1: \begin{abstract}

2: Model reduction attempts to guarantee a desired ``model quality'', e.g. given in terms of

3: accuracy requirements, with as small a model size as possible. This article highlights some recent

4: developments concerning this issue for the so called Reduced Basis Method (RBM) for models based

5: on parameter dependent families of PDEs. In this context the key task is to sample the {\em solution manifold}

6: at judiciously chosen parameter values usually determined in a {\em greedy fashion}.

7: The corresponding {\em space growth} concepts are closely related to so called {\em weak greedy} algorithms in Hilbert and Banach spaces

8: which can be shown to give rise to convergence rates comparable to the best possible rates, namely the {\em Kolmogorov $n$-widths} rates.

9: Such algorithms can be interpreted as {\em adaptive sampling} strategies for approximating compact sets in Hilbert spaces.

10: We briefly discuss the results most relevant for the present RBM context.

11: The applicability of the results for weak greedy algorithms has however been confined so far essentially to

12: well-conditioned coercive problems. A critical issue is therefore an extension of these concepts to a wider range of problem classes for which the conventional methods

13: do not work well. A second main topic of this article is therefore to outline recent developments of RBMs that do realize $n$-width rates

14: for a much wider class of variational problems covering indefinite or singularly perturbed unsymmetric problems.

15: A key element in this context is the design of {\em well-conditioned variational formulations}

16: and their numerical treatment via saddle point formulations. We conclude with some remarks concerning the relevance

17: of uniformly approximating the whole solution manifold also when the {\em quantity of interest} is only of a {\em functional} of the parameter dependent solutions.

18: \end{abstract}

19: