Why does the Mean Value Theorem assume a closed interval for continuity and an open interval for differentiability?

The MVT says: Let $f$ be a continuous function on $[a,b]$ that is differentiable on $(a,b)$, then....

Is there any example where one of them isn't true so that the MVT is not true?

  • 7
    $\begingroup$ Because it isn't necessarily true if these conditions are relaxed. $\endgroup$ – carmichael561 Jun 11 '17 at 0:50
  • 4
    $\begingroup$ You may also wonder why we assume differentiability on the open interval $(a,b)$ rather than (perhaps more naturally) assuming that "$f$ is differentiable on $[a,b]$" The reason is that functions like $f(x) = \sqrt{x}$ defined on $[0,1]$ are not differentiable at an endpoint, but nonetheless the MVT holds. $\endgroup$ – MathematicsStudent1122 Jun 11 '17 at 0:59
  • $\begingroup$ @MathematicsStudent ..sorry but I didn't understand why derivative doesn't exist at endpoint..are u talking about 0??but RHD is $+\infty$..so if derivative is infinite can we say function is not differentiable at that point $\endgroup$ – Believer Jan 8 '19 at 11:13
  • $\begingroup$ @Believer Saying that the derivative of $\sqrt{x}$ at $0$ is $+\infty$ is the same thing as saying that the derivative doesn't exist or that the graph of $\sqrt{x}$ is completely vertical there. MathematicsStudent1122 must have been talking about the endpoint at $0$ since the derivative at the other endpoint, $1$, is $1$. The point is that it's a counterexample to the natural assumption that you could define the MVT with functions differentiable on $[a,b]$. $\endgroup$ – Jam Jul 17 '19 at 14:32

Relax the first constraint: Let $f:[0,1] \to \mathbb R$ so that $f(0) = 1,f(x) = 0$ for $x \in \left]0,1\right]$. Then $(f(1) - f(0))/(1-0) = -1$ but $f'(x) = 0$ on $]0,1[$.

Relax the second contraint: Let $f(x) = |x|$ on $[-1,1]$, then $(f(1)-f(-1))/(1-(-1)) = 0$ but $f'(x) = 0$ nowhere.


The MVT is a consequence of Rolle's Theorem. you need continuity at $[a,b] $ to be sure that the function is bounded. if its extremum is attained at $x=c\in (a,b) $ you use differentiability at $(a,b) $ to get $f'(c)=0.$

  • 1
    $\begingroup$ You need continuity for more than just boundedness. Henricus W.'s first example is bounded, but still fails. $\endgroup$ – Paul Sinclair Jun 11 '17 at 1:36
  • $\begingroup$ @PaulSinclair and to attain extrema. $\endgroup$ – hamam_Abdallah Jun 11 '17 at 1:44
  • $\begingroup$ Without continuity at the end-points, the MVT is obviously false, isn't it? It's nothing to do with boundedness. $\endgroup$ – TonyK Jun 11 '17 at 11:20
  • $\begingroup$ @TonyK In the counterexample above, the function is not continuous but still bounded. $\endgroup$ – hamam_Abdallah Jun 11 '17 at 11:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.