Êtes-vous un étudiant de l'EPFL à la recherche d'un projet de semestre?
Travaillez avec nous sur des projets en science des données et en visualisation, et déployez votre projet sous forme d'application sur GraphSearch.
We propose and analyse randomized cubature formulae for the numerical integration of functions with respect to a given probability measure μ defined on a domain Γ⊆ℝ^d, in any dimension d. Each cubature formula is conceived to be exact on a given finite dimensional subspace V_n⊂L^2(Γ,μ) of dimension n, and uses pointwise evaluations of the integrand function φ:Γ→ℝ at m>n independent random points. These points are distributed according to a suitable auxiliary probability measure that depends on V_n. We show that, up to a logarithmic factor, a linear proportionality between m and n with dimension-independent constant ensures stability of the cubature formula with very high probability. We also prove error estimates in probability and in expectation for any n≥1 and m>n, thus covering both pre-asymptotic and asymptotic regimes. Our analysis shows that the expected cubature error decays as √(n/m) times the L^2(Γ,μ)-best approximation error of φ in V_n. On the one hand, for fixed n and m→∞ our cubature formula can be seen as a variance reduction technique for a Monte Carlo estimator, and can lead to enormous variance reduction for smooth integrand functions and subspaces V_n with spectral approximation properties. On the other hand, when we let n,m→∞, our cubature becomes of high order with spectral convergence. Finally we show that, under a more demanding (at least quadratic) proportionality between m and n, the weights of the cubature are positive with very high probability. As an example of application, we discuss the case where the domain Γ has the structure of Cartesian product, μ is a product measure on Γ and the space V_n contains algebraic multivariate polynomials.