An extremal limit theorem for the argmax process of Brownian motion minus a parabolic drift (Q1297901)

Let \(V(t)\) be a location of the maximum of Brownian motion minus a parabolic drift, i.e. \(V(t)=\hbox{argmax}\{x:W(x)-(x-t)^2\}\), where \(W(t)\) is a standard Brownian motion. The extremal behavior of \(\xi(t)=V(t)-t\) is studied. Note that \(\xi(t)\) is a stationary Markov process but its sample paths are not a.s. continuous. {Main Theorem.} If \(u_t\to\infty\) as \(T\to\infty\), then \[ \Pr\{\sup_{t\in[0,T]}\xi(t)\leq u_T\}\to e^{-\tau}\quad\text{and}\quad\Pr\{\sup_{t\in[0,T]}| \xi(t)| \leq u_T\}\to e^{-2\tau} \] iff \(Tf(u_T)\to\tau>0\), where \(f\) is the density of the r.v.\(V(0)\). The tail behavior of \(f\) is given by \(f(x)\sim\lambda| x| \exp(-2/3| x| ^3- \kappa| x|)\) as \(| x| \to\infty\), \(\lambda=4^{1/3}\text{Ai}'(a_1)\), \(\kappa=2^{1/3}| a_1| \), \(a_1\approx -2.3381\) is the largest zero of the Airy function Ai.