On the balancing principle for some problems of Numerical Analysis

Abstract

We discuss a choice of weight in penalization methods. The motivation for the use of penalization in computational mathematics is to improve the conditioning of the numerical solution. One example of such improvement is a regularization, where a penalization substitutes an ill-posed problem for a well-posed one. In modern numerical methods for PDEs a penalization is used, for example, to enforce a continuity of an approximate solution on non-matching grids. A choice of penalty weight should provide a balance between error components related with convergence and stability, which are usually unknown. In this paper we propose and analyze a simple adaptive strategy for the choice of penalty weight which does not rely on a priori estimates of above mentioned components. It is shown that under natural assumptions the accuracy provided by our adaptive strategy is worse only by a constant factor than one could achieve in the case of known stability and convergence rates. Finally, we successfully apply our strategy for self-regularization of Volterra-type severely ill-posed problems, such as the sideways heat equation, and for the choice of a weight in interior penalty discontinuous approximation on non-matching grids. Numerical experiments on a series of model problems support theoretical results.

On The Analog Of The Monotone Error Rule For Parameter Choice In The (Iterated) Lavrentiev Regularization

“…Another analog [23] of the modified discrepancy principle uses the sequence α i = α i−1 q with q > 1 as the popular balancing principle [12,13,17] and chooses α(δ) = α i , where i is the first index, for which…”

Analysis of Finite Element Domain Embedding Methods for Curved Domains using Uniform Grids

“…We first establish the estimates under the mere assumption that the original domain is Lipschitz. We then assume the domain boundary is smoothly curved and establish sharp estimates, and use the aforementioned regularization/penalty error estimate to determine the balance between the mesh-size and regularization/penalty parameter such that the overall accuracy of the finite element domain embedding method is optimized [20,19]. We shall see that the balance is rather delicate in some cases.…”

On Estimating Differential Conductance from Noisy I-V Measurements in Delineating Device Parameters

“…Literature abounds in techniques of choosing optimal value of regularization parameters; however an optimal regularization parameter which makes a trade-off or in other words, balances the smoothness constraint and the data fitting is possibly the suitable one. Recently, selection of regularization parameter based on balancing principle is getting an increased attention [9][10][11]. However, as demonstrated in [11], the method requires at least an ad-hoc estimation of the noise level in the data.…”