Math 414 Professor Lieberman May 12, 2003 SOLUTIONS TO FINAL EXAM 1. First we check the case n = 1: F1 = 1 and 1 √ 5 " √ !n 1+ 5 − 2 For n = 2, we have √ !2 1 1+ 5 √ − 2 5 " √ # √ !n # √ 1− 5 1 1+ 5 1− 5 = 1. − =√ 2 2 2 5 " ! !# √ !2 √ √ 1− 5 1 1+ 5 1− 5 =√ +1 − +1 2 2 2 5 " √ √ # 1 1+ 5 1− 5 =√ − =1 2 2 5 = F2 . Now we suppose that 1 Fm = √ 5 " √ !m 1+ 5 − 2 √ !m # 1− 5 2 for some natural number m ≥ 2. Then Fm+1 = Fm + Fm−1 " √ !m−1 √ !m−1 √ !m √ !m # 1+ 5 1− 5 1 1+ 5 1− 5 1 − − +√ =√ 2 2 2 2 5 5 ! ! √ !m−1 √ √ !m−1 √ 1+ 5 1+ 5 1− 5 1− 5 1 +1 − +1 =√ 2 2 2 2 5 √ !m−1 √ !2 √ !m−1 √ !2 1+ 5 1− 5 1− 5 1 1+ 5 − =√ 2 2 2 2 5 √ !m+1 √ !m+1 1 1+ 5 1− 5 . − =√ 2 2 5 2. There is no maximum or minimum. The supremum is 1 and the infimum is −1. 1 2 3. FALSE. Take f (x) = ( 1 1 x if x = 0, if x > 0, and define xn = 1/n. Then lim f (xn ) = ∞, but lim f (x) = 0. n→∞ x→∞ 4. Take xn = nπ and tn = xn + 1/n. Then |xn − tn | → 0 as n → ∞ but 1 |f (xn ) − f (tn )| = tn sin . n Because lim sin h/h = 1, it follows that there is a natural number N such that h→0 sin(1/n) ≥ 1/(2n) if n ≥ N . Hence, if n ≥ N , then π 1 1 ) ≥ . n 2n 2 Therefore f is not uniformly continuous because a theorem in the book says that, if f is uniformly continuous on D, then for any two sequences (xn ) and (tn ) of points in D, with lim |xn − tn | = 0, we have lim f (xn ) − f (tn ) = 0. |f (xn ) − f (tn )| ≥ (nπ + n→∞ n→∞ 5. We write f (x) = x/(x + 1), g(x) = ln(x + 1) and h(x) = x. Then f (0) = g(0) = h(0) = 0, so we have equality if x = 0. In addition, f 0 (x) = 1 , (x + 1)2 g 0 (x) = 1 , x+1 h0 (x) = 1. For x > 0, we have f 0 (x) < g 0 (x) < h0 (x), so f − g and g − h are strictly decreasing functions on (0, ∞). It follows that x < ln(x + 1) < x x+1 for x > 0, so the inequality is true and it’s an equality only if x = 0. 6. We first use the definition to see that an is the Riemann sum for the integral corresponding to the partition Pn = {x0 , . . . , xn } with xj = j/n. Therefore L(Pn , f ) ≤ an ≤ U (Pn , f ). Because the mesh of Pn (the mesh is just the length of the longest subinterval in the partition) is 1/n, it follows that Z 1 lim L(Pn , f ) = lim U (Pn , f ) = f (x) dx, n→∞ n→∞ 0 and the squeeze theorem says that Z 1 f (x) dx = lim an . 0 n→∞ 3 7. Method 1 (The root test) s k 1 1 1 = ≤ k (ln k) ln k ln 3 for k ≥ 3 and 1/ ln 3 < 1, so the series CONVERGES. Method 2 (The comparison test) For k ≥ 3, 1 1 ≤ , k (ln k) (ln 3)k and ∞ X k=2 1 (ln 3)k is a convergent geometric series, so the original series CONVERGES. 8. (a) When n = 1, we have n X ak bk = a1 b1 = s1 b1 = s1 b2 − s1 (b2 − b1 ) k=1 = s1 b 2 − 1 X sk (bk+1 − bk ). k=1 If the equation is true for n = m and m ∈ N, then m+1 X ak b k = k=1 m X ak bk + am+1 bm+1 k=1 = sm bm+1 − m X sk (bk+1 − bk ) + am+1 bm+1 k=1 m X = sm+1 bm+1 − sk (bk+1 − bk ) k=1 = sm+2 bm+1 − sm+2 (bm+2 − bm+1 ) − m X sk (bk+1 − bk ) k=1 = sm+1 bm+2 − m+1 X sk (bk+1 − bk ). k=1 (b) From (a), we have n X ak bk = sn bn+1 − k=1 n X sk (bk+1 − bk ) ≤ M bn+1 − m X M (bk+1 − bk ) k=1 k=1 because bn+1 ≥ 0 and (bk+1 − bk ) ≤ 0. Next, M bn+1 − m X k=1 M (bk+1 − bk ) = M bn+1 − n X k=1 ! (bk+1 − bk ) = M b1 , 4 so n X ak b k ≤ M b 1 . k=1 The other inequality is proved similarly. 9. (a) It is not enough to show that the partial sums are bounded! Instead, start with a given ε > 0 and write L for the bound on (sn ), so |sn | ≤ L for all n ∈ N. Then there is a natural number N such that bN ≤ ε/L. If m ≥ n ≥ N , then m X ak bk ≤ bm L ≤ bN L < ε. k=n P Therefore the series ak bk converges. (b) Set b = lim bn . Suppose first that (bn ) decreases to b, and set cn = bn − b. Then n→∞ P (cn ) is a decreasing sequence which converges to zero, and the partial sums of ak P are bounded, so the series ak ck converges by part (a). Next, we have X X X ak b k = ak c k + ak b, and we have just shown that the first sum converges. The second sum converges because it’s a constant b times a convergent series. When (bn ) increases to b, a similar argument applies (with dn = b − bn defining the sequence which decreases to zero). 10. In both parts, we will use the interval (a) Use 1 2 f (x) = x 1 [0, 1]. 2 if x = 0, if 0 < x < 1, if x = 1. 1 f (x) = x 0 if x = 0, if 0 < x < 1, if x = 1. (b) Use 11. Here, we need a δ−ε type proof. Given ε > 0, there is a δ > 0 such that |f 0 (x)−A| < ε if 0 < |x − c| < δ and a < x < b. If |h| < min{δ, c − a, b − c}, then there is a number x between c and c + h such that f (c + h) − f (c) = f 0 (x), h so f (c + h) − f (c) < ε. − A h Since ε > 0 is arbitrary, it follows that f (c + h) − f (c) lim = A, h→0 h which means that f 0 (c) exists and equals A. 5 √ 12. The substitution u = x (so x = u2 and dx = 2u du) gives √ Z 1 Z 1 exp( x) √ dx = eu 2 du = 2(e1 − e0 ) = 2e − 2. x 0 0 To be technically correct, you need to recognize that the original integral is an im√ √ proper integral (because the integrand exp( x )/ x ) is unbounded near x = 0) and take some appropriate limits. Nobody noticed this fact, so no points were deducted for missing it.