Solution.
Since f is continuous, it is in particular integrable, so let c=∫01f(x)dx. Consider g(x)=ex−(ec+ec(x−c)), i.e., the error between ex and its tangent line at x=c. Observe that
g′(x)=ex−ec=0⟺x=candg′′(x)=ex>0.
Thus, g is convex, so it has a unique global minimum at x=c, which is g(c)=0. This proves that ex≥ec+ec(x−c) for x∈R. In particular, ef(x)≥ec+ec(f(x)−c) for x∈[0,1], so integrating both sides yields
∫01ef(x)dx≥ec+ec(∫01f(x)dx−c)=exp(∫01f(x)dx),
since the Lebesgue measure on [0,1] has total mass 1 and because c=∫01f(x)dx by definition.
Observe that in our proof, ef(x)>ec+ec(f(x)−c)⟺g(x)>0. By compactness of [0,1] and continuity of g, g attains its minimum m>0, so the strict inequality holds in integration:
∫01g(x)dx≥m>0⟹∫01ef(x)dx>exp(∫01f(x)dx).
Thus, equality occurs if and only if ef(x)=ec+ec(f(x)−c)⟺g(f(x))=0 everywhere. But because g has a unique global minimum at x=c, it follows that f(x)=c everywhere, which completes the proof.