Approximation of \(k\)-monotone functions (Q1268710): Difference between revisions

The author makes an interesting and important observation that a polynomial of degree \(k-1\) which interpolates a \(k\)-monotone function \(f\in L_p[-1,1]\), \(0<p\leq\infty\) at \(k\) points in a subinterval \([a,b]\), which is far from the end points \(\pm 1\), approximated \(f\) well, in adjacent intervals of the length \(b-a\) on each side of \([a,b]\), no matter what the distances between the interpolation points are. This phenomenon is in complete contrast to the situation with interpolating polynomials of a general continuous function, where the distance between the points of interpolation influences the degree of approximation in a major way. He then applies this to proving results on simultaneous approximation to \(f\) and its derivatives provided \(f\in W^r_p\) is \(k\)-monotone. This is especially interesting when \(0<p<1\) since we know that this range has all kinds of pathologies including the impossibility of simultaneous approximation. He also applies that to obtaining generalizations of estimates on the degree of approximation of convex functions to higher order \(k\)-monotonicity.