Hitting times of interacting drifted Brownian motions and the vertex reinforced jump process (Q784160)

At the beginning of the paper, the authors recall some classic facts about hitting times of standard Brownian motion \({(B(t))_{t \ge 0}}\). If \(X(t) = \theta + B(t) - \eta t\) is a drifted Brownian motion with negative drift \( - \eta < 0\) starting at \(\theta > 0\), then \(T = \inf \{ t \ge 0:X(t) = 0)\) has the inverse Gaussian distribution with parameters \((\frac{\theta }{\eta },{\theta ^2})\), that is, \(T\) has density \(f(t) = \frac{\theta }{{\sqrt {2\pi {t^3}} }}\exp ( - \frac{1}{2}(\frac{{{\theta ^2}}}{t} + {\eta ^2}t - 2\eta \theta ))\), \(t > 0\). Conditionally on \(T\), \({({X_t})_{0 \le t \le T}}\) has the law of a three-dimensional Bessel bridge from \(\theta \) to 0 on time interval \([0,T]\). The paper gives a generalization of these results on a conductance network, namely, for a family of Brownian motions with interacting drifts indexed by the vertices of the network. The distribution of hitting times of these processes are given by a multivariate exponential family of distributions appearing in the context of self-interacting processes and random Schrödinger operators.