I've been studying infinitesimals and came upon the idea of uniformly microcontinuous functions. My question is: if a function $f^*: \mathbb{R}^* \to \mathbb{R}^*$ the natural extension of $f: \mathbb{R} \to \mathbb{R}$ is microcontinuous on $A\subset\mathbb{R}$ does that imply that it is continuous with the $\epsilon-\delta$ definition on $A$? If this is not true does the (I think)weaker statement hold that if the function is uniformly microcontinuous on all of $\mathbb{R^*}$ does that imply that it is continuous with the $\epsilon-\delta$ definition on $\mathbb{R}$?

Just a note that the definition I'm using for microcontinuous at $x$ is $\forall x' (x \approx x') \implies (f(x) \approx f(x'))$ where $a \approx b$ means that the standard part of $a-b$ is $0$. From this a function is micrcontinuous on some domain $S$ means that $\forall x\in S\, \forall x' (x \approx x') \implies (f(x) \approx f(x'))$

  • $\begingroup$ @tomasz A microcontinuous function is a function that is microcontinuous for all points in its domain I believe. I'll edit that into the definition though. $\endgroup$ – ruler501 Mar 29 '14 at 3:04
  • 1
    $\begingroup$ Yeah, I see that now. Sorry, I'm little tired. $\endgroup$ – tomasz Mar 29 '14 at 3:04
  • $\begingroup$ Understandable. I should've made sure it was stated clearly since it doesn't seem to be a common concept for people to know. $\endgroup$ – ruler501 Mar 29 '14 at 3:06
  • $\begingroup$ @ruler501: What book are you using? $\endgroup$ – Jose Antonio Mar 29 '14 at 3:11
  • 1
    $\begingroup$ @JoseAntonio I have been reading from math.wisc.edu/~keisler/calc.html and math.wisc.edu/~keisler/foundations.html but I got the definition of microcontinuity from wikipedia. $\endgroup$ – ruler501 Mar 29 '14 at 3:14

Yes, it does. To see that, choose any (standard) real $x_0\in A$. Now for any $x$ such that $$\models \{x_0-1/n<x<x_0+1/n\mid n\in {\bf N}\}$$ we have $$\models \{ f(x_0)-1/m<f(x)<f(x_0)+1/m\mid m\in {\bf N}\}$$ This means that in particular, for each $m\in {\bf N}$, we have $$\{x_0-1/n<x<x_0+1/n\mid n\in {\bf N}\}\vdash f(x_0)-1/m<f(x)<f(x_0)+1/m$$ and by compactness there is some $n\in {\bf N}$ such that $$x_0-1/n<x<x_0+1/n\vdash f(x_0)-1/m<f(x)<f(x_0)+1/m$$ but then $f$ is clearly ($\varepsilon$-$\delta$) continuous at $x_0$ when considered as a function on real numbers, so it's also just continuous, since $x_0$ was arbitrary. Since $\varepsilon$-$\delta$ continuity is first order, it is also $\varepsilon$-$\delta$ continuous on any given nonstandard model.

Note that it only applies if $f$ is a standard real function (a symbol of the language) to begin with. Otherwise the part about restricting to reals does not make sense and neither does the compactness argument, and indeed we can find a counterexample: choose an infinitesimal $\varepsilon$ and let $f(x)=0$ for $\lvert x\rvert >\varepsilon$, $f(0)=0$, and $f(x)=\varepsilon$ for $x\in [-\varepsilon,0)\cup (0,\varepsilon]$. Then $f$ is microcontinuous but not continuous.

  • $\begingroup$ I added to the question that it should be an $f^*$ that is the natural extension of an $f: \mathbb{R} \to \mathbb{R}$ so that the question makes more sense(addressing your note). $\endgroup$ – ruler501 Mar 29 '14 at 4:15
  • $\begingroup$ More generally the notion of microcontinuity (it should be pointed out that the term seems to have been introduced by Martin Davis in his 1976 book) is still a meaningful notion for internal functions. $\endgroup$ – Mikhail Katz Mar 31 '14 at 9:08

It is not clear what you mean exactly by "uniformly microcontinuous". If you simply mean "microcontinuous", then microcontinuity of $f^*$ at every point of a real set $A$ implies continuity of $f$ on $A$. If you assume that $f^*$ is microcontinuous on all of $\mathbb{R}^*$ then it follows that $f$ is not merely continuous on $\mathbb{R}$ but is actually uniformly continuous on $\mathbb{R}$.


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.