All news

Bayesian neural operators for PDEs augmented with residual-based error estimates

The RI-BNO neural operator learns operator inadequacy from a posteriori error estimates of the governing PDE and folds the correction and its uncertainty into Bayesian training and prediction. Local error bounds are established for the residual-informed estimates under stability and smoothness assumptions. The motivation is repeated PDE solves for digital twins, parameter inference and optimal design, where direct simulation is too costly.

Why it matters

Calibrated uncertainty in surrogate solvers is critical where asset-management decisions are built on their output.

Relevant to DT products

Original headline
Learning inadequacy of Bayesian neural operators for PDE-constrained inference and design
Read the original: arXiv API — полнотекстовый поиск "digital twin"

Translation and summary are machine-generated from the source. The full article is not republished; its rights belong to the source.