Approximate Optimal Designs for Multivariate Polynomial Regression - LAAS - Laboratoire d'Analyse et d'Architecture des Systèmes Access content directly
Journal Articles Annals of Statistics Year : 2019

Approximate Optimal Designs for Multivariate Polynomial Regression


We introduce a new approach aiming at computing approximate optimal designs for multivariate polynomial regressions on compact (semi-algebraic) design spaces. We use the moment-sum-of-squares hierarchy of semidefinite programming problems to solve numerically the approximate optimal design problem. The geometry of the design is recovered via semidefinite programming duality theory. This article shows that the hierarchy converges to the approximate optimal design as the order of the hierarchy increases. Furthermore, we provide a dual certificate ensuring finite convergence of the hierarchy and showing that the approximate optimal design can be computed numerically with our method. As a byproduct, we revisit the equivalence theorem of the experimental design theory: it is linked to the Christoffel polynomial and it characterizes finite convergence of the moment-sum-of-square hierarchies.
Fichier principal
Vignette du fichier
AoSv2 (2).pdf (1018.79 Ko) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-01483490 , version 1 (06-03-2017)
hal-01483490 , version 2 (17-10-2017)



Yohann de Castro, Fabrice Gamboa, Didier Henrion, Roxana Hess, Jean-Bernard Lasserre. Approximate Optimal Designs for Multivariate Polynomial Regression. Annals of Statistics, 2019, 47 (1), pp.127-155. ⟨10.1214/18-AOS1683⟩. ⟨hal-01483490v2⟩
638 View
79 Download



Gmail Mastodon Facebook X LinkedIn More