“4 1. Basic Concepts Indeed, the notion of pointwise convergence of sequences of functions on [a,b], ie lim fy =f if and only if lim fy(x) = f(x) Vx € [a,b], cannot be described by any metric on a nontrivial class of functions (Dugundji (8; p.273]), (ii) Let X = Cla. b], and d(f.g) = sup {|f (se) — g()|: x € (a. b]}. Let falar be a sequence of elements of C[a, b] that converges to, say, f € Cla, bl; ie., lim dfn f) = lim sup {\fulx) — f(2)f:x € (a, b]} = 0. This means that for ¢ > 0 there exists an integer mp(e) such that sup {{fy(x) — f(x)|:x € [a,b]} f uniformly on [a,b], then for any > 0 there exists an integer nng(«) such that sup {[fa(x) — f(2)|:x € [a,b}} 0, there exists an integer m(@) such that [xy —2,|<@ whenever m= m(e) and n= ng(e). The Cauchy principle states that a sequence in R or C is convergent if and only if it is Cauchy.) The principle enables us to prove the convergence of a sequence without prior knowledge of its limit. The real sequence1.4, Cauchy Sequences 45 is such that for m= m the distance between the terms is given by I-s)-( which tends to zero as m, 1 tend to infinity. In other words, the real sequence {el 1» Where xy = 1 — 1/28, satisfies the Cauchy criterion and hence converges by Cauchy's principle of convergence. A similar situation arises with sequences of functions; in fact, it comes up more often than with real or complex sequences. An extension of the idea of Cauchy sequences to metric spaces turns out to be useful. [sn =m Definition 1.4.1. Let dbe a metric on a set X. A sequence {x,,,~ , in the set Xis said to bea Cauchy sequence if, for every & > 0, there exists a natural number 1 such that A(X Xn) <6 whenever n= ny and m= no. Remark 1. A sequence {x,} in R or C is a Cauchy sequence in the sense familiar from elementary analysis if and only if it is a Cauchy sequence according to Definition 1.4.1 in the sense of the usual metric on R or C. Remark 2. Itis cumbersome to keep referring to a “sequence in a set X with metric d”, especially if itis understood which metric is intended and no symbol such as d has been introduced to denote it. We shall therefore adopt the standard phrase “sequence in a metric space X”. Examples 1.4.2. (i) The sequence {x,},>, where x, = 1+1/2+...+1/m, does not satisfy Cauchy's criterion of convergence. Indeed, Su, it iy Hot the cave here tat [ay — am] 0 for lange me acl re Gi) In C{0,1], the sequence fi, fis fis--- given by f=, xe (011 is Cauchy in the uniform metric. For mm the function mx me (m= nx? mtx ntx (mtx)(n+x)" being continuous on [0, 1], assumes its maximum at some point 29 € [0,1]. So, fon» fa) = sup {lfn(x) — fad]: € [0, 1]} ames (manta) nex46 1. Basic Concepts for large m and nt. Morcover, the sequence {f,},~1 converges to some limit. For fax (fo) — Fl0)| = | x = Therefore, {faln=1 converges to the limit f; where f(x) = x for all x € [0,1]. Proposition 1.4.3, A convergent sequence in a metric space is a Cauchy sequence, Proof. Let {xn} be a sequence in a set X with metric d, and let x be an element of X such that lim)... 4, =x. Given any ¢ > 0, there exists some natural number tm such that d(x,,x) < €/2 whenever n> 7. Consider any natural numbers 1 and m such that n= 1 and m2 1. Then d(x) < €/2 and d(x x) < €/2. Therefore (Xs Xm) S A Xns X) + A(X X) < 6/2 + 6/2 = B. a Does the converse of Proposition 1.4.3 hold? Ifa sequence {q},~1 in a metric space (X, ) fulfills the Cauchy condition of Definition 1.4.1, does it follow that the sequence converges? Examples 1.4.4. (i) Let X denote the set of all rational numbers with the usual metric, namely, d(sx,y) = |x — y| for x,y € X. It is well known that the sequence 14, L4l, L414, converges to V2. It is therefore Cauchy. However, it does not converge to a point of X. So, a Cauchy sequence need not converge to a point of the space. (ii) Another example of a Cauchy sequence that does not converge to a point of the space is the following: Let X = C[0, 1] with metric d defined by aGf.g) =| [flo s0olds, fx € C101 A Let {fy}n=2 be the sequence in C[0,1] defined by °, 1 Sul) = (x4) +1, Clearly, {fabn=2 is a sequence in C{0,1]. Moreover, it is Cauchy because : ay A) = [39 ~ floes 7 = [. folsas+ | 1/2-1/m et Sul x)ee = waa) 1/21) ‘The measure d(fy,» f,) is the shaded area in Fig. 1.5.1.4, Cauchy Sequences a7 (0,1) Suppose now that there is a continuous function fsuch that d(fj, f) —» 0. It will be shown that this leads to a contradiction. Since v2 1 [fa(x) — f(x)|dx +] [1 —f(x)|dx Jia fe ironta+ | Jiprajn ago | lo and d(fipf) —* U-as n> 00, it Follows that 1/2 | [Foo|de=0 and | ll —f@)|dx =o. 0 he Since f is continuous, we sce that f(x) =0 for 0=x=1/2 and f(x) =1 for 1/2 < x31, which is impossible. ‘Thus, the metric spaces in which Cauchy sequences are guaranteed to converge are special and we need a name for them. Definition 1.4.5. A metric space (X,d) is said to be complete if every Cauchy sequence in X is convergent. It follows from Cauchy's principle of convergence that R, C and R" equipped with their standard metrics (namely, d(x,y) = |x — y| for x,y © R; d(z,w) = |z— w| for 2,w € Cand d(x, y) = (S22, (4) — yi?) for x = (1.x). - 24%) and y= -+Jn) in") are complete metric spaces. The metric space (X,d), where X denotes the set of rationals and d(x, y) = |x — y| for x, y € X, has been observed to be an incomplete metric space (see Example 1.4.4(i)). That the metric space (X, d) of rationals is incomplete also follows on considering the sequence {xy}, Where48 1. Basic Concepts 14 malt ytgytet nl as this is a Cauchy sequence but it converges to the irrational number e. In our next proposition, we need the following definition. Definition 1.4.6. Let {x)},=1 be a given sequence in a metric space (Xd) and let {nu} be a sequence of positive integers such that m < 1m < m <.... Then the sequence {x J,=1 is called a subsequence of (x,],=1. If {xq,]e=1 converges, its limit is called a subsequential limit of {2] =. It is clear that a sequence {%,},=1 in X converges to x if and only if every subsequence of it converges to x. Proposition 1.4.7. If a Cauchy sequence of points in a metric space (X,4) contains a convergent subsequence, then the sequence converges to the same limit as the subsequence. Proof. Let {xy = be a Cauchy sequence in (X,d). Then for every positive number & there exists an integer np(€) such that A(X Xq) < whenever m, n= noe) Denote by {»,} a convergent subsequence of}, = 1 and its limit by x It follows that AUSngs x) <€ Whenever m, r= no(C)s since {ng} is a strictly increasing sequence of positive integers. Now, Ax, %) S d(x, Xyq) + ACLs Xe) SAL, oq) + & whenever m, n= role). Letting m — 00, we have A(x, x4) Se whenever n= noe). So, the sequence {~, 1 converges to x. o We next show that the spaces (R", dp), fy, (S) and C[a,b] are complete. Proposition 1.4.8. The metric space X =R" with the metric given by (= Ie >t) "yeh p(x, y) where x = (x1,.%2)...4%n) and y = (yi, ¥25-+ +» Yn) areinR", isa complete metric space. Proof, Let {x Jmy=15 0 = (ai, x6",...,.2™), denote a Cauchy sequence in (Xd), ies, p(x”, x!") — 0 as m,n! + oo. Then, for a given & > 0 there exists an integer 1() such that1.4, Cauchy Sequences 9 vp r) 0, there exists a positive integer (2) such that ~ Mp gta, x) = (= lag? - wr) 0, therefore by the Cauchy property of {f,}, there exists some ry such that (fur fn) <1 whenever n= ny and m= ny. In particular, d(fs fiy) <1, and hence [fi(s) fuy(s)| <1 for all » € S, whenever n= ny. Since fy, € BS), there exists some M > 0 such that |fy(s)|=M for all s€ S. Therefore, Lfa(S)155 [faCS) — fay + Ufg()[S1-+M Vs © S whenever n= no. It follows that [f(s)| = limy oo fy(s)| 1+ M for every s € S. Thus, f € B(S), as claimed, Now consider any & > 0. By the Cauchy property of {fy]s there exists some np such that dfurfn) <€ — whenever n= rp and m= mw Therefore, als) —fals)| <@ VSE Swhenever nz np and m= 1. It follows upon letting m — oo that (als) —f(s)| Se whenever s € S and n= 1 This says that lim,—o0 dfn f) = a Proposition 1.4.13. Let X = Cla,b] and d(f,g) = supi|f(x) ~ g(x)|:aS x= b} be the associated metric. Then (Xd) is a complete metric space. Proof. Let [ful,= 1 be a Cauchy sequence in Cla,b]. Then for every & > 0 there exists an integer rp(e) such that m,n= no(e) implies (fn fo) = suptlfn(x) — flx)|: a=x=bj 00. We have thus defined a function f with domain [a,b]. It remains to show that f € Cla,b] and that lim, d(fmf) =0.52 1. Basic Concepts Since for every x € [4,6], Voile) — fi(2)] < provided that m, n= no(e), it follows upon letting m—> 00 that [filx) — fx) Se (1.18) for all n= ng(e) and all x € (a,b). ‘To see why fis continuous, consider any x € [a,b] and any 1 > 0. According to what has been noted in the preceding paragraph, there exists an integer m (n) such that, for every x € [a,b], we have |fu(x)—f(x)| < n/3 provided that m= mn). Select m= n(x). Then Uns) —fia(39| <3 for all x € [a,b] (1.19) ‘Now use the continuity of f, to obtain 8 > 0 such that Ut) ~ fo] <2 for |x ~ 29] <8. (1.20) Since Uf) = f%0)] = LF) = fu ®)] + U2) — fe 20)| + [fn 80) = f2%0)), it follows from (1.19) and (1.20) that |f(x) — f(xo)| <9 whenever |x — x9| <8. Therefore, f € Cla, b]. Moreover, (1.18) says that limy.>c d(fu»f) = 0. As already noted, this completes the proof. o Examples 1.4.14, (i) Lot X be any nonempty set and let d be defined by _f0 ifx=y, ats = {4 tx yy. ‘Then (X, d) is a complete metric space. Indeed, if {x,},~1 is a Cauchy sequence, then for 0 < ¢ < 1 there exists a positive integer no(e) such that dp %) <€ for all m, n= ng(e). So for n= np(e), we have %y =X. Thus, any Cauchy sequence in (X,d) is of the form (Gases Fas tas---), which is clearly convergent to the limit x. (ii) Let N denote the set of natural numbers. Define ls) + - 4, mnen. ma (m,n) Then (N, d) is an incomplete metric space. That (N, d) is a metric space is clear. The sequence {11}, = can be shown to be Cauchy by arguing as follows. Let ¢ > 0 and let 1 be the least integer greater than fe. If rym ry then1.4, Cauchy Sequences 33 od + pS <6. 1 (mn) = Thus, the sequence is Cauchy. To see why it does not converge, suppose that it were to converge if possible to, say, p € N. Let m be any integer greater than 2p. Then n= n implies that ul op" (p,m) = pea This shows that the sequence cannot converge to p and therefore does not converge atall. (iii) Let X = CU foo} and let d be given by lai — a 2a al when 4,2) €C, L dln) = Vit lahyt+ lap /1+ la Then (X,d) is a complete metric space. Let {z,J,=1 be a Cauchy sequence in (X,d). If the sequence {z,},,=1 contains the point at infinity infinitely often, then it contains a convergent subsequence, namely the subsequence each of whose terms is 00. In this case, the sequence {2}, converges to 00 by Proposition 1.4.7. On the other hand, if the point at infinity appears only finitely many times in the sequence, then we may assume without loss of «generality that the sequence consists of points of C only, as the deletion (or insertion) of finitely many terms does not alter the convergence behaviour of a sequence. when 2) € Cand 2 = 0. (ase, If the sequence {|Z,|},— is unbounded, then for every natural number k there exists a term z,, of the sequence such that |z,,| > k, where these terms can be chosen so that mg. > mk = 1,2,....Now, 2 2 vVitlenP VFR We thus have lim... 2, =00 in (Xd). By Proposition 1.4.7, it follows that Tim ysco Zn = 00. A240) 0 ask os Case I, The sequence {|2}n=1 is bounded, say by M > 0. Let > 0 be given. There exists 1m € N such that m1. 1 implies Since |zq] oo. Thus, (Xd) is a complete metric space.34 1. Basic Concepts (iv) Let X = Rand d:R x R— BR be defined by |= yl “Vise Viee We shall show that (R, d) is a metric space which is not complete. ‘As the proof that (R, d) is a metric space is no different from that of (xiii) of Example 1.2.2, it is not included. We shall only show that (R, d) is not complete. Consider the sequence {1}, =; of natural numbers, Observe that [a= m| eal Vise VIF eo d (x,y) qa) (n,m) = The right hand side of the above inequality can be made arbitrarily small by choosing m and msufficiently large. Thus {i} = is a Cauchy sequence in the metric space (R, d). We claim that it does not converge in R: Suppose, if possible, that it were lo Converge lo some limit p € R. Ther —-—_ln-al An, p) = Fe Vie so that lim d(m,p) = This shows that the Cauchy sequence {n},, does not converge to p € R. Since pis acbitrary, the argument is complete. 1.5. Completion of a Metric Space Let (X, d) be a metric space that is not complete. It is always possible to construct a larger space which is complete and contains just enough points so that every Cauchy sequence in X has a limit in the larger space. In fact, we need to adjoin new points to (% d) and extend d to all these new points in such a way that the formerly nonconvergent Cauchy sequences find limits among these new points and the new points are limits of sequences in X. Definition 1.5.1. Let (X, d) be an arbitrary metric space. A complete metric space (X*,d") is said to be a completion of the metric space (X, d) if1.5. Completion of a Metric Space 35 (i) X is a subspace of X* (ii) every point of X* is the limit of some sequence in X. For example, the space of all real numbers is the completion of the space of rationals. Also, the closed interval [0,1] is the completion of (0,1), (0,1), (0,1] and itself. In fact, any complete metric space is its own completion. We note that the Weierstrass approximation theorem (Proposition 0.8.4) shows that the metric space Cala, b] of Example 1.2.2(ix) is the completion of its subset consisting of polynomials. Definition 1.5.2. Let (X,d) and (X’, d’) be two metric spaces. A mapping fof X into X’ isan isometry if af (x), FO) = d(x y) for all x,y € X. The mapping fis also called an isometric embedding of X into X’. If, however, the mapping is onto, the spaces X and X" themselves, between which there exists an isometric mapping, are said to be isometric. It may be noted that an isometry is always one-to-one. Theorem 1.5.3. Every metric space has a completion and any two completions are isometric to each other. Proof. Let (X,d) be a metric space and let X denote the set of all Cauchy sequences in X. We define two Cauchy sequences {x,} and {yy} in X to be equivalent if lime d(p, Yq) = 0 and write this in symbols as {x} ~ {y,}. We shall now show that this is an equivalence relation in X, i.e., the relation ~ is reflexive, symmetric and transitive. Reflexivity: {x} ~ {a}, since d{xy,x4)=0 for every Time dlp, Xq) = 0. Symmetry: If {x4} ~ {ys then limyo dl Ju) (Xu, ¥n) = A%,) for every n, and, therefore, lim. dq) =0, so that Un} ~ {an}. Transitivity: If {x0} ~ {yn} and {yn} ~ {Znfs then lim so d(%n Yu) = 0 and limy 2, d( 9») = 0. We shall show that limy 3. d(x,» 2») = 0. Since OS d(x, Zn) Xn Yn) + Any Zn) for all n, it follows that OS lim d(Xny 20) S lim dm yn) + lim dyn 2n) = 0, ies, Limy oe A(X Z) = 0. ‘Thus, ~ is an equivalence relation and X splits into equivalence classes. Any two members of the same equivalence class are equivalent, and no member of an ‘equivalence class is equivalent to a member of any other equivalence class. Let X denote the set of all equivalence classes; the elements of X will be denoted by %;j, ete. Observe that if a Cauchy sequence {x,} has a limit xX, and if {y,) is56 1. Basic Concepts equivalent to [ ing inequality: J}, then lim)... x, =: This follows immediately from the follow- AU Ypy X) = ACY Xp) + A(X 2). Moreover, if {x9} and {yq} are two nonequivalent sequences, then limy-. x, # Timo. Yu For, if limy sy == lim. yy then the inequality 0S (5 Yu) S Alan, x) + d(x Jn) leads t0 limy- so l(a yx) = 0, contradicting the fact that {x,} and {y,} are two nonequivalent sequences. The constant sequence (4, 4 ...,%,...) is evidently Cau- chy and has limit Define a mapping f:X — X as follows: f(x) = %, where % denotes the equiva- lence class each of whose members converges to x. Thus the constant sequence (X,3)...5)...) isa representative of %. In view of the observations made above, the mapping fis one-to-one. We next define a metric p in X. For x,y € X, set PHF) = Him dy ya), where (49) € ¥ and {yy} € 7 Observe that [Cn yu) ~ A(x Yu)| Ae Sn) + Ans Yu)s where {xq} © & and {y,} © 7 and so {d(%j.¥x)}n=1 is a Cauchy sequence of real numbers. Hence, lity: d( 9, yn) exists, for R is a complete metric space. We first show that p is well defined. Indeed, if {x7} ~ {xn} and {4} ~ {yj} then fim (xXx) = 0 and lim d(y»y/,) = 0. From the relation (xy ¥/,) Sd(x5 Xn) + A(X» Yn) + dur Yy) and the relation (Sp Yn) SAC» X,) + A ¥,) + Ao Yu)» it follows that Tim (Xn Yn) = lim d(xjo J, ‘We next show that p is a metric on X. Since d(%pYq) =O for all_m it follows that limy-oe d(x yq)=0. ‘Thus, 0(% 7) ZO. IF =F, then pH P) =limy co dl su Yu)» where [x4] €% [yal €F and {x,} ~ {y}- But this means (by definition of ~) that lim. d(x ¥4) = 0. Therefore p(%,¥) = 0. Conversely, if pl% 7) =0, then limy sae dln Yu) = 0 (by definition of p) and hence {xy} ~ {ys}, 80 that 5 = 7. Since d(x, yn) = d(Yas Xa) for all n, we have p(%,) — p(7,%). Finally,1.5. Completion of a Metric Space 37 0(%2) — lim d(x Zn) = Tim d(x ya) + Tim dyn Zn) — 0 J) + PZ) where Lan} € % (Yul € and (2) € 2 ‘The metric p defined in X has the property that p(%, of F(x) fly) = diy) for all x,y € X, ie, fis an isometric embedding of X into X. Next we prove that every X€ X is the limit of a sequence in f{X). For this purpose, suppose {2%} is a Cauchy sequence in the equivalence class %. For any k EN, there exists a positive integer ny such that d(x), %,) <1/k for n= m, Let ye be the equivalence class containing all Cauchy sequences converging to Xq, ies Jk = fn) Then 6%, F%n,)) Thus, ¥ = lime ooo f(%n,). Finally, we show that X is complete. Let {%'"!} be a Cauchy sequence in ince each X”) is the limit of a sequence AX), we can find 7!" € F(X) such that p'"), 7) < 4 Then the sequence {7} in X can be shown to be Cauchy by arguing as follows: 69, 7h) < pp, 2) + pe”, 3) + pH, 7m) 1 hy qin gm Sy tee x to The right hand side can be made as small as desired by choosing mand nlargeenough, for {%"} is Cauchy. Since } isin f{X), thereexistssome yn € X such that f(y) = 7", ‘The sequence {y,} in X must be Cauchy because {j/"} isa Cauchy sequence in X and fis an isometry. Therefore, {y,} belongs to some equivalence class % € X. We shall show that lim... pC”, 2) = 0. To this end, we take any & > Oand observe that 12,3) = pl, Ff) + 063) <2 + 9H.) and 069", 3) = p(n) 5)) = lim dm yp) $6 for sufficiently large 1, since {yy} is a Cauchy sequence in X. This implies that Time pC", X) = 0, thereby completing the proof that X is complete. Uniqueness: Let (X*,d*) and (X**,d**) be any two completions of (X,d). We shall show that (X*,d*) and (X**,d"*) are isometric. Tet x* © X* he arbitrary. By the definition af completion, there exists a sequence {qhn=1 in X such that lim, ox, = x*. The sequence {x,},=1 may be assumed to belong to X**. Since X** is complete, {x,},=1 converges in X** to x**, say, ie., limy-oo ay = x**. Define g: X* — X** by setting g(x*) = x"*. It is clear that the mapping © is one-to-one and does not depend on the choice of the sequence {lie converging to x*. Moreover, by construction, @(x) = x for x ¢ X and a*(o(x*), @(9")) = d*(x*, y*) for all x*,y* € X*. ‘Thie completes the proof, a58 1. Basic Concepts Remarks. (i) The proof explicitly assumes the completeness of R. Hence, the above method of completion cannot be employed for constructing the real number system from the rational number system. (ii) There exist other methods of completion of an incomplete space. One such method will be provided in Example. 17 of Chapter 3 (Section 3.8). 1.6. Exercises 1. For a,b > 0 and 0 < p< 1, show that (a + b)? sa? + bP. Hint: The function g(x) = 4 =1, has derivative ae which is positive when x > I. So, g(x) is increasing for x> 1. Moreover, g(x) > 1 as x — 00, and hence, g(x) <1 for x= 1, Put x = a/b or bla, whichever is = 1. 2. Let X = C" and 0 < p< 1. Define dy by ‘i Up d(z,w) = ( la- nt) 5 +) are in C", Does dp define a where 2 = (2525-524) and w= (05 Ws. metric on C"? Hint: No, For z=(1,1,0,0,...,0),¢=(0,1,0,0,...,0) and w=(0,0,0,0,...,0), dy(2,w) = 24? dyl 2,0) = 1= dy( Sw), $0 that dp(z,w) > do(2s6) + dol sw). 3. Let X be the set of all sequences {2}, of numbers that are p-summable, i.e. Zs lil” < 00, with d:X x X — R defined by d(z,w)= 7 \a— wi)’, where 0 0, 0 < p< and Exer-1.6. Exercises 59 (MS4*) If x,y,z € X are distinct, then d(x, 2) = d(x, y) + diy). Hint: For x= y,d(x,z) = d(y,z) = d(y,y) + dly,z) = d(x y) + d(yz). The argment for y = 2 is similar. For x = 2,d(x,2) = 05 d(x y) + d(y,2). 5. Show that d(x, y) = y/|x — y| defines a metric on the set of reals. 6. (a) Let X = Rand for x,y ER, define d(xy) by aflx-v41 if exactly one among x and y is strictly positive. — { yl otherwise. Prove that (Xd) is a metric space. (b) In any space X with metric d, let f be a self map that is one-to-one. Set (xy) = d(f(cd.f(v) ). Prove that Dis a metric on X. (©) Inany space X with metric d, let Uand V be disjoint nonempty subsets with union X; define D(xsy) to be d(x, y) + 1 if exactly one among x and y belongs to U, and to be d(xy), otherwise. Prove that D is a metric on X. Hint: (a) Ifx > Oand 250, then d(x,z) = |x — 2) +15 |x—y|+|y—2] +1. 1F ¥=0, this equals d(x, y) + dly,2) because |x— y| +1 = d(xy). If y > 0, this again equals d(x, y) + d(y, 2) because now |y — z| +1 = d(y,2). (b) Each of properties (MS1),(MS3) and (MS4) for D is an immediate conse- ‘quence of the corresponding property for d. Property (MS2) follows for D from the corresponding property for d in conjunction with the fact that fis ‘one-to-one, Note that for R with the usual metric and fdefined as f(x) = x for x <0 and f(x) =x-+1 for x=0, the metric D is precisely the one of Exercise 6(a). (6) Imitate the argument for (a). 7. (a) Let X = R4, and for x, y € R* define d(x) by Ia — nl ife=y> (xy) = dC (x15 %2)s y2)) = et ain yal-+bil if otherwise Show that (Xd) ia a metrie space. (b) Let X =R, and for x, y € R define d(x, y) = |x| + |x — y| + |y| when x 4 y, and d(x, y) = 0 when x = y. Show that d is a metric on R. Hint: (a) Let x = (a5), = (yi, y2) and z = (zi,22) be in X. We note firstly that Jar — yi] S(x,y). If x9 = yo, then dix,y) = lar — i] S[mi — alt la — yi] S (x2) + d(z,y)- fx # yoy then 2» cannot be equal to both x2 and yp; so assume 2 # %. Then Cx) = ail + [x2 — yal + [yt] aa] + fae — 22] + [ee — vel + |] 2{talth altel th nl ifpea, (jail + ba — 29] + lal) + (ail +l — alt la) if Am = d(x,2) +dlz,y). (b) Straightforward.60 1. Basic Concepts Ficure 16 8, Let X = C(0,1], and for x,y € X define d(x, y) = supf|x(¢) — y(1)|:0= r= 1). Calculate the distance between x(t) = and y(t) =. Hint: |x(t) — y(t)| = +t- =-(/2— 0)? +1/4 so sup {| x(t) — y(e)]: 0=t a min Ulsy — nls 1h = Prove that (X,d) is a metric space. Let {x!""},=1 = {{x!"})= bea sequence in X. Show that x") — x in (X,d) if and only if x," — » for all k. Hint: min {|x — yl, 1} SL. If min {|xq— yq[) 1} = 1 or min {ly — nl, LP = 1, then min {|x — zp] 1} min {lay — ynfy 1} + min {|yn — Zely 1}. So, 1 1 “1 amin (lx, — ze, SD gy min lng — yale M+ SG min (lye — zeh 1 & a 1 1 < Spemn fixe — yb + 2 & min {I — ze 1 f f and hence, d(xz)Sd(xy)+d(y,2). In the case min{[x)—yylp1}<1 and min{|ye—Zaly1}<1, min {lacy —yuls 1] min {yy —Zyls 11a Fal + [Ya boe—Zyl2 min {|xp— Zeb 1. Assume x\" — x for all k. For a fixed ¢ > 0, choose an integer p such that TR pu Ve < (1/2)e. There exists an integer 1p such that = 1g implies1.6. Exercises 6 3 kay J in th — mn a=Ptt ‘The converse is left to the reader. 10. Let X = C(R), the set of real-valued continuous functions on R, and let sy € X, Define dy(x, y) = sup{|x(t) — y(t)|:¢ € [n,n] and C1 daly) HOD DWTS dile9) oat Prove that d is a metric on X. Hint: Cleary, diy = 1,2,-.. i a pseudometric on X. If dy(xsy) =0, then 1,2,...5 is also a pseudometric on X. Since for all sy € Xspy x uni- formly and therefore x(W) = a(0) = i ysdds, ie. (1) = y(1). Since each x, is continuous and x, — y uniformly, therefore y is continuous. Hence, x € C'[0, 1]. Show that the sequence {xy},=1, Where x» = (1, 1/2,..-51/1,0,0,...) con- verges to x = (1,1/2,1/3,...51/m...) in Hint: d(xpx) = (Saga (1/@)) 2+ 0 as n+ 00. Let {u)y=1 be a sequence in a metric space (Xd) such that the three sub- sequences {x2mbn=11 [Xonet}nz1 and {25u}21 axe all convergent. Show that {ular i8 convergent. Hint: By hypothesis, there exists an integer 1 such that d(x x») < e/3 for n and m even and =n; d(x» %n) <£/3 for nm and m odd and 19; dy Sn) < 6/3 for wand m multiples of 3 and = ny. For n even and m odd, choose | even and h odd. each a multiple of 3 and hh =m. Then Ap Xp) = Ay 84) + AChy X4) + ACR An) m, [50 — Xin] = een = Sint Xn — Xz Xun tt mp — Knl 5 [sy — Beil + [Sica — tacal +o. Hitt — Sal 1 + spl al 1 =z |b— al. 15. Describe the convergent sequences in the metric spaces of Exercises 6 and 7. 16. Let (Xd) be a metric space and C(X,R) denote the set of all continuous bounded real-valued functions defined on X, equipped with the uniform metric A(f,g) = sup{|f(2) — g(x)| zx € X} Show that C;(X,R) is a complete metric space. 17, Let X be the set of complex numbers and d:X x X — R be defined by (2,2) = 0 when 2 = 2 but |z4| + |z9| when 2; # 2). Show that (Xd) is a metric space which is complete. Hint: If z1,z, and zs are distinct, then d(zi,zs) =z] +|z]<_ |al+ lea] + [za] + [zs] = dlainze) + dlz,z:). If 21 = 2, and zs is distinct from 2 (and hence from 2), then d(z1,25)=|a|+|al= lal +z (23525) = d(21, 2) + d(z, 23), since d(21, 22) = 0. The other cases are trivial. We note that |z| = d(z,0) whether z= 0 or not. Now let {z,] be a Cauchy sequence. If there is some p such that zy = 2, whenever m, = p, then {2} converges to zp, If there is no such p, then there is some subsequence {4} all points of which are distinct, so that d(z,1) Zui) = |2na)| + [Zan] when j # ke ‘We argue that the subsequence converges to zero: Consider any & > 0; there exists my such that d(zyy2,)< whenever mnZ, $0 that lens] + [zc] = €(n4)s 2uj)) < € whenever k,j=no and j # k. In particular, lzu)| <€ whenever k= ny, However, as noted earlier, |2x)| = d(2yt),0)- Therefore the subsequence converges to zero. The rest follows by Proposition 147.