Figure 8 Figure ‘7 3- y y = x .3 = - 2 .t S 1 x ‘ x Fixed-Point Algorithm — — = g(x,) = (2Vx,, + 1)1/2 = (x + 1)’/ = 2 cos x using the Fixed-Point Algorithm. I SOLUTION Note first that solving this equation is equivalent to solving the pair 2 cos x. Thus, to get our initial value, we graph these x and y of equations y two equations (Figure 8) and observe that the two curves cross at approximately x 1. Taking x 11 = 2 cos x,, we obtain the 1 = 1 and applying the algorithm x, results in the following table. Solve x 4 5 6 • EXAMPLE 5 1.8350868 1.8350867 1.8350867 1.8350867 9 10 11 12 1.8392994 1.8357680 1,8351969 1.8351045 3 1.8350871 8 1.8612097 2 1.8350896 2.0 7 n 1 n Figure 7 suggests that the point of intersection of the curves y x and 1/4 ‘+ occurs between 1 and 2, probably (x closer to 2, so we i) take x 1 = 2 y ‘V as our starting point. This leads to the following table. The solution is approxi mately 1.8350867. 1 x÷ 2 = 2Vx -1- 1, which leads to x ± (2VJ)h/2. SOLUTION We write x Since we know the solution is positive, we take the positive square root and write the iteration as • EXAMPLE 4 Use the Fixed-Point Algorithm to approximate the solution of 2ViJ = 0. f(x) 1. — 1 be an initial approximation to the Let g(x) be a continuous function, and let x root r of x g(x). Let E denote a bound for the error x,I. 1, 2,... until Ix, Repeat the following step for n i x,j < E: 1 Al orithm — The Fixed-Point Algorithm The Fixed-Point Algorithm is simple and straightforward, but it often works. Suppose that an equation can be written in the form x g(x). To solve this equation is to find a number r that is unchanged by the function g. We call such a number a fixed point of g. To find this number, we propose the following algo . Then let x 1 2 = g(x rithin. Make a first guess x ), x 1 3 = g(x ), and so on. If we are 2 Do. lucky, x, will converge to the root r as n 1 is wrong (see also Problem 22). For the function in Figure 6, the difficulty is that x not close enough to r to get a convergent process started. Other difficulties arise if f’(x) is zero or undefined at or near r. When Newton’s Method fails to produce approximations that converge to the solution, then you can retry Newton’s Method with a different starting point, or use a different method such as the Bisection Method. n— OC Newton’s Method creates a sequence of successive approximations to the root. (We mentioned sequences briefly in Section 1.5.) It is often the case that Newton’s Method produces a sequence {x} that converges to the root of f(x) 0, that is, urn .x, = r. This is not always the case, however. Figure 6 illustrates what can go Chapter 3 Applications of the Derivative Figure 6 194 r — 3. 2 cos x sin x = 0; [1,2] 4. x—2+2cosx=0;[1,2] 8 9 10 1.4394614 0.2619155 1.9317916 —0.7064109 1.5213931 = 2 x,, + 2 cos x,, 7 8 9 10 11 12 1 1.0403023 1.0261107 1.0312046 1.0293881 1.0300374 1 2 3 4 5 6 1.0298054 1.0298883 1.0298588 1.0298693 1.0298655 1.0298668 x 13 14 15 16 n 1.0298665 1.0298666 1.0298665 1.0298666 X,, = x is called a of g. — + 5 + — 0 + 2 cos x = + 9x 2 6x 6. The real root of 7x 3 + x 7. The largest root of x 2 5. Thelargestrootofx 3 = 0 (see Problem 4) 1=0 1 In Problems 5—14, use Nesvtonr Method to approximate the indicated root of the given equation accurate to five decimal places. Begin by sketching a graph. 4. A point x satisfying g(x) ...,,.__; 3. The Bisection Method, Newton’s Method, and the FixedPoint Algorithm are examples of that is, they provide a fi nite sequence of steps that, if followed, will produce a root of an equation to desired accuracy. Now we raise an obvious question. Why did the second algorithm yield a con vergent sequence, whereas the first one failed to do so? Whether or not the FixedPoint Algorithm works depends on two factors. One is the formulation of the equation x g(x). Example 5 demonstrates that an equation such as x 2 cos x can be rewritten in a form that yields a different sequence of approximations. In Example 5 the reformulation was x = (x + 2 cos x)/2. In general, there may be many ways to write the equation and the trick is to find one that works. Another factor that affects whether the Fixed-Point Algorithm converges is the closeness of the starting point x 1 to the root r. As we suggested for Newton’s Method, if the Fixed-Point Algorithm fails with one starting point, you can try a different one. n xn ii In Problems 1—4, use the Bisection Method to approximate the real root of the givelz equation on the given interval. Each answer should be accurate to two decimal places. 1. x 3 +2x—6=0;[1,2] 2. 3 -1-5x 1 = 0;[—1,0] 4 x + Problem Set 3.7 . 6 7 x This process produces a convergent sequence, shown in the following table. (The oscillation in the last digit is probably due to round-off errors.) 1. The virtues of the Bisection Method are its simplicity and reliability; its vice is its 2. If f is continuous on [a, bi, and f(a) and f(b) have oppoSite signs, then there is a of f(x) = 0 between a and b. This follows from the Theorem. Concepts Review 1 1.0806046 0.9415902 1.1770062 0.7673820 n 195 Quite clearly the process is unstable, even though our initial guess is very close to the actual root. Let’s take a different tack. Rewrite the equation x = 2 cos x as x = (x -i- 2 cos x)/2 and use the algorithm 5 1 2 3 4 n Section 3.7 Solving Equations Numerically -I- — 1 — 24x = 0 = 2 ± 24.v 2.v sinx + — 14. The smallest positive root of 2 cot x — 3 6x 2 3 H- 22x 8x = 2 13. The positive rootof 2x 4 12. Alireal roots ofx 11. All real roots of 4 — 8 -f 8 = sin x 0 0 = 0 (see to five decimal = = = f(s) f(s) 18. 19. 20. f(x) H- + 3 x +- 2 x 2 sin; [0,4ir] x -;[1T,31rj 4 x • [—1,1] 1 (l+i)k H- 1)24 — (1 + i) 24 -1- 1 = 0 L 500i — 4 23 E20i H- 19i —1 -4- (1 ± i,j— (c) Find i accurate to five decimal places starting with 0.012, and then give the annual rate r as a percent (r 1200i). 1,, 24. In applying Newton’s Method to solve f(x) = 0, one can ,x 1 ,x 2 , 3 usually tell by simply looking at the numbers x whether the sequence is converging. But even if it converges, say to T, can we be sure that ? is a solution? Show that the answer is 0. yesprovidedfandf’ are continuousatT andf’() E1 = (b) Show that Newton’s Method for this equation reduces to 20i(1 where i is the interest rate per month. Tom bought a used car for $2000 and agreed to pay for it with $100 payments at the end of each of the next 24 months. (a) Show that i satisfies the equation i 23. In installment buying. one would like to figure out the real interest rate (effective rate), but unfortunately this involves solving a complicated equation. If one buys an item worth $P today and agrees to pay for it with payments of $R at the end of each month for k months, then 173 Obviously, its only x x 22. Sketch the graph of y = . intercept is zero. Convince yourself that Newton’s Method fails to converge. Explain this failure, 21. Kepler’s equation x = in -I- E sin x is important in as tronomy. Use the Fixed-Point Algorithm to solve this equation forx when in = 0.8 and E = 0.2. L1 = 17. f(x) In Problems 17—2 approximate the values of x that give maximnuni and niininiumn values of the function on the indicated intervals. E i. Use Newton’s Method to calculate places. F1 15. Use Newton’s Method to calculate “/ to five decimal places. Flint: Solve x 3 6 0. E1 2x 10. Therootof2x —sin x 9. The root of cos x — Chapter 3 Applications of the Derivative S. ‘The smallest positive root of 2 cos x Problem 3) 196 = = 1 \/.2 + x; x \/2.7 + x; x 1 = = 47 1 — = 5(x — + 1 + 33. Consider x = + 1 + + + = 1 + 1 S —. 1 x 1 x = = = 1 to 0 to 0 to ) 2 x — 1+ 1+ 1 1 1+• 1 = 1 2x, — ax using several different values of a. (a) Make a conjecture about what this algorithm computes. (b) Prove your conjecture. x+ 35. Experiment with the algorithm 34. Consider the equation x = x — f(x)/f’(x) and sup pose that f’(x) 0 in an interval [a, b]. (a) Show that if r is in [a, b] then r is a root of the equation x = x f(x)/f’(x) if and only if f(r) = 0. (b) Show that Newton’s Method is a special case of the FixedPoint Algorithm, in which g’(r) = 0. 1+ (c) Evaluate the following expression. (An expression like this is called a continued fraction.) (b) Algebraically solve for x in x (a) Apply the Fixed-Point Algorithm starting with , x 3 , and x 4 . 5 find x ,x 2 E (c) Evaluate 32. Consider x = (a) Apply the Fixed-Point Algorithm starting with find 2, x , x 3 , and x 4 . 5 (b) Algebraically solve for x in x = 1 (c) Evaluate 1 (a) Apply the Fixed-Point Algorithm starting with x 1 find x . 5 ,x 2 ,x 3 , and x 4 (b) Algebraically_solve for x in x = 1 31. Consider x = 30. Follow the directions of Problem 29 for x g(x). E ) = g(x). 2 x 29. Consider the equation x = 2(x (a) Sketch the graph of y = x and y g(x) using the same co ordinate system, and thereby approximately locate the posi tive root of x = g(x). (b) Try solving the equation by the Fixed-Point Algorithm start ing with x 1 0.7. (c) Solve the equation algebraically. 11 28. x 27. x 2 x=2—sinx;x = 26. 1 25. x=cosx;x =1 1 In Problems 25—28, use the Fixed-Point Algorithm with x 1 as indicated to solve the equations to five decimal places. L1 Figure 1 Figure 9 3.8 Figure 10 F’(x) = 3 4x x Antiderivatives 1050 Answers to Concepts Review: 1. slowness of convergence 2. root: Intermediate Value 3. algorithms 4. fixed point (a) Find the position of the object when it is closest to the observer. (b) Find the position of the object when it is farthest from the observer. stands 3 feet from the bottom of the cliff. —-- 39. An object thrown from the edge of a 42-foot cliff follows 2 2x + x + 42 (Figure 10). An observer the path given by y = 197 = f(x) on I, that is, if I Find an antiderivative of the function f(x) = 3 4x on A moment’s thought will suggest other solutions to Example 1. The function F(x) = x 4 + 6 also satisfies F’(x) = 4x ; it too is an antiderivative of 3 + (x) . 3 4x where In fact, F(x) = 4 x C is any constant, is an antiderivative C, f 3 on (—, no) (see Figure 1). of 4x SOLUTION We seek a function F satisfying F’(x) = 4x 3 for all real x. From our experience with differentiation, we know that F(x) = x 4 is one such function. (-oo, Do). • EXAMPLE 1 We said an antiderivative rather than the antiderivative in our definition. You will soon see why. We callFan antiderivative off on the interval Ill DF(x) F’(x) f(x)forallxinl. Definition Most of the mathematical operations that we work with come in inverse pairs: addition and subtraction, multiplication and division, and exponentiation and root taking. In each case, the second operation undoes the first, and vice versa. One rea son for our interest in inverse operations is their usefulness in solving equations. For example, solving x 3 8 involves taking roots. We have been studying differen tiation in this chapter and the previous one. If we want to solve equations involving derivatives we will need its inverse, called antidifferentiation or integration. E1 After differentiating and setting the result equal to zero, many practical max—mm problems lead to an equation that cannot be solved exactly. For the following problems, use a numerical method to approximate the solution to the problem. 36. A rectangle has two corners on the x-axis and the other . What are the 2 two on the curve y = cos x, with —n’/2 < x < iT/ dimensions of the rectangle of this type with maximum area? (See Figure 24 of Section 3.4.) 37. Two hallways meet in a right angle as shown in Figure 6 of Section 3.4, except the widths of the hallways are 8.6 feet and 6.2 feet. What is the length of the longest thin rod that can be carried around the corner? 38. An 8-foot-wide hallway makes a turn as shown in Fig ure 9. What is the length of the longest thin rod that can be car ried around the corner? Section 3.8 Antiderivatives