Math 414 Professor Lieberman April 28, 2003 HOMEWORK #12 SOLUTIONS Section 6.4 5. (c) Integrate by parts (with u = sin(ln x) and dv = dx) to obtain Z Z Z cos(ln x) dx = x sin(ln x) − cos(ln x) dx. sin(ln x) dx = x sin(ln x) − x x Another integration by parts (with u = cos(ln x) and dv = dx) then gives Z Z Z − sin(ln x) cos(ln x) dx = x cos(ln x) − x dx = x sin(ln x) + sin(ln x) dx. x Therefore Z Z sin(ln x) dx = x sin(ln x) − x cos(ln x) − sin(ln x) dx, so Z 1 sin(ln x) dx = [x sin(ln x) − x cos(ln x)]. 2 (i) The integrand (x−2 ) is undefined (and unbounded) at x = 0, so the integral does not exist. 7. (c) F 0 (x) = 2x arctan(x4 ) − 2 arctan(4x2 ). 8. We have x Z F (x) − F (t) = f (s) ds t for a ≤ x < t ≤ b, and there are numbers m and M such that m ≤ f (s) ≤ M for s ∈ [a, b]. It follows from Theorem 6.3.2 that Z x Z x Z x m(t − x) = m ds ≤ f (s) ds ≤ m ds ≤ M (t − x), t t t and therefore |F (x) − F (t)| ≤ max{|m|, |M |}|x − t|, so F is Lipschitzian. 10. (d) If we rewrite n −3/2 √ 1 k= n r k , n then Exercise 6.2.10 gives lim n−3/2 n→∞ n √ X Z k= 0 k=1 1 1√ 2 x dx = . 3 2 (f) Now we have n 1 1 = , n2 + k 2 n 1 + (k/n)2 Z 1 n 1 π X n 1 = dx = arctan x lim = . 2 2 2 n→∞ n +k 4 0 0 1+x so k=1 22. To make the formulas easier to read, define Z x F (x) = f (t) dt. 0 Now we integrate by parts with u = [f (x)]2 and dv = f (x) dx (so v = F (x)) to obtain Z 1 1 Z 1 3 2 [f (x)] dx = [f (x)] F (x) − 2f (x)f 0 (x)F (x) dx 0 0 0 Z 1 = [f (1)]2 F (1) − 2f (x)f 0 (x)F (x) dx 0 Z 1 2 = [f (1)] F (1) − 2f (x)f 0 (x)[F (1) − F (x)] dx. 0 f0 Now we use the hypothesis 0 < to conclude that f ≥ 0 (because f is increasing and f (0) = 0). It also follows that F ≥ 0 by similar reasoning. Therefore f (x)f 0 (x)[F (1) − F (x)] ≤ f (x)[F (1) − F (x)] because f 0 ≤ 1. From our previous calculation, we conclude that Z 1 Z 1 Z 1 3 [f (x)] dx ≤ 2f (x)[F (1) − F (x)] dx = 2f (x)F (1) − 2f (x)F (x) dx 0 0 0 Z 1 2 2 = 2F (x)F (1) − [F (x)] = F (1) = 0 Section 6.5 3. (a) p > 1 because Z 2 1 f (x) dx 0 b b 1 −p+1 dx = (1 − p)x = p − 1 − (p − 1)b1−p , p 1 1 x and this expression has a limit as b → ∞ if p > 1. (b) p < 1 because Z b 1 dx = p − 1 − (p − 1)b1−p , p x 1 and this expression has a limit as b → 0 if p < 1. R1 7. (a) The Cauchy principal value for −1 (1/x) dx is Z −ε Z 1 1 1 lim dx + dx = lim (ln ε − ln ε) = 0, ε→0+ ε→0+ −1 x ε x but Z 0 1 1 dx = lim x ε→0+ Z ε 1 1 dx = lim − ln ε = ∞, x ε→0+ . 3 so 1/x is not improper Riemann integrable on [−1, 1]. R1 (b) The Cauchy principal value for −1 (1/x) dx is Z −ε Z 1 1 1 dx + dx = lim (− ln ε − ln ε) = ∞. lim ε→0+ ε→0+ ε |x| −1 |x| (c) the Cauchy principal value is Z 1−ε Z 2 4 4 lim dx + dx = lim (4 ln ε − 4 ln 2 − [4 ln 1 − 4 ln ε]) = 4 ln 2. + x−1 ε→0 ε→0+ −1 1+ε x − 1 9. (i) First, notice that limx→∞ | ln x/(1+x1/2 )| = 0, so | ln x/(1+x2 )| ≤ x−3/2 for x sufficiently large and therefore Z ∞ ln x dx 1 + x2 1 converges. Also, Z 1 Z 1 ln x − ln t −1 dx = dt 2 1 + x 1 + (1/t)2 t2 k 1/k (using the change of variables t = 1/x), so Z 1 Z 1/k ln x ln x dx = − dx. 2 1 + x2 k 1+x 1 Therefore Z 1 ln x dx 1 + x2 0 also converges and Z ∞ Z 1 ln x ln x dx = − dx. 2 2 1+x 1 0 1+x It follows that Z ∞ ln x dx = 0. 1 + x2 0 19. If 0 < x < y, then integration by parts with u = 1/t and dv = sin t dx (so v = 1 − cos t) gives y Z Z y y 1 − cos t 1 − cos t sin t dt = dt. + t t t2 x x x Note that (1 − cos x)/x → 0 as x → 0+ , and (1 − cos y)/y → 0 as y → ∞. In addition, |(1 − cos t)/t2 | ≤ 2/(1 + t2 ), so the integral Z y 1 − cos t dt t2 x converges as x → 0+ and y → ∞. Therefore Z ∞ sin x dx x 0 converges and Z ∞ Z ∞ sin x 1 − cos x dx = dx. x x2 0 0 4 To see that the integral is only conditionally convergent, we first observe that Z (n+1)π Z (n+1)π | sin x| 1 2 dx ≥ | sin x| dx = . x (n + 1)π nπ (n + 1)π nπ It follows that Z nπ n X 2 | sin x| dx ≥ . x kπ 0 k=1 This sum diverges to infinity as n → ∞, so the integral diverges. 22. Since f is continuous on [0, ∞) and f (x) ≤ 1/x2 , the integral is convergent. Now we use Exercise 19 to see that Z ∞ Z ∞ Z ∞ sin x 1 − cos x sin2 (x/2) dx = dx = 2 dx x x2 x2 0 0 0 from the trig identity 1 − cos x = sin2 (x/2). Now the change of variables u = x/2 yields Z ∞ Z ∞ Z sin2 (x/2) sin2 u 1 ∞ sin2 u dx = 2 du = dx. x2 4u2 2 0 u2 0 0 Section 6.7 R∞ Rb 1. TRUE. 0 sin x dx diverges but 0 sin x dx is between 0 and 2 for any positive b. 8. TRUE. (This is similar to the argument for Dirichlet’s function.) First notice that L(P, f ) = 0 for any partition P . Now let q be an integer greater than 1 and set n = q 2 − q + 1. If P is the partition {x0 , . . . , xn } with x1 = 1/q and xi = (q + i − 1)/q 2 for i = 2, . . . , n, then the maximum of f is 1 on [x0 , x1 ] and also on at most 2q intervals [xi−1 , xi ] with i ≥ 2. Therefore 1 1 3q + 1 U (P, f ) ≤ + 2q = 2 . q n q −q+1 Now we note that lim (3q + 1)/(q 2 − q + 1) = 0 to conclude that we can make U (P, f ) less q→∞ than any given ε > 0 by taking q large enough.