Abstract

We prove a-posteriori error-estimates for reduced-order modeling of quasilinear parabolic PDEs with non-monotone nonlinearity. We consider the solution of a semi-discrete in space equation as reference, and therefore incorporate reduced basis-, empirical interpolation-, and time-discretization-errors in our consideration. Numerical experiments illustrate our results.

Highlights

  • In the present paper we are concerned with a-posteriori error estimation for model order reduction applied to a semi-discrete in space quasilinear parabolic partial differential equation (PDE) with non-monotone nonlinearity

  • In case that hyperreduction of the nonlinearity is done by empirical interpolation (EIM) we provide estimates that include the additional empirical interpolation method (EIM)-error in Sect

  • In order to allow for an efficient offline-online splitting, the evaluation of nonlinearities in the reduced-order model for (Eq) needs to be done by methods of hyperreduction, e.g. the Empirical Interpolation Method (EIM, [8])

Read more

Summary

Introduction

In the present paper we are concerned with a-posteriori error estimation for model order reduction applied to a semi-discrete in space quasilinear parabolic partial differential equation (PDE) with non-monotone nonlinearity. We keep the summary of results concerning (Eq) on the continuous level rather short: We refer for instance to [9, 36, 50] for a discussion of Assumption 2.1 and only recall two regularity results from the literature that might be seen as motivation for exploiting the respective regularity of the semi-discrete in space solution lateron: Regularity in case of right hand sides in the slightly more regular Bessel potentialspaces H− ,p and appropriate initial regularity has been addressed in [9, section 3.2]. Applying their setting to (Eq) yields C0, (I, W1,∞)-regularity of the solutions with some α > 0 , cf. [36, Corollary 5.4]

Semi‐discretization in space and RB‐MOR
Outlook towards optimal control
A‐posteriori RB‐error‐estimates
Some preliminary calculations
A‐posteriori estimates—Approach I
A‐posteriori estimates—Approach II
A‐posteriori RB‐ and EIM‐error‐estimates
Empirical interpolation of A
A‐posteriori RB‐error‐estimates including the EIM‐error
Numerical illustration for POD‐MOR
Test problems and technical details
Discussion of the results
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call