Math Problem Statement

Let f : X → R be a differentiable function with X ⊆ R being a convex set. Prove that f′ is nondeccreasing iff f(x2) ≥ f(x1) + f′(x1)(x2 − x1), ∀x1, x2 ∈ X

Solution

To prove the equivalence between the two conditions for a differentiable function f:XRf: X \to \mathbb{R}, where XRX \subseteq \mathbb{R} is a convex set, we will establish the two implications:

  1. f(x)f'(x) is nondecreasing implies f(x2)f(x1)+f(x1)(x2x1)f(x_2) \geq f(x_1) + f'(x_1)(x_2 - x_1) for all x1,x2Xx_1, x_2 \in X.

  2. f(x2)f(x1)+f(x1)(x2x1)f(x_2) \geq f(x_1) + f'(x_1)(x_2 - x_1) for all x1,x2Xx_1, x_2 \in X implies f(x)f'(x) is nondecreasing.


Part 1: f(x)f'(x) is nondecreasing implies the inequality

Assume that f(x)f'(x) is nondecreasing on the convex set XX, meaning that for any x1,x2Xx_1, x_2 \in X with x1x2x_1 \leq x_2, we have: f(x1)f(x2).f'(x_1) \leq f'(x_2). Now, we want to show that: f(x2)f(x1)+f(x1)(x2x1).f(x_2) \geq f(x_1) + f'(x_1)(x_2 - x_1).

Let g(x)=f(x)f(x1)f(x1)(xx1)g(x) = f(x) - f(x_1) - f'(x_1)(x - x_1). The goal is to show that g(x2)0g(x_2) \geq 0. Note that: g(x1)=f(x1)f(x1)f(x1)(x1x1)=0.g(x_1) = f(x_1) - f(x_1) - f'(x_1)(x_1 - x_1) = 0. The derivative of g(x)g(x) is: g(x)=f(x)f(x1).g'(x) = f'(x) - f'(x_1). Since f(x)f'(x) is nondecreasing, for xx1x \geq x_1, we have f(x)f(x1)f'(x) \geq f'(x_1), which implies that g(x)0g'(x) \geq 0 for xx1x \geq x_1. Therefore, g(x)g(x) is nondecreasing for xx1x \geq x_1, and since g(x1)=0g(x_1) = 0, it follows that g(x2)0g(x_2) \geq 0 for x2x1x_2 \geq x_1. Thus, we have: f(x2)f(x1)+f(x1)(x2x1),f(x_2) \geq f(x_1) + f'(x_1)(x_2 - x_1), as required.

Part 2: The inequality implies f(x)f'(x) is nondecreasing

Now assume that for all x1,x2Xx_1, x_2 \in X, we have: f(x2)f(x1)+f(x1)(x2x1).f(x_2) \geq f(x_1) + f'(x_1)(x_2 - x_1). We want to show that f(x)f'(x) is nondecreasing.

Let x1<x2x_1 < x_2 be any two points in XX, and define the function: ϕ(x)=f(x)f(x1)xx1.\phi(x) = \frac{f(x) - f(x_1)}{x - x_1}. The function ϕ(x)\phi(x) represents the slope of the secant line connecting (x1,f(x1))(x_1, f(x_1)) and (x,f(x))(x, f(x)). By the assumption, we have: f(x2)f(x1)+f(x1)(x2x1),f(x_2) \geq f(x_1) + f'(x_1)(x_2 - x_1), which can be rearranged as: f(x2)f(x1)x2x1f(x1).\frac{f(x_2) - f(x_1)}{x_2 - x_1} \geq f'(x_1). This means that ϕ(x2)f(x1)\phi(x_2) \geq f'(x_1).

Now, consider the limit as x2x1+x_2 \to x_1^+ (i.e., approaching x1x_1 from the right): limx2x1+f(x2)f(x1)x2x1=f(x1).\lim_{x_2 \to x_1^+} \frac{f(x_2) - f(x_1)}{x_2 - x_1} = f'(x_1). Thus, ϕ(x2)f(x1)\phi(x_2) \geq f'(x_1) for x2>x1x_2 > x_1, and in the limit, the inequality holds with equality. This implies that the derivative f(x)f'(x) is nondecreasing.

Conclusion

We have shown that:

  • If f(x)f'(x) is nondecreasing, then f(x2)f(x1)+f(x1)(x2x1)f(x_2) \geq f(x_1) + f'(x_1)(x_2 - x_1) for all x1,x2Xx_1, x_2 \in X.
  • If f(x2)f(x1)+f(x1)(x2x1)f(x_2) \geq f(x_1) + f'(x_1)(x_2 - x_1) for all x1,x2Xx_1, x_2 \in X, then f(x)f'(x) is nondecreasing.

Thus, the two conditions are equivalent.


Would you like more details on any part of the proof or have any questions?

Here are 5 related questions you can explore:

  1. How does convexity of the set XX play a role in this proof?
  2. What other equivalent conditions can describe a function's convexity?
  3. Can the inequality in the problem statement be used to define convex functions?
  4. How can we use second-order derivatives to verify that f(x)f'(x) is nondecreasing?
  5. What is the geometric interpretation of the inequality involving f(x1)f'(x_1)?

Tip: When proving implications between statements, breaking the problem into two separate cases (if and only if) can clarify the structure of the proof.

Ask a new question for Free

By Image

Drop file here or Click Here to upload

Math Problem Analysis

Mathematical Concepts

Differentiation
Convexity
Inequalities
Monotonicity

Formulas

f(x2) ≥ f(x1) + f'(x1)(x2 - x1)
f'(x1) ≤ f'(x2) for x1 ≤ x2

Theorems

Mean Value Theorem
First Derivative Test
Properties of Convex Functions

Suitable Grade Level

Undergraduate Level (Calculus III / Real Analysis)