The Nowhere-Zero Derivative Paradox
Let be differentiable everywhere, and suppose for all .
Must have constant sign? That is, must either for all , or for all ?
Prove or find a counterexample.
Answer: Nowhere-Zero Derivative Paradox
Key Idea / Intuition
Your first instinct might be: "If is never zero, it can't switch sign, because by the Intermediate Value Theorem it would have to pass through zero." And this is exactly right — but the key insight is subtle: the derivative need not be continuous, yet it still satisfies the Intermediate Value Property (Darboux's theorem). So even though might be discontinuous, it cannot skip over zero. Therefore yes: must have constant sign.
Formal Proof / Solution
Claim: If is differentiable everywhere and for all , then has constant sign.
Step 1: State Darboux's Theorem.
Darboux's Theorem: If is differentiable on , then satisfies the Intermediate Value Property: for any and any value strictly between and , there exists with .
This is remarkable because need not be continuous — yet it cannot "jump" over any value. (The proof uses that achieves its extremum at an interior point where .)
Step 2: Apply Darboux to conclude constant sign.
Suppose for contradiction that takes both positive and negative values. Then there exist with and .
By Darboux's theorem applied on the interval , since lies strictly between and , there exists between and such that:
This contradicts the assumption that for all .
Conclusion: cannot change sign. Since is never zero, we must have either for all , or for all .
Why this is surprising:
One might think: "Maybe oscillates wildly enough to avoid zero while still changing sign — like ." But Darboux blocks this entirely. The IVP for derivatives is a hidden rigidity that survives even in the absence of continuity.
A classic example of a discontinuous derivative: (with ) has but oscillates near . This shows derivatives can be wild — but Darboux guarantees they can never skip a value.