Linear Algebra for Computing Gröbner Bases of Linear Recursive Multidimensional Sequences

International audienceSakata generalized the Berlekamp -- Massey algorithm to $n$ dimensions in~1988. The Berlekamp -- Massey -- Sakata (BMS)algorithm can be used for finding a Gröbner basis of a $0$-dimensionalideal of relations verified by a table. We investigate this problem usinglinear algebra techniques, with motivations such as accelerating change ofbasis algorithms (FGLM) or improving their complexity.We first define and characterize multidimensional linear recursive sequencesfor $0$-dimensional ideals.Under genericity assumptions, we propose a randomized preprocessing of thetable that corresponds to performing a linear change of coordinates on thepolynomials associated with the linear recurrences. This technique thenessentially reduces our problem to using the efficient $1$-dimensional Berlekamp -- Massey (BM)algorithm.However, the number of probes to the table in this scheme may be elevated.We thus consider the table in the \emph{black-box} model: we assume probing thetable is expensive and we minimize the number of probes to the table in ourcomplexity model.We produce an FGLM-like algorithm for finding the relations in thetable, which lets us use linear algebra techniques. Under some additionalassumptions, we make this algorithm adaptive and reduce further the numberof table probes.This number can be estimated by counting the number of distinct elements in amulti-Hankel matrix (a multivariate generalization of Hankel matrices); we canrelate this quantity with the \emph{geometry} of the final staircase. Hence,in favorable cases such as convex ones, the complexity is essentially linear inthe size of the output. Finally, when using the \textsc{lex} ordering, we canmake use of fast structured linear algebra similarly to the Hankelinterpretation of Berlekamp -- Massey