A note on integral points on elliptic curves (Q2642785): Difference between revisions

Let \(E: y^2 = x^3 + Ax + B\) be an elliptic curve with integral coefficients \(A\), \(B\). For measuring the size of an integral point \((X,Y)\) on \(E\) one uses \(\rho = \log(X)/\log(\max( | A| ^{1/2},| B| ^{1/3}))\). A conjecture going back to Lang and Stark states that \(\rho \leq 10 + o(1)\) except for finitely many parametric families of elliptic curves. In 1988, Elkies (in a letter to Zagier, which is included as an appendix) constructed infinitely many examples with \(\rho \geq 12 - o(1)\). Elkies' approach leads to four extremal cases, which give rise to polynomial systems of equations. Elkies solved one of them using resultants; the other three cases are too complicated to solve even with the help of Gröbner bases, and Elkies suggested using multidimensional \(p\)-adic Newton iteration for attacking the remaining three cases. This is done by the author, who manages to find solutions in the second and the third case, leading to parametric families of elliptic curves defined over number fields of degree \(4\) and \(9\), respectively. The question whether there exist other solutions remains open.