Signed QoI Gaps for Goal-Oriented Error Estimation: Reliability and Sharpness from the Same First-Order System Residuals
Shun Zhang
Abstract
There are two common ways to estimate the error in a quantity of interest (QoI). The product of primal and adjoint error estimators gives a reliable but pessimistic bound. The dual-weighted residual method gives a sharp signed estimate from a more accurate approximation of the adjoint solution. We write the primal and adjoint problems as first-order systems and obtain both a reliable bound and a sharp estimate from the same residuals. The first paper of this series (arXiv:2607.23850) uses two consistent evaluators of a quantity of interest, one built from the potentials and one in which the gradients are expressed through the fluxes. We show that they differ exactly by the A-1 inner product of the primal and adjoint constitutive residuals, and we call this difference the signed QoI gap. The identity holds for arbitrary primal and adjoint approximations. In our numerical experiments, the signed QoI gap is of the size of the larger of the errors of the two evaluators, while the product of the estimators exceeds this error by factors of up to several hundred. We therefore use the least-squares estimators to refine the mesh and the signed QoI gap to decide when to stop. We also introduce a third evaluator, built from the fluxes alone, which restores the reaction term. The resulting reaction-restored QoI gap estimates the error when reaction dominates. With the non-intrusive least-squares flux recovery, we can apply the method to any discretization that produces a conforming potential, including finite difference and finite volume schemes.
Create a lesson
Related papers
Stochastic counterpart of Nesterov's method for sampling and provable non-acceleration of samplers with heavy-ball structure
A. Sharma
A time-domain adversarial example demonstrating the failure of H1 elements in a-ϕ electromagnetic simulation
Timothy Carlson, Sanjay Govindjee
Perturbation Algorithm and Iterative Realization of a Four-Layer Semi-Discrete Scheme for an Evolutionary Problem with a Variable Operator
David Gulua, Jemal Rogava, Zurab Vashakidze
Regularity transfer for the Ritz projection
Lars Diening, Julian Rolfes, Abner J. Salgado
High-Order Area-Preserving Runge--Kutta Methods for Surface Diffusion Flow of Curves
Lei Cui, Gengen Zhang
Time-independent equivalent matrix representation for time-domain boundary element analysis of large-scale 3D acoustic problems
Weiyu Zhou, Yongqiang Chen