11. Z 1 0 dx =2 √ x(1 − x) Z Z 1/2 s 0 1/2 dx 1 1 2 − x− 4 2 dx c→0+ c 1 2 1 − x− 4 2 1/2 = 2 lim sin−1 (2x − 1) = π. = 2 lim s c→0+ The integral converges. c 14. Z π/2 0 = C sec x d x = lim ln | sec x + tan x| C→(π/2)− 0 lim C→(π/2)− ln | sec C + tan C| = ∞. This integral diverges to infinity. √ √ 2 x x x 38. Since 0 ≤ 1 − cos x = 2 sin2 = , for x ≥ 0, therefore ≤ 2 2 2 2 Z π2 Z π2 dx dx , which diverges to infinity. √ ≥2 x 1 − cos x 0 0 √ 37. Since sin x ≥ 2x on [0, π/2], we have π Z ∞ 0 Z π/2 sin x dx x2 0 Z 2 π/2 d x ≥ = ∞. π 0 x | sin x| dx ≥ x2 The given integral diverges to infinity. y y=sin x 2x y= π π 2 Fig. 5-37 x 40. Since ln x grows more slowly than any positive power of x, therefore we haveZln x ≤ kx 1/4 ∞ 1 dx 1 ≥ 3/4 for x ≥ 2 and for some constant k and every x ≥ 2. Thus, √ √ kx x ln x 2 Z x ln x 1 ∞ dx . diverges to infinity by comparison with k 2 x 3/4 4. Z ∞ dx √ + x +1 1 t2 1 2 dt dx = − 3 t Z 0 1 2 dt = − 3 2 r t 1 1 1 + 1 + t2 t2 Z 1 t dt . =2 4 3 0 t +t +1 x2 Let x = 5. Z π/2 dx √ sin x 0 =2 Z 1 0 1 =2 Z 1 =4 Z =4 Z 0 0 0 1 Let sin x = u 2 √ 2u du = cos x d x = 1 − u 4 d x u du √ u 1 − u4 du p (1 − u)(1 + u)(1 + y 2 Let 1 − u = v 2 −du = 2v dv v dv p v (1 + 1 − v 2 )(1 + (1 − v 2 )2 ) dv p . (2 − v 2 )(2 − 2v 2 + v 4 ) 3. One possibility: let x = sin θ and get I = Z 1 −1 ex d x = √ 1 − x2 Z π/2 esin θ dθ. −π/2 Another possibility: I = Z 0 −1 ex d x + √ 1 − x2 Z 1 0 ex d x = I1 + I2 . √ 1 − x2 In I1 put 1 + x = u 2 ; in I2 put 1 − x = u 2 : so I = 2 Z 0 1 eu 2 −1 I1 = Z I2 = Z + e1−u √ 2 − u2 2 du. 1 0 1 0 Z 1 2 2eu −1 u du =2 √ u 2 − u2 0 Z 1 2 1−u 2e u du =2 √ u 2 − u2 0 eu √ 2 −1 du 2 − u2 2 e1−u du √ 2 − u2 8. Let ∞ 1 t 1 dt dx = − 2 t Z 0 Z 1 −1/t 2 1 e −(1/t)2 e − 2 dt = = dt. t t2 1 0 I = Observe that Z 2 e−x d x Let x = 2 e−1/t t −2 h ∞ i lim = lim 2 t→0+ t→0+ e1/t t2 ∞ −3 −2t = lim t→0+ e1/t 2 (−2t −3 ) 1 = lim = 0. 2 t→0+ e1/t Hence, i 1 1 h −4 −1 0 + 4(4e ) + e S2 = 3 2 ≈ 0.1101549 1 1 0 + 4(16e−16 ) + 2(4e−4 ) S4 = 3 4 16 −16/9 −1 e +e +4 9 ≈ 0.1430237 64 64 1 1 0 + 4 64e−64 + e−64/9 + e−64/25 + S8 = 3 8 9 25 64 −64/49 16 −16/9 −16 −4 −1 e + 2 16e + 4e + e +e 49 9 ≈ 0.1393877. Hence, I ≈ 0.14, accurate to 2 decimal places. These approximations do not converge very 2 quickly, because the fourth derivative of e−1/t has very large values for some values of t near 0. In fact, higher and higher derivatives behave more and more badly near 0, so higher order methods cannot be expected to work well either. 14. Let y = f (x). We are given that m 1 is the midpoint of [x0 , x1 ] where x1 − x0 = h. By tangent line approximate in the subinterval [x0 , x1 ], f (x) ≈ f (m 1 ) + f ′ (m 1 )(x − m 1 ). The error in this approximation is E(x) = f (x) − f (m 1 ) − f ′ (m 1 )(x − m 1 ). If f ′′ (t) exists for all t in [x0 , x1 ] and | f ′′ (t)| ≤ K for some constant K , then by Theorem 11 of Section 4.9, K |E(x)| ≤ (x − m 1 )2 . 2 Hence, K | f (x) − f (m 1 ) − f ′ (m 1 )(x − m 1 )| ≤ (x − m 1 )2 . 2 We integrate both sides of this inequlity. Noting that x1 − m 1 = m 1 − x0 = 21 h, we obtain for the left side Z x1 Z x1 f (x) d x − f (m 1 ) d x x0 Z x1 ′ − f (m 1 )(x − m 1 ) d x x0 x0 Z x1 2 x1 (x − m ) 1 = f (x) d x − f (m 1 )h − f ′ (m 1 ) x0 2 x0 Z x 1 = f (x) d x − f (m 1 )h . x0 Integrating the right-hand side, we get K (x − m 1 )3 x1 K 2 (x − m 1 ) d x = 2 3 x0 2 x0 3 K h h3 K 3 = + = h . 6 8 8 24 Z Hence, Z x1 x0 Z = x1 f (x) d x − f (m 1 )h x1 x0 [ f (x) − f (m 1 ) − f ′ (m 1 )(x − m 1 )] d x K 3 h . ≤ 24 2 1 1 1 1 1 1 x d x = . M1 = 13. I = . (1) = . The actual error is I − M1 = − = 3 2 4 3 4 12 0 Since the second derivative of x 2 is 2, the error estimate is Z 1 2 |I − M1 | ≤ 2 1 (1 − 0)2 (12 ) = . 24 12 Thus the constant in the error estimate for the Midpoint Rule cannot be improved; no smaller constant will work for f (x) = x 2 . 46. Ŵ(x) = Z ∞ t x−1 e−t dt. 0 a) Since limt→∞ t x−1 e−t/2 = 0, there exists T > 0 such that t x−1 e−t/2 ≤ 1 if t ≥ T . Thus Z ∞ Z ∞ x−1 −t e−t dt = 2e−T /2 t e dt ≤ 0≤ T T and Z ∞ t x−1 e−t dt converges by the comparison theorem. T If x > 0, then 0≤ Z 0 T t x−1 −t e dt < Z T t x−1 dt 0 converges by Theorem 2(b). Thus the integral defining Ŵ(x) converges. Z ∞ t x e−t dt b) Ŵ(x + 1) = 0 Z R t x e−t dt = lim c→0+ R→∞ c d V = e−t dt U = tx dU = xt x−1 d x V = −e−t ! R Z R t x−1 e−t dt = lim −t x e−t + x c→0+ c c R→∞ Z ∞ t x−1 e−t dt = xŴ(x). =0+x 0 c) Ŵ(1) = Z ∞ 0 e−t dt = 1 = 0!. By (b), Ŵ(2) = 1Ŵ(1) = 1 × 1 = 1 = 1!. In general, if Ŵ(k + 1) = k! for some positive integer k, then Ŵ(k + 2) = (k + 1)Ŵ(k + 1) = (k + 1)k! = (k + 1)!. Hence Ŵ(n + 1) = n! for all integers n ≥ 0, by induction. Z ∞ 1 d) Ŵ t −1/2 e−t dt Let t = x 2 = 2 0 dt =Z2x d x Z ∞ ∞ √ 1 −x 2 2 e−x d x = π e 2x d x = 2 = 0 0 x 3 1 1 1√ Ŵ = Ŵ = π. 2 2 2 2