Exact \(L_{2}\)-small ball asymptotics of Gaussian processes and the spectrum of boundary-value problems (Q842397)

Let \(X=(X(t))_{0\leq t\leq 1}\) be a centered Gaussian process. The aim of the present paper is determine the asymptotic behavior of the small deviation probability \[ \mathbb P\left(\int_0^1|X(t)|^2\mathrm{d} t\leq\varepsilon^2\right) \] as \(\varepsilon\to 0\). Due to the Karhunen--Loève expansion for Gaussian processes the above small deviation probability coincides with \[ \mathbb P\left(\sum_{n=1}^\infty\lambda_n\xi_n^2\leq \varepsilon^2\right) \] where the \(\xi_n\) are i.i.d.~standard normal and the \(\lambda_n\) are the eigenvalues of the integral equation \[ \lambda y(t)=\int_0^1 G_X(t,s) y(s)\mathrm{d} s \] with covariance function \(G_X(t,s)=\mathbb E X(t) X(s)\). If \(G_X\) coincides with the Green function of a suitable boundary value problem, then, as mentioned in 2004 by \textit{A. I. Nazarov} and \textit{Ya. Yu. Nikitin} [Probab. Theory Relat. Fields 129, No. 4, 469--494 (2004; Zbl 1051.60041)] the behavior of the small deviation probability of \(X\) is completely determined by the corresponding boundary value problem. In the present paper the author extends this connection to a broader class of boundary value problems. This allows him to treat a larger class of Gaussian processes including those as the integrated generalized Slepian process, the integrated centered Wiener process or the integrated centered Brownian bridge. In all this cases the small deviation behavior of is determined with precise constants.