Efficient Rare-event Simulation for the Maximum of Heavy-tailed Random Walks (with P. Glynn).

Summary:

This paper is the first to propose the use of Lyapunov bounds for controlling the error of importance sampling estimators. The technique is applied to solving the problem of estimating first passage time probabilities of in heavy-tailed random walks (basically assuming only subexponential distributions). The technique has then been applied to other heavy-tailed problems.

Bibtex:

@Article{BlanGly08HTRW,

author = {Blanchet, J. and Glynn, P.},

title = {Efficient rare-event simulation for the maximum of heavy-tailed random walks},

journal = {Annals of Applied Probability},

year = {2008},

volume = {18},

pages = {1351-1378}

}