Compressed Multiscale Kernel Collocation and Physics-Informed Kernel Learning for Elliptic BVPs
Abstract
Kernel collocation is a meshfree method for partial differential equations whose accuracy depends on the smoothness of the kernel and of the solution. For globally supported kernels, however, the collocation matrices are dense and severely ill-conditioned. We address both issues by combining a multiscale collocation scheme with samplets, a multiresolution basis of discrete signed measures with vanishing moments. In the samplet basis, the collocation matrix has only O(N log N) relevant entries, while the multiscale scheme adapts the kernel width to the spacing of the points on every level. Since the differential operator scales the equations at the interior points differently from those at the boundary points, the rescaling alone does not bound the condition number of the level systems. We therefore solve them by a Schur-complement reduction that is invariant under this scaling. For noisy data, we consider physics-informed kernel learning, which replaces the collocation system by a regularized least-squares fit. We show that it reduces to collocation as the regularization vanishes, and we solve it by a deflated conjugate gradient method that reuses the Schur-complement reduction. Numerical experiments in two and three dimensions, with exact and noisy boundary data, show that the method handles significantly more points than dense collocation.
Related
- High-order smoothness and high-order collocation approximation for Volterra integral equation with multiscale and nonlinear exponent kernel
- Efficient and Fast Tensor-Product Multinode Shepard Collocation for Elliptic PDEs on Cartesian Grids
- Well-Conditioned Birkhoff-Collocation Methods for Elliptic-type Problems in Multiple Dimensions
- Residual-Based Adaptive Refinement in Hybrid Variational-Collocation PINNs for Singular Elliptic PDEs with Mixed Dirichlet-Neumann Boundary Conditions
- Physics-informed traditional and generative Mori-Zwanzig-reduced models for multiscale systems
BibTeX
@misc{avesani2026compressed,
title = {{Compressed Multiscale Kernel Collocation and Physics-Informed Kernel Learning for Elliptic BVPs}},
author = {Sara Avesani and Manel Babay and Angela Kunoth and Michael Multerer},
howpublished = {arXiv},
year = {2026},
month = nov,
doi = {10.48550/arxiv.2611.02832},
eprint = {2611.02832},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2611.02832},
}
RIS
TY - UNPB
TI - Compressed Multiscale Kernel Collocation and Physics-Informed Kernel Learning for Elliptic BVPs
AU - Sara Avesani
AU - Manel Babay
AU - Angela Kunoth
AU - Michael Multerer
PY - 2026
DA - 2026/11/02
PB - arXiv
DO - 10.48550/arxiv.2611.02832
UR - https://arxiv.org/abs/2611.02832
AB - Kernel collocation is a meshfree method for partial differential equations whose accuracy depends on the smoothness of the kernel and of the solution. For globally supported kernels, however, the collocation matrices are dense and severely ill-conditioned. We address both issues by combining a multiscale collocation scheme with samplets, a multiresolution basis of discrete signed measures with vanishing moments. In the samplet basis, the collocation matrix has only O(N log N) relevant entries, while the multiscale scheme adapts the kernel width to the spacing of the points on every level. Since the differential operator scales the equations at the interior points differently from those at the boundary points, the rescaling alone does not bound the condition number of the level systems. We therefore solve them by a Schur-complement reduction that is invariant under this scaling. For noisy data, we consider physics-informed kernel learning, which replaces the collocation system by a regularized least-squares fit. We show that it reduces to collocation as the regularization vanishes, and we solve it by a deflated conjugate gradient method that reuses the Schur-complement reduction. Numerical experiments in two and three dimensions, with exact and noisy boundary data, show that the method handles significantly more points than dense collocation.
ER -