1. 0 ≤ fn(x) ≤ an and 0 ≤ gn(x) ≤ an.. 3. . Chapter 8. we have 0 ≤ f’n(x) ≤ Mn.. we can say that because cos(1+(x/n)) ∈ [1. thus proving that the series Σfn(x) converges for at least one point x0 in E1. Answer: Given any numbers n and x. and because sin(z) ∈ [1. 1] for any real number z. Applying Weierstrass’ Mtest. Let us define fn(x) = ansin(nx) and define gn(x) = ancos(nx). Looking at Σf’n(x) = Σn=1∞ ((1)n/n×√(n))cos(1+(x/n)). Therefore. and that gn(x) ≤ an×1 = an. Therefore. we use the Integral Test (see Theorem 1223 and Thomas & Finney. knowing that each fn is a realvalued function defined on E1 (if n = 1. we see that fn(x) ≤ an×1 = an. By Weierstrass’ Mtest.. recall that sin(nx) ≤ 1 and cos(nx) ≤ 1. we have the conclusion that Σf’n(x) converges uniformly on E1. . we see that Σn=1∞ fn(x) = Σn=1∞ ansin(nx) converges uniformly on E1 if Σn=1∞ an converges. 2. Show that Σn=1∞ ansin(nx) and Σn=1∞ ancos(nx) are uniformly convergent on E1 if Σn=1∞ an converges. Therefore. In other words. Therefore. we can split it up into two parts: Σf’n(x) = Σn=1∞((1)n/n×√(n))cos(1+(x/n)) = Σn=1∞((1)n/n×√(n))×cos(1+(x/n)) Now the blue bit is bounded above by 1 for all possible x and n as cos(z) ∈ [1. we can say that the series Σfn(0) converges (Σn=1∞ ((1)n/√(n)) converges so sin(1)Σn=1∞ ((1)n/√(n)) converges as well). 1319. To do this. and knowing that Σf’n(x) converges uniformly on E1. QED. 1] for all x and for all n. f’n(x) cannot possibly be larger than this if cos(1+(x/n)) ∈ [1. In other words. using this theorem.. knowing that the derivative f’n(x) exists for each x in E1. 2.Theorem 1216 says that if {an} is a decreasing sequence converging to zero. we can apply Theorem 1314 to say that the series Σfn(x) converges uniformly on E1. then the alternating series Σ(1)nan converges.. then each term f’n(x) is bounded above by ((1)n/n3/2)×1 or simply by (1/n3/2) (to repeat. and that Σn=1∞ gn(x) = Σn=1∞ ancos(nx) converges uniformly on E1 if Σn=1∞ an converges. knowing that the series Σfn(x0) converges for x0 = 0.). Example 3: The pseries Σn=1∞ 1/np converges if p > 1 and diverges if p ≤ 1). which our series {an} is. to show that Σf’n(x) converges uniformly on E1. where Mn = 1/n3/2. we have shown that because ΣMn converges. we shall use Weierstrass’ Mtest (Theorem 137).
. all we need do is to show that ΣMn converges. 1]). Section 8.4. To show that ΣMn = 1/n3/2 converges.
.n→∞ fn = f on [a.(0/2+sin(0)/4)] = π/2.. we find that ∫cos²ax = x/2 + sin(2ax)/4a + C. We can apply the result from Thomas & Finney iteratively... p3 = p2(7/8). leaving us to prove that
π lim n→∞ 0
∫ (cosnx)²dx = 0. Assume that f ∈ R on [a.
Looking at the back of Thomas & Finney in the Table of Integrals. b].. etc. and we write l..( / )( / )( / ). b]. ∫0πcos²xdx = [x/2+sin(2x)/4]π0 = [(π/2+sin(2π)/4) . (a) (b) Show that l.m..m. b]. p2 = 3/4×5/6. Therefore. Show that {fn} converges pointwise but not uniformly on [0..1325. Examples: p1 = ¾. or that limn→∞ ∫0π cos2nxdx = 0.. Let fn(x) = cosnx if 0 ≤ x ≤ π. π] but that {fn(π)} does not converge..n→∞ fn = 0 on [0. if limn→∞ ∫ba fn(x)f(x)²dx = 0.( / )( / )( / )∫ cos xdx = ( / )( / )( / ).( / )( / )( / )( / ). so we obtain lim 2n1 /2n)∫0π cos2nxdx n→∞ ( = limn→∞ (2n1/2n)(2n3/2n2)(2n5/2n4). we need to show that limn→∞ ∫0π (cosnx0)²dx = 0. recall the following definition from page 5 of the notes: Definition 1318. b]. the above reduces to lim 2n1 /2n)∫0π cos2n2xdx. Now as sin(0) = 0 and as sin(π) = 0. =(/) ( / )( / )( / ). We can remove the modulus signs because we are squaring one term.i. n ∫ cos2nxdx = ([(cos2n1(x)sin(x))/n]π0 + (2n1/2n)∫0π cos2n2(x)dx).i. Looking at the Table of Integrals again.
Answer: Before we start on the answer..
.(7/8)(5/6)(3/4)∫0π cos2xdx. (a) Looking at the above definition. The sequence {fn} is said to converge in the mean to f on [a. π/2]. n→∞ ( Therefore. It follows that / )( / )( / ). we find that n−1 ax ∫cosnaxdx = cos na sin ax + n−1 ° cos n−2 axdx. =(/) Π ( / ) = ( / )Π ( / )
π lim 2n1 2n3 2n5 7 5 3 2 n→∞ 2n 2n2 2n4 8 6 4 0 π lim 2n1 2n3 2n5 7 5 3 n→∞ 2n 2n2 2n4 8 6 4 2 π lim 2n1 2n3 2n5 7 5 3 2 n→∞ 2n 2n2 2n4 8 6 4 π lim n 1+2k 2 n→∞ k=1 2+2k π ∞ 1+2k 2 k=1 2+2k π lim n→∞ 0 lim n→∞
(
Now let us define pn = Πk=1n(1+2k/2+2k).. Let {fn} be a sequence of Riemannintegrable functions defined on [a.
Note that this will only work if y ∈ [0. we see that {pn} is a strictly monotonically decreasing sequence bounded above by ¾.
. I will not prove that it converges to zero and assume that this is the case. It remains to show that {fn} does not converge uniformly on [0. 1]. and to zero if x ∈ (0. QED. 1] but that l. 1].From the above. π]. then because cos(0) = 1. we have yn < ε. (b) We now want to show that {fn} converges pointwise on [0. We can do this simply by observing that fn(π) = (1)n for all n. and it follows as π π ∞ 1+2k lim n /2+2k) = 0. It follows that cosn(x) converges pointwise to 1 if x = 0. limn→∞ pn = 0. a series that alternates between 1 and 1 and so does not converge. Therefore. Answer: To show that {fn} converges pointwise to 0 on [0. 1) (which is all right!) and if ε ∈ (0. we need to find an N such that if n > N. and when x ∈ (0. It follows that fn(x) converges pointwise for x ∈ [0. given an arbitrary ε > 0. then yn < ε as required. It follows that if we set y = cos(x) (which we can do because cos(x) ∈ [0. we have just proved that limn→∞ yn = 0. To show this. n→∞ ∫0 (cos x0)²dx = ( /2)Πk=1 ( Our next task is to show that {fn(π)} does not converge. I will skip this! 1326. π/2].n→∞fn = 0 on [0. Therefore. it is bounded). 1). There are two cases to deal with here: when x = 0. In order to prove this claim.m. we have cosn(0) = 1n = 1 for all n. then n = logε/logy.n→∞ fn ≠ 0 on [0. 1). Show that {fn} converges pointwise to 0 on [0. then cos(x) ∈ [0. for every x ∈ [0. we must show that limn→∞ fn(x) = 0 for all x ∈ [0. 1]. But if ε > 1. and so if n > ceiling(logε/logy). then limn→∞ cosnx = 0. and let fn(x) = n if 1/n < x < 2/n. then yn < ε holds automatically for y ∈ [0.i. we must find an N so that if n > N. 1). 1]. 1). π/2]. π/2]. we can say that l.i. we can say that the sequence {pn} converges. Let fn(x) = 0 if 0 ≤ x ≤ 1/n or if 2/n ≤ x ≤ 1. 1). If x = 0. But if yn = ε. and only if. π/2]. π/2]: to one if x = 0. But what does it converge to? It must converge to zero because (intuitively) we are constantly multiplying together numbers in the range (0. Claim: limn→∞ yn = 0. but just note that a proof can be obtained by modifying the proof in Exercise 45 or the proof that appears below in part (b). 1) for x ∈ (0. Therefore. and for a given ε > 0. and so (π/2)Πk=1∞ (1+2k/2+2k) = (π/2)×0 = 0. If x ∈ (0. we have fn(x) < ε. so we don’t have to worry about this case. π/2]. π/2]). Now consider an arbitrary y ∈ [0. applying Theorem 126 (A monotonic sequence converges if.m.
then we have x ∈ [2/n. then for all n > N.Now if x = 0.
z−z 0  an lim ndº  a n+1

. if N = ceiling(2/x). and so fn(x) = 0 < ε for all n > N. all we need do is to find an N so that if n > N. Comparing the above to the definition of radius of convergence in lim a n Theorem 1321. we have x ∈ [2/n. we have found an N (dependent on x) so that if n > N. we have 2/n < x.i. QED. the series converges absolutely when r < 1. To do this. If this is the case. If x ∈ (0. and z−z 0  diverges when r > 1.. then in order to prove that limn→∞ fn(x) = 0. By the ratio test (not the one shown in Theorem 1225 — see the one for example in Thomas & z−z 0  Finney. 1].n→∞ fn ≠ 0 on [0. 1] so that fn(x) = 0 < ε for all n > N. then 2/n < x. that is.. different to the definition that forms part of Theorem 1321 (that derives from the root test. we get ∫ fn(x)0²dx = ∫10 fn(x)²dx = [∫01/n0²dx + ∫1/n2/n n²dx + ∫2/n1 0²dx] = ∫1/n2/n n²dx = [n²x]1/n2/n = [n²(2/n1/n)] = [n(21)] = n = ∞ ≠ 0. then we see that
. Section 8. then 2/n < x as required. If r = ∞. then fn(x) = 0 (by definition) for all n. 2.6). we must show that lim 1 n→∞∫ 0 fn(x)0²dx ≠ 0. and so limn→∞ fn(x) = 0 for all x ∈ (0. For each n > 1.e. the test shows divergence except for z = z0 (I lim a n will not prove this). Claim: r is the radius of convergence of the series Σan(zz0)n. that is. Therefore. define rn =  n+1 (z−z 0 ) n  =  a n z − z 0 . If we define r =
lim a n ndº  a n+1
. zz0 < r. Claim: For a particular x ∈ (0. limn→∞ fn(x) = 0 if x = 0. Consider the a (z−z ) n+1 a n+1 power series Σan(zz0)n. 1]. zz0 > r. If r = 0. the test shows that the series converges absolutely for all z.)
1327. then for all n > N. an 0 Now
lim ndº r n z−z  lim ndº r n
=
lim a n+1 ndº  a n
z − z 0  =
= r 0 . Proof of claim: if n > 2/x. It automatically follows that if n > ceiling(2/x). then we can manipulate to give 2/n < x as required (the inequality doesn’t change because we have n > 0 and x > 0). it follows that the r defined above as r = ndº  a n+1 is the radius of convergence of the series Σan(zz0)n. Theorem 1226). 1].
Answer: The important factor about this question is to show that the radius of convergence can be defined in another way (under certain conditions). It remains to show that l. Therefore. Proof of Claim. Consider that we derive a definition for r by considering the ratio test. and so limn→∞ fn(x) = 0 for this particular x ∈ (0. Manipulating the left hand side of this equation. fn(x) < ε for all ε > 0 and all n if x = 0. show that lim sup a n lim inf a n ndº  a n+1  [ r [ ndº  a n+1 . End of Proof.m. i.
If r is the radius of convergence of Σan(zz0)n.
lim 1 n→∞ 0 lim n→∞ lim n→∞ lim n→∞ lim n→∞ lim n→∞ lim n→∞ lim n→∞
(because n = 1. 1]. 1]. . so that ndº  a n+1  = 0. 1].
2 −1/x 2 e −0 x3 2 −1/x 2 e x3
Case n = 2.
.
lim f(x)−f(c) xdc x−c . f’(0) = xd0 x−0 = xd0 e x−0−0 = xd0 1 e −1/x . f(n)(0) exists for every n — and we have f(n)(0) = 0 for every n. no matter which n is chosen. then we shall always have 2 lim f(n)(0) = tdº tf(t)e −t = 0. then t → ∞ as x → 0+. If x ≠ 0. 2 2 lim lim lim 1 −1/x 2 = xd0 − 1 e −1/x = tdº te −t = 0. we have
lim xd0 lim xd0
2 −1/x 2 e x3 2
lim xd0
x
. QA303. in particular. lim sup a n an or lim inf  a n+1  [ r [ ndº  a n+1  as required. QA303. then f’’(x) = ( x46 − x
6 −1/x 2 . Therefore. 1982. Note: The above solution was taken from “Advanced Calculus with Applications” by Nicholas J. we have found an alternative definition for the radius of convergence: lim a n r = ndº  a n+1 .
lim = tdº 2t 4 e −t = 0. However. we always get (for x ≠ 0) f(n)(x) = g(1/x)e1/x². 1982. x xd0 x e Hence f’(0) = 0.D4: see section 7. if the limit does not exist. Hence f’’(0) = 0. (a) (b) Show that f(n)(0) exists for all n > 1. ndº 1330. Further. I shall use f’(c) =
f(x)−f(0)
−1/x 2
lim lim lim Case n = 1. QED. Theorem 730. and t → ∞ as x → 0. then f’(x) = (2/x³)e1/x².9. x 1 Let t = /x.8. Show that the Taylor’s series about x0 = 0 generated by f converges everywhere on E1 but that it represents f only at the origin. then there is no problem and we have lim sup a n lim a n lim inf a n ndº  a n+1  = ndº  a n+1  = ndº  a n+1 . f(0) = 0. f’’(0) = = x−0 1 If we use t = /x again. where g(1/x) is some polynomial in 1/x. x 4 )e
We may continue calculating f(n)(0) and f(n)(x) for every positive integer in this manner. then we have (using Theorem 123) lim sup a n lim a n lim inf a n ndº  a n+1  [ ndº  a n+1  [ ndº  a n+1 . Thus we conclude that f ∈ C∞ everywhere in E1.
2
Answer: In this question. DeLillo.D4: see section 7. Exercise 738. since et² approaches 0 more quickly than any polynomial in t approaches infinity. and. However. DeLillo. if we continue to apply t = 1/x.Note: Portions of the above proof were taken from “Advanced Calculus with Applications” by Nicholas J. QED. If x ≠ 0. If this limit exists. Let f(x) = e1/x² if x ≠ 0. Therefore.
e. we see k=0 that f is zero only at x = 0 (e1/x² > 0 for all x ≠ 0).(b) The Taylor’s series about x0 = 0 generated by f is given by
º 0 f(x) = S º k! (x − 0) k = S k=0 k! x k = 0 for all x. so that the Taylor series only represents f at the origin i. QED. f (k) (0)
. But looking at the definition of f. at x = 0.
and gn(x) = 0 for 11/n < x ≤ 1.Possible Further Work / Evaluation
1−1/n Exercise 1314: Further explanation of the result limn→∞ ∫10 fn(x)dx = limn→∞ ° 0 fn(x)dx.
. Possible method: define gn(x) = fn(x) for 0 ≤ x ≤ 11/n. Then 1− 1 1 ° 0 n f n (x) = ° 0 g n (x) and we need to show that gn(x) → f(x) uniformly.
Chapter 15
.
g)]½ [(f. φ0)½ =
cos nx sin nx 1 2o 1 2o o dx = o 0 cos(nx) sin(nx)dx = 2o 0 sin(2nx)dx o 1 −cos(2nx) 2o 1 1 ] 0 = 4no [− cos(4on) + cos(0)] = 4no [−1 + 1] = 0. φ2n1)½ = 1. g) + (g. (φ2n. g)½ + (g. Derive the Minkowski inequality f+g ≤ f + g from the CauchySchwarz inequality. φ2n) = 0. f)½ + (g.
1 o 2
(a) (φ0. (φ2n1.
2o °0
= (c) (φ2n1. φ2n1(x) =
cos nx o .
(c) (e)
(φ2n1.). Now f+g = = = ≤ = = = (f+g.
°
2o °0
= (d) (φ0. Answer: The trigonometric system in question is as follows: φ0(x) =
1 2o
. f) + 2(f. 2π]. 2π]. g)½ f + g as required. Answer: The CauchySchwarz inequality assumes the form (f. f) + (g. f) + (f.. φ2n) = 1. φ2n) = 0. φ2n) =
1 sin nx o dx 2o −cos(2on) 1 [ n o 2
= +
2o 1 −cos nx 2o 1 0 sin(nx)dx = o 2 [ n ] 0 o 2 cos(0) 1 n ] = o 2 [−1 + 1] = 0. Verify that the trigonometric system in equation (6) is orthonormal on [0. and (2) φn = 1 for all n. f)½ + (g.
Now in order to prove that this system is orthonormal on [0. φ2n) =
1 cos nx 1 o dx = o 2 2o sin(2on) sin(0) 1 [ n − n ] o 2
2o ° 0 cos(nx)dx =
[ sinnnx ] 2o 0
=
1 o 2
[0 − 0] = 0. g)]½ [(f.
φ2n(x) =
sin nx o
(n = 1. we must prove the following: (a) (d) (φ0. φm) = 0 whenever m ≠ n. 2o [ 2n 1 2o 1/2 1/2 1 = [ 2o [x] 2o ] 1/2 0
°
°
1 2o ) 2 dx = 2o ° 0 dx 1 = [ 2o [2o − 0] ] 1/2 = 1 = 1. φ2n1) = 0. (φ0. . g) ≤ f g. we must show that the following holds: (1) (φn. 2. φ2n1) =
= (b) (φ0. (by definition) (using elementary property 3) (using elementary property 2) (using CauchySchwarz)
152.
2o °0
(b) (d)
(φ0. g)]½ [ {(f.. QED. f)½(g. f+g)½ [(f. g)½}² ]½ (f. f) + 2(f. we see that in order to prove that it is orthonormal.Chapter 15: Fourier Series and Fourier Integrals
151. Looking at the system in question. g) + (g. φ0)½ = 1.
2o °0 (
.
π/2(1)n . bn = 2/π∫0π xsin(nx)dx = 2/π[1/n²sin(nx) . which can be extended to π ≤ x ≤ π because f(π) = f(π) for a an even function. For f(x) = x on π < x < π. if f has period 2π. where an = 2/π∫0π t²cos(nt)dt.
. so that we can apply the result from Exercise 156 which said that if f ∈ R on [π. and if f(x) = f(x) when 0 < a x < π.x/ncos(nx)]π0 = 2/π[1/n²sin(nπ) . Show that the expansions below are valid in the ranges indicated. then f(x) ~ Σn=1∞bnsin(nx). φ2n)½ =
1 2o o[ 2
+
sin(4on) 4n
−
−
1 o
1 = [ o [o + 0 − 0 − 0] ] 1/2
sin(2nx) 2o 1/2 4n ] 0 1/2
= 1 = 1. QED.0 + 0] = 2/π(π/2(1)n) = 2/n(1)n = 2/n(1)n1. where bn = 2/π∫0π tsin(nt)dt.
Answer: (a) The function f(x) = x is an odd function. For f(x) = x² on π < x < π. if π ≤ x ≤ π. φ2n1)½ =
2o ° 0 ( cos(nx) ) 2 dx o
1/2
=
0 2
1 o
2o ° 0 cos 2 (nx)dx 1/2 sin(0) 4n ]
1/2
=
1 x o[2
+
= (f) (φ2n. where an = 2/π∫0π f(t)cos(nt)dt. we have x² ~ 20 +Σn=1∞ancos(nx).(e) (φ2n1. if f has period 2π. n2
. π].π/ncos(nπ) . n (b) The function f(x) = x² is an even function. we have x ~ Σn=1∞bnsin(nx). then f(x) ~ 20 +Σn=1∞ ancos(nx). so that we can apply the result from Exercise 156 which said that if f ∈ R on [π. (a) (b)
º x = 2 S n=1 n 2 o º x² = 3 + 4 S n=1 (−1) n cos nx (−1) n cos nx . where bn = 2/π∫0π f(t)sin(nt)dt. 159. if π < x < π. Note: I have used some entries from the Table of Integrals at the back of Thomas & Finney to do some of the calculations above. and if f(x) = f(x) when 0 < x < π. π].1/n²sin(0) + 0] = 2/π[0 .
(Result taken from Thomas & Finney)
So for f(x) = x on π < x < π. Now a0 = 2/π∫0π x²cos(0)dt = 2/π[x³/3]π0 = 2/π[π³/30³/3] = 2π³/3π = 2π²/3.
2o ° 0 ( sin(nx) ) 2 dx o
1/2
=
0 2
2o ° 0 sin 2 (nx)dx 1/2 sin(0) 4n ]
=
1 x o[2
−
=
1 2o o[ 2
−
sin(4on) 4n
−
+
1 = [ o [o − 0 − 0 + 0] ]
sin(2nx) 2o 1/2 4n ] 0 1/2
= 1 = 1.
QED. Let us now calculate bn for every n > 1. we have x = Σn=1∞bnsin(nx) = Σn=1∞ 2/n(1)n1sin(nx) n º (−1) cos nx = 2 S n=1 as required.
Let us now calculate an for every n > 1.2π/n .x/ncos(nx)]2π0 =2/nπ[1/n²sin(2πn) . we have a0 = 1/π∫02π x²cos(0)dt = 1/π[x³/3]2π0 = 1/π[(2π)³/3(0)³/3] = 8π³/3π = 8π²/3. we let u = x² so that du/dx = 2x. and let dv/dx = cos(nx) so that v = 1/nsin(nx).∫π0 2x/nsin(nx)dx} = 2/π∫π0 2x/nsin(nx)dx = 4/nπ∫0π xsin(nx)dx =4/nπ[1/n²sin(nx) . where a0 = 2π²/3 and an = 4/n²(1)n.
1
(Result taken from Thomas & Finney)
. QED.2π/ncos(2nπ) . Show that the expansion x² = x < 2π. It follows that /π∫02π x²cos(nx)dx = 1/π{[x²/nsin(nx)]2π0 . It follows that /π∫0π x²cos(nx)dx = 2/π{[x²/nsin(nx)]π0 . Integrating by parts.1/n²sin(0) + 0] =4/nπ[0 .0 + 0] = 4π/n²π(1)n = 4/n²(1)n.π/ncos(nπ) .0 + 0] = 4π/n²π = 4/n². n2 1510.
2
(Result taken from Thomas & Finney)
Using x² ~
a0 ∞ 2 +Σn=1
ancos(nx). Integrating by parts.
4 2 3o º + 4 S n=1 ( cos2nx − n o sin nx n ) is
valid in the range 0 <
Answer: Recall that the Fourier Series generated by f is given by
º f(x) ~ 20 + S n=1 (a n cos nx + b n sin nx).∫02π 2x/n sin(nx)dx} = 1/π{[4π²/nsin(2πn)0] . we find that
x² = (2π²/3)½ + Σn=1∞ 4/n²(1)ncos(nx) n 2 º (−1) cos nx = o3 + 4 S n=1 as required.1/n²sin(0) + 0] =2/nπ[0 .∫2π0 2x/nsin(nx)dx} = 1/π∫2π0 2x/nsin(nx)dx = 2/nπ∫02π xsin(nx)dx =2/nπ[1/n²sin(nx) . Further. an = 1/π∫02π x²cos(nx)dx.∫0π 2x/n sin(nx)dx} = 2/π{[π²/nsin(πn)0] . and bn = 1/π∫02π f(t)sin(nt)dt. and let dv/dx = cos(nx) so that v = 1/nsin(nx).x/ncos(nx)]π0 =4/nπ[1/n²sin(πn) . we let u = x² so that du/dx = 2x. a
For f(x) = x².π/n(1)n . We want to calculate an = 2/πè02 x²cos(nx)dx. where an = 1/π∫02π f(t)cos(nt)dt.
x+δ]. we a º can apply the formula f(x) ~ 20 + S n=1 (a n cos nx + b n sin nx) to say that for 0 < x < 2π. t t
Then we have the formula f(x+)+f(x−) 1 º º = o ° 0 [° −º f(u) cos(v(u − x))du ]dv.2/nπ[1/n²cos(nx)+x/nsin(nx)]2π0 (Result taken from Thomas & Finney) 4π 2 1 2π 1 0 = . that an = 4/n² for n > 1.1/π∫2π0 2x/ncos(nx)dx = 4π/n . and that bn = 4π/n for n > 1. then f(x+)+f(x−) º 2 º = o ° 0 cos(vx)[° 0 f(u) cos(vu)du ]dv./n²cos(0) . +∞)./nsin(0)] = 4π/n . QED. then f(x+)+f(x−) º 2 º = o ° 0 sin(vx)[° 0 f(u) sin(vu)du ]dv. that is. and let dv/dx = sin(nx) so that v = 1/ncos(nx). we let u = x² so that du/dx = 2x. if f(t) = f(t) for every t. or else both limits f(x+) and f(x) exist and both improper integrals d d ° 0+ f(x+t)−f(x+) dt and ° 0+ f(x−t)−f(x−) dt are absolutely convergent. Suppose there is a point x in E1 and an interval [xδ.∫2π0 2x/ncos(nx)dx} = 1/π(4π²/n) .2/nπ∫2π0 xcos(nx)dx = 4π/n .4π/nsin(nx)) º = 4 o 2 + 4 S n=1 ( cos2nx − o sin nx ) . x+δ] about x such that either (a) (b) f is of bounded variation on [xδ. Integrating by parts again./n . bn = 1/π∫02π x²sin(nx)dx. show that (a) If f is even. 2 If f is odd.Similarly. n 3 n 1522: If f satisfies the hypothesis of the Fourier integral theorem. 2
(b)
Answer: Let us first recall the Fourier Integral Theorem: Assume that f ∈ R*(∞. that is. we have x² = (8π²/3)(½) + Σn=1∞ (4/n²cos(nx) .
1
Now that we know that a0 = 8π²/3./nπ[ /n²cos(2πn) + /nsin(2πn) . It follows that /π∫02π x²sin(nx)dx = 1/π{[x²/ncos(nx)]2π0 + ∫02π 2x/n cos(nx)dx} = 1/π{[4π²/ncos(2πn)0] . f(t) = f(t) for every t.0] = 4π/n. 2
.2/nπ[1/n² + 0 .1/n² .
if g(x) is an even function.If we manipulate the formula on the previous page. 1 o
1528: Verify the entries in the following table of Laplace transforms:
f(t) eαt cos(αt) sin(αt) tpeαt F(z) = ° 0 eztf(t) dt. p > 0)
. we see that
f(x+)+f(x−) 1 º º = o 0 [ −º f(u) cos(v(u − x))du ]dv 2 1 º º = o 0 [ −º f(u) cos(vu − vx))du ]dv 1 º º = o 0 [ −º f(u)(cos(vu) cos(vx) + sin(vu) sin(vx))du ]dv º 1 º º = o 0 [ −º f(u) cos(vu) cos(vx)du + −º f(u) sin(vu) sin(vx)du ]dv º º 1 º = o 0 [cos(vx) −º f(u) cos(vu)du + sin(vx) −º f(u) sin(vu)du ]dv
° ° ° ° ° ° °
° °
°
°
°
((1)). then f(u)cos(vu) will be even (as cos(vu) is even). then f(u)cos(vu) will be odd (as cos(vu) is even). QED. and f(u)sin(vu) will be odd (as sin(vu) is odd). ⇒ f(t)g(t) = f(t)g(t) so that f(x)g(x) is an odd function. (a) If f is even. if f(x) is an odd function. We can use the above information to simplify expression (1) in the cases where f is odd or where f is even. Finally. Therefore. then ° −º f(x)dx = 0. and f(u)sin(vu) will be even (as sin(vu) is odd). then g(t) = g(t) ⇒ g(t)² = g(t)g(t) = g(t)g(t) = g(t)² so that g²(x) is an even function.
º ° 0 [cos(vx) ° º f(u) cos(vu)du + sin(vx) ° º f(u) sin(vu)du ]dv −º −º º 1 º ° 0 [cos(vx)(0) + sin(vx) % 2 ° 0 f(u) sin(vu)du ]dv =o º 2 º = o ° 0 [sin(vx) ° 0 f(u) sin(vu)du ]dv. then f(t) = f(t) and g(t) = g(t). then f(t) = f(t) ⇒ f(t)² = f(t)f(t) = (f(t))(f(t)) = f(t)² so that f²(x) is an even function. º º and if f(x) is an even function.
º ° 0 [cos(vx) ° º f(u) cos(vu)du + sin(vx) ° º f(u) sin(vu)du ]dv −º −º º 1 º = o ° 0 [cos(vx) % 2 ° 0 f(u) cos(vu)du + sin(vx)(0) ]dv º 2 º = o ° 0 [cos(vx) ° 0 f(u) cos(vu)du ]dv. Therefore. 1 o
º
(b) If f is odd. QED. Further. then ° −º f(x)dx = 2 ° 0 f(x)dx. z = x + iy (zα)1 z/(z²+α²) α/(z²+α²) Γ(p+1)/(zα)p+1
º
(a) (b) (c) (d)
(x > α) (x > 0) (x > 0) (x > α.
Now if f(x) is an odd function. If f(x) is odd and if g(x) is even.
It follows that limN→∞[∫0N e(αz)ttpdt] = limN→∞{[(tp/(αz))e(αz)t]N0 . setting u = tp so that du/dx = ptp1. and setting dv/dx = e(αz)t so that v = 1/(αz)e(αz)t.(e0/z²+α²)(zcos(0)+αsin(0)) 2 = 0 . QED.(1/z²+α²)(0α) (for x > 0) α = /z²+α² for x > 0.Answer: (a) F(z) = ∫0∞ ezteαtdt = ∫0∞ e(αz)tdt = limN→∞[∫0N e(αz)tdt] 1 = limN→∞[ a−z e (a−z)t ] N 0 1 (a−z)N 1 lim = N→∞[ a−z e − a−z e 0 ] = 1/zα + limN→∞[1/αze(αz)N] = 1/(zα) + 0 = (zα)1 for x > α. QED.(1/z²+α²)(z+0) (for x > 0) z = /z²+α² for x > 0. F(z) = ∫0∞ eztsin(αt)dt = limN→∞[∫0N eztsin(αt)] e −zt = limN→∞[ z 2 +a 2 (−z sin at − a cos at)]N0 (Result taken from Thomas & Finney) −zN = (limN→∞ ze +a 2 (−z sin aN − a cos aN) . we arrive at the following: [ / ∫ e t dt] = [ / ∫ e t dt] = .(e0/z²+α²)(zsin(0)αcos(0)) 2 = 0 .
. F(z) = ∫0∞ ezttpeαtdt = limN→∞[∫0N e(αz)ttpdt]
(c)
(d)
Now let us integrate the above by parts.∫0N (ptp1/(αz))e(αz)tdt} = limN→∞{[00] . (b)
(z = x + iy)
(for x > α)
F(z) = ∫0∞ eztcos(αt)dt = limN→∞[∫0N eztcos(αt)] e −zt = limN→∞[ z 2 +a 2 (−z cos at + a sin at)]N0 (Result taken from Thomas & Finney) −zN = (limN→∞ ze +a 2 (−z cos aN + a sin aN) . = [(p!/(zα) )∫ e dt]
lim p N (αz)t p1 N→∞ zα 0 lim p(p1) N (αz)t p2 N→∞ (zα)² 0 lim p N (αz)t N→∞ 0
(if p > 0). QED. and assuming that p > 0.∫0N (ptp1/(αz))e(αz)tdt} = limN→∞[p/zα∫0N e(αz)ttp1dt] We can now integrate by parts repeatedly...
QED. we see that Γ(n+1) = n! for any integer n > 0 so that p!/(zα)p+1 = Γ(p+1)/(zα)p+1 if p > 0 and x > α as required.
.We can now use the information gained by doing part (a) to say that [(p!/(zα)p)∫0N e(αz)tdt] p! 1 = (z−a) p ( z−a ) = p!/(zα)p+1.
lim N→∞
(if p > 0 and x > α)
Now looking at Exercise 1431.
Chapter 16
.
Let Γ be a rectifiable Jordan curve such that both Γ and its inner region lie within S.
Answer: Let us first remind ourselves of a few definitions. We now need to use Theorem 960: If Γ is a rectifiable curve of length Λ(Γ) described by a complexvalued continuous function z. for every integer n > 1. f(z) f(z) n! n! n! Therefore. .
Because C(r) is a circle with radius r centred at z0. R). Definition.. and if ∫Γ[z] F(z)dz exists. then we can write f(z) (z−z 0 ) n+1 dz. Assume that f is analytic on an open region S in E2.
if we define the path Γ[z] to be the circle
f (n) (z 0 ) =
n! 2oi
°C(r)
f(z) (z−z 0 ) n+1 dz .
.). Theorem (Cauchy’s integral formula). Further. Let f = u + iv be a complexvalued function defined on an open set S in E2.
again provided that the path Γ[z] is positively oriented. Then. the derivative f(n)(z0) exists and is given by the integral f (n) (z 0 ) =
n! 2oi
°G[z]
f(z) (z−z 0 ) n+1 dz. for every point z0 inside Γ. we have f(z 0 ) =
1 2oi
°G[z]
f(z) z−z 0 dz. let C(r) denote the circle of radius r with z0 as centre and let M(r) denote the maximum of f on C(r). 2oi °C(r) (z−z 0 ) n+1 dz = 2oi °C(r) r n+1 dz = (2oi)r n+1 °C(r) f(z)dz ..
where M is such that F(z) ≤ M for all z on Γ.Chapter 16: Cauchy’s Theorem and the Residue Calculus
Exercise 161: Assume that f is analytic on a neighbourhood N(z0. 2. 1. then we have the inequality
°G[z] F(z)dz [ ML(G). If 0 < r < R.
provided that the path Γ[z] is positively orientated. Deduce Cauchy’s inequalities: f (n) (z 0 ) [
M(r)n! rn
(n = 0. then we can write (zz0)n+1 = rn+1.
n! Now knowing that f (n) (z 0 ) = 2oi °G[z] C(r) defined above. The function f is said to be analytic on S if the derivative f’ exists and is continuous at every point of S.
Putting in n = 1. (Note: I am not sure about the last bit. let us manipulate our expression to get it into the desired form: f (n) (z 0 ) =
n! (2oi)r n+1 n! °C(r) f(z)dz u f (n) (z 0 ) =  (2oi)r n+1 °C(r) f(z)dz = n! (2o)r n+1  C(r)
°
f(z)dz.
1 1 f’(z0) = 2or ° 0 u(z 0 + re ih )e −ih dh + 2or ° 0 u(z 0 + re ih )e −ih dh 1 2o = or ° 0 u(z 0 + re ih )e −ih dh as required. we encountered the formula f(z) f(z) 1 n! f (n) (z 0 ) = 2oi °C(r) (z−z 0 )n+1 dz. where C(r) is a curve of length 2πr). we obtain f’(z0) = =
1 2o
ih 2o f(z 0 +re ih )(i(z(h)−z 0 )) 1 2o f(z 0 +re ) 1 dh = 2o 0 z(h)−z 0 dh 2 2oi 0 (z(h)−z 0 ) 2o f(z 0 +re ih ) 2o 1 dh = 2or 0 f(z 0 + re ih )e −ih dh . show that f’(z0) = 1/πr∫02π u(z0+reiθ)eiθdθ. R). Therefore. we can write f’(z0) =
1 2oi
° 2o 0
f[z(h)] (z(h)−z 0 ) 2 dz(h)
=
1 2oi
2o °0
f(z 0 +re ih ) Â (z(h)−z 0 ) 2 z (h)dh. then we have u = v. Answer: Recall that in the previous exercise. 2o 2o
.
Now because we are integrating over a circle. If 0 < r < R. we get the formula f’(z0) = 2oi °C(r) (z−z 0 ) 2 dz.Before applying the above theorem. if 0 ≤ θ ≤ 2π. 0 re ih
°
°
°
°
But as f = u + iv.
Since z’(θ) = ireiθ = i(z(θ)z0).
but the conclusion reached is clearly the one we want). the above can be written as f’(z0) =
1 2or
° 2o u(z 0 + re ih )e −ih dh + 0
i 2or
2o ° 0 v(z 0 + re ih )e −ih dh.
Thus f (n) (z 0 ) [
as required. Now let Γ be a circle with centre at z and radius r. QED. described by the function z.
Applying Theorem 960 to  °C(r) f(z)dz (knowing that f(z) ≤ M(r) on C(r). Using this information. and so
1 2or
° 2o u(z 0 + re ih )e −ih dh = 0
i 2or
2o ° 0 v(z 0 + re ih )e −ih dh. where z(θ) = z0+reiθ. QED. we obtain the following: f (n) (z 0 ) =
n! (2o)r n+1  C(r) n!M(r) rn
°
f(z)dz [
n! 2or n+1 (M(r)
% 2or) =
n!M(r) r n+1 .
Exercise 163: Let f = u + iv be analytic on a neighbourhood N(z0.
) = (a0² + (a0a 1 r)eiθ + (a0 a 2r²)e2iθ + .
(b)
Use (a) to deduce the inequality Σn=0∞ an²r2n ≤ M(r)².. But knowing that ∫02π einθdθ = 0 for any integer n ≠ 0.. + a2r²e2iθa 0 + a2r²e2iθa 1 reiθ + a2r²e2iθa 2 r²e2iθ + ..g. + a2r4 + (a2a 0 r²)e2iθ + (a2a 1 r³)eiθ + . Similarly. B1(r)..e. valid in N(z0.....2. deduce Parseval’s identity:
1 2o 2o º ° 0 f(z 0 + re ih ) 2 dh = S n=0 a n  2 r 2n . where M(r) is the maximum of f on the circle zz0 = r. where A1(r).) = Σn=0∞ an²r2n + A1(r)eiθ + B1(r)eiθ + ... Exercise 4..) = (a0² + a0a 1 reiθ + a0 a 2 r²e2ιθ + ... + ... A1(r) = a1a 0 r + a2a 1 r³ + .
.
Answer: Note: Portions of the following solution have been taken from “Problems in Complex Variable Theory” by Jan G. 1971... e. Krzyz... + . Thus
1 2o 2o ° 0 f(z 0 + re ih ) 2 dh = 1 ∞ 2o (Σn=0
an²r2n)2π = Σn=0∞ an²r2n as required. then we can integrate the expression for f² termbyterm with respect to θ from θ = 0 to θ = 2π.. The vital key to answering this question is to notice that f² = f . i..6..... + a1reiθa 0 + a1reiθa 1 reiθ + a1reiθ a 2r²e2iθ + .)dθ.. f(z 0 + re ih ) = Σn=0∞ a n rneinθ. + a1²r² + (a1a 0 r)eiθ + (a1a 2 r³)eiθ + . are functions of r and the constants an.. f Now f(z0+reiθ) = Σn=0∞ an(z0+reiθz0)n = Σn=0∞ anrneinθ. (a) If 0 ≤ r < R. R).Exercise 169: Assume that f has the Taylor expansion f(z) = Σn=0∞ an(zz0)n.. we can find ∫02π f²dθ = ∫02π (Σn=0∞ an²r2n + A1(r)eiθ + B1(r)eiθ + . Because we are only dealing with values of r inside the circle of convergence of the Taylor series. QED. The product f f = f² is calculated as follows: ff = (Σn=0∞ anrneinθ)(Σn=0∞ a n rneinθ) = (a0+a1reiθ+a2r²e2iθ+. the above integral reduces to ∫02π (Σn=0∞ an²r2n)dθ = (Σn=0∞ an²r2n)∫02π dθ = (Σn=0∞ an²r2n)2π..)( a 0+ a 1reiθ+ a 2r²e2iθ+. etc.
with the exception of a finite number of isolated singularities (“poles”). applying all these substitutions. 1−a Answer: Let us first introduce some theory to be used in the solution.
cos(2t)dt 2o 0 1−2a cos(t)+a 2
If z = eiθ.
. 0
Exercise 1627: Evaluate the following integral by means of residuals: 2 cos(2t)dt 2o ° 0 1−2a cos(t)+a 2 = 2oa 2 . Then we have
n °G[z] f(z)dz = 2oi S k=1 Res z=z k f(z).
1 2o 1 1 It follows that 2o ° 0 f(z 0 + re ih ) 2 dh ≤ 2o M(r) 2 2o = M(r)². But knowing this. zn.
Let us now find the poles of f(z). if a² < 1. 0) d m−1 lim 1 lim d z 4 +1 = zdz 0 (m−1)! dz m−1 (z − z 0 ) m f(z) = (m = 2.. In this question. we see that g(θ) ≤ M(r)² on the circle zz0 = r. By definition. we want to apply the above theorem to the integral given by ° . but there are no singularities on Γ itself. so it follows that 2o
S º a n  2 r 2n = n=0
1 2o
° 2o f(z 0 + re ih ) 2 dh ≤ M(r)² as required. and we do this by defining z = eiθ for 0 ≤ θ ≤ 2π. So b1 = Res(f(z). The Cauchy Residue Theorem.
provided that the path Γ[z] is positively orientated. The first thing to do is to reparametrise the integral to see the poles.(b) Let us now define a new integral I = ∫02π g(θ)dθ. But we know from part (a) that 2o º ° 0 f(z 0 + re ih ) 2 dh = S n=0 a n  2 r 2n . z 0 = 0) = zd0 [ dz (z − 0) 2 ( z 2 (z−a)(z−a −1 ) ) ] =
1 −1 3 4 lim (z−a)(z−a )(4z )−(z +1)(2z− a −a) −1 )) 2 zd0 ((z−a)(z−a
=
0−1(− 1 −a) a (1) 2
=
1 a
+ a. if a² < 1. Because we know that f ≤ M(r) on the circle zz0 = r. say z1. . Let Γ be a rectifiable Jordan curve such that both Γ and its interior lie within S. then it follows that f² ≤ M(r)² on the circle zz0 = r. where f(z) is defined as f(z) =
z 4 +1 z 2 (z−a)(z−a −1 ). then dz = izdθ and (because e±iθ = cosθ ± isinθ) cosθ = ½(z+1/z) and cos2θ = ½(z²+z2). we can apply Theorem 960 (see the solution for Exercise 161 to see this) to say that ∫02π g(θ)dθ ≤ M(r)²×2π. we see that
h=2o ° h=0 cos(2t)dt 1−2a cos(t)+a 2
1 2 −2 a (z +z ) z 2 a −z −1+az
= ° z=0
z=1 1 2i
=
1 2i
1 °0
dz =
1 2 −2 2 (z +z ) dz 1 1 2 iz 1−2a( 2 (z+ z ))+a 1 4 1 a (z +1) z 0 z 2 ( a −z 2 −1+az)
=
1 i
1 °0 1 2ai
°
dz =
°0
1 2 −2 2 (z +z ) (1−a(z+ 1 )+a 2 )z z 1 z 4 +1
dz =
1 2i
1 °0 −1 2ai
z 2 +z −2 z−az 2 −a+a 2 z dz 1 °0 z 2 (z−a)(z−a −1 ) z 4 +1
z 2 (a−z)(z−a −1 )
dz =
dz. QED.. Assume that Γ contains a certain number of singularities of f in its interior. Let f be analytic at all points on an open region S in E2. where g(θ) = f(z0+reiθ)². Therefore.
The first pole to consider is the pole of order 2 at z = 0..
z = 2 −1+i 3 upper halfplane.e.
Exercise 1630: Evaluate the following integral by means of residuals: º 1 ° −º x 2 +x+1 dx = 2o3 3 .The second pole to consider is the simple pole at z = a. Then P(z) = 1 and Q(z) = z²+z+1.
. we can say that
º ° −º 1 x 2 +x+1 dx n = 2oi S k=1 Res z=z k f(z)
= 2oi
3 3i
=
2o 3 3
as required. Answer: To do this question. i. and hence the properties of the above theorem are satisfied. we ignore it. we need some further theory:
Res lim n Theorem: Rd+º ° −R f(x)dx = 2oi S k=1 z=z k f(z). Of these. The poles of f are given by the roots of the −1!i 3 −1! 1−4 = 2 . So b2 = Res(f(z). we can apply Cauchy’s Residue Theorem to our problem:
2o °0 cos(2t)dt −1 1 −1 1 −1 Res z 4 +1 n 1−2a cos(t)+a 2 = 2ai 0 z 2 (z−a)(z−a −1 ) dz = 2ai 0 f(z) = ( 2ai ) % 2oi k=1 z=z k f(z) 2 4 2 +a 4 2 a 2 −1+a 2 (a 2 −1)+a 4 +1 −1 a4 ( 2ai )(2oi)( 1 + a + a(a 2+1 ) = ( −o )( ) = −o( a −1+a(a−a−1) +1 ) = −2oa2 . But because 1/a is outside the area of integration (a² < 1. a a 2 2 2 −1) −1) a 1−a a(a
°
°
S
=
QED.
The final pole to consider is the simple pole at z = 1/a. The residue at this pole (at z = 2 ) is given by −1+i 3 2 −1+i 3 ( 2 lim
−1+i 3 zd 2
º
Res(f(z). only the ‘positive’ root lies in the equation z²+z+1 = 0. R 1 1 For ° −º x 2 +x+1 dx . This is automatically satisfied if f is the quotient of two polynomials.
Therefore. =
lim
−1+i 3 zd 2
)=
zd z+ 1 − 2
i 3 2
lim
−1+i 3 2
(z −
)
))f(z) =
1
i 3 z+ 1 + 2 2
lim zd
−1+i 3 2
z+ 1 − 2
i 3 2
z 2 +z+1 1
i 3 (z+ 1 − 2 2
i 3 )(z+ 1 + 2 2
=
=
−1+i 3 2
i 3 +1+ 2 2
=
1
i 3 2 i 3 + 2
=
1 i 3
=
3 3i
. using the above theorem. and 0 < z < 1). provided that the degree of Q exceeds the degree of P by at least 2. a) 4 +1 z 4 +1 lim z 4 +1 lim lim = zdz 0 [(z − z 0 )f(z) ] = (z 0 = a) = zda (z − a)( z 2 (z−a)(z−a −1 ) ) = zda z 2 (z−a −1 ) = a 2a(a− 1 ) =
a
a 4 +1 a(a 2 −1) . QED. say f = P/Q. Now that we have found all of the poles. let f(z) = z 2 +z+1 .
draq .) Consider that f is defined as above and we also have g defined as g(z) = rz+s whenever rz+s ≠ 0.cpbs . But (ap+br)(cq+ds) . Manipulating. and if psrq ≠ 0. Let us now consider the composite function fg = f(g(z)). b. We can also define fg everywhere on the extended plane E2* by setting cq+ds ap+br fg(− cp+dr ) = ∞ and fg(∞) = cp+dr .) So we have shown that fg is a Möbius transformation (defined whenever (cp+dr)z+(cq+ds) ≠ 0).bc(psrq) = (adbc)(psrq) ≠ 0 (as (adbc) ≠ 0 and as (psrq) ≠ 0.drbs = apds + brcq . We define g everywhere on the extended plane E2* as above.cpaq . r and s are complex numbers such that psrq ≠ 0.draq = ad(psrq) . where p. these last two equations are to be replaced by the single equation fg(∞) = ∞.cpbs . q. f(g(z)) =
pz+q f( rz+s pz+q
)=
a( rz+s )+b
pz+q c( rz+s )+d
pz+q
=
a(pz+q)+b(rz+s) rz+s c(pz+q)+d(rz+s) rz+s
=
a(pz+q)+b(rz+s) c(pz+q)+d(rz+s)
=
(ap+br)z+(aq+bs) (cp+dr)z+(cq+ds) . show that the composite function fg is also a Möbius transformation. these last two equations are to be replaced by the single equation f(∞) = ∞. then we have (ap+br)(cq+ds) (cp+dr)(aq+bs) ≠ 0.Exercise 1638: If f and g are Möbius transformations.)
. c and d are complex numbers such that adbc ≠ 0. we define f(z) = az+b whenever cz+d ≠ 0.(cp+dr)(aq+bs) = apcq + apds + brcq + brds . Answer: Let us first remind ourselves of the definition for a Möbius transformation: they are functions f defined as follows: if a. (If c = 0. (If cp+dr = 0. It is convenient to define f everywhere on the cz+d extended plane E2* by setting f(d/c) = ∞ and f(∞) = a/c.
We now need to show that if adbc ≠ 0.
we translate the point z to the point z+b. where a in this case is given by a = 1/p²+q². In other words. i.
Answer: (a) Assuming that b is a complex number. the Möbius transformation in this case in effect changes the scale of the axes in an Argand diagram. where a > 0 (c) f(z) = eiαz. to x’ = x×a and to y’ = y×a. where α is real (d) f(z) = 1/z
y z+b b z (a) y 2z x
(Translation). f(z) = ei(α+θ)r.
iθ
½z (b) y
z x
az = re α (c) y
i(θ+α)
r θ
z = reiθ x
z x 1/z (d)
(d) Finally. then if z = x+iy. the transformation moves the point along the vector defined in an Argand diagram by z by an amount specified by the number a. (Stretching).
. Geometrically. say b = p+qi. (Rotation). Geometrically. (b) Assuming that a is a real number. then the Möbius transformation in this case moves the point z to the point (x+p)+(q+y)i. recall that if z = p+iq. then the Möbius transformation in this case rotates the point z by an angle of α around the origin (in the anticlockwise direction). then 1/z = z1 = piq/p²+q². from the original origin O to the new origin b. and then apply a “Stretching” transformation (as in part (b)). we reflect the point z in the xaxis of an Argand diagram. (c) Recalling that a complex number can be written in the form z = re .Exercise 1639: Describe geometrically what happens to a point z in E2 when it is carried into f(z) by the following special Möbius transformations: (a) f(z) = z+b (b) f(z) = az. (Inversion). This is useful if a change of origin is required.e.
Possible Further Work / Evaluation
i 1 Exercise 163: To get the result 2or ° 0 u(z 0 + re ih )e −ih dh = 2or ° 0 v(z 0 + re ih )e −ih dh.e. we could try using the analyticity of f (i. 2o 2o
. using the Cauchy Riemann equations in polar form).