Remarks on the functional equation $f(x+1)=g(x)f(x)$ and a uniqueness theorem for the gamma function

TL;DR

This paper investigates the functional equation related to gamma functions, showing how asymptotic conditions can be simplified, and establishes new uniqueness theorems that extend classical results like the Bohr-Mollerup theorem.

Contribution

It generalizes existing uniqueness theorems for solutions of the gamma functional equation by relaxing conditions and introduces new criteria involving log-convexity and log-concavity.

Findings

Asymptotic condition can be reduced to convergence of g(n+1)/g(n) to 1.

Uniqueness of solutions extends to those that are eventually log-convex or log-concave of second order.

A new gamma function uniqueness theorem replaces log-convexity with log-concavity of order two.

Abstract

The topic of gamma type functions and related functional equation $f(x+1)=g(x)f(x)$ has been seriously studied from first half of the twentieth century till now. Regarding unique solutions of the equation the asymptotic condition $\displaystyle{\lim_{x \to \infty}}\frac{g(x+w)}{g(x)}=1$, for each $w>0$, is considered in many serial papers including R. Webster's article (1997). Motivated by the topic of limit summability of real functions (introduced by the author in 2001), we first show that the asymptotic condition in the Webster's paper can be reduced to the convergence of the sequence $\frac{g(n+1)}{g(n)}$ to $1$, and then by removing the initial condition $f(1)=1$ we generalize it. On the other hand, Rassias and Trif proved that if $g$ satisfies another appropriate asymptotic condition, then the equation admits at most one solution $f$, which is eventually $\log$-convex of the second order. We also sow that without changing the asymptotic condition for this case, a uniqueness theorem similar to the Webster's result is also valid for eventually $\log$-convex solutions $f$ of the second order. This result implies a new uniqueness theorem for the gamma function which states the $\log$-convexity condition in the Bohr-Mollerup Theorem can be replaced by $\log$-concavity of order two. At last, some important questions about them will be asked.