Instrumental Transfer Learning for Solution Operators
Abstract
Data-driven learning for the solution operator of a differential equation has broad scientific importance, but observations from a target system are often scarce and costly to collect. Related source systems may provide abundant observations under similar physical conditions, whereas even small changes in the governing equation can induce complex changes in the solution operator, making naive data pooling ineffective. We propose a structured transfer learning method that accounts for unknown sparse differences in the governing equations from a known library of candidate terms. Using the resolvent identity and a whitened instrumental variable approach, we construct unbiased source estimating equations for the target operator. Building on these equations, we develop a three-stage estimator that consists of a target-only ridge pilot, source-wise sparse calibration, and a projected refit designed to remove the leading calibration error inherited from the pilot. An iterative recalibration scheme further reduces the remaining error. We establish non-asymptotic prediction bounds showing when transfer achieves the pooled all-source rate, together with minimax lower bounds that characterize both rate optimality and fundamental identifiability limits under common shifts. Simulations and an electronic band-pass filter application demonstrate clear gains over target-only learning, naive pooling and existing transfer methods.
Related
- Stochastic Latent Twins: Learning Stochastic Solution Operators
- Learning PDE solution operators with variable initial conditions via Latent Dynamics Networks
- Instrumental Variable Analysis in Underrepresented Subpopulations Powered by Knowledge Transfer
- On the dynamics of self-consistent transfer operators for interval maps with singularities
- Repulsive Mean-Field Couplings and Self-Consistent Transfer Operators
BibTeX
@misc{li2026instrumental,
title = {{Instrumental Transfer Learning for Solution Operators}},
author = {Tongyu Li and Zhuoyuan Li and Ruoxu Tan},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.41262},
eprint = {2610.41262},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.41262},
}
RIS
TY - UNPB
TI - Instrumental Transfer Learning for Solution Operators
AU - Tongyu Li
AU - Zhuoyuan Li
AU - Ruoxu Tan
PY - 2026
DA - 2026/10/30
PB - arXiv
DO - 10.48550/arxiv.2610.41262
UR - https://arxiv.org/abs/2610.41262
AB - Data-driven learning for the solution operator of a differential equation has broad scientific importance, but observations from a target system are often scarce and costly to collect. Related source systems may provide abundant observations under similar physical conditions, whereas even small changes in the governing equation can induce complex changes in the solution operator, making naive data pooling ineffective. We propose a structured transfer learning method that accounts for unknown sparse differences in the governing equations from a known library of candidate terms. Using the resolvent identity and a whitened instrumental variable approach, we construct unbiased source estimating equations for the target operator. Building on these equations, we develop a three-stage estimator that consists of a target-only ridge pilot, source-wise sparse calibration, and a projected refit designed to remove the leading calibration error inherited from the pilot. An iterative recalibration scheme further reduces the remaining error. We establish non-asymptotic prediction bounds showing when transfer achieves the pooled all-source rate, together with minimax lower bounds that characterize both rate optimality and fundamental identifiability limits under common shifts. Simulations and an electronic band-pass filter application demonstrate clear gains over target-only learning, naive pooling and existing transfer methods.
ER -