🧮 Brain Teaser

The Monotone Function That Isn't

Let f:[0,1]Rf: [0,1] \to \mathbb{R} be a function that is differentiable everywhere on [0,1][0,1], with f(x)0f'(x) \geq 0 for all x[0,1]x \in [0,1].

Must ff be monotone (non-decreasing)?

Now suppose instead we only know that f(x)>0f'(x) > 0 for all x[0,1]x \in [0,1].

Must ff be strictly increasing?

For each case, either prove it or give a counterexample.

mean value theoremmonotonicitydifferentiabilityreal analysiscounterexample

Answer: The Monotone Function That Isn't

Key Idea / Intuition

The two cases have surprisingly different answers. When f0f' \geq 0 everywhere, the Mean Value Theorem forces ff to be non-decreasing — you cannot go down without having a negative derivative somewhere. When f>0f' > 0 everywhere, you might expect strict monotonicity, and indeed this is true too, but the subtlety is that "strictly positive derivative everywhere" is stronger than it sounds: there is no room for f(a)=f(b)f(a) = f(b) with a<ba < b, since the MVT would force f=0f' = 0 somewhere in between.

But here is the real conceptual trap: can a function be everywhere differentiable with f(x)>0f'(x) > 0 for all xx, yet fail to be strictly increasing globally? The answer is no — but the proof requires care. The interesting "near-miss" is that we can construct functions where ff' is positive but arbitrarily small (approaching zero), which are still strictly increasing, just not uniformly so.


Formal Proof / Solution

Case 1: f0f' \geq 0 implies ff is non-decreasing

Claim: Yes, ff must be non-decreasing.

Proof: Take any a<ba < b in [0,1][0,1]. By the Mean Value Theorem, there exists c(a,b)c \in (a,b) such that f(b)f(a)=f(c)(ba).f(b) - f(a) = f'(c)(b - a). Since f(c)0f'(c) \geq 0 and ba>0b - a > 0, we get f(b)f(a)0f(b) - f(a) \geq 0, i.e., f(b)f(a)f(b) \geq f(a).

So ff is non-decreasing. \square


Case 2: f>0f' > 0 implies ff is strictly increasing

Claim: Yes, ff must be strictly increasing.

Proof: Take any a<ba < b in [0,1][0,1]. By the same MVT argument: f(b)f(a)=f(c)(ba)f(b) - f(a) = f'(c)(b-a) for some c(a,b)c \in (a,b). Now f(c)>0f'(c) > 0 and ba>0b - a > 0, so f(b)f(a)>0f(b) - f(a) > 0.

Hence ff is strictly increasing. \square


The Punchline and Subtlety

Both claims are true, and the proofs are almost identical. So where is the interesting mathematics?

The subtlety lives in the converse direction. Consider:

A strictly increasing, differentiable function can have f(x)=0f'(x) = 0 at some points.

A famous example: f(x)=x3f(x) = x^3 is strictly increasing on R\mathbb{R}, but f(0)=0f'(0) = 0.

More dramatically, one can construct a strictly increasing CC^\infty function whose derivative vanishes on a Cantor set of positive measure — so f=0f' = 0 on a fat set, yet ff is still strictly increasing!

So the logical structure is: f>0    f strictly increasingf' > 0 \implies f \text{ strictly increasing} f strictly increasing̸    f>0f \text{ strictly increasing} \not\implies f' > 0 f0    f non-decreasingf' \geq 0 \implies f \text{ non-decreasing} f non-decreasing̸    f0 (it may not even be differentiable)f \text{ non-decreasing} \not\implies f' \geq 0 \text{ (it may not even be differentiable)}

The deeper lesson: positivity of the derivative is a sufficient but not necessary condition for monotonicity, and the MVT is the clean bridge between the local (derivative) and the global (monotone behavior).

Type: analysisEdit on GitHub ↗