A conditional limit theorem for generalized diffusion processes

Abstract

Let $\mathbf{X} = \{ X(t) :t \leq 0\}$ be a one-dimensional generalized diffusion process with initial state $X(0) > 0$, hitting time $\tau _{\mathbf{X}}(0)$ at the origin and speed measure m which is regularly varying at infinity with exponent $1/\alpha -1 > 0$. It is proved that, for a suitable function $u(c)$, the probability law of $\{ u(c)^{-1}X(ct) : 0 < t \leq 1\}$ conditioned by $\{\tau _{\mathbf{X}}(0) > c \}$ converges as $c \to \infty$ to the conditioned $2(1-\alpha )$-dimensional Bessel excursion on natural scale and that the latter is equivalent to the $2(1-\alpha )$-dimensional Bessel meander up to a scale transformation. In particular, the distribution of $u(c)^{-1}X(c)$ converges to the Weibull distribution. From the conditional limit theorem we also derive a limit theorem for some of regenerative process associated with $\mathbf{X}$.