Encoding algebraic power series (Q1656381)

Let \(K[x]:=K[x_1,\dots,x_n]\) be the polynomial ring over a field \(K\) in \(n\) variables and \(K[[x]]\) the formal power series ring in the same variables. Although the division in \(K[[x]]\) is ``an infinite algorithm in the infinitely many coefficients of the series'', the authors of this paper show that, when the series belong to the algebraic closure of \(K[X]\) in \(K[[X]]\), a much better situation can be obtained. More precisely, the authors face and give a solution to the very interesting problem of proving that a division algorithm for ideals of algebraic power series on a field \(K\) can be performed in a finite number of steps, in the further hypotheses that the algebraic power series satisfy the Hironaka's box condition (Theorem 11.1). The intriguing theoretical aspects of this algorithm are carefully highlighted. Although actual computation is unfeasible in general, some examples are provided. This result can be described as an effective version of Hironaka's Henselian Weierstrass Theorem (see Section 54.5 of the book [\textit{T. Mora}, Solving polynomial equation systems. Vol. IV. Buchberger theory and beyond. Cambridge: Cambridge University Press (2016; Zbl 1362.12001)]). It generalizes the Weierstrass division theorem (that is, for principal ideals), which has been computationally solved by \textit{M. E. Alonso} et al. [J. Pure Appl. Algebra 77, No. 1, 1--38 (1992; Zbl 0749.13017)], where a computational model for algebraic formal power series is developed, with several significant results for the theory of algebraic power series. One of the main tools that are used here is the fact that the algebraic power series that are involved in the division can be expressed in terms of defining polynomial codes. Codes of algebraic power series have been introduced in the paper of Alonso, Mora and Raimondo, as well, basing on an effective version of the Artin-Mazur theorem which characterize algebraic power series by the assumption of the implicit function theorem. In this paper, also modules of algebraic power series are described by polynomial codes and all the operations to be performed are guaranteed to be described in the same terms. The importance of the polynomial codes is that they determine distinctively the algebraic power series, differently from a minimal polynomial, which gives a weaker information, indeed. In this paper, an explicit construction of these codes is described and, in particular, of the codes of the reduced standard basis of the ideal. The proof proceeds by induction on the number of the variables. The main result is first investigated in the so-called \(x_n\)-regular case, that is the case ''where the initial module of the given module of algebraic power series vectors is generated by monomial vectors depending only on the last variable \(x_n\)''. Then, the induction step can be performed thanks to the Hironaka's box condition, which is a special decomposition request on the initial module. This condition needs the more general use of modules, instead of ideals. The paper consists of several sections, that in many cases are quite short and help, in this way, the reader in following the many details that are needed to achieve the final result, and highlight the role of every ingredient in the steps of the algorithm. The Introduction gives a framework of the rich background and results. The article ends with an explanatory example that takes the last entire section.