0

Let $f: [a, b] \to \Bbb R$ be a convex continuous function such that $f(a) < 0 < f (b)$. Prove that there is only one point $c \in (a, b)$ such that $f(c) = 0$.

TIP: The resolution to be made is similar to the Bolzano theorem used for the demonstration of the Intermediate Value Theorem. However, the idea is to use derivatives to solve this.

mdcq
  • 1,658
Nosda
  • 17
  • 1
    The reason your questions aren't getting answers is that you're supposed to add more into your questions than just the exercise as written on your homework assignment. Show us some effort and we'll help you finish it. Otherwise you're not likely to get answered. –  Nov 02 '16 at 23:19
  • 2
    I look in vain a "question" in your unique sentence. Besides, where is the connection with derivatives your mention in your title ? It looks nonsense. Improve your text ! – Jean Marie Nov 02 '16 at 23:34
  • The tittle is like that because is from a online classe and list of derivativas, then it is suggestive that the resolution involves derivatives. I'm having to answer lists that did not even have classes, because the university is on strike. I am catching it to all of you for make possible be helped with ideas and suggestions so that I can study and understand the problem. Calm down, I'm beginner, I do not yet mastered well. thanks for the comprehension. :-) – Nosda Nov 02 '16 at 23:52
  • @AdsonSarinho your sentence doesn't make sense. Have you copied the question correctly? – user251257 Nov 03 '16 at 00:27
  • @user251257 and another. There was a formatting error that omitted a part of the text. I made the correction. Thanks for letting. – Nosda Nov 03 '16 at 02:24
  • @user251257 Yeah. The correct is "c in (a, b)" and not "a in the (a, b). Again, sorry and thank you. – Nosda Nov 03 '16 at 02:33
  • @Bye_World Now I stop to read it, and yeah, there was a lot of erros in writing. Sorry. – Nosda Nov 03 '16 at 02:34
  • The translator tool sometimes eats many letters. – Nosda Nov 03 '16 at 02:35
  • Assuming there are $a<c<d<b$ such $f(c)=f(d)=0$. Compare the differential quotients on the 3 subintervals. Spot the contradiction. – user251257 Nov 03 '16 at 02:36

1 Answers1

2

Preface: There's at least one such point by the Intermediate Value Theorem. And depending how you define "convexity", it usually implies that your function is at least twice differentiable.

Now suppose there are two values $c_1, c_2 \in (a, b)$ which meet your conditions. In other words, $f(c_1) = f(c_2) = 0$. The order is arbitrary, so define $c_1 < c_2$.

Since $c_1 > a$ and $f(c_1) > f(a)$, $\frac{f(c_1)-f(a)}{c_1 -a} > 0$. So by the Mean Value Theorem, there exists a point $x \in (a, c_1)$ such that $f'(x) > 0$.

Since $c_2 > c_1$ and $f(c_2) = f(c_1)$, $\frac{f(c_2)-f(c_1)}{c_2 - c_1} = 0$. So by the Mean Value Theorem, there exists a value $y \in (c_1, c_2)$ such that $f'(y) = 0$.

Since $b > c_2$ and $f(b) > f(c_2)$, $\frac{f(b)-f(c_2)}{b-c_2} > 0$. So by the Mean Value Theorem, there exists a value $z \in (c_2, b)$ such that $f'(z) > 0$.

This means that $f'$ has decreased between $x$ and $y$, but increased between $y$ and $z$. Thus the concavity of $f$ is not consistent. Since we assumed $f$ was convex, this is a contradiction.

So there can be only one such point. QED

Draconis
  • 1,443
  • Thank you very much! Through his explanation I got rid of some doubts! – Nosda Nov 03 '16 at 02:49
  • You don't actually need differentiability for your proof. One can elementarily show that the differential quotient of a convex function is increasing. – user251257 Nov 03 '16 at 03:04
  • @user251257 Very true; that would probably be the more elegant way to solve it (with fewer dependencies). My flimsy excuse is that this sounded like the sort of question designed to demonstrate the Mean Value Theorem. :P – Draconis Nov 03 '16 at 03:06
  • What you mean with consistent concavity? @Draconis – Nosda Nov 03 '16 at 06:33
  • @user251257 Please, feel free to show what would be your most effective demonstration proposal. :) – Nosda Nov 03 '16 at 06:35
  • @Nosda You defined $f$ to be a convex function. That means its concavity can't change—if it did, it would end up being concave somewhere. – Draconis Nov 03 '16 at 14:26