l36 Dijj et eutiutiu u n Secttn,oll 32. ilimiits For a more advanced example consider Contrinulty alad M i once again di-rect subsiitution rvill not do, Evidently symbolized j. Direct substirution yierds closer io as 3,/(x) becomes closer = 13 , or/(x) -+ 13 as -r + 32'1. we shalr say jrg:/(-y) = r tf f(x) can be made arbitrarily crose to ll::::l --^- ]tg53x then.fair{y.easy see that -to limit is 10, value of the if-r approaches (") fim Yrt' (3-r + L z Direct subsiiruiion yierds ,i (which is yery near to zero), then * (whatever * means). J{r..ft"l rla s(x)] = /(x) S(x) , lim =x_)a '-" !g"f@)/g(x) = = - 106, rvhicl; is iiot even I l#ge number of Limits vrithout r rucrr flraboLu. =Land _t r g(x) = IttI. Tlrcn t_o x+a" r- ,{;/trl r 1im s(x) = L + kt. ,f(;r) .tim s&) = LM x+o".. 4-a ;tg.f$)/fi"!{x) = (- . t,* "mF, --,-sqat ,provit)ed tc +0 Et,. The proofof this theolem is son-rewhat technical aud we shall not consi,ter.it hele. Theonerm 32.3. (a) then 3x + 4 approaches 10. Thus the Example lin x=a x4o (b1 lf f(.t)=c *r"r,J'urr"*, ;;r';J; ;l i'J,nriclr then tim /(x)=c. , , x)a" L Evaluate lirn (x2 + 4.r + 3). x -+2 lVe have jE 'l}"' ff t * However, using rhis rheorem ana ure two-rairiy curailed. "-u.ri"", 4) = 10. fi- ;' -.t ,Jt.r_l the nreaningles, ro* -lim x)o ': [[(x) F- ri C. one mighr be.remprerr ro thi,,,r rhar (dl lim /r/(x) = k lim /(-r) = ld- for ail k e X-+a x )a' Sometimes. however, matters ate not so obvious. For example consider observ Lig , . ofcourse, one's reac{.ion to this may.be that it is al] utteriy obvious jndeed in u'I@rr manv cases --- 'limits can be obrained bv direct ruuirii"tio* rrri;."""";;u;;;#, : Iim (x2 - 4) =21 and lim^ (x + 3) = 1. - r+5' ,8. (b) 0. z, 10-6 lfheorery! 32.2, Suppose thatx should be arrowed to ser croseland cioser to aTt"^y::Lyhat,foilows it any fashion, but that it shourd neven actuarny nEacru o.-+ - (32.1) 3. io-a, uut without actuarly alrowing -r to be equar to a. "nough Consider for example x.= The following theorem ailows us ro compure a to 13. This wi_ri be (c) whe'ever rve choose x crore f However, lt IImIts Deillmteion ror- j . rrre undefined "91';= positive, let al6ne near above argumenr oniy makes (32.1) pJa-usibre - it does not prove it. rn order ro give a formal proof, we should need a.fblmar 'a.ri"ition. ihi;-;;;rd";; ;riiifi'airi"-ii i, ,rri, slage, so we shall malce do wirh $re following .i,jiirftf"* "Ai.j IT 9--- !-. (xz + 4r + 3) =iim^ xr =ltun^.r * ,b+ $r =4+B-r'3 . This problem ca,n be overcome by = ='tg1G+1)=2. I h"' ff F-;' X ..t T'e ,l ,*{}{tl'i:i:f* 1 x+Q r ,$,lttl ;: is far. _. lrm * Consider then = il f(4\ =15 "fA) f\?.s-) : 12 rb'.st =14 12.3 f.\?.\ = f\.l) = 13.2 f(z.ee) = t2.e8 7it.oi> = 13.02 lt would appear that as -r becomes ri bult fi.om obvious hor,,, to rroceed. Indeeditisnor.evencreartharr'eIimit.niir.utoii-dui*.i"i,',i,iprl'ir"i"i"r] .,ii"',i.i,*, It is obvious that/(3) = 13. However, the quesrion'we are interested in is not what is the value of /.v'rhel r:3, but what is the vaiqe orlrr,.n.r is ciose to 3. consider a few i'e' sin ;r does not exist. of -r as x approaches a deFrnile ualue. Let us consider a ve-ry simple exar4ple. Suppose/(x) Lr + = It is ,. igb , As we saw in the last section,.it-is.frequendy necessary to consider the iaiue of some ' function values. (* imits &nd rContinuitl, | 15. 4" 3 , (Theorern 32.2(a)) "J!1 * 4lim .r + 3 (by 'Iher:r.ems 32.2(b)(ti1, 32.3(b)) iby'Xheorern 3e.S(t,j) t, Differentiation i,,,:.auz,t|" :i, ,l' 1r .l .l Liwits and Continwity -r2 + * 6uutuui" 3i.+ i s 3tL.2,and3?,3 repeatedly; rve obtain Eremple x2+3x+1 ,, lllTl- ,$" : Lranplet such liry ' -a as *iese are covered ,b11r x ilm ta DU + lir ,+ bg+ P(l:,:Sp:.nt* + .:. + anxn an:d +O , then ull-=' ttm {Q = P-ftl , +a Q.(x) O(a\ bP Example + bza2 S. e@) = b6 + b1x + ., + :rmilar fashion to example 3. vhat fris f r. tlteorem savs is ."i,r.i ,rjir"ii!'J'""Ji:fi a The orems 32.2 and 32.3 n [til,Ti: iit*s lffitolll -* Example f@=++ ,l la: !c euough. rvi rny i.orLrlts "tim lirnirl r,f this : r' A --. () , rhen l<ind z + rJm x- ,ll+t a --7j fef,inaition large enough. fairly easy to evaluate. _. -.rst Fi we note the follo.aing simpie = 0" (.r _ needs t). anot'er definition. 32,6."lg1,f(x)=*ifw.1 canmalce/(r)aslargeaqwepleasebytakingr :_ If ii > 0 then Please .lge;tT7 =.,{|G-7tr lim =x+€ we sho'Id like ro say trrat this rimit is-infiiite. This by choosing.r r +1 "ry_#l_ The following results are fairly obvious a-r.e =o ;;"-;;;;i;i,i"til,g Dividingbyxwehave_tim Dividing byx we have show'l]:!ll)-, 1.5, f (?) -2.4, (r,o) f =z.g.t[3and /(r00) =2.9997. $ :::"#jil::{ ?:Hlf)o$qr"s .r*"..i,"J'croill,'o'i u, -r. becomes rarger and rarger. would YernltrQn obscu'e,the ittu., ro r.-i',ur["ti"'i'Jn'o*i'g sf,"ewi;;;;ii;;;;:' be made arbitrarily crose to 'ir 7. Evatuare Lj recr colnPutaLions /(r) = L ii Ik)can =-B- Please note that rhe methoa of the last three examples is only appricabre when evaruating a (or _r --;. rt isoi a iimtt as tends ro some finire "o given by ri;ion, 32'5' (32.2). ,t. ')*x-1+3x of a fr function it is frequently-usetut vely largeJ}tr'ffi%T jl. xample the function ro know what happens ro values ofx. Consia", i^..1 a#i,..,:??# ::N =hh+=*.tu J r+6 1 lim'r rorm can be evaluared bv direct subsritution sketch_ing the rhe €raph gaph _lre,r (ll'i,."::::!.n'ns rot' 5 a xla ) t. Dividing throughout by x we obtain a -l 1r ruu (32.2)) = . 6, Evaluate lim =" ' x-)-xz-X+3 he caoe. p(n) =.1 is of particutar inteiesr In that case Theorem 32.4 says *rar if p is any p p. ,r],'lortuat rl],'nornial dren ilren lim p(l) = p(d. t;,| I Dividing throughout by 12, we have x+-" Jt--:, -l xx4 of rirn to an artifice. Divid.ing the -2. 2x + 7 !'-x]*3x2+x-5' b^tm.If Exanple ' 50 w€ "', ' resorr Eva-luate 1;rn ,r;: -= P(a) alter repeared applications d.O , . t-- by the following theorem. lilaeoremr 32.4'.. Let .J@) b2xz = I t ., x+ .I '-:-L/ Irm; , =lim--=l= _ I x+6n r+_ r I + btx +'tb2xz), where,Ds, 01, b2 are constanrs. * b2x\ = fao + b.tx - and denominalor by x, we .c,xa:nple 3. tfn . _r_ig1 +.d V/e have, ,. - ++ clearly we cannot merery substirute.r e, -have = numerator J Eya.luale [m ;-ge; 19 L or @ is replaced by 4. Eva.luate x+l r+3 the results of Theorem 32.2 hold wher.e a (bur nor f"}t:i.tt Usin6 Tlreorem ,: t39 2. r2..\ pi ,, = r+e -F,t) = *. tim (-r,, . ]* i;. *), = _t+a t+e qote that jl-", :tlP:t "*" is only used when raUcing abour rcau \J'.LJ ilumner a limit. Z .i l$nrr*. ) I)iJfere nliatio n t90 ![!xampl] n 8. . r5 -2.r2 r 5.r t Evaluatc - -2r't'l "'*"**.'i-lirn '- Jtz -This is the sarne as ,,."' _ lr! r- Contifi miu,orus lFrnnatuont I Defimitiom 32,3. A function/is comtinuous if, for iury (. F-1.'+t'-*4 re- ----n-+ 3_Z+ E^ imits end Cantin*itjt . - xxL Exarnple ln general we can maire ttre following statement:- 9. Consider P(*) = ao + atx + ... + anxn QQ) = bs +bF :nd (a) if n>n thenlim =- hn f(x):J@) Geometrically-this means that the graph of./has no'lurnps" or "holesl'in ir. Tl,e r:oflcepl. ' may be most cJearly unde'stood byionsidering some lis..onilnuous f.nctions. =llm-=@ Suppose a, +'.. + b,&n lx+l when-r>b when-r<0^ Ix f@ :1 The graph of/is shorvn beior whate an, b^ > 0' ffi (b) P -" =O *ii0(x) ifri<rr (c) if n=m then iim **"*Lffi =fi X. . 'l'hus if the highest power of x in the denorninator is lesftfan the highest power in the numerator, thd ratio tends to infinity. If the reverse holds it tends to zero, while if the highest power in each case is equai, then the iimit is the ratio of the coefficients of the highest order terms. One further kjnd of limit arises natuaIIy. A typical case is given by .. Itm I , we note that 32.7. hnt x+a "f(;r) = if we can make/(.r) - as large as we please because the graph a It is knportant to note that we are choosing x to be near a, but .. ltm l1.r)=, 0.1, -0.0t ai.l .{).0{) ol/has a"jump" atx=0. E,xatnple 10, by taking ;r Consider close enough (but not equal to) a. whetlrerx > a ot x < a. For example . )+0" f(0.1) = 100, /(0.01) = I 04 ,/(-0. i) = i00 and/(- 0.01) = 1s+. lVe should therefore like to say the the iimit is infinite. The following definition is then natural combination of the ideas of Definitions 32.\ and32.6. lDefirnirtiom that lirn / (-r) = i x+0- different value frorn that obtained by approaching it irorn rhe righr. Tltur. lar fiorn being equal to/(0), JrJrl" /(x) is nreartingless,-i.e. f is nor cdniinuor_r:. ,l-his is 1 * One might be tempted to thinlc Flowever, tatdngx=-0.1, -0.01, -0.001 we have wrucn are not near lo l. Thus lirn/(x.)' doesno[evenexist,ibrifyouapproach0lromtheleftyo,,t:hlaina -, x -+0 XL Leuinc f(x) = Ey definition/(0) = 1. Now what is iim /(jr)? Talcingx=0.1, 0.01, 0.001 we have/(x)=1.1., 1.01, and 1.00i. fx2 when.r+ l /P; -{ t0 r,vhen -r = I we are not specifying r 1 r+0I does not exist. For when x:0.01 we obtain 100, but whenx = - 0.01 we have irlow rcca.ll (see remarir after Theorem 32.4) that Iim P(x) )t1562 100. ts, L' Lt' if P(-r) is a polynomial then = P(c). + 1) = 4 - 4 +7 = l( i For exantple - 2-x - (4.t'o )t' t Now/(i) =S by definition. htowever,/1l.1 ) lim, 7. This means that, in this case, one can merely substitute in when evaluating the Limit. This pluDcrly is so pleasant that we give it a special name, .LJl /{x) - =-.Zt,ltl.0t; Il'r1 1'l - | I +l = 1.9261 u,,O . i I lirr.rit lin / (-r) rve clci .ot co'sider the r allie ol f' )tl -I citself.) -Thus, since .l(1) * lim i(rr), I i:t n()t (,rcntinurru-r. This ir br:.rLrsre @emember, when evaluating the the graoh of/has a I "holc" irr il I {fiiH}mr l(- I iri rr*.{f:ilrff