5
$\begingroup$

Suppose $\lambda^{*}$ is the Lebesgue outer measure.

Question:

Does there exist an explicit $f:\mathbb{R}\to\mathbb{R}$, where:

  1. The range of $f$ is $\mathbb{R}$
  2. For all real $x_1,x_2,y_1,y_2$, where $-\infty<x_1<x_2<\infty$ and $-\infty<y_1<y_2<\infty$: $$\lambda^{*}(\left([x_1,x_2]\times[y_1,y_2]\right)\cap\left\{(x,f(x)):x\in\mathbb{R}\right\})>0$$

Attempt: I'm not sure how to answer this question but heard of Conway’s Base-13 function.

I have also asked a similar question here with an answer suggesting that no function satisfies the question.

Edit: I changed the question since there is a simpler question that has to be answered.

$\endgroup$

1 Answer 1

8
$\begingroup$

The answer is yes.

We construct $f$ by transfinite recursion using a well ordering of the reals. (So this may not be explicit enough for you.)

In fact, we can make the function $f$ bijective, with the graph of $f$ having full outer measure in every rectangle.

Theorem. There is a bijective function $f:\mathbb{R}\to\mathbb{R}$ with full outer measure in every rectangle in the plane.

Proof. There are only continuum many Borel sets, and so we may enumerate a list $(N_\alpha,R_\alpha)$, for $\alpha<\frak{c}$ in type continuum, where $N_\alpha\subset R_\alpha$ is a set with less than full measure in rectangle $R_\alpha$ in the plane, and all such combinations arise in the list.

We now define the function $f$ in stages. At any stage $\alpha<\frak{c}$, we will have defined $f$ on fewer than continuum many points. At stage $\alpha$, we consider the set $N_\alpha\subset R_\alpha$ sitting inside that rectangle. Because it has has less than full measure, the complement $R_\alpha\setminus N_\alpha$ has positive measure, and so there must be continuum many $x$ on whose section in $R_\alpha$ there are points in $R_\alpha\setminus N_\alpha$. So there must some such $x$ on which $f$ is not yet defined, and we may define $f(x)$ so that $(x,f(x))$ is one of those points in $R_\alpha\setminus N_\alpha$. Continue this process for continuum many stages, and then extend $f$ on the remaining points arbitrarily. Since on any rectangle, the graph of $f$ is not contained in any measurable set of less than full measure, the graph must have full outer measure on all such rectangles.

We can arrange that $f$ is onto by changing its values on a size continuum measure zero set, which will not affect the outer measure property of its graph.

We can actually make $f$ bijective, since at each stage in the construction, there will be a not-yet-defined section $x$ on which $N_\alpha$ omits continuum many points in $R_\alpha$, and so at that stage we can let $f(x)$ be a totally new value, while still avoiding $N_\alpha$. And then at the end, we can change $f$ on measure zero set so as to hit all the other values, so $f$ will be bijective. $\Box$

Regarding the question of "explicit", we have seen on the other question that there can be no measurable function with a postive outer measure graph. In particular, there can be no Borel function, and this is a common sense of explicitness. Meanwhile, it is consistent with ZFC that there is a projectively definable well-ordering of the reals, and in this model the function $f$ I provide can be projectively definable, at a fairly low level of complexity. Meanwhile, it is also relatively consistent with ZFC that every projective set is measurable. So the question of whether there is an explicit function $f$ as desired, if this is taken to mean projectively definable, is independent of ZFC.

$\endgroup$
5
  • 2
    $\begingroup$ Can the graph have full outer measure intersected with each rectangle? $\endgroup$ Jul 25 at 17:27
  • 3
    $\begingroup$ Good idea. Yes, I believe so, this is what the argument is going to show. I'll edit. $\endgroup$ Jul 25 at 17:33
  • 2
    $\begingroup$ This answers the question in the title, about "any function ... at all". The question in the text below asks for "an explicit $f$", and that's independent of ZFC. $\endgroup$ Jul 25 at 17:38
  • 2
    $\begingroup$ I have edited to get full measure. And @AndreasBlass, yes, I had stated in my answer that because the construction uses a well-ordering of the reals, it may not be explicit enough for the OP. But if projective counts as explicit, then it is independent of ZFC. $\endgroup$ Jul 25 at 17:41
  • 1
    $\begingroup$ I think a further refinement will make $f$ have positive outer measure in every rectangle, but never full outer measure. One should fix a set with positive but not full measure in every rectangle, and then make the choices for $f$ avoid that set. $\endgroup$ Jul 25 at 17:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.