Lengthscale-informed sparse grids for kernel methods in high dimensions
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866911260260433920 |
|---|---|
| author | Addy, Elliot J. Latz, Jonas Teckentrup, Aretha L. |
| author_facet | Addy, Elliot J. Latz, Jonas Teckentrup, Aretha L. |
| contents | Kernel interpolation, especially in the context of Gaussian process emulation, is a widely used technique in surrogate modelling, where the goal is to cheaply approximate an input-output map using a limited number of function evaluations. However, in high-dimensional settings, such methods typically suffer from the curse of dimensionality; the number of required evaluations to achieve a fixed approximation error grows exponentially with the input dimension. To overcome this, a common technique used in high-dimensional approximation methods, such as quasi-Monte Carlo and sparse grids, is to exploit functional anisotropy: the idea that some input dimensions are more 'sensitive' than others. In doing so, such methods can significantly reduce the dimension dependence in the error. In this work, we propose a generalisation of sparse grid methods that incorporates a form of anisotropy encoded by the lengthscale parameter in Matérn kernels. We derive error bounds and perform numerical experiments that show that our approach enables effective emulation over arbitrarily high dimensions for functions exhibiting sufficient anisotropy. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_07797 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Lengthscale-informed sparse grids for kernel methods in high dimensions Addy, Elliot J. Latz, Jonas Teckentrup, Aretha L. Numerical Analysis 41A25, 41A63, 62G08, 65B99, 65C20, 65D12, 65D15, 65D32, 65D40 Kernel interpolation, especially in the context of Gaussian process emulation, is a widely used technique in surrogate modelling, where the goal is to cheaply approximate an input-output map using a limited number of function evaluations. However, in high-dimensional settings, such methods typically suffer from the curse of dimensionality; the number of required evaluations to achieve a fixed approximation error grows exponentially with the input dimension. To overcome this, a common technique used in high-dimensional approximation methods, such as quasi-Monte Carlo and sparse grids, is to exploit functional anisotropy: the idea that some input dimensions are more 'sensitive' than others. In doing so, such methods can significantly reduce the dimension dependence in the error. In this work, we propose a generalisation of sparse grid methods that incorporates a form of anisotropy encoded by the lengthscale parameter in Matérn kernels. We derive error bounds and perform numerical experiments that show that our approach enables effective emulation over arbitrarily high dimensions for functions exhibiting sufficient anisotropy. |
| title | Lengthscale-informed sparse grids for kernel methods in high dimensions |
| topic | Numerical Analysis 41A25, 41A63, 62G08, 65B99, 65C20, 65D12, 65D15, 65D32, 65D40 |
| url | https://arxiv.org/abs/2506.07797 |