Chung's law for homogeneous Brownian functionals (Q976557)

For a linear Brownian motion \(\{B(t),\,t\geq 0\}\) starting at \(0,\) consider the functional \(X=\{X_t,\,t\geq 0\}\) defined by \[ X_t=\int_0^t V(B_s)ds,\quad t\geq 0, \] where \(V(x)=x^\alpha\) if \(x\geq 0\) and \(V(x)=-\lambda|x|^\alpha\) if \(x\leq 0,\) for some \(\alpha,\lambda >0.\) The authors' main theorem proves a first-order exponential large deviation rate for the two-sided exit time \(T_{ab}=\inf\{t>0,~X_t\notin(-a,b)\},\) with \(a,b>0,\) i.e., there exists a finite positive constant \(\mathcal K\) such that \[ \lim_{t\to\infty} t^{-1}\log {\mathbf P}\left[T_{ab}>t\right]=-{\mathcal K}.\tag{1.1} \] By self-similarity, for \(a=b=1,\) the latter rate results in a small ball probability estimate showing that there exists a positive constant \(\mathcal K'\) such that \[ \lim_{\varepsilon \to 0}\varepsilon^{-2/(\alpha +2)}\log {\mathbf P} \left[\|X\|_\infty <\varepsilon\right]=-{\mathcal K}',\tag{1.2} \] where \(\|\cdot\|\) denotes the supremum norm over \([0,1].\) As a consequence, a Chung-type law of the iterated logarithm can be deduced from (1.2), that is, one has \[ \liminf_{t\to\infty}\frac{\sup\{|X_s|,\,s\leq t\}}{\left(t/\log\log t\right)^{(\alpha +2)/2}} ={\mathcal K}_1^{(\alpha +2)/2}\quad\text{a.s.,} \] where \({\mathcal K}_1\) is the constant appearing in (1.1) when \(a=b=1.\)