Some Generalized Fermat-type Equations via Q-Curves and Modularity

Abstract

[eng] The main purpose of this thesis is to apply the modular approach to Diophantine equations to study some Fermat-type equations of signature (r; r; p) with r >/= 5 a fixed prime and “p” varying. In particular, we will study equations of the form x(r) + y(r) = Cz(p), where C is an integer divisible only by primes “q” is non-identical to 1; 0 (mod “r”) and obtain explicit arithmetic results for “r” = 5, 7, 13.

We start with equations of the form x(5) + y(5) = Cz(p). Firstly, we attach two Frey curves E; F defined over Q(square root 5) to putative solutions of the equation.

Then by using the work of J. Quer on embedding problems and on abelian varieties attached to Q-curves we prove that the p-adic Galois representations attached to E, F can be extended to p-adic representations E), (F) of Gal(Q=Q). Finally, we apply Serre's conjecture to the residual representations  (E), (F) and using Siksek's multi-Frey technique we conclude that the initial solution can not exist.

We also describe a general method for attacking infinitely many equations of the form x(r) + y(r) = Cz(p) for all r>/= 7. The method makes use of elliptic curves over totally real fields, modularity and irreducibility results for representations attached to elliptic curves and level lowering theorems for Hilbert modular forms. Indeed, for each fixed “r” we produce several Frey curves defined over K+, the maximal totally real subfield of Q(xi-r). Moreover, if “r” is of the form 6k + 1 we prove the existence of a Frey curve defined over K(0) the subfield of K(+) of degree k. We prove also an irreducibility result for the mod “p” representations attached to certain elliptic curves and a modularity statement for elliptic curves over totally real abelian number fields satisfying some local conditions at 3. Finally, for r = 7 and r = 13 we are able to compute the required spaces of (Hilbert) newforms and by applying our general methods we obtain explicit arithmetic results for equations of signature (7; 7; p) and (13; 13; p).

We end by providing two more Frey k-curves (a generalization of Q-curve), where “k” is a certain subfield of K(+), when “r” is a fixed prime of the form 4m+1.

[spa] En esta tesis, utilizaremos el método modular para profundizar en el estudio de las ecuaciones de tipo (r; r; p) para r un primo fijado. Empezamos por utilizar la teoría de J. Quer sobre variedades abelianas asociadas con Q-curvas y embedding problems para producir dos curvas de Frey asociadas con hipotéticas soluciones de infinitas ecuaciones de tipo (5; 5; p). Después, utilizando la conjetura de Serre y el método multi-Frey de Siksek demostraremos que las hipotéticas soluciones no pueden existir. Describiremos también un método general que nos permite atacar un número infinito de ecuaciones de tipo (r; r; p) para cada primo “r” mayor o igual que 7. El método hace uso de curvas elípticas sobre cuerpos de números, teoremas de modularidad, teoremas de bajada de nivel y formas modulares de Hilbert. Además, para ecuaciones de tipo (7; 7; p) y (13; 13; p) calcularemos los espacios de formas modulares relevantes y demostraremos que una familia infinita de ecuaciones no admite cierto tipo de soluciones. Además, demostraremos un nuevo teorema de modularidad para curvas elípticas sobre cuerpos totalmente reales abelianos. Finalmente, para primos congruentes con 1 módulo 4 propondremos dos curvas de Frey más. Demostraremos que son “k-curves” (una generalización de Q-curva) y también que satisfacen las propiedades necesarias para que pueda ser útiles en la aplicación del método modular.