Catalan's conjecture (Q2482911): Difference between revisions

The Catalan Conjecture (1844) asserts that the only consecutive integers which are both pure powers are \(8\) and \(9\); in other words, the only solution of the Diophantine equation \[ x^p - y^q = 1 \tag{E} \] in integers \(x\), \(y\), both \({}>1\), and \(p\) and \(q\) are prime numbers, is given by \(3^2 - 2^3 = 1\). It appeared in a letter of Eugène Catalan to the Editor of Crelle's Journal. This conjecture was proved in 2002 by \textit{P. Mihăilescu} [``Primary cyclotomic units and a proof of Catalan's conjecture'', J. Reine Angew. Math. 572, 167--195 (2004; Zbl 1067.11017)], see also the Bourbaki lecture given by \textit{Yu. Bilu} [Catalan's conjecture (after Mihăilescu). Bourbaki seminar. Volume 2002/2003. Exposés 909--923. Paris: Société Mathématique de France. Astérisque 294, 1--26, Exp. No. 909 (2004; Zbl 1094.11014)]. In this booklet the author gives a complete proof of this theorem in less than 120 pages. First, the author treats the case where \(pq\) is even, solved elementarily by V. A. Lebesgue (1850) for \(q=2\), and by Ko Chao (1965) for \(p=2\), respectively. From now on we assume that \(p\) and \(q\) are odd. Then the author goes on with Cassels' fundamental result (1960) -- that is, \(p\) divides \(y\) and \(q\) divides \(x\) -- which uses Diophantine approximation; more precisely he shows that Cassels' proof belongs to Runge's method, which is rather surprising for a divisibility result. This result is used in all the proofs of Mihăilescu's theorems. In particular, it easily implies that the solutions of (E), if any, must be very large. In the two following chapters, using the elementary theory of cyclotomic fields and especially the minus class group, by an argument of Mihăilescu, it is proved -- without any computer verification -- that if there is another solution than \(9-8=1\) then \(p\) and \(q\) are \({}\geq 41\). After a chapter devoted to Stickelberger's theorem, the first of the main three steps of \textit{P. Mihăilescu}'s proof is obtained: his nice class number free criterion involving the double Wieferich pairs [J. Number Theory 99, No. 2, 225--231 (2003; Zbl 1049.11036)]: Theorem 1. If \(p\) and \(q\) are odd primes such that (E) has a non trivial solution then \[ p^{q-1}\equiv 1 \pmod{q^2}\quad \text{and}\quad q^{p-1}\equiv 1 \pmod{p^2}. \] The next chapter contains a proof of the following result: Theorem 2. -- (Mihăilescu, 2003) If \(p\) and \(q\) are odd primes such that (E) has a non trivial solution then \[ p < 4q^2 \quad \text{and}\quad q < 4 p^2. \] The proof uses subtle cyclotomic properties and an ``Archimedean'' argument (namely that \(x\) and \(y\) must be very large, as noticed above). Then, to conclude the proof, following Mihăilescu, it remains to prove that \[ p \equiv 1 \pmod q\quad \text{and}\quad q \equiv 1\pmod p. \] [It is easy to see that this result contradicts the inequalities of Theorem 2, proving that Catalan's equation has no other solution than \(9-8=1\).] This step is by far the most difficult one, its proof uses the deep Thaine's theorem and a very tricky application of Runge's method. This nice booklet is very well written, with a lot of useful remarks, comments and exercises.