Math 539 Math 539 Homework #1 due Monday, September 26, 2005 at 10 AM Reality check problems. Not to write up; just ensure that you know how to do them. I. Bateman and Diamond, p. 16, #2.3 II. Bateman and Diamond, p. 32, #2.16 III. Bateman and Diamond, p. 35, #2.18 IV. Justify the chain of reasoning ∑ p ∞ 1 1 1 1 1 ∞ + + < < 2 4 k∑ 4 2 k∑ p2 =1 (2k + 1 ) =1 1 1 − 2k 2k + 2 1 = . 2 V. Let f and g be functions defined on the interval (0, ∞) that take values in (0, ∞) as well. Suppose that c(t) is an increasing continuous function such that lim c(t) = 0 and t→0+ lim c(t) = ∞. t→∞ Prove that f ( x) g( x) if and only if f (c(t)) g(c(t)). Homework problems. To write up and hand in. I. (a) Show that 2 x = 1 + O( x) for x in any bounded range. (b) What is wrong with the following “proof” that 2 N N 2 ? 2N = 2 + 2N − 2 = 2 + ∑ 2n n< N = O(1) + ∑ (1 + O(n)) n< N = O(1) + ∑ n< N 1+O ∑ n n< N N ( N − 1) = O(1) + N − 1 + O 2 = O( N 2 ). II. Let Q( x) denote the number of squarefree integers not exceeding x. We know that Q( x) ∼ π62 x, but the following elementary argument has merit as well. (a) Prove that Q( N ) ≥ N − ∑ p b N / p2 c for every positive integer N. Conclude that Q( N ) > N /2 for every N. (b) Prove that every integer greater than 1 can be written as the sum of two squarefree numbers. III. For this problem, all functions are defined on the interval (0, ∞) and take values in (0, ∞) as well (other than log). (a) Suppose that a( y) is an increasing concave function. Prove that h( x) k( x) implies a(h( x)) a(k( x)). Show via a counterexample that the concavity hypothesis is necessary. (b) Suppose that b( y) is an increasing convex function. Prove that b( f ( x)) b( g( x)) implies f ( x) g( x). Show via a counterexample that the convexity hypothesis is necessary. (c) For all real numbers A > 0 and 0 < b < 1 and ε > 0, show that log A x exp(logb x) xε uniformly for x ≥ 1. IV. (a) Prove the “Cauchy Condensation Test”: if { an } is a decreasing sequence of positive numbers, then the two series ∞ ∑ an ∞ and n=1 ∑ 2 n a 2n n=1 either both converge or both diverge. (b) Let logk n denote the k-fold iterated logarmithm log log · · · log n. Prove that for every k ≥ 1 and every ε > 0, ∑ n : logk 1 <∞ n(log n)(log log n) · · · (logk−1 n)(logk n)1+ε n>0 while ∑ n : logk 1 = ∞. n(log n)(log log n) · · · (logk−1 n)(logk n)1−ε n>0 V. Let k > 1 be a fixed squarefree integer. Are integers that are divisible by k more or less likely to be squarefree than those that are not divisible by k? Turn this into a rigorous question and resolve it. VI. (a) In one of my papers, I have as a lemma the inequality 2ω(n) ≤ 2 log 16 4− log 11 log 2 n log 11 for all n ∈ N. Prove this lemma (can you improve the constant?) and generalize. (b) Prove that τ (n) ε nε for all ε > 0. Why is it impossible for τ (n) nε uniformly for all ε > 0? (c) Find the smallest constant C such that τ (n) ≤ Cn1/4 for all n ∈ N. VII. Let ν p (n) denote the exponent of p in the prime factorization of n, so that there exists an integer m with p - m and n = pν p (n) m. Compute the average value of ν p (n); of ν p (n)2 . Is the second answer the square of the first? Should it be? VIII. (a) What is wrong with the following beginning of an attempt to find the average order of ω2 ? 2 x 2 ∑ ω(n) = ∑ ∑ 1 = ∑ ∑ ∑ 1 = ∑ ∑ ∑ 1 = ∑ ∑ p1 p2 . p1 ≤ x p2 ≤ x n≤ x p |n p |n p1 ≤ x p2 ≤ x n≤ x n≤ x n≤ x p|n 1 2 p1 p2 |n (b) Correct this beginning and carry through the evaluation of the average order of ω2 . D EFINITION . The (natural) density δ ( S) of a set S of positive integers is the average value of its indicator function, that is, 1 1 δ ( S) = lim #{n ≤ x : n ∈ S} = lim ∑ 1, x→∞ x x→∞ x n≤ x n∈ S if the limit exists (if it doesn’t exist, we say the set does not have a density). For example, the set of odd integers has density 21 , and the set of primes has density 0. IX. (a) Pick a number between 1 and 9 (inclusive). (b) Show that the set of positive integers whose first digit is the number you chose in part (a) does not have a density. X. Let f be an arithmetic function that takes only finitely many values v1 , . . . , vm . Suppose that for each 1 ≤ j ≤ m, the density δ j of the set {n ∈ N : f (n) = v j } exists. (a) Prove that ∑m j=1 δ j = 1. (b) Prove that the average value 1 α Aα = lim ∑ f (n) x→∞ x n≤ x exists for any α ∈ R (with the stipulation that none of the v j equals 0 if α < 0). (c) Suppose the values are indexed so that δ1 ≥ δ2 ≥ · · · ≥ δm . Show that if the average values A1 , A2 , . . . , Am are known, then all of the densities δ1 , . . . , δm can be recovered.