Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Uniform continuity and Bounded Derivative

  1. Jul 16, 2011 #1
    Hi, All:

    Let f R-->R be differentiable. If |f'(x)|<M< oo, then f is uniformly continuous, e.g.,

    by the MVTheorem. Is this conditions necessary too, i.e., if f:R-->R is differentiable

    and uniformly continuous, does it follow that |f'(x)|<M<oo ?

  2. jcsd
  3. Jul 16, 2011 #2
    Hi Bacle! :smile:

    This is not true. Consider the function

    [tex]f:]0,1[\rightarrow\mathbb{R}: x\rightarrow \sqrt{x}[/tex]

    then this is uniform continuous (continuous on compact interval). But the derivative [itex]f^\prime(x)[/itex] grows very large if x gets closer to 0.

    In fact, the condition you mention is equivalent to the Lipschitz-condition (for differentiable functions). That is, if [itex]f:[a,b]\rightarrow \mathbb{R}[/itex] is continuous and differentiable on ]a,b[, then the following are equivalent
    • [itex]|f~{}^\prime(x)|\leq k~[/itex] for all x in ]a,b[
    • f is k-Lipschitz, that is [itex]|f(x)-f(y)|\leq k|x-y|~~[/itex] for all x,y in [a,b]

    The proof uses the mid-value theorem.

    It is easy to see that being Lipschitz is stronger than uniform continuity.
  4. Jul 16, 2011 #3


    User Avatar
    Homework Helper

    Unless I'm reading something wrong, the statement in the original post requires differentiability on all R. This example isn't applicable.
  5. Jul 16, 2011 #4
    Fine, consider

    [tex]f:[0,1]\rightarrow \mathbb{R}:x\rightarrow x^{3/2}\sin(1/x)[/tex]

    this extends to an uniform continuous, differentiable function on [itex]\mathbb{R}[/itex], but it's derivative is not bounded.
  6. Jul 16, 2011 #5
    Thanks, micromass; I agree, because continuous+compact implies uniformly-continuous;

    I was thinking more of what if the extension outside of

    [a,b] is non-trivial (e.g., a constant function, or without exhausting , by covering

    the remainder by compact sets and pasting/smoothing at the endpoints), if the condition on

    the derivative must be satisfied in (-00,a)\/(b,oo), i.e., if the function is not defined

    piecewise in countably-many compact intervals.
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook