ous Functions


Let us note that if the topology of the range space Y is given by a basis £, then to prove continuity of / it suffices to show that the inverse image of every basis element is open: The arbitrary open set V of Y can be written as a union of basis elements

V = (J Ba.


r1(v) = \jr\Ba),

so that f~ (V) is open if each set f~ (Ba) is open. If the topology on Y is given by a subbasis S, to prove continuity of / it will even suffice to show that the inverse image of each subbasis element is open: The arbitrary basis element B for Y can be written as a finite intersection S\ n ‡ ‡ ‡ n Sn of subbasis elements; it follows from the equation /-l(B) = / - 1 ( Si ) n - - - n / - 1 ( 5 Ä ) that the inverse image of every basis element is open. EXAMPLE 1. Let us consider a function like those studied in analysis, a "real-valued function of a real variable," / : R ²> R. In analysis, one defines continuity of / via the "e-S definition," a bugaboo over the years for every student of mathematics. As one would expect, the e-S definition and ours are equivalent. To prove that our definition implies the e -<5 definition, for instance, we proceed as follows: Given xo in R, and given e > 0, the interval V = (/(*o) ² e, /(*o) + e) is an open set of the range space R. Therefore, /_1(V) is an open set in the domain space R. Because /_1 (V) contains the point xo, it contains some basis element (a, b) about x$. We choose 8 to be the smaller of the two numbers xo ² a and b ² xo- Then if |JC ² xo\ < 8, the point x must be in (a, b), so that f(x) e V, and \f(x) ² /(*o)l < C as desired. Proving that the e-8 definition implies our definition is no harder; we leave it to you. We shall return to this example when we study metric spaces. EXAMPLE 2. In calculus one considers the property of continuity for many kinds of functions. For example, one studies functions of the following types:



(curves in the plane) (curves in space) (functions f(x,y) of two real variables) (functions f(x, y, z) of three real variables) (vector fields V(JC, y) in the plane).

Each of them has a notion of continuity defined for it. Our general definition of continuity includes all these as special cases; this fact will be a consequence of general theorems we shall prove concerning continuous functions on product spaces and on metric spaces.

Assume that / is continuous. one has /(A) c f(A). as desired. If the condition in (4) holds for the point x of X. Let R denote the set of real numbers in its usual topology. Then f~x (V) is an open set of X containing x.2 E XAMPLE 3. and they are considered in the theorems that follow. Let / : R ²> Re be the identity function. Then r\B) = f-\Y) ~ f-\v) = x. Let B be closed in Y and let A = f~ x(B). Set B = Y . Therefore.104 Topological Spaces and Continuous Functions Ch. In analysis. it must intersect A in some point y. b) is itself. so that A = A. so that f(x) e /(A). then f(x) e f(A). Now B is a closed set of Y. there is a neighborhood U of x such that f(U) C V. By elementary set theory. the identity function g : Re ² R is continuous. we say that / is continuous at the point x. Let X and Y be topological spaces. as desired. so that x e f~x{B) = A. (4) For each x e X and each neighborhood V of f(x). which is open in R^. we show that A ² A. Then the following are equivalent: (1) f is continuous. Let V be an open set of Y. Thus A C A. we have f(A) = f{f~\B)) c B. Let A be a subset of X.V. let f : X -> Y. the set /~'(B) is closedin X. We show that if x ¼ A. they will be treated when we study metric spaces.r\v). Let V be a neighborhood of f(x). The familiar "e-6" definition and the "convergent sequence definition" do not generalize to arbitrary spaces. Proof.1. (2) For every subset A ofX. the inverse image of the open set [a. and let R^ denote the same set in the lower limit topology. (3) For every closed set B ofY. Then f~l(B) is closed in X by hypothesis. Then / is not a continuous function. because the inverse image of (a. Some of these generalize to arbitrary spaces. (1) =>‡ (2). as desired. so that /""' (V) is open in X. b) of R^ equals itself. On the other hand. if x e A. fix) ¼ /(A) c J\A) CB = B. one studies several different but equivalent ways of formulating the definition of continuity. (2) = (3). We show that (1) =» (2) => (3) => (1) and that (1) => (4) => (1). which is not open in R. . (3) = (1). Then V intersects f(A) in the point f(y). Theorem 18. f(x) ² x for every real number x. We wish to prove that A is closed in X.

any property of X that is entirely expressed in terms of the topology of X (that is. Let V be an open set of 7. It follows that f~l (V) can be written as the union of the open sets Ux. so that it is open. But the inverse image of £/ under the map /-1 is the same as the image of U under the map /. If /' happens to be a homeomorphism of X with Z.1 This remark shows that a homeomorphism / : X -* Y gives us a bijective correspondence not only between X and Y but between the collections of open sets of X and of Y. the inverse image of U under the map f~l : Y -> X is open in F. Now suppose that / : X ²> Y is an injective continuous map. Figure 18. Then the set U = f~l(V) is a neighborhood of JC such that f(U) c V. The condition that /_1 be continuous says that for each open set U of X. See Figure 18. If both the function / and the inverse function f~l:Y^X are continuous. or simply an imbedding. let* be a point of /~'(V). considered as a subspace of Y\ then the function /' : X ²» Z obtained by restricting the range of / is bijective. via the correspondence /. Let x e X and let V be a neighborhood of f(x). As a result. let / : X -> Y be a bijection. it is a bijective correspondence that preserves the topological structure involved. So another way to define a homeomorphism is to say that it is a bijective correspondence / : X -> Y such that /([/) is open if and only if U is open. in terms of the open sets of X) yields. (4) =» (1). The analogous concept in topology is that of homeomorphism.1. Homeomorphisms Let X and Y be topological spaces. An isomorphism is a bijective correspondence that preserves the algebraic structure involved.§18 Continuous Functions 105 (1) =$ (4). the corresponding property for the space Y. You may have studied in modern algebra the notion of an isomorphism between algebraic objects such as groups or rings. . where X and Y are topological spaces. Then/(x) e V. Then Ux c /-1(V). Such a property of X is called a topological property of X. Let Z be the image set /(X). of X in Y. then / is called a homeomorphism. we say that the map / : X ²> Y is a topological imbedding. so that by hypothesis there is a neighborhood Ux of * such that f(Ux) C V.

See Figure 18. F carries a basis element for the order topology in (² 1. A bijective function f : X -* Y can be continuous without being a homeomorphism. As a result. ¤ ¥£  ¢ ¡  . The functi n / : R -» R gi en by f = 3x + 1 i a homeomorphi m. F is automatically a homeomorphism of (²1. 3x + 1 F(x) 1-x2 Figure 18.2. Since the order topology on (² 1. It follows that / is bijecti e and that g = / 1. We have already noted in Example 9 of 3 that F is a bijective order-preserving correspondence. 1) withR. E XAM E 5. If we define g : R -* R by the equation   © ¨ then one can check easily that f g(y)) = y and g(f(x)) = x for all real numbers x and y.1) and the usual (subspace) topology agree. One such function is the identity map g : Re -» R considered in Example 3. 1) onto a basis element for the order topology in R and vice versa. EXAM E 6. These are familiar facts from calculus.1) -* R defined by   © §¦ *"(*) = l-*2 is a homeomorphism.  [x *2 + y2 = l}. F is a homeomorphism of (-1. See Figure 18.3 A second way to show F a homeomorphism is to use the continuity of the algebraic functions and the square-root function to show that both F and G are continuous. Another is the following: Let Sl denote the unit circle. One way is to note that because F is order preserving and bijective.3. its inverse is the function G defined by  G(y) = y l + (l+4y2)l/2- The fact that F is a homeomorphism can be proved in two ways.106 Topologi l paces and Continuous Functions   §¦ C EXAM E 4. The function F : (-1. the continuity of / and g is a familiar result from calculus.2 Figure 18. 1) with R (both in the order topology).

(c) (Composites) If f : X ²> Y and g : Y ² Z is continuous. (a) (Constant function) If f : X ²> Y maps all of X into t e single point yo ofY.4 E XAM LE 7. T orem 18. Z are continuous. But the function f l is not continuous. is not open in S1.2 (Rules for constructing continuous functions). Consider the function g : [0.4. The image under / of the open set U = [0.§18 considered as a subspace of the plane R2. t en f is continuous.1) ² > S1 Continuous Functions 107 be the map defined by /(/) = (cos 2nt. of which some generali e to arbitrary topological spaces and others do not. and let F : [0.Y. The map g is an example of a continuous injective map that is not an imbedding. for instance. |) of the domain. sin 2nt). See Figure 18. for the point p ² /(0) lies in no open set V of R 2 such that V n S1 C f(U).  U oi 4 Figure 18. We study first some constructions that do hold for general topological spaces. Let X. and Z be topological spaces. (b) (Inclusion) If A is a subspace ofX. t e inclusion function j : A ²* X is continuous. t en t e map g o f : X ->‡ ' ' & % $ #" ! . Constructing Continuous Functions How does one go about constructing continuous functions from one topological space to another? There are a number of methods used in analysis. deferring consideration of the others until later. The fact that / is bijective and continuous follows from familiar properties of the trigonometric functions. 1) ² M2 obtained from the function / of the preceding example by expanding the range.

depending on whether V contains yo or not. But rl(V) = \J(f-l(V)nua). Since f~l(U) is open.Y be continuous. Let f : A -* Y and g : B . (c) If U is open in Z. then the restricted function f \ A : A -> Y is continuous. because both expressions represent the set of those points x lying in Ua for which f(x) e V. we can write X as a union of open sets Ua. this set is open in Ua. If f(X) c Z C Y. If f(x) = g(x) for every x ¼ An B. defined by setting h(x) = f(x) ifx ¼ A. such that f\Ua is continuous for each a. (b) If U is open in X. andh(x) = g(x) ifx e B. . then f and g combine to give a continuous function h : X ²>‡ Y'. Proof. 2 (d) (Restricting the domain) If f : X -> Y is continuous. Since f\Ua is continuous. by elementary set theory. we show that the function g : X -> Z obtained from / is continuous. then the function h : X -> Z obtained by expanding the range of f is continuous. (a) Let f(x) = yo for every x in X.3 (The pasting lemma). Let V be open in Y. by elementary set theory. In either case. Then f-l(V)nua = (f\uarl(V). then g~l(U) is open in Y and f~l (g~l(U)) is open in X. If Z is a subspace of Y containing the image set f(X). Let X = A U B. Because Z contains the entire image set f(X). (f) (Local formulation of continuity) The map f : X ²* Y is continuous ifX can be written as the union of open sets Ua such that f\Ua is continuous for each a. and hence open in X. IfZ is a space having Y as a subspace. so is g~l (B). Let V be an open set in Y. (f) By hypothesis. The set f~l (V) equals X or 0. l Theorem 18. note that h is the composite of the map / : X -> Y and the inclusion map j : Y -* Z. which is open in A by definition of the subspace topology. then j~l (U) ² U D A.108 Topological Spaces and Continuous Functions Ch. Let B be open in Z. To show h : X -> Z is continuous if Z has Y as a subspace. But r1(g~Hu)) = (gofri(u). r\u) = g-l(B). Then B = Z n U for some open set U of Y. (d) The function f\ A equals the composite of the inclusion map j : A ²>‡ X and the map / : X ²> Y. where A and B are closed in X. and if A is a subspace ofX. (e) Let / : X -> Y be continuous. then the function g : X -> Z obtained by restricting the range off is continuous. it is open. both of which are continuous. (e) (Restricting or expanding the range) Let f : X ² Y be continuous. a so that f~ (V) is also open in X.

this is just a special case of the "local formulation of continuity" rule given in preceding theorem. Similarly.5 . Let C be a closed subset of Y. closed in X. do not define a function. x/2 for * > 0.2 for x < 0. One needs the "pieces" of the function to agree on the overlapping part of their domains in order to have a function at all. But / is not continuous. f~l(C) is closed in A and. On the other hand. 3). Now Continuous Functions 109 h-1(C) = rHC)Ug-\C). 1).5. and both of the pieces are continuous. V. ( for instance. Each of the "pieces" of this definition is a continuous function. The equations k(x) = x . See Figure 18. Let us define a function h h(x) = R by setting x for x < 0. by elementary set theory. / Figure 18. for instance. the function h is continuous. Their union h~l (C) is thus closed in X. x + 2 for x > 0. This theorem also holds if A and B are open sets in X. EXAMPLE 8. Since their domains are closed in K. therefore. The equations Kx) x ² 2 for x < 0.§18 Proof. g~x(C) is closed in B and therefore closed in X. do define a function / mapping R into IR. for instance. which is the one-point set {0 . the inverse image of the open set (1. Since / is continuous. is the nonopen set[0. one needs some limitations on the sets A and B to guarantee continuity. x + 2 for x > 0. and they agree on the overlapping part of their domains.

or quotients of continuous real-valued functions. Proof. r1(t/xV) = /1-1(£/)n/2-'(V).y)) is said to be continuous if both P and Q are continuous functions. and / ‡ g are continuous. then f + g. y(t)). then f\ and fa are composites of continuous functions and therefore continuous. These maps are continuous.2 Theorem 18.y) = P(x. or equivalently.y). respectively. differences. A point a is in f~l(U x V) if and only if f(a) e U x V. For Tt^x(JJ) = U x Y and ^'(V) = X x V." but this conjecture is not true. suppose that f\ and fa are continuous. . In calculus. so is their intersection. There is no useful criterion for the continuity of a map / : A x B ²> X whose domain is a product space. b] -* R2. X and fa:A-+Y If the function / is continuous.y)i+Q(x.Q(x. Both of these statements are simply special cases of the preceding theorem. a vector field in the plane \(x. Then f is continuous if and only if the functions /i : A ² are continuous. Conversely. products. (See Exercise 12.y)j = (P(x./2(a)). a parametrized curve in the plane is defined to be a continuous map / : [a. Similarly. Therefore. that is. Since both of the sets /f' (£/) and /2_ l (V) are open.) EXAMPLE 9. f ² g. The maps fa and fa are called the coordinate functions of /. It is often expressed in the form /(f) = (x(t). Note that for each a e A. and these sets are open if U and V are open. if v is continuous as a map of R2 into K2.110 Topological Spaces and Continuous Functions Ch. We shall consider this theorem in §21. and f/g is continuous if g(x) ^ 0 for all x. Let f : A -+ X xY be given by the equation /(fl) = (/i(o). Let n\ : X x Y ² X and n2 : X x Y ²> Y be projections onto the first and second factors. Ma) = nx{f{a)) and fa(a) = n2(f(a)). We show that for each basis element U x V for the topology ofXxF. It is a standard theorem that if /. g : X -> M are continuous. One way of forming continuous functions that is used a great deal in analysis is to take sums. One might conjecture that / is continuous if it is continuous "in each variable separately. its inverse image f~l (U x V) is open.4 (Maps into products). and one frequently uses the fact that / is a continuous function of t if both x and y are. if and only if f\(a) e U and fa(a) ¼ V.

§18 Continuous Functions 111 .

for instance. This theorem generalizes to a theorem about maps of an arbitrary topological space X into a metric space Y. then the limit function is necessarily continuous. There is a theorem to the effect that if a sequence of continuous real-valued functions of a real variable converges uniformly to a limit function. to demonstrate the continuity of the trigonometric functions. We shall prove it in §21. when one defines these functions rigorously using the infinite series definitions of the sine and cosine. .Yet another method for constructing continuous functions that is familiar from analysis is to take the limit of an infinite sequence of functions. It is used. This theorem is called the Uniform Limit Theorem.