I want to show that Hölder continuous functions are uniformly continuous using $\epsilon-\delta$. Is it sufficient to find a $\delta>0$ that does not depend on $"x"$ and depends only on $"\epsilon"$? Furthermore is my method correct? Here is the context:

Let $(X,d_x)$ and $(Y,d_y)$ be metric spaces.

Let $\quad f:X\rightarrow Y \quad$ be a Holder continuous function.

Let $\,\epsilon>0,\quad\alpha>0,\quad C>0, \quad \delta =\frac\epsilon{2C}, \quad \gamma>0,\quad x_1,x_2\in X$

Let $\gamma=\delta^{\frac1 \alpha}<1$

As $f$ is continuous we have: $\,\,d_x(x_1,x_2)<\gamma\implies d_y(f(x_1),f(x_2))<\epsilon$

So in particular: $\,\,d_x(x_1,x_2)^{\alpha}<\delta\implies d_y(f(x_1),f(x_2))<\epsilon$

As $f$ is Hölder continuous, we have:

$\,\,d_y(f(x_1),f(x_2)\le C*d_x(x_1,x_2)^{\alpha}<C*\delta=C*\frac\epsilon{2C}<\epsilon$

As $\delta$ does not depend on $"x"$, and this is true for all $"\epsilon>0"$, then $f$ is uniformly continuous. $CQFD.$


1 Answer 1


As $f$ is continuous we have: $d_x(x_1,x_2)<\gamma\implies d_y(f(x_1),f(x_2))<\epsilon$

There are two problems here. First, why is this true for this specific $\gamma$ you chose? Second, if it's true for some choice of $\gamma$ then you already have uniform continuity, so why continue the proof?

Correct approach. The input is: $\alpha, C$ such that $d_y(f(x_1),f(x_2)\le C\,d_x(x_1,x_2)^{\alpha}$, and an arbitrary $\epsilon>0$.

From this we must come up with $\delta$. I would use $\delta = (\epsilon/C)^{1/\alpha}$; then the verification of $$d_x(x_1,x_2)<\delta\implies d_y(f(x_1),f(x_2))<\epsilon$$ should be easy.

  • $\begingroup$ Thank you for your reply! So to be sure, you are saying my $\gamma$ could still be dependent on $"x"$ and $"\epsilon"$ which means my $"\delta"$ could still be dependent on $"x"$ and $"\epsilon"$ when we want it to depend ONLY on $"\epsilon"$. @bro $\endgroup$
    – Kam
    Apr 23, 2018 at 21:37
  • 1
    $\begingroup$ The way you defined $\gamma$, it does not depend on $x$, but you haven't explained how you know $d_x(x_1,x_2)<\gamma\implies d_y(f(x_1),f(x_2))<\epsilon$. If the reason is "because $f$ is continuous" (which is given in that sentence) then $\gamma$ would have to be chosen based on $x_1$ or $x_2$. $\endgroup$
    – user357151
    Apr 23, 2018 at 23:34
  • $\begingroup$ I see exactly where I went wrong!! Instead of supposing I had $d_x(x_1,x_2)<\gamma$ and tried to prove $d_y(f(x_1),f(x_2))<\epsilon$ to be true, I assumed the implication to be true already! (I hope I am correct in this realization haha) Thanks a lot anyway for the help!! $\endgroup$
    – Kam
    Apr 24, 2018 at 10:02

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.