Searching for extensible Korobov rules (Q2465292)

In this paper the authors consider the extensible Korobov lattice rules proposed by \textit{F. J. Hickernell}, \textit{H. S. Hong}, \textit{P. L'Écuyer} and \textit{C. Lemieux} [SIAM J. Sci. Comput. 22, No. 3, 1117--1138 (2000; Zbl 0974.65004)] both from a theoretical and practical point of view, emphasizing on two-dimensional projections. They prove the existence of ``good'' extensible Korobov lattice rules with respect to a quality measure that considers two-dimensional projections. That is, for any prime \(p\) and for any \(\alpha> 1\), \(v\in[1,\alpha)\) and \(\varepsilon> 0\), there exists a generator \(a\) such that the order-2 weighted \({\mathcal P}_\alpha(a)={\mathcal M}_{\alpha,n,s}(a)\) defined by \textit{F. J. Hickernell} [Random and quasi-random point sets, Lect. Notes Stat. 138, 109--166 (1998; Zbl 0920.65010)] satisfies \[ {\mathcal M}_{\alpha, n,s}(a)\leq c(s,\varepsilon)(\log(\log n+ 1))^{v(1+ \varepsilon)}(\log n)^v n^{-v}, \] for \(n= p,p^2,\dots\), and \(s= 1,2,\dots\), where \(c(s,\varepsilon)> 0\) is a constant. The numerical results of experiments made on different problems are also given (application to the pricing of Asian and digital options), suggesting that the rules found here can outperform previously published rules.