1. Limited time only! Sign up for a free 30min personal tutor trial with Chegg Tutors
    Dismiss Notice
Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Homework Help: Convex function

  1. Feb 28, 2010 #1
    1. The problem statement, all variables and given/known data
    Hey, the original question is not in english, so I am translating. So just to make sure I'm understood, i take convex to mean that the graph of the function is below the tangent.

    The question:
    Let F be a convex function and F is bounded from above by some number C, prove that F is static (again, my translation, by static I mean that for every x F(X)=a)



    3. The attempt at a solution

    I don't think I'm close, but I am stumped, some mild hin tto point me in the right direction would be great.
    we will write F as a taylor expansion:
    [tex] f(x_0)+f'(x_0)(x-x_0) > f(x_0)+f'(x_0)(x-x_0) + \frac {f''(c)(x-x_0)^2}{2} <a \iff [/tex]
    [tex]
    0 > \frac {f''(c)(x-x_0)^2}{2} <a - f(x_0)+f'(x_0)(x-x_0) \iff [/tex]
    [tex]
    0 >\frac {f''(c)}{2} < \frac {a}{(x-x_0)^2} - \frac {f(x_0)+f'(x_0)}{x-x_0} [/tex]
    1. The problem statement, all variables and given/known data



    2. Relevant equations



    3. The attempt at a solution
    1. The problem statement, all variables and given/known data



    2. Relevant equations



    3. The attempt at a solution
    1. The problem statement, all variables and given/known data



    2. Relevant equations



    3. The attempt at a solution
    1. The problem statement, all variables and given/known data



    2. Relevant equations



    3. The attempt at a solution
    1. The problem statement, all variables and given/known data



    2. Relevant equations



    3. The attempt at a solution
    1. The problem statement, all variables and given/known data



    2. Relevant equations



    3. The attempt at a solution
     
  2. jcsd
  3. Feb 28, 2010 #2
    I don't think Taylor's theorem is required here. The other flaw with your approach is that you are assuming some value for a beforehand in trying to show the function is constant. I think a better approach is to show that f(x) = f(y) for every real x,y.

    I think it's a lot easier if you consider two points x and y and suppose x < y. If f(y) > f(x), what can you say about the behavior of f as x approaches infinity? Similarly, what if f(y) < f(x)? Remember in working with these two cases, you are trying to obtain a contradiction if your goal is to show the function is constant. But there is really only one statement in the hypothesis you can directly contradict.
     
  4. Feb 28, 2010 #3
    The definition I learned of convex is that the graph lies above the tangent. I don't think the problem works with your definition: for example, f(x)=-x2 is bounded above by 0, and lies below its tangent.
     
  5. Feb 28, 2010 #4
    The function doesn't need to be strictly convex.
     
  6. Mar 1, 2010 #5
    Ok, it took me a few days to work this over and I'm still not there. Here's what I have so far:

    Since F is convex then every point has one sided deriviatives such that
    [tex] x>y f_{-}'(y) \leq f_{+}'(y) \leq f_{-}'(x) \leq f_{+}'(x) [/tex]
    which means that the function is monotonic increasing.
    let x>y then [tex] lim_{x-> \infty} f(x) = c [/tex] because the function is monotonic increasing and bounded from above. But this means that
    [tex] lim_{x-> \infty} \frac {f(x) -f(y)}{x-y}= lim_{x-> \infty} \frac {c -f(y)}{x-y} =0 [/tex] Which means that there exists a point y<"a"<x where [tex] f'(a)=0 [/tex] but this means that for any x<a [tex] f'(x)=0 [/tex]

    Now take y>x>a. Then [tex] lim_{x-> \infty} f(x) = c and f(y)=c [/tex] so for all x after a certain point [tex] f'(x)=0 [/tex] then all points before that point must also have [tex] f'(x)=0 [/tex] because of the monotonity of f. Then f is constant.

    Is that correct?
    Thanks
     
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook