Inhomogeneous order 1 iterative functional equations with applications to combinatorics

Abstract: We show that if a Laurent series $f\in\mathbb{C}((t))$ satisfies a particular kind of linear iterative equation, then $f$ is either a rational function or it is differentially transcendental over $\mathbb{C}(t)$. This condition is more precisely stated as follows: We consider $R,b\in \mathbb{C}(t)$ with $R(0)=0$, such that $f(R(t))=f(t)+b(t)$. If either $R'(0)=0$ or $R'(0)$ is a root of unity, then either $f$ is a rational function, or $f$ does not satisfy a polynomial differential equation. More generally a solution of a functional equation of the form $f(R(t))=a(t)f(t)+b(t)$ will be either differentially trascendental or the solution of an inhomogeneous linear differential equation of order $1$ with rational coefficients. We illustrate how to apply these results to deduce the differential transcendence of combinatorial generating functions by considering three examples: the ordinary generating function for a family of complete trees; the Green function for excursions on the Sierpinski graph; and a series related to the enumeration of permutations avoiding the consecutive pattern 1423. The proof strategy is inspired by the Galois theory of functional equations and relies on the property of the dynamics of $R$.