8
$\begingroup$

Is there a map $s: \mathbb{Q}\times\mathbb{Q} \to \mathbb{N}$ with the following properties?

  1. For all $z\in\mathbb{Z}$, the restriction $s|_{[z,z+1)\times [z,z+1)}: [z,z+1)\times [z,z+1) \to \mathbb{N}$ is bijective, and

  2. For all $q\in\mathbb{Q}$, the restrictions $s|_{\{q\} \times \mathbb{Q}}: \{q\} \times \mathbb{Q} \to \mathbb{N}$ and $s|_{\mathbb{Q} \times \{q\}}: \mathbb{Q} \times \{q\} \to \mathbb{N}$ are bijective.

$\endgroup$
12
  • 10
    $\begingroup$ Isn’t this more or less trivial? Just enumerate the countably many conditions saying that something should be in the range, and build a countable chain of finite maps obeying the injectivity conditions to meet them all. $\endgroup$ Commented Jan 7 at 12:52
  • 3
    $\begingroup$ Perhaps more fun to prove: define the obvious forcing notion (with finite condition), it is ccc and does not change cardinals, nor it changes the meaning of the natural or rational numbers. Then, the existence of this function is sufficiently simple ($\Sigma_1^1$ if I counted right, but certainly below $\Pi^1_2$), so using Shoenfield's absoluteness, there's a solution in the ground model. $\endgroup$ Commented Jan 7 at 13:41
  • 2
    $\begingroup$ @AsafKaragila: I'm trying to imagine what kind of "fun" you would have playing Chutes and Ladders. $\endgroup$ Commented Jan 7 at 13:48
  • 5
    $\begingroup$ @Lee: The fun that was always intended. Learning about virtues, morality, and karma. $\endgroup$ Commented Jan 7 at 14:03
  • 2
    $\begingroup$ So, what is the smallest collection of starting values needed to determine a unique solution? $\endgroup$ Commented Jan 7 at 14:58

3 Answers 3

16
$\begingroup$

All sets involved are countably infinite, so there is a "step-by-step" construction which achieves the desired result.

Specifically, let us consider the following collection of conditions:

  • $A(z,n)$ for $z\in\Bbb Z,n\in\Bbb N$: the square $[z,z+1)\times[z,z+1)$ has a point with assigned value $n$,
  • $B(q,n)$ for $q\in\Bbb Q,n\in\Bbb N$: the line $\Bbb Q\times\{q\}$ has a point with assigned value $n$,
  • $C(q,n)$ for $q\in\Bbb Q,n\in\Bbb N$: the line $\{q\}\times\Bbb Q$ has a point with assigned value $n$,
  • $D(q,r)$ for $q,r\in\Bbb Q$: the point $(q,r)$ is assigned some value.

We can enumerate this set of all conditions as a single list $P_i,i\in\Bbb N$, and our goal is to successively assign values to points in $\Bbb Q^2$ to satisfy all conditions them all while making sure that the restrictions to each square or line is injective. The key is that at every step only finitely many values will have been assigned, leaving us with infinitely many possibilities to satisfy a condition and not break injectivity.

If $P_i$ is a condition of the form $A(z,n),B(q,n)$ or $C(q,n)$ which has not been satisfied yet, then we can pick some point in the relevant square or line which is not in any of the other squares of lines to which a value has been assigned yet, and give it value $n$. If $P_i$ is of the form $D(q,r)$, then we can assign to $(q,r)$ a value different from any values so far assigned to points on the square or lines which $(q,r)$ lies on.

$\endgroup$
7
  • $\begingroup$ This answer, like MANY-MANY other MO-answers, features dense English (would a translation into Inuit help, with its $\ \aleph_0\ $ descriptions of snow? :) ) but not enough of formal mathematical notation. $\endgroup$ Commented Jan 8 at 13:46
  • 10
    $\begingroup$ @WlodAA Because I, like MANY-MANY mathematicians on MO and elsewhere, believe that explanations in plain English are more understandable than trying to encode everything in formal mathematical notation. There is time and place to write things in formal notation, but its excessive use damages readability. $\endgroup$ Commented Jan 8 at 14:10
  • 8
    $\begingroup$ I don't think this particular answer would benefit from having significantly more mathematical notation. You are welcome to disagree. $\endgroup$ Commented Jan 8 at 14:47
  • 1
    $\begingroup$ I hope I gave you a reason to come back to. this discussion. $\endgroup$ Commented Jan 11 at 11:33
  • 6
    $\begingroup$ @WlodAA If you are referring to posting of your own answer - no, I do not believe you have. $\endgroup$ Commented Jan 11 at 11:53
4
$\begingroup$

We can do this with permutations.

Let $ \def\N{\mathbb{N}} \def\Q{\mathbb{Q}} \def\Z{\mathbb{Z}} \def\fq{\lfloor q \rfloor} \def\fr{\lfloor r \rfloor} \def\fx{\lfloor x \rfloor} \def\fy{\lfloor y \rfloor} \def\fz{\lfloor z \rfloor} \Q_k$ denote $\Q\cap[k,k+1)$, and let $c:\Q_0^2\to\N$ be a bijective function counting off the elements of $\Q_0^2$.

Let superscripts from $\Z$ denote a simply transitive action of $\Z$ on $\Q_0$, and define $p:\Q^2\to\Q_0^2$ piecewise by $$p(x,y)=(\{x\}^{\fx-\fy},\{y\}^{\fx-\fy}).$$

Then the claim is satisfied with $s=c\circ p$.

Proof: Since $c$ is bijective, it is enough to show that $p$ is bijective on $\Q_m\times\Q_n$ for integer $m,n$, and bijective on horizontal and vertical lines. The bijectivity on $\Q_m\times\Q_n$ is immediate from the inverse of the group action, and the proofs for horizontal and vertical lines are the same. So it suffices to prove bijectivity on vertical lines.

Injectivity on Vertical Lines: When $p(x,y)=p(x,z)$, the first coordinates are $$\{x\}^{\fx-\fy}=\{x\}^{\fx-\fz}.$$ By the simple transitivity, $\fy=\fz$. So $(x,y)$ and $(x,z)$ are both in $\Q_{\fx}\times\Q_{\fy}$, on which $p$ is bijective, and $y=z$.

Surjectivity on Vertical Lines: Suppose $q \in \Q$ and $x,y\in\Q_0$. Let $k$ be an integer such that $\{q\}^k=x$. Let $r=\fq - k + y^{-k}$. Then $$\fq-\fr=\fq-(\fq-k)=k$$ and as desired $$p(q,r)=(\{q\}^k,\{r\}^k)=(x,(y^{-k})^k) = (x,y).$$

$\endgroup$
0
1
$\begingroup$

$\qquad\qquad$ Countable SUDOKU

I'll provide three parts:

  • Sudoku countable plate definition;
  • Sudoku solution definition;
  • Sudoku solution

A Sudoku plate is an infinite countable family $\ \mathbf A\ $ of infinite countable sets such that for arbitry finite $\ T\subseteq\bigcup\mathbf A,\ $ and $\ X\in\mathbf A,\ $ we have $$ X\setminus\bigcup\{Y\in\mathbf A\setminus\{X\}: Y\cap T\ne\emptyset\}\,\ \ne\ \ \emptyset $$


Sudoku solution definition:   a solution is an arbitrary function $\ f:\bigcup\mathbf A\to\mathbb N\ :=\ \{1\ 2\ \ldots\} $ such that $\ f|X:X\to\mathbb N\ $ is a bijection for every $\ X\in\mathbf A.$


Solution: Define $$ \forall_{n\in\mathbb N}\qquad B_n:=\{n\in\mathbb N: n\equiv 2^{n-1}\mod 2^n\} $$

Let $\ \mathbf A=(X_n: n\in\mathbb N).$

Let $\ \pi:\bigcup\mathbf A\to\mathbb N\ $ be defined piecewise as a common extension of certain bijections

$$ \pi_n:\ X_n\setminus\bigcup_{m<n}X_m\ \to\ B_n $$ for every $\ n\in\mathbb N\ $ (thus, $\ \pi:\bigcup\mathbf A\to\mathbb N\ $ is a bijection).

Now, consider certain bijections:

$$ \rho_n:\ B_n\ \to\ \mathbb N\setminus \bigcup_{m<n}\left\{\rho_m(\pi(x)): x\in X_m\cap X_n\right\} $$ -- Solution -- $\ f:\bigcup\mathbf A\to\mathbb B\ $ can be given by:

$$ \forall_{n\in\mathbb N}\qquad \left( x\in X_n\setminus\bigcup\{X_m:m<n\}\quad\ \Rightarrow\quad f(x)\ =\ (\rho_n\circ\pi_n)(x) \right) $$

! Enjoy !




REMARK: The subtle moment is hidden in the selection of the ordering of $\ \bigcup\mathbf A$.

$\endgroup$

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.