Coincidences of division fields (Q6040262): Difference between revisions

The authors bring up two questions for division fields \(F_m=\mathbb Q(E[m])\) of a fixed elliptic curve \(E\) defined over \(\mathbb Q\); (1) Are there distinct integers \(n,m\ge 2\) such that \(F_n=F_m\) ? (2) Are there distinct prime numbers \(p,q\), and \(k\ge 1\) such that \(F_p\cap\mathbb Q(\zeta_q^k)\ne \mathbb Q\) ? Here \(\zeta_n\) is a primitive \(n\)-th root of unity. These questions are related to the size of the index \(d_E\) of the adelic Galois representation of \(E\). In fact, let \(\Delta_E\) is the minimal discriminant of \(E\). then \(F_2\cap\mathbb Q(\zeta_{4|\Delta_E|})\supset \mathbb Q(\sqrt{\Delta_E})\). This implies that \(d_E\ge 2\). Since Serre's result shows that for almost all \(m\), \(\mathrm{Gal}(F_m/\mathbb Q)\simeq \mathrm{GL}_2(\mathbb Z/m\mathbb Z)\), they point out that the affirmative answers of the questions are rare. Studying \(\mathrm{Gal}(F_m/\mathbb Q)\) as a subgroup of \(\mathrm{GL}_2(\mathbb Z/m\mathbb Z)\) and by using the fact that \(\zeta_m\in F_m\), they prove the following results. (1) For a prime number \(p\), if \(F_{p^n}=F_{p^{n+1}}\), then \(p=2,n=1\), and \(E\) is obtainable as a one parameter family. If \(F_{p^n}\supset\mathbb Q(\zeta_{p^{n+1}})\), then \(p=2\). For an elliptic curve \(y^2=x^3-11x-14\), \(F_{p^n}\supset\mathbb Q(\zeta_{p^{n+1}})\) for all \(n\). (2) For primes \(p<q\), if \(F_{p^n}=F_{q^m}\), then \(p^n=2,q^m=3\) and \(E\) is obtainable as a one parameter family. If \(F_p\supset \mathbb Q(\zeta_{q^{n}})\), then either \(\mathbb Q(\zeta_{q^{n}})=\mathbb Q,\mathbb Q(i)\) or \(\mathbb Q(\zeta_3)\), or \(E\) has a rational \(p\)-isogeny, \(p=,2,3,5,7,11,13,17, 19, 37,43,67\) or \(163\), and \(q^{n-1}(q-1)\) divides \(p-1\). Further for abelian division fields, (3) If \(F_n/\mathbb Q\) is abelian and \(F_n=F_m,n>m\ge 2\), then either \(m=2,n=4, F_4=\mathbb Q(i)\), or \(m=3,n=6\). In both cases, \(E\) is obtainable as a one parameter family. For a prime \(p\) such that \(F_p\) is abelian, if \(q\ne p\) is prime, then, \(F_p\cap\mathbb Q(\zeta_{q^k})\) can be trivial, quadratic, cubic (for \(p=2\)), or cyclic quartic (for \(p=5\)). Further, restricting to the range \(2\le m<n\le 10\), they show that if \(F_m=F_n\), then \((m,n)\in \{(2,3),(2,4),(2,6),(3,6),(4,6),(6,8),(6,9),(5,10)\}\).