A remark on conditions that a diffusion in the natural scale is a martingale (Q1750373)

Let \(-\infty\leq a< b\leq+\infty\), \(m\) a Borel measure with support \((a,b)\), and \((\{X_t\}_{g\geq 0},\;\{P_x\}_{x\in(a,b)})\) the minimal diffusion on \((a,b)\). Define \(\tau_c:=\text{inf}\{t\geq 0\mid X_t= c\}\), \(\tau(x):= \lim_{c\to x}\tau_c\), \(\tau:= \tau(a)\wedge\tau(b)\), and \(\text{inf\,} X^\sigma_t:X_{t\wedge\sigma}\). Let \(f_-\) denote the positive increasing solution of \(d^2f/dmdx=\lambda f\), \(\lambda>0\) (unique up to constants). The authors give comparatively simple, purely analytic proofs for the following two known facts: (1) \(\{X^\tau_t\}\) is a \(P_x\)-submartingale if and only if \(\lim_{z\to\infty} f_-(z)=\infty\). (2) Given that \(-\infty< a< b< +\infty\), \(\{X^{\tau(a)}_t)\) is a strict \(P_x\)-supermartingale if and only if for all \(x\in (a,b)\) \(\lim_{t\to\infty} X_{t\wedge\tau(a)}= a\). For the probabilistic approach see \textit{H. Hulley} and \textit{E. Platen}, Banach Cent. Publ. 83, 139--157 (2008; Zbl 1153.60381)] and \textit{A. Gushchin} et al. [Proc. Steklov Inst. Math. 287, 122--132 (2014; Zbl 1322.60161)].