Discrete Generating Series and Linear Difference Equations

Abstract

We define discrete generating series for arbitrary functions \( f \colon \mathbb{Z}^n \rightarrow \mathbb{C} \) and derive functional relations that these series satisfy. For linear difference equations with constant coefficients, we establish explicit functional equations linking the generating series to the initial data, and for equations with polynomial coefficients, we introduce an analogue of Stanley's \( D \)-finiteness criterion, proving that a discrete generating series is \( D \)-finite if and only if the corresponding sequence is polynomially recursive. The framework is further generalized to multidimensional settings, where we investigate the interplay between discrete generating series and solutions to Cauchy problems for difference equations. Key structural properties are uncovered through the introduction of polynomial shift operators and projection techniques. The theory is illustrated with concrete examples, including the Tribonacci recurrence and Schr\"oder's second problem.