Approximation of weak adjoints by reverse automatic differentiation of BDF methods

DOI10.1007/S00211-013-0570-4MaRDI QIDQ2454029zbMATH OpenOpenAlexFDO

Authors Dörte Beigel, M. S. Mommer, Leonard Wirsching, Hans Georg Bock

Publication date 12 June 2014

Published in Numerische Mathematik (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1109.3061

zbMATH Keywords

backward differentiation formula (BDF) methodsBanach spaceconstrained variational problemconvergencediscrete adjointsHilbert spacesinfinite-dimensional constrained variational problemnumerical resultPetrov-Galerkin discretizationreverse automatic differentiation

Mathematics Subject Classification ID

Nonlinear ordinary differential equations and systems (34A34) Sensitivity, stability, well-posedness (49K40) Multistep, Runge-Kutta and extrapolation methods for ordinary differential equations (65L06) Stability and convergence of numerical methods for ordinary differential equations (65L20) Finite element, Rayleigh-Ritz, Galerkin and collocation methods for ordinary differential equations (65L60)

Abstract: With this contribution, we shed light on the relation between the discrete adjoints of multistep backward differentiation formula (BDF) methods and the solution of the adjoint differential equation. To this end, we develop a functional-analytic framework based on a constrained variational problem and introduce the notion of weak adjoint solutions. We devise a finite element Petrov-Galerkin interpretation of the BDF method together with its discrete adjoint scheme obtained by reverse internal numerical differentiation. We show how the finite element approximation of the weak adjoint is computed by the discrete adjoint scheme and prove its asymptotic convergence in the space of normalized functions of bounded variation. We also obtain asymptotic convergence of the discrete adjoints to the classical adjoints on the inner time interval. Finally, we give numerical results for non-adaptive and fully adaptive BDF schemes. The presented framework opens the way to carry over the existing theory on global error estimation techniques from finite element methods to BDF methods.

Review text

The authors discuss the approximation of weak adjoints by reverse automatic differentiation of backward differentiation formula (BDF) methods. They first recall the adjoint differential equation as part of optimality conditions of an infinite-dimensional constrained variational problem in Hilbert spaces. They embed the constrained variational problem into the Banach space of continuously differentiable functions and extend the settings to the space of all functions that are continuous and piecewise continuously differentiable. They prove the convergence of the discrete adjoints to the classical solution of the adjoint differential equation on the inner time interval. Numerical results are provided.

Recommendations

• scientific article; zbMATH DE number 5353260
• On the discrete adjoints of adaptive time stepping algorithms
• Efficient goal-oriented global error estimation for BDF-type methods using discrete adjoints.
• Efficient goal-oriented global error estimators for BDF methods using discrete adjoints
• On the Properties of Runge-Kutta Discrete Adjoints

Cites work

• A Posteriori Error Estimation and Global Error Control for Ordinary Differential Equations by the Adjoint Method
• Adjoint sensitivity analysis for differential-algebraic equations: algorithms and software
• Adjoint-based algorithms and numerical methods for sensitivity generation and optimization of large scale dynamic systems.
• Automatic differentiation of explicit Runge-Kutta methods for optimal control
• Convergence rates for adaptive approximation of ordinary differential equations
• Error Estimates and Adaptive Time-Step Control for a Class of One-Step Methods for Stiff Ordinary Differential Equations
• Functional analysis
• scientific article; zbMATH DE number 1805738 (Why is no real title available?)
• scientific article; zbMATH DE number 5353260 (Why is no real title available?)
• scientific article; zbMATH DE number 4009690 (Why is no real title available?)
• scientific article; zbMATH DE number 42145 (Why is no real title available?)
• scientific article; zbMATH DE number 3470718 (Why is no real title available?)
• scientific article; zbMATH DE number 3541848 (Why is no real title available?)
• scientific article; zbMATH DE number 3634008 (Why is no real title available?)
• scientific article; zbMATH DE number 780776 (Why is no real title available?)
• scientific article; zbMATH DE number 2208228 (Why is no real title available?)
• scientific article; zbMATH DE number 3273462 (Why is no real title available?)
• scientific article; zbMATH DE number 3281211 (Why is no real title available?)
• scientific article; zbMATH DE number 3337815 (Why is no real title available?)
• scientific article; zbMATH DE number 3347071 (Why is no real title available?)
• Numerical Solution of Ordinary Differential Equations
• On Global Error Estimation and Control for Initial Value Problems
• Optimal control theory
• Ordinary differential equations.
• Solving Ordinary Differential Equations I

Cited in

(6)

• On the discrete adjoints of adaptive time stepping algorithms
• Discrete adjoint implicit peer methods in optimal control
• Efficient goal-oriented global error estimators for BDF methods using discrete adjoints
• Efficient goal-oriented global error estimation for BDF-type methods using discrete adjoints.
• Implicit peer triplets in gradient-based solution algorithms for ODE constrained optimal control
• Variable-stepsize implicit peer triplets in ODE constrained optimal control

Describes a project that uses

Uses Software

• DAESOL-II

This page was built for publication: Approximation of weak adjoints by reverse automatic differentiation of BDF methods

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2454029)