Let L be a linear ordinary differential equation with coefficients in C (x). This thesis presents algorithms to solve L in closed form. The key part of this thesis is 2-descent method, which is used to reduce L to an equation that is easier to solve. The starting point is an irreducible L, and the goal of 2-descent is to decide if L is projectively equivalent to another equation L˜ that is defined over a subfield C (f) of C (x).;Although part of the mathematics for 2-descent has already been treated before, a complete implementation could not be given because it involved a step for which we do not have a complete implementation. Our key novelty is to give an approach that is fully implementable. We describe and implement the algorithm for order 2, and show by examples that the same also work for higher order. By doing 2-descent for L, the number of true singularities drops to at most n/2 + 2 (n is the number of true singularities of L). This provides us ways to solve L in closed form (e.g. in terms of hypergeometric functions).
展开▼
