1 paper · 1 filter
Milan Bacchetta, Tobias Ehring, Bernard Haasdonk
This paper considers kernel-based approximation methods for nonlinear partial differential equations. To this end, the problem is formulated as an optimal-recovery generalized inte…