Sur les fonctions entières à double pas récurrent

Nicolas Brisebarre, Laurent HabsiegerAnnales 1999 · DOI

Abstract

This paper generalizes the study of entire functions that satisfy systems of two linear difference equations. Building on the foundational work of J.-P. Bézivin and F. Gramain, the authors characterize these functions as exponential polynomials under broad conditions—specifically when the ratio of the recurrence steps is a real irrational number. In addition to this theoretical extension, the paper provides a constructive three-step algorithm to find explicit solutions using symbolic computation techniques.

Introduction

In the fields of complex analysis and number theory, a recurring challenge is the characterization of functions that satisfy specific functional equations. A classic problem, originally posed by D.W. Masser, asks whether an entire function $f(z)$—a function complex-differentiable across the entire plane—must be an exponential polynomial if it satisfies two distinct linear difference equations with polynomial coefficients.

An exponential polynomial takes the form $\sum P_k(z)e^{\delta_k z}$, where $P_k$ are polynomials. These functions represent the most structured solutions to linear recurrences. While previous research by Bézivin and Gramain established that such functions are at least quotients of exponential polynomials by polynomials, a complete characterization remained elusive for cases where the ratio of the recurrence steps, $\alpha$ and $\beta$, is a real irrational number. This paper bridges that gap, moving the problem from existence proofs to constructive algorithmic solutions.

Theoretical Framework and Methodology

The authors utilize a combination of algebraic analysis, operator theory, and functional equations to address the "irrational ratio" problem.

Density and Differential Equations

The primary hurdle in previous literature was the case where $\alpha/\beta \in \mathbb{R} \setminus \mathbb{Q}$. The authors resolve this by leveraging the fact that the subgroup $\mathbb{Z}\alpha + \mathbb{Z}\beta$ is dense in $\mathbb{R}$. By applying density arguments, they demonstrate that any entire function satisfying the two recurrence relations must also satisfy a differential equation with constant coefficients. This property effectively forces the function to take the form of an exponential polynomial.

Operator Theory and Rational Characterization

The difference equations are treated as operators of the form $P(e^{\alpha D})$ acting on the space of functions. A key technical contribution is Lemma 3.1, which characterizes pairs of rational functions satisfying a specific "cross-shift" identity. This lemma is instrumental in reducing complex systems with polynomial coefficients into simpler systems with constant coefficients.

Algorithmic Integration

To move beyond pure theory, the authors adapt symbolic computation tools to provide a constructive method for finding solutions:

  • Abramov’s Algorithm: Used to determine the denominators of potential rational solutions.
  • Petkovšek’s "Poly" Algorithm: Employed to identify the polynomial components of the solutions.

Key Insights

  1. The Irrational Bridge: The most significant theoretical insight is that the difficulty of real irrational ratios can be overcome via the density of the underlying group. This extends results that were previously restricted to complex (non-real) ratios.
  2. Reduction to Constant Coefficients: The authors establish a "vanishing criterion" (Criterion 4.1). If a system satisfies this criterion, the entire system can be transformed into a constant-coefficient system by dividing the function by an appropriate rational function.
  3. Frequency Matching: The search space for solutions is narrowed by the requirement that the "frequencies" (the $\delta$ in $e^{\delta z}$) must simultaneously satisfy the characteristic equations of both difference equations.

The Three-Step Algorithm

The paper formalizes a procedure for explicitly determining the solutions:

  1. Determine Frequencies: Solve for $\delta_s$ by identifying the common roots of the characteristic polynomials of both recurrence equations.
  2. Determine Rational Coefficients: Apply Abramov and Petkovšek’s methods to find the $x_s(z)$ terms that serve as coefficients for the exponential parts.
  3. Ensure Global Integrity: Solve a system of linear equations to ensure the resulting sum is an entire function. This step is crucial to ensure that any poles present in the individual rational parts cancel each other out globally.

Limitations and Future Directions

While the algorithm is robust for non-real ratios, the authors note that the first step—comparing frequencies—presents computational challenges when the ratio is a real irrational number. Because the subgroup is dense, determining if two frequencies are equivalent requires high-precision Diophantine approximation (such as the LLL algorithm), for which universal bounds are not always available in this context.

Future work may involve extending these results to "P-recursive" functions, which satisfy equations involving both shifts and derivatives, potentially broadening the scope of automated discovery of functional solutions in complex analysis.

Summary updated

Abstract

Nous proposons une nouvelle approche et une généralisation d’un problème résolu par J.-P. Bézivin et F. Gramain, dont l’objet est de caractériser les fonctions entières solutions de systèmes de deux équations aux différences finies. De plus, nous donnons un algorithme qui permet de trouver la forme explicite des solutions.
BibTeX
@inproceedings{brisebarre1999sur,
  author = {Nicolas Brisebarre and Laurent Habsieger},
  title = {{Sur les fonctions entières à double pas récurrent}},
  booktitle = {Annales},
  year = {1999},
  doi = {10.5802/aif.1685},
}

Loading related content...

Figure