{"id":"arxiv/2608.07210","url":"https://arxiv.org/abs/2608.07210","tags":["math.NA"],"score":0,"title":"Sparse-grids-like surrogate models enhanced with gradient information","domain":"math.NA","source":"arxiv","authors":["Andrea Bressan","Sofia Imperatore","Francesca Locatelli","Lorenzo Tamellini"],"reviews":0,"subject":"math.NA","summary":"This work concerns surrogate modeling for quantities of interest (QoI) arising from parametric non-linear partial differential equations (PDEs). More specifically, we consider extending the sparse-grids surrogate modeling approach to incorporate derivatives of the QoI with respect to the PDE parameters. We discuss why this operation is not straightforward and propose a hybrid approach in which a sparse-grid scheme provides the collocation points in the parameter domain and a suitable polynomial space, but the surrogate model is built with a least-squares approach. We showcase our approach on…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-08-07","reactions":{}}