Squares in arithmetic progressions (Q1196391)

In this high-level paper the authors prove that the number of squares in the arithmetic progression \(a+nq\) (\(1\leq n\leq N\), \(q\neq 0\)) is at most \(c_ 1 N^{2/3}(\log N)^{c_ 2}\) (\(c_ 1\), \(c_ 2\) are effectively computable, absolute constants). Even the proof of \(o(N)\) seems to be very difficult. One possible approach is via Szemeredi's theorem on sets without 4-term arithmetic progressions. The authors' goal is to apply Faltings' theorem on the number of rational points on a curve of genus \(g\geq 2\). Let \(qn_ i+a=m_ i^ 2\) (\(i=0,\dots,4\); \(n_ i\) distinct), \({\mathbf n}=(n_ 0,\dots,n_ 4)\). Then \({\mathbf n}\) determines a rational point \({\mathbf m}=(m_ 0,\dots,m_ 4)\) on the algebraic projective curve of genus 5 \[ C_ 3: (n_ i-n_{i+1})x_{i-1}^ 2+(n_{i+1}- n_{i-1}) x_ i^ 2+(n_{i-1}-n_ i)x_{i+1}^ 2=0 \qquad (i=1,2,3). \] A detailed study of such curves, in particular of the rank of the Mordell-Weil group of rational points of the Jacobian, together with the first author's simplified proof of Faltings' theorem [Ann. Sc. Norm. Super. Pisa, Cl. Sci., IV. Ser. 17, No. 4, 615-640 (1990; Zbl 0722.14010)] lead to an explicit upper bound for the number of rational points on \(C_ 3\). Some combinatorial arguments, which, as the authors remark, could possibly be improved, finally lead to the main result. It is the authors' hope that the similar problem with higher powers in arithmetic progressions can be treated in an analogue way, however there are still some complications in the study of the Mordell-Weil group of the Jacobian of the associated curves.