Search Knowledge

© 2026 LIBREUNI PROJECT

Introduction to Calculus/Limits and derivatives
Page 4 of 12

Problem Set: Limits and Derivatives

Challenging problems, synthesis proofs, and computational exercises spanning limits, differentiability, and the Mean Value Theorem.

Chapter problem set

Deep and Challenge Problems

Estimated time: 120 min

Target Outcomes

  • Prove non-linear limit existence and denominator boundedness from epsilon-delta definitions
  • Construct counterexamples demonstrating the necessity of hypotheses for differentiability and the Mean Value Theorem
  • Derive truncation error bounds for finite-difference derivative approximations
  • Apply Cauchy Mean Value Theorem to establish functional inequalities and justify L'Hopital rules
Levels: recall · application · transfer · synthesis★ = Challenge problem

Practice problems bridging the first three lessons. These problems demand proof synthesis, hypothesis auditing, and computational tool use across multiple theorems.

★

Problem 1.P.1: Algebra of Limits: Bounding the Denominator

Proofsynthesis

Complete the proof of the Algebra of Limits (Theorem 1.1.2): assuming lim⁡x→cg(x)=M≠0\lim_{x \to c} g(x) = M \ne 0 prove there is a δ0>0\delta_0 > 0 such that ∣g(x)∣>∣M∣/2|g(x)| > |M|/2 whenever 0<∣x−c∣<δ00 < |x - c| < \delta_0 then use this to derive the quotient rule from the product and reciprocal rules.

HintExpand

First pin gg away from 00 with ϵ=∣M∣/2\epsilon = |M|/2 in the limit definition; the preliminary bound then controls the quotient ∣1/g(x)−1/M∣|1/g(x) - 1/M|.

Worked solutionExpand

Step 1 (pin away from zero). Apply Definition 1.1.1 with ϵ=∣M∣/2\epsilon = |M|/2 there is δ0>0\delta_0 > 0 such that 0<∣x−c∣<δ00 < |x-c| < \delta_0 implies ∣g(x)−M∣<∣M∣/2|g(x) - M| < |M|/2 Then by the reverse triangle inequality,

∣g(x)∣=∣M+(g(x)−M)∣≥∣M∣−∣g(x)−M∣>∣M∣−∣M∣2=∣M∣2>0.|g(x)| = |M + (g(x) - M)| \ge |M| - |g(x) - M| > |M| - \tfrac{|M|}{2} = \tfrac{|M|}{2} > 0.

Step 2 (reciprocal). For 0<∣x−c∣<δ00 < |x - c| < \delta_0

∣1g(x)−1M∣=∣M−g(x)∣∣g(x)∣ ∣M∣<∣g(x)−M∣(∣M∣/2) ∣M∣=2∣M∣2 ∣g(x)−M∣.\left|\frac{1}{g(x)} - \frac{1}{M}\right| = \frac{|M - g(x)|}{|g(x)|\,|M|} < \frac{|g(x) - M|}{(|M|/2)\,|M|} = \frac{2}{|M|^2}\,|g(x) - M|.

Given ϵ>0\epsilon > 0 choose δ1\delta_1 so that ∣g(x)−M∣<ϵ ∣M∣2/2|g(x) - M| < \epsilon\,|M|^2/2 for 0<∣x−c∣<δ10 < |x - c| < \delta_1 and set δ=min⁡(δ0,δ1)\delta = \min(\delta_0, \delta_1) Then ∣1/g(x)−1/M∣<ϵ|1/g(x) - 1/M| < \epsilon throughout, so lim⁡x→c1/g(x)=1/M\lim_{x \to c} 1/g(x) = 1/M

Step 3 (quotient). The product rule of Theorem 1.1.2 applied to ff and 1/g1/g gives lim⁡f/g=L/M\lim f/g = L/M ∎

Verification criteria. A complete proof requires all three steps: the preliminary pin ∣g(x)∣>∣M∣/2|g(x)| > |M|/2 (with ϵ=∣M∣/2\epsilon = |M|/2 explicit), the reciprocal estimate with the 2/∣M∣22/|M|^2 constant appearing, and the final composition with the product rule.

Problem 1.P.2: Numerical Epsilon-Delta Neighborhood Check

Tool / Simulationapplication

Complete the verifier below so it checks the ϵ\epsilon-δ\delta condition numerically. Test inputs: f(x)=3x−1f(x) = 3x - 1 c=2c = 2 L=5L = 5 ϵ=0.3\epsilon = 0.3 δ=0.1\delta = 0.1 (must pass), and the same with δ=0.15\delta = 0.15 (must fail — the sample points falsify it).

Code Exercise

1 token to supply

Complete the Epsilon-Delta check

def verify_limit(f, c, L, epsilon, delta, samples=1000, seed=0):
  # Check random points in the delta-neighborhood of c (excluding c itself)
  import numpy as np
  rng = np.random.default_rng(seed)
  x_vals = c + rng.uniform(-delta, delta, samples)
  x_vals = x_vals[x_vals != c]

  # The condition is: 0 < |x - c| < delta implies |f(x) - L| < epsilon
  satisfied = np.all( < epsilon)
  return satisfied
HintExpand

Inspect the vectorized evaluation in NumPy: the test asserts that all points in the punctured neighborhood satisfy the epsilon inequality.

Worked solutionExpand

The blank must compute the elementwise absolute deviation np.abs(f(x_vals) - L). Passing δ=0.1\delta = 0.1 verifies that for all 10001000 points ∣f(x)−5∣=3∣x−2∣<3(0.1)=0.3|f(x) - 5| = 3|x - 2| < 3(0.1) = 0.3 For δ=0.15\delta = 0.15 points near the boundaries x=1.86x = 1.86 or x=2.14x = 2.14 yield deviations ∣f(x)−5∣≈0.42>0.3|f(x) - 5| \approx 0.42 > 0.3 causing np.all to return False.

★

Problem 1.P.3: Quantifier Perturbation: The 'Verconges' Definition

Perturbed Definitiontransfer

Suppose a mathematician proposes an altered definition of limit: a function ff is said to verconge to LL at cc if: There exists δ>0\delta > 0 such that for every ϵ>0\epsilon > 0, for all xx 0<∣x−c∣<δ  ⟹  ∣f(x)−L∣<ϵ.0 < |x - c| < \delta \implies |f(x) - L| < \epsilon.

Characterize all functions that verconge to LL at cc What does this reveal about why the order of quantifiers in Definition 1.1.1 matters?

HintExpand

If ∣f(x)−L∣<ϵ|f(x) - L| < \epsilon for every ϵ>0\epsilon > 0 on a fixed interval, what is the only non-negative number that can satisfy that?

Worked solutionExpand

Characterization: A function ff verconges to LL at cc if and only if f(x)=Lf(x) = L identically on the entire punctured neighborhood (c−δ,c+δ)∖{c}(c - \delta, c + \delta) \setminus \{c\}.

Proof: Suppose ff verconges to LL at cc so there exists a fixed δ>0\delta > 0 such that for every x∈(c−δ,c+δ)∖{c}x \in (c - \delta, c + \delta) \setminus \{c\} the distance ∣f(x)−L∣<ϵ|f(x) - L| < \epsilon for every ϵ>0\epsilon > 0 The only non-negative real number that is strictly less than every positive real number ϵ\epsilon is 00 Therefore, for every xx in the punctured window, ∣f(x)−L∣=0|f(x) - L| = 0 which implies f(x)=Lf(x) = L Conversely, if f(x)=Lf(x) = L on (c−δ,c+δ)∖{c}(c - \delta, c + \delta) \setminus \{c\} then ∣f(x)−L∣=0<ϵ|f(x) - L| = 0 < \epsilon holds for all ϵ>0\epsilon > 0

Significance: In Definition 1.1.1, δ\delta is chosen after ϵ\epsilon ∀ϵ>0,∃δ>0\forall \epsilon > 0, \exists \delta > 0 This allows δ\delta to shrink as the tolerance ϵ\epsilon tightens, capturing local sensitivity and convergence. Swapping the quantifiers forces one static window δ\delta to satisfy all tolerances at once, which destroys the entire theory of continuous change and restricts the concept to trivial locally constant functions.

Problem 1.P.4: Epsilon-Delta Derivative Existence and Lipschitz Bound

Proofsynthesis

Let f(x)=x3f(x) = x^3

  1. Using Definition 1.1.1, give a rigorous ϵ\epsilon-δ\delta proof that lim⁡h→0(x+h)3−x3h=3x2\lim_{h \to 0} \frac{(x+h)^3 - x^3}{h} = 3x^2 at an arbitrary fixed x∈Rx \in \mathbb{R}
  2. Using Theorem 1.3.2, show that for all x1,x2∈[1,3]x_1, x_2 \in [1, 3] ∣x13−x23∣≤27∣x1−x2∣|x_1^3 - x_2^3| \le 27 |x_1 - x_2|
HintExpand

Expand (x+h)3(x+h)^3, factor out hh, and freeze the variable factor in the difference quotient with a preliminary bound ∣h∣≤1|h| \le 1.

Worked solutionExpand

Part 1: ϵ\epsilon-δ\delta proof of the derivative. For h≠0h \ne 0 (x+h)3−x3h=3x2h+3xh2+h3h=3x2+3xh+h2.\frac{(x+h)^3 - x^3}{h} = \frac{3x^2 h + 3xh^2 + h^3}{h} = 3x^2 + 3xh + h^2. The difference from the candidate limit 3x23x^2 is: ∣(3x2+3xh+h2)−3x2∣=∣3xh+h2∣=∣h∣ ∣3x+h∣.\left|\left(3x^2 + 3xh + h^2\right) - 3x^2\right| = |3xh + h^2| = |h|\,|3x + h|. Impose the preliminary bound ∣h∣≤1|h| \le 1 Then by the triangle inequality, ∣3x+h∣≤3∣x∣+∣h∣≤3∣x∣+1|3x + h| \le 3|x| + |h| \le 3|x| + 1 Given ϵ>0\epsilon > 0 choose: δ=min⁡(1,ϵ3∣x∣+1).\delta = \min\left(1, \frac{\epsilon}{3|x| + 1}\right). Whenever 0<∣h∣<δ0 < |h| < \delta we have ∣h∣≤1|h| \le 1 and: ∣(x+h)3−x3h−3x2∣≤∣h∣(3∣x∣+1)<ϵ3∣x∣+1(3∣x∣+1)=ϵ.\left|\frac{(x+h)^3 - x^3}{h} - 3x^2\right| \le |h|(3|x| + 1) < \frac{\epsilon}{3|x| + 1} (3|x| + 1) = \epsilon. This establishes that f′(x)=3x2f'(x) = 3x^2 directly from Definition 1.1.1.

Part 2: Lipschitz bound via MVT. Let x1,x2∈[1,3]x_1, x_2 \in [1, 3] with x1<x2x_1 < x_2 By Theorem 1.3.2 applied to f(t)=t3f(t) = t^3 on [x1,x2][x_1, x_2] there exists c∈(x1,x2)c \in (x_1, x_2) such that: x23−x13=f′(c)(x2−x1)=3c2(x2−x1).x_2^3 - x_1^3 = f'(c)(x_2 - x_1) = 3c^2(x_2 - x_1). Since c∈(x1,x2)⊂[1,3]c \in (x_1, x_2) \subset [1, 3] c≤3c \le 3 so 3c2≤3(32)=273c^2 \le 3(3^2) = 27 Taking absolute values: ∣x23−x13∣=3c2∣x2−x1∣≤27∣x2−x1∣.|x_2^3 - x_1^3| = 3c^2 |x_2 - x_1| \le 27 |x_2 - x_1|.

★

Problem 1.P.5: Differentiable Function with Discontinuous Derivative

Example / Impossibletransfer

Define f:R→Rf: \mathbb{R} \to \mathbb{R} by f(x)=x2sin⁡(1/x)f(x) = x^2 \sin(1/x) for x≠0x \ne 0 and f(0)=0f(0) = 0 Prove that ff is differentiable everywhere on R\mathbb{R} but the derivative function f′f' is discontinuous at x=0x = 0

HintExpand

Use the limit definition directly at x=0x = 0. For x≠0x \ne 0, use standard differentiation rules and inspect the limit of f′(x)f'(x) as x→0x \to 0.

Worked solutionExpand

Differentiability at 00 By Definition 1.2.1: f′(0)=lim⁡h→0f(h)−f(0)h=lim⁡h→0h2sin⁡(1/h)−0h=lim⁡h→0hsin⁡(1/h).f'(0) = \lim_{h \to 0} \frac{f(h) - f(0)}{h} = \lim_{h \to 0} \frac{h^2 \sin(1/h) - 0}{h} = \lim_{h \to 0} h \sin(1/h). Since ∣sin⁡(1/h)∣≤1|\sin(1/h)| \le 1 we have ∣hsin⁡(1/h)∣≤∣h∣→0|h \sin(1/h)| \le |h| \to 0 By the Squeeze Theorem, f′(0)=0f'(0) = 0 Thus ff is differentiable at 00

Derivative for x≠0x \ne 0 By Theorem 1.2.3 and Theorem 1.2.4: f′(x)=2xsin⁡(1x)+x2cos⁡(1x)⋅(−1x2)=2xsin⁡(1x)−cos⁡(1x).f'(x) = 2x \sin\left(\frac{1}{x}\right) + x^2 \cos\left(\frac{1}{x}\right) \cdot \left(-\frac{1}{x^2}\right) = 2x \sin\left(\frac{1}{x}\right) - \cos\left(\frac{1}{x}\right).

Discontinuity of f′f' at 00 As x→0x \to 0 2xsin⁡(1/x)→02x \sin(1/x) \to 0 but −cos⁡(1/x)-\cos(1/x) oscillates between −1-1 and 11 in every neighborhood of 00 Therefore, lim⁡x→0f′(x)\lim_{x \to 0} f'(x) does not exist. Since lim⁡x→0f′(x)≠f′(0)=0\lim_{x \to 0} f'(x) \ne f'(0) = 0 f′f' is discontinuous at 00

Conclusion: Differentiability guarantees ff is continuous (Theorem 1.2.2), but does not guarantee f′f' is continuous.

★

Problem 1.P.6: The False Proof and Flaw in the Chain Rule

Proofsynthesis

A proposed derivation offers the following proof of Theorem 1.2.4: f(g(x+h))−f(g(x))h=f(g(x+h))−f(g(x))g(x+h)−g(x)⋅g(x+h)−g(x)h.\frac{f(g(x+h)) - f(g(x))}{h} = \frac{f(g(x+h)) - f(g(x))}{g(x+h) - g(x)} \cdot \frac{g(x+h) - g(x)}{h}. Taking h→0h \to 0 the first term approaches f′(g(x))f'(g(x)) and the second approaches g′(x)g'(x) giving f′(g(x))g′(x)f'(g(x))g'(x) Diagnose the logical flaw in this argument. How does Carathéodory’s definition or an auxiliary function resolve it?

HintExpand

What happens if g(x+h)=g(x)g(x+h) = g(x) for values of hh arbitrarily close to 00? Can we divide by zero in the difference quotient?

Worked solutionExpand

The Flaw: The proof divides by g(x+h)−g(x)g(x+h) - g(x) However, gg may be constant in a neighborhood of xx or worse, oscillate infinitely often such that g(x+h)=g(x)g(x+h) = g(x) for points hn→0h_n \to 0 (for instance, g(t)=x+t2sin⁡(1/t)g(t) = x + t^2 \sin(1/t) For such hh the fraction f(g(x+h))−f(g(x))g(x+h)−g(x)\frac{f(g(x+h)) - f(g(x))}{g(x+h) - g(x)} is undefined 0/00/0 One cannot divide by a quantity that may vanish along the limit sequence.

The Resolution (Carathéodory’s approach): Define the auxiliary difference-quotient slope function Φ\Phi on the domain of ff Φ(y)={f(y)−f(u)y−uif y≠uf′(u)if y=u\Phi(y) = \begin{cases} \frac{f(y) - f(u)}{y - u} & \text{if } y \ne u \\ f'(u) & \text{if } y = u \end{cases} where u=g(x)u = g(x) Because ff is differentiable at uu lim⁡y→uΦ(y)=f′(u)=Φ(u)\lim_{y \to u} \Phi(y) = f'(u) = \Phi(u) so Φ\Phi is continuous at uu Notice that f(y)−f(u)=Φ(y)(y−u)f(y) - f(u) = \Phi(y)(y - u) holds unconditionally for all yy even when y=uy = u Now substitute y=g(x+h)y = g(x+h) f(g(x+h))−f(g(x))h=Φ(g(x+h))⋅g(x+h)−g(x)h.\frac{f(g(x+h)) - f(g(x))}{h} = \Phi(g(x+h)) \cdot \frac{g(x+h) - g(x)}{h}. Since gg is differentiable at xx it is continuous at xx so lim⁡h→0g(x+h)=g(x)=u\lim_{h \to 0} g(x+h) = g(x) = u Because Φ\Phi is continuous at uu lim⁡h→0Φ(g(x+h))=Φ(u)=f′(g(x))\lim_{h \to 0} \Phi(g(x+h)) = \Phi(u) = f'(g(x)) Multiplying limits: (f∘g)′(x)=f′(g(x))⋅g′(x).(f \circ g)'(x) = f'(g(x)) \cdot g'(x). This proof never divides by g(x+h)−g(x)g(x+h) - g(x) and works unconditionally for all differentiable functions.

Problem 1.P.7: Central Difference Truncation Error Analysis

Tool / Simulationapplication

Suppose ff is three times differentiable on an open interval containing xx with f′′′f''' continuous at xx Complete the numerical central difference formula below and prove using Theorem 1.3.5 that its truncation error satisfies lim⁡h→0f(x+h)−f(x−h)2h−f′(x)h2=f′′′(x)6,\lim_{h \to 0} \frac{\frac{f(x+h) - f(x-h)}{2h} - f'(x)}{h^2} = \frac{f'''(x)}{6}, establishing that the central difference error is O(h2)\mathcal{O}(h^2) rather than O(h)\mathcal{O}(h)

Code Exercise

1 token to supply

Complete the Numerical Derivative (Central Difference)

def central_difference(f, x, h=1e-5):
  # Central difference formula provides O(h^2) accuracy compared to forward O(h)
  return ( - f(x - h)) / (2 * h)
HintExpand

Rewrite the ratio as f(x+h)−f(x−h)−2hf′(x)2h3\frac{f(x+h) - f(x-h) - 2hf'(x)}{2h^3} and apply L'Hôpital's Rule with respect to hh three times.

Worked solutionExpand

Derivation of Truncation Error via L’Hôpital’s Rule: Combine the numerator over the common denominator 2h32h^3 E(h)=f(x+h)−f(x−h)2h−f′(x)h2=f(x+h)−f(x−h)−2hf′(x)2h3.E(h) = \frac{\frac{f(x+h) - f(x-h)}{2h} - f'(x)}{h^2} = \frac{f(x+h) - f(x-h) - 2hf'(x)}{2h^3}.

  1. First application of Theorem 1.3.5: As h→0h \to 0 both numerator and denominator approach 00 Differentiating with respect to hh (using Theorem 1.2.4 on f(x−h)f(x-h) lim⁡h→0f(x+h)−f(x−h)−2hf′(x)2h3=lim⁡h→0f′(x+h)+f′(x−h)−2f′(x)6h2.\lim_{h \to 0} \frac{f(x+h) - f(x-h) - 2hf'(x)}{2h^3} = \lim_{h \to 0} \frac{f'(x+h) + f'(x-h) - 2f'(x)}{6h^2}.
  2. Second application of Theorem 1.3.5: At h=0h = 0 the numerator is f′(x)+f′(x)−2f′(x)=0f'(x) + f'(x) - 2f'(x) = 0 and the denominator is 00 Differentiating with respect to hh again: lim⁡h→0f′′(x+h)−f′′(x−h)12h.\lim_{h \to 0} \frac{f''(x+h) - f''(x-h)}{12h}.
  3. Evaluate via Definition 1.2.1 (or a third L’Hôpital step): lim⁡h→016⋅f′′(x+h)−f′′(x−h)2h=16f′′′(x).\lim_{h \to 0} \frac{1}{6} \cdot \frac{f''(x+h) - f''(x-h)}{2h} = \frac{1}{6} f'''(x). Thus f(x+h)−f(x−h)2h−f′(x)=f′′′(x)6h2+o(h2)=O(h2)\frac{f(x+h) - f(x-h)}{2h} - f'(x) = \frac{f'''(x)}{6}h^2 + o(h^2) = \mathcal{O}(h^2) whereas the one-sided forward difference f(x+h)−f(x)h−f′(x)\frac{f(x+h) - f(x)}{h} - f'(x) has first-order error f′′(x)2h+o(h)=O(h)\frac{f''(x)}{2}h + o(h) = \mathcal{O}(h)

Problem 1.P.8: Carathéodory Equivalence Formulation

Formalisationtransfer

Carathéodory’s Theorem states: ff is differentiable at cc if and only if there exists a function ϕ\phi continuous at cc such that f(x)−f(c)=ϕ(x)(x−c)f(x) - f(c) = \phi(x)(x - c) for all xx Prove that if such a ϕ\phi exists, then f′(c)=ϕ(c)f'(c) = \phi(c)

HintExpand

Take the limit as x→cx \to c of (f(x)−f(c))/(x−c)(f(x) - f(c))/(x - c) and use the continuity of ϕ\phi.

Worked solutionExpand

Proof: Suppose there exists a function ϕ\phi continuous at cc such that: f(x)−f(c)=ϕ(x)(x−c)for all x.f(x) - f(c) = \phi(x)(x - c) \quad \text{for all } x. For any x≠cx \ne c we can divide by x−cx - c f(x)−f(c)x−c=ϕ(x).\frac{f(x) - f(c)}{x - c} = \phi(x). Now evaluate the limit as x→cx \to c using Definition 1.2.1: f′(c)=lim⁡x→cf(x)−f(c)x−c=lim⁡x→cϕ(x).f'(c) = \lim_{x \to c} \frac{f(x) - f(c)}{x - c} = \lim_{x \to c} \phi(x). Because ϕ\phi is assumed continuous at cc Definition 1.1.4 guarantees that lim⁡x→cϕ(x)=ϕ(c)\lim_{x \to c} \phi(x) = \phi(c) Therefore, the limit exists and: f′(c)=ϕ(c).f'(c) = \phi(c). Conversely, if ff is differentiable at cc defining ϕ(x)=f(x)−f(c)x−c\phi(x) = \frac{f(x)-f(c)}{x-c} for x≠cx \ne c and ϕ(c)=f′(c)\phi(c) = f'(c) yields a function continuous at cc that satisfies the equation for all xx ∎

Problem 1.P.9: Second-Order Remainder Bound via Cauchy Mean Value Theorem

Constructionsynthesis

Let ff be twice differentiable on [a,b][a, b] and let L(x)=f(a)+f′(a)(x−a)L(x) = f(a) + f'(a)(x - a) be the linear approximation at aa Prove that for any x∈(a,b]x \in (a, b] there exists c∈(a,x)c \in (a, x) such that: f(x)−L(x)=f′′(c)2(x−a)2.f(x) - L(x) = \frac{f''(c)}{2}(x - a)^2.

HintExpand

Apply Cauchy's Mean Value Theorem to F(t)=f(x)−f(t)−f′(t)(x−t)F(t) = f(x) - f(t) - f'(t)(x - t) and G(t)=(x−t)2G(t) = (x - t)^2 on [a,x][a, x].

Worked solutionExpand

Define two auxiliary functions on [a,x][a, x] F(t)=f(x)−f(t)−f′(t)(x−t),F(t) = f(x) - f(t) - f'(t)(x - t), G(t)=(x−t)2.G(t) = (x - t)^2. Evaluate at t=xt = x F(x)=f(x)−f(x)−f′(x)(0)=0F(x) = f(x) - f(x) - f'(x)(0) = 0 and G(x)=(x−x)2=0G(x) = (x - x)^2 = 0 Evaluate at t=at = a F(a)=f(x)−f(a)−f′(a)(x−a)=f(x)−L(x)F(a) = f(x) - f(a) - f'(a)(x - a) = f(x) - L(x) and G(a)=(x−a)2G(a) = (x - a)^2 Now differentiate with respect to tt F′(t)=−f′(t)−[f′′(t)(x−t)−f′(t)]=−f′′(t)(x−t),F'(t) = -f'(t) - [f''(t)(x - t) - f'(t)] = -f''(t)(x - t), G′(t)=−2(x−t).G'(t) = -2(x - t). Note that for t∈(a,x)t \in (a, x) G′(t)=−2(x−t)≠0G'(t) = -2(x - t) \ne 0 By Cauchy’s Mean Value Theorem (Theorem 1.3.4), there exists c∈(a,x)c \in (a, x) such that: F(x)−F(a)G(x)−G(a)=F′(c)G′(c).\frac{F(x) - F(a)}{G(x) - G(a)} = \frac{F'(c)}{G'(c)}. Substituting the values: 0−(f(x)−L(x))0−(x−a)2=−f′′(c)(x−c)−2(x−c)=f′′(c)2.\frac{0 - (f(x) - L(x))}{0 - (x - a)^2} = \frac{-f''(c)(x - c)}{-2(x - c)} = \frac{f''(c)}{2}. Multiplying by (x−a)2(x - a)^2 f(x)−L(x)=f′′(c)2(x−a)2.f(x) - L(x) = \frac{f''(c)}{2}(x - a)^2. This establishes Lagrange’s form of the Taylor remainder for degree 11

★

Problem 1.P.10: Circular Reasoning in Differentiation of Sine

Negationtransfer

Explain why using Theorem 1.3.5 to evaluate lim⁡x→0sin⁡xx=lim⁡x→0cos⁡x1=1\lim_{x \to 0} \frac{\sin x}{x} = \lim_{x \to 0} \frac{\cos x}{1} = 1 is circular reasoning when defining the derivative of sine.

HintExpand

How is the derivative of sin⁡x\sin x proved in the first place? Inspect the difference quotient of sine.

Worked solutionExpand

The Circularity: To apply Theorem 1.3.5 to lim⁡x→0sin⁡xx\lim_{x \to 0} \frac{\sin x}{x} one must differentiate sin⁡x\sin x at x=0x = 0 to get cos⁡0\cos 0 However, by Definition 1.2.1, the derivative of sin⁡x\sin x at 00 is defined as: sin⁡′(0)=lim⁡h→0sin⁡(0+h)−sin⁡0h=lim⁡h→0sin⁡hh.\sin'(0) = \lim_{h \to 0} \frac{\sin(0 + h) - \sin 0}{h} = \lim_{h \to 0} \frac{\sin h}{h}. Thus, knowing that ddx[sin⁡x]=cos⁡x\frac{d}{dx}[\sin x] = \cos x already requires knowing that lim⁡h→0sin⁡hh=1\lim_{h \to 0} \frac{\sin h}{h} = 1! Using L’Hôpital’s Rule to prove that lim⁡x→0sin⁡xx=1\lim_{x \to 0} \frac{\sin x}{x} = 1 assumes the very truth of the fact being evaluated. The limit must be proved independently using geometry and the Squeeze Theorem (Theorem 1.1.3).

Problem 1.P.11: Failure of Equality in Vector-Valued MVT

Descriptiontransfer

Let f:[0,2π]→R2\mathbf{f}: [0, 2\pi] \to \mathbb{R}^2 be given by f(t)=(cos⁡t,sin⁡t)\mathbf{f}(t) = (\cos t, \sin t) Explain why there is no point c∈(0,2π)c \in (0, 2\pi) satisfying f′(c)=f(2π)−f(0)2π−0\mathbf{f}'(c) = \frac{\mathbf{f}(2\pi) - \mathbf{f}(0)}{2\pi - 0} What inequality replaces the MVT in higher dimensions?

HintExpand

Compute the secant vector and the velocity vector ∥f′(t)∥\|\mathbf{f}'(t)\|.

Worked solutionExpand
  1. The secant vector is: f(2π)−f(0)2π=(1,0)−(1,0)2π=(0,0).\frac{\mathbf{f}(2\pi) - \mathbf{f}(0)}{2\pi} = \frac{(1, 0) - (1, 0)}{2\pi} = (0, 0).
  2. The derivative vector is f′(t)=(−sin⁡t,cos⁡t)\mathbf{f}'(t) = (-\sin t, \cos t)
  3. Its Euclidean norm is: ∥f′(t)∥=(−sin⁡t)2+(cos⁡t)2=1=1for all t∈[0,2π].\|\mathbf{f}'(t)\| = \sqrt{(-\sin t)^2 + (\cos t)^2} = \sqrt{1} = 1 \quad \text{for all } t \in [0, 2\pi].
  4. Since the velocity vector is everywhere of length 11 it is never equal to the zero vector (0,0)(0, 0) Thus no such cc exists.
  5. In vector and multivariable analysis, equality is replaced by the Mean Value Inequality: ∥f(b)−f(a)∥≤(b−a)sup⁡c∈(a,b)∥f′(c)∥.\|\mathbf{f}(b) - \mathbf{f}(a)\| \le (b - a) \sup_{c \in (a, b)} \|\mathbf{f}'(c)\|.
★

Problem 1.P.12: The Open Interval Hypothesis and Disconnected Domains

Perturbed Definitionsynthesis

In Corollary 1.3.3, the hypothesis states that f′(x)=0f'(x) = 0 on an open interval II. Construct an example of a function ff whose derivative is identically zero everywhere on its domain, but ff is not constant. Why does this not contradict Corollary 1.3.3?

HintExpand

Consider a domain that consists of two disjoint intervals, such as (0,1)∪(2,3)(0, 1) \cup (2, 3).

Worked solutionExpand

Construction: Let the domain be D=(0,1)∪(2,3)D = (0, 1) \cup (2, 3) Define f:D→Rf: D \to \mathbb{R} by: f(x)={5if x∈(0,1)9if x∈(2,3)f(x) = \begin{cases} 5 & \text{if } x \in (0, 1) \\ 9 & \text{if } x \in (2, 3) \end{cases}

  1. On (0,1)(0, 1) ff is constant, so f′(x)=0f'(x) = 0
  2. On (2,3)(2, 3) ff is constant, so f′(x)=0f'(x) = 0 Thus f′(x)=0f'(x) = 0 for all x∈Dx \in D However, ff is not constant on DD f(0.5)=5≠9=f(2.5)f(0.5) = 5 \ne 9 = f(2.5)

Why Corollary 1.3.3 is not contradicted: Corollary 1.3.3 requires the domain to be a single connected interval II If x1∈(0,1)x_1 \in (0, 1) and x2∈(2,3)x_2 \in (2, 3) the line segment [x1,x2][x_1, x_2] is not contained in DD (it contains the gap [1,2][1, 2] where ff is undefined). Because the domain is disconnected, Theorem 1.3.2 cannot be applied across the gap. The proof of Corollary 1.3.3 strictly requires [x1,x2]⊆D[x_1, x_2] \subseteq D

Problem 1.P.13: Rigorous Proof of L'Hôpital's Rule for the 0/0 Form

Proofsynthesis

Assume ff and gg are differentiable on (c,b)(c, b) with g′(x)≠0g'(x) \ne 0 and lim⁡x→c+f(x)=lim⁡x→c+g(x)=0\lim_{x \to c^+} f(x) = \lim_{x \to c^+} g(x) = 0 Assume further that lim⁡x→c+f′(x)g′(x)=L∈R\lim_{x \to c^+} \frac{f'(x)}{g'(x)} = L \in \mathbb{R} Prove that lim⁡x→c+f(x)g(x)=L\lim_{x \to c^+} \frac{f(x)}{g(x)} = L

HintExpand

Extend ff and gg continuously to cc by defining f(c)=g(c)=0f(c) = g(c) = 0, then apply Cauchy's Mean Value Theorem on [c,x][c, x].

Worked solutionExpand

Extend ff and gg to [c,b)[c, b) by defining f(c)=0f(c) = 0 and g(c)=0g(c) = 0 Since lim⁡x→c+f(x)=0=f(c)\lim_{x \to c^+} f(x) = 0 = f(c) and lim⁡x→c+g(x)=0=g(c)\lim_{x \to c^+} g(x) = 0 = g(c) both functions are continuous on [c,x][c, x] for every x∈(c,b)x \in (c, b) Both functions are differentiable on (c,x)(c, x) and g′(t)≠0g'(t) \ne 0 on (c,x)(c, x) By Rolle’s Theorem, g(x)−g(c)=g(x)≠0g(x) - g(c) = g(x) \ne 0 (otherwise g′g' would vanish). By Cauchy’s Mean Value Theorem (Theorem 1.3.4), there exists some ξx∈(c,x)\xi_x \in (c, x) such that: f(x)g(x)=f(x)−f(c)g(x)−g(c)=f′(ξx)g′(ξx).\frac{f(x)}{g(x)} = \frac{f(x) - f(c)}{g(x) - g(c)} = \frac{f'(\xi_x)}{g'(\xi_x)}. Now consider the limit as x→c+x \to c^+ Since c<ξx<xc < \xi_x < x the Squeeze Theorem (Theorem 1.1.3) forces lim⁡x→c+ξx=c+\lim_{x \to c^+} \xi_x = c^+ Therefore: lim⁡x→c+f(x)g(x)=lim⁡x→c+f′(ξx)g′(ξx)=lim⁡ξ→c+f′(ξ)g′(ξ)=L.\lim_{x \to c^+} \frac{f(x)}{g(x)} = \lim_{x \to c^+} \frac{f'(\xi_x)}{g'(\xi_x)} = \lim_{\xi \to c^+} \frac{f'(\xi)}{g'(\xi)} = L. This completes the proof of Theorem 1.3.5 for the right-sided 0/00/0 form. ∎

Problem 1.P.14: Root Interlacing Between Function and Derivative

Prooftransfer

Let f:R→Rf: \mathbb{R} \to \mathbb{R} be twice differentiable.

  1. Prove that between any two real roots of ff there exists at least one real root of f′f'
  2. Deduce that if f′′(x)>0f''(x) > 0 for all x∈Rx \in \mathbb{R} then ff can have at most two real roots.
HintExpand

Apply Rolle's Theorem to consecutive roots of ff, then examine monotonicity of f′f' from the sign of f′′f''.

Worked solutionExpand

Part 1: Interlacing of roots. Let r1<r2r_1 < r_2 be two distinct real roots of ff so f(r1)=f(r2)=0f(r_1) = f(r_2) = 0 Since ff is differentiable on R\mathbb{R} it is continuous on [r1,r2][r_1, r_2] and differentiable on (r1,r2)(r_1, r_2) By Rolle’s Theorem (Theorem 1.3.1), there exists at least one c∈(r1,r2)c \in (r_1, r_2) such that f′(c)=0f'(c) = 0

Part 2: Bound on number of roots. Suppose for contradiction that ff has three or more distinct roots r1<r2<r3r_1 < r_2 < r_3 By Part 1, there exists c1∈(r1,r2)c_1 \in (r_1, r_2) with f′(c1)=0f'(c_1) = 0 and there exists c2∈(r2,r3)c_2 \in (r_2, r_3) with f′(c2)=0f'(c_2) = 0 Since r1<c1<r2<c2<r3r_1 < c_1 < r_2 < c_2 < r_3 we have c1<c2c_1 < c_2 Now apply Rolle’s Theorem to f′f' on [c1,c2][c_1, c_2] since f′f' is continuous on [c1,c2][c_1, c_2] and differentiable on (c1,c2)(c_1, c_2) with f′(c1)=f′(c2)=0f'(c_1) = f'(c_2) = 0 there exists ξ∈(c1,c2)\xi \in (c_1, c_2) such that: f′′(ξ)=0.f''(\xi) = 0. However, the hypothesis states that f′′(x)>0f''(x) > 0 for all x∈Rx \in \mathbb{R} This contradiction proves that ff cannot have three or more roots; it can have at most two real roots.

References & Further Reading

  • Michael Spivak (2008). Calculus (4th ed.), Chapters 5, 9, 11. Publish or Perish.
  • Walter Rudin (1976). Principles of Mathematical Analysis (3rd ed.), Chapters 4 and 5. McGraw-Hill.
  • Stephen Abbott (2015). Understanding Analysis (2nd ed.), Chapters 4 and 5. Springer.