On the conditional distributions and the efficient simulations of exponential integrals of Gaussian random fields (Q2511562)

The authors consider a Gaussian random field \(f(t)\) on a compact set \(T\). The main aim is to study its tail behavior, more precisely, \(p_b=P(\int_{T}\exp\{f(t)\}\,dt>b)\). The main result is an algorithm that runs in polynomial time of \(\log\,b\) and estimate \(p_b\) with prescribed relative accuracy. Central to the analysis is the construction of a change of measure on the space \(C(T)\)c of continuous functions on \(T\). The proposed change of measure is not a classical exponential tilting form. The newly proposed measure has several appealing features both from the theoretical and the practical point of view. Another contribution lies in the numerical evaluation of \(v(b)=P(\int_T \exp\{\sigma f(t)+\mu(t)\}\,dt>b)\), where \(\sigma>0\) and \(\mu(t)\) is a deterministic function. The key requirement is the twice differentiability of \(f(t)\).