Formal reduction of singular linear differential systems using eigenrings: a refined approach (Q2200310)

Let \(F=\mathbb{C}\)((x)) and \(\delta=\frac{d}{dx}\). The paper discusses reduction algorithms for systems of linear differential equations of the form (1) \(\delta Y=AY,\:A\in M_{F}(n)\) by means of linear transformation of variables \(Y\longrightarrow Z=TY,T\in GL_{F}(n)\). The normal form of the fundamental matrix of system (1) with respect to such transformations was presented in [\textit{W. Balser} et al., Funkc. Ekvacioj, Ser. Int. 22, 197--221 (1979; Zbl 0434.34002)]. Linear transformations of the fundamental matrix of system (1) induce pseudo-linear transformations of the matrix of coefficients of this system \(A\longrightarrow B=TAT^{-1}+\delta TT^{-1},\quad\delta Z=BZ\). In the paper, the authors present a new, efficient, based on the use of the eigenring techniques, an algorithm for reducing a matrix of coefficients to the normal form. The algorithm is implemented in the Maple computer algebra system and examples are given to illustrate it. We note a slightly different approach for the classification of systems (1) based on the use of nonsplit skew polynomials in [\textit{V. V. Sergeĭchuk}, Differ. Equations 25, No. 10, 1201--1206 (1989); translation from Differ. Uravn. 25, No. 10, 1721--1727 (1989; Zbl 0701.34018)], as well as the classification of the associated with these systems Picard-Vessiot extensions of field \(\mathbb{C}((x))\) (see the reviewer [Math. USSR, Sb. 37, 327--335 (1980); translation from Mat. Sb., N. Ser. 109(151), 355--364 (1979; Zbl 0416.12013)]).