Effective error estimation for model reduction with inhomogeneous initial conditions

TL;DR

This paper introduces a new error estimator for model reduction that effectively handles inhomogeneous initial conditions without restrictive subspace constraints, applicable to various reduction methods and demonstrated to be more effective.

Contribution

The paper proposes a versatile error estimator for model reduction with inhomogeneous initial conditions, removing previous restrictions and improving estimation effectiveness.

Findings

The estimator applies to a broad class of reduction methods.

Numerical results show improved error estimation accuracy.

The approach eliminates the need for low-dimensional initial condition subspace.

Abstract

A priori error bounds have been derived for different balancing-related model reduction methods. The most classical result is a bound for balanced truncation and singular perturbation approximation that is applicable for asymptotically stable linear time-invariant systems with homogeneous initial conditions. Recently, there have been a few attempts to generalize the balancing-related reduction methods to the case with inhomogeneous initial conditions, but the existing error bounds for these generalizations are quite restrictive. Particularly, it is required to restrict the initial conditions to a low-dimensional subspace, which has to be chosen before the reduced model is constructed. In this paper, we propose an estimator that circumvents this hard constraint completely. Our estimator is applicable to a large class of reduction methods, whereas the former results were only derived for certain specific methods. Moreover, our approach yields to significantly more effective error estimation, as also will be demonstrated numerically.