Kolmogorov criterion for uniform approximation
ID: kolmogorov-criterion-for-uniform-approximation
Let be a compact Hausdorff space, a real linear subspace of , and . Put and . Then is a best uniform approximation from precisely when every has some with .[ref]
Math/CS 887: Approximation Theory by Carl de Boor. Kolmogorov criterion (47), printed p.29, real specialization
If a direction has strictly matching error signs everywhere on , compactness makes a uniformly better approximant for sufficiently small . Conversely, for any candidate , an active point with has , preventing improvement. The linear subspace hypothesis ensures that differences of candidates are admissible directions; for a general nonlinear family, the quantified direction set would have to change. This criterion yields the Chebyshev alternation theorem.
New to topics? Read the docs here!