No label defined (Q4713402)

The author finds the log-convex solutions of the functional equations (1) \(G(x+1)=\Gamma(x)G(x)\) and (2) \(F(x+1)=x^ xF(x)\). The unique log-convex solution of (1) with \(G(1)=1\) is the Barnes function \[ G(x)=\lim_{n\to\infty}{\Gamma(1)\Gamma(2)\cdots \Gamma(n)[\Gamma(n+1)\Gamma(n+x)]^{x/2}\over \Gamma(x)\Gamma(x+1)\cdots \Gamma(x+n)}. \] The unique log-convex solution of (2) with \(\Gamma(1)=1\) is given by \[ F(x)=\lim_{n\to\infty}{(1^ 12^ 2\cdots n^ n)(n+1)^{x(n+1)/2}(n+x)^{x(n+x)/2}\over x^ x(x+1)^{x+1}\cdots(x+n)^{x+n}}. \] As a Corollary \(F(x)=\Gamma^{x- 1}(x)/G(x)\). As the author comments these two functional equations are of the form \(f(x+1)=g(x)f(x)\) considered by W. Krull but the condition \(\lim_{x\to\infty}g(x+1)/g(x)=1\) fails for \(g(x)=\Gamma(x)\) or \(g(x)=x^ x\). Thus, W. Krull's results do not apply.