An absorption problem for several Brownian motions (Q4027481)

[For the entire collection see Zbl 0745.00038.] Let \(B_ 0,B_ 1,\dots,B_ N\) be independent Brownian motions starting at \(0,b_ 1,\dots,b_ N\), respectively. Denote by \(P_ b\) the probability measure of \(B_ 0,B_ 1,\dots,B_ N\), given that \(B_ i(0)=b_ i\). The author proposes a number of upper and lower estimations for the tail distribution of stopping time \[ \tau=\inf\{t>0:B_ i(t)=B_ 0(t)\quad\text{for some } 1\leq i\leq N\}. \] For instance, if \(0<b_ i<a\), \(1\leq i\leq N\), then there exists some constant \(\alpha>0\) such that for large \(t\) \[ P_ b\{\tau>t\}\leq(a^{-2}t)^{-\alpha\log N}. \] On the other hand, if \(b_ i>c\geq 1\), then for each \(\varepsilon>0\) there exists an \(N_ 0(\varepsilon)\) such that \(P_ b\{\tau>t\}\geq(e^{-2}t)^{-(1+\varepsilon){1\over 4}\log N}\) for \(N>N_ 0(\varepsilon)\), \(e^{-2}t>t_ 1=t_ 1(N)\).