Strong uniqueness in nonlinear approximation

In this paper two conditions, one analytic and one geometric, are shown to be equivalent to the so-called strong Kolmogorov criterion. These conditions are shown to be sufficient for a best approximation from an arbitrary set K to be strongly unique. The set K is called a strongly Kolmogorov set if the strong Kolmogorov criterion is also a necessity for a best approximation to be strongly unique. Some characterizations of strongly Kolmogorov sets are given; however an intrinsic characterization is still lacking. It is also shown that in a smooth space a set which is strongly Chebyshev and strongly Kolmogorov must be convex; but in general, the class of strongly Kolmogorov sets is strictly larger than the class of convex sets. As an application of some of the results, it is shown that best \(L_ 1\)- approximation of a continuous function by monotone polynomials need not be strongly unique. Meanwhile, the author has obtained further results characterizing strongly Kolmogorov sets, some of which parallel those on Kolmogorov sets. The intrinsic characterization is still elusive.

Recommendations

Cites work

Cited in

(9)

This page was built for publication: Strong uniqueness in nonlinear approximation

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q791788)