18.014 Problem Set 8 Solutions Sam Elder November 4, 2015 Problem 1. Let f : R → R be a differentiable function satisfying f 0 (x) = f (x) for all x ∈ R. Prove that f (x) = cex for some c ∈ R. Solution. Let g(x) = e−x f (x). By the product and chain rules, g 0 (x) = −e−x f (x) + e−x f 0 (x) = e−x (f 0 (x) − f (x)) = 0 since f 0 (x) = f (x). Therefore, g(x) = c is constant, and f (x) = cex , as desired. Problem 2. Prove that limx→0 (1 + x)1/x = e. Solution. First, we show the following lemma: Lemma. limx→0 log(1+x) x = 1. Proof. This is the definition of the derivative of log(x) applied at x = 1. Since log(x) = second Fundamental Theorem of Calculus, log0 (x) = x1 , so log0 (1) = 11 = 1, as desired. Rx 1 1 t dt, by the Now, we notice that e log(1+x) x x1 1 = elog(1+x) = (1 + x) x , since exp and log are inverse functions, and eab = (ea )b . Now since f (x) = ex is differentiable, it is continuous, so log(1+x) log(1+x) log(1 + x) lim = lim e x = elimx→0 x = e1 = e, x→0 x→0 x as desired. Problem 3. Compute the derivative (for x > 0) of the function x f (x) = xx . Solution. For x > 0, we have x f (x) = (elog x )x = ex x log x = ee x log x log x . For clarity, define g(x) = ex log x log x, so f (x) = eg(x) . By the chain rule, 1 0 x log x 1 x log x +e log x + x log x g (x) = e x x 1 = ex log x + log2 x + log x x 1 2 x + log x + log x =x x x 1 f 0 (x) = eg(x) g 0 (x) = xx xx + log x + log2 x . x 1 Problem 4. The goal of this problem is to construct the sine function. In other words, we want to find a twice-differentiable function f : R → R such that f (0) = 0, f 0 (0) = 1, and f 00 (x) = −f (x) for all x ∈ R. Part 4.1. Let g : [− 21 , 21 ] → R be defined by the integral Z x 1 √ g(x) = dt. 1 − t2 0 Prove that g is continuous and strictly increasing. 1 is defined and continuous, since 1 − t2 = 1 − |t|2 ≥ 1 − (1/2)2 = Solution. On [− 21 , 21 ], the function √1−t 2 1 3/4 > 0. Therefore, as an indefinite integral, g is continuous. Moreover, g 0 (x) = √1−x > 0, so g is strictly 2 increasing, as desired. Part 4.2. Let a = g( 12 ) > 0. Let h : [−a, a] → [− 21 , 12 ] be the inverse function of g. Show that h0 (x) = p 1 − h(x)2 for all x ∈ [−a, a]. Differentiate this again to obtain h00 (x) = −h(x) for all x ∈ [−a, a]. Solution. By Theorem 6.7 (derivatives of inverse functions), h0 (x) = 1 g 0 (h(x)) = 1 √ 1 2 1−h(x) = p 1 − h(x)2 , as desired. Taking another derivative, h00 (x) = 1 h0 (x) (1 − h(x)2 )−1/2 (−2h(x)h0 (x)) = −h(x) p = −h(x), 2 1 − h(x)2 as desired, because h0 (x) = p 1 − h(x)2 . Part 4.3. Check that h(0) = 0 and h0 (0) = 1. R0 1 Solution. We have g(0) = 0 √1−t dt = 0, so h(0) = 0 as the inverse function. Then by the previous part, 2 p h0 (0) = 1 − h(0)2 = 1. Part 4.4. Suppose now that b > 0 and j : [−b, b] → R satisfies j(0) = 0, j 0 (0) = 1, and j 00 (x) = −j(x) for all x ∈ [−b, b]. Prove that the function k : [−2b, 2b] → R defined by k(x) = 2j(x/2)j 0 (x/2) satisfies k(0) = 0, k 0 (0) = 1, and k 00 (x) = −k(x) for all x ∈ [−2b, 2b]. Solution. First, k(0) = 2j(0/2)j 0 (0/2) = 2 · 0 · 1 = 0 and by the product rule, k 0 (x) = j 0 (x/2)2 + j(x/2)j 00 (x/2) = j 0 (x/2)2 − j(x/2)2 =⇒ k 0 (0) = j 0 (0/2)2 − j(0/2)2 = 12 − 02 = 1, as desired. Finally, since j 00 (x/2) = −j(x/2), we can differentiate one more time: k 00 (x) = j 0 (x/2)j 00 (x/2) − j(x/2)j 0 (x/2) = −j 0 (x/2)j(x/2) − j(x/2)j 0 (x/2) = −2j(x/2)j 0 (x/2) = −k(x), as desired. Part 4.5. Explain how to define the sine function by repeatedly applying the previous part to the function h. 2 Solution. Since k(x) and j(x) both satisfy the criteria defining sin, as we proved in class, we must have k(x) = j(x) for all x ∈ [−b, b]. Therefore, we can define a sequence of functions hi : [−2i a, 2i a] → R by h0 = h and hi+1 (x) = 2hi (x/2)h0i (x/2). By induction, hi (x) = hj (x) wherever both are defined, and h00i (x) = −hi (x). Moreover, ∀x ∈ R, ∃i ∈ Z+ such that |x| ≤ 2i a. Therefore, we can define the function h∞ (x) : R → R by hi (x) where i = min{j : |x| ≤ 2j a}. We now verify the derivative properties, taking advantage of the fact that derivatives are local, so we only need to consider a small interval containing each point. First, we focus on 0: since 0 ∈ (− 21 , 12 ), h∞ (x) = h0 (x) = h(x) for x near 0, so h∞ (0) = h(0) = 0 and h0∞ (0) = h0 (0) = 1. Finally, for any x ∈ R, there is some interval (x − δ, x + δ) and i ∈ Z+ such that h∞ (y) = hi (y) or h∞ (y) = hi+1 (y) for y ∈ (x − δ, x + δ). (In fact, we guarantee this by taking δ = a/2.) But by the previous part, hi+1 (y) = hi (y) wherever hi (y) is defined, so in fact, h∞ (y) = hi+1 (y) for all y ∈ (x − δ, x + δ). Therefore, h00∞ (y) = h00i+1 (y) = −hi+1 (y) = −h∞ (y), as desired. 3