1
$\begingroup$

Let $(M,\rho)$ be a metric space, $F\subset M$ a closed subset, $\varepsilon>0$.

On page 8 of the book "Convergence of Probability measures" by Patrick Billingsley one says that the function $$f:M\to [0,1]\\ x\mapsto(1-\rho(x,F)/\varepsilon)^+$$ is uniformly continuous since it holds $|f(x)-f(y)|\leq \rho(x,y)/\varepsilon$.

Actually I'm not able to prove this inequality: I tried to solve all the possible cases ($x,y\in F, \not\in F, \in F^\varepsilon, \not\in F^\varepsilon$) but it doesn't looks elegant at all and I even could't prove all the possibilities.

$(.)^+:=max(0,.)$

Thanks

$\endgroup$
1
  • $\begingroup$ It looks like a triangle inequality application. Perhaps it would help if you identified one of the cases you solved as well as one you could not. $\endgroup$ – hardmath Jan 14 '17 at 20:18
0
$\begingroup$

If $x$ is in the metric space and $z \in \mathrm{F}$ then $d(x, \mathrm{F}) \leq d(x, z) \leq d(x, y) + d(y ,z)$ so taking the infimum on $z$ on the left-most and right-most gives $d(x, \mathrm{F}) - d(y, \mathrm{F}) \leq d(x,y)$ and by symmetry, you can put absolute value. So $f(x) - f(y) = \dfrac{\rho(x, \mathrm{F}) - \rho(y, \mathrm{F})}{\varepsilon}.$ Is the rest clear from here?

$\endgroup$
1
  • $\begingroup$ The rest is clear, thank you! $\endgroup$ – Demetrio Masciurett Jan 14 '17 at 21:09

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.