Computing splitting fields using Galois theory and other Galois constructions (Q2100056): Difference between revisions

This paper is an algorithmic paper. The paper is concerned with computing splitting fields using Galois theory instead of resorting to polynomial factorization. This paper depends upon algorithms computing Galois groups, developed previously in papers written by \textit{C. Fieker} and \textit{J. Klüners} [LMS J. Comput. Math. 17, 141--158 (2014; Zbl 1326.11070)] for number fields and \textit{N. Sutherland} [J. Symb. Comput. 71, 73--97 (2015; Zbl 1393.11073)]; \textit{D. Krumm} and \textit{N. Sutherland} [J. Symb. Comput. 103, 108--126 (2021; Zbl 1461.11152)] for function fields. These algorithms are much less limited compared to the ones that were proposed before them. In a context, where one considers a polynomial that is solvable by radicals, the authors compute algorithmically a tower of radical extensions as a splitting field thanks to the algorithms mentioned above. The first algorithm is the basic step toward their goal as it computes the fixed field of a subgroup of a Galois group by giving back the polynomial defining the extension defined by this fixed field. The authors have some restrictions on the base field they are working over, indeed this base field is either a number field, or an algebraic extension of \(\mathbb{F}_q(t) (\alpha)\), or \(\mathbb{Q} (t)\). The second algorithm of the paper computes a splitting field through a chain of subgroups of its Galois group. The base field is further restricted to number fields or \(\mathbb{F}_q(t)\). The third algorithm computes the chain of subgroups necessary to compute a splitting field as a tower of extensions and its last step calls the second algorithm to complete the computation of a splitting field. The second algorithm calls the fourth algorithm in the step where computing the resolvent of the polynomial is needed. Their fourth algorithm requires the computation of certain bounds that are necessary to compute a precision to use for the computation of the roots such that the computations of the resulting defining polynomials are accurate. The authors dedicate section 4.1 to discussing these bounds. Algorithm 5 computes a tower of radical polynomials in which a polynomial splits from the tower of cyclic extensions obtained in Algorithm 2, for polynomials with solvable Galois groups. The transition from Algorithm 2 to Algorithm 5 relies on handling the necessary roots of unity. This is done in step 6 of Algorithm 5 which requires extracting a radical polynomial from the cyclic extensions building up to the tower. This step calls Algorithm 6. The author finishes the paper by turning their attention to the computation of the geometric Galois group offered by Algorithm 7 for polynomials with coefficients over \(F(t)\) where \(F\) is a number field. Their motivation relies on the connection between those objects and the Galois inverse theory as well as the alternative approach to computing absolute factorization they provide. Throughout the paper, the reader can enjoy examples illustrated using Magma.