9
$\begingroup$

When I was in high school I learned about an elementary proof that there exist irrational numbers $p$ and $q$ such that $p^q$ is rational. Put $p = q = \sqrt{2}$; if $p^q$ is rational, we are done. Otherwise take $p = \sqrt{2}^\sqrt{2}$and $q = \sqrt{2}$.

I am sure that by now much more is known about this phenomenon. Maybe the answer can be described in terms of graph theory. Define a graph $G$ on the positive irrationals. Two irrationals $p$ and $q$ are adjacent in $G$ if and only if at least one of the two powers $p^q$ or $q^p$ is a rational number that is not an integer power of another rational. A similar graph $G^*$ could be defined on the positive real numbers. I would like to know what can be said about $G$ and $G^*$. For instance, what is the cardinality of the edge set $E(G)\ ?$ What is the maximum degree of $G\ ?$

Suggestions for proper tagging are most welcome.

$\endgroup$
2
  • 3
    $\begingroup$ I can hardly understand what is this strange construction for. BTW, {\sqrt{2}}^{\sqrt{2}} is known to be transcendental. I have seen some speculations about what can be said about the "simultaneous" arithmetic of p, q, p^q and q^p in a recent preprint of Diego Marques and Jonathan Sondow (not available from their webs nor from the arxiv, sorry). I wonder whether you can find their (conjectural) answers to what you are looking for. $\endgroup$ Commented Apr 6, 2010 at 8:31
  • 1
    $\begingroup$ The URL is ami.ektf.hu/uploads/papers/finalpdf/AMI_37_from151to164.pdf for our paper D. Marques and J. Sondow, Algebraic and transcendental solutions of some exponential equations, Annales Mathematicae et Informaticae 37 (2010) 151-164. $\endgroup$ Commented Jan 11, 2011 at 4:55

1 Answer 1

14
$\begingroup$

The answers are that $|E(G)|=2^{\aleph_0}$ and that every vertex of $G$ has degree $\aleph_0$.

Proof: The positive real solutions to $x^y=2$ form a curve of cardinality $2^{\aleph_0}$, and at most $\aleph_0$ of these have $x$ or $y$ rational, so $|E(G)|=2^{\aleph_0}$. (It cannot be larger, because this is also the number of pairs of vertices.)

For $g \in G$ with $g>1$, the set of positive real numbers $x$ such that $g^x$ is a prime number contains at most one rational number since if there were two, then we would obtain an equation $p^q=p'$ with $p,p'$ distinct primes and $q$ rational, which is impossible by unique factorization. Thus $g$ has infinite degree. On the other hand, the degree of $g$ is at most countable since $g^x = q$ and $x^g=q$ have at most one solution $x$ each for each $q \in \mathbf{Q}$. Finally, the same arguments apply when $g<1$, with reciprocals of primes in place of primes.

Remark: The same statements hold for $G^*$ except that one should exclude the vertex $1$.

$\endgroup$
2
  • $\begingroup$ Wow, that is very nice. Is it possible to say anything about the lengths of paths in G, existence of cycles, number and sizes of connected components, existence of non-trivial automorphisms, etc. $\endgroup$ Commented Apr 6, 2010 at 12:34
  • 1
    $\begingroup$ Since the degree of each vertex is at most countable, each component is countable, so the number of components is the cardinality of the continuum. $\endgroup$
    – gowers
    Commented Apr 6, 2010 at 16:43

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .