A problem about the difference of functions with equidistant knots

From MaRDI portal





The authors extend a result due to \textit{A. G. Azpeitia} [Am. Math. Mon. 89, 311--312 (1982; Zbl 0597.41033)]. If \(f\) exists in a neighborhood of \(a\), if \(f\) is continuous at \(a\), if \(f(a) \neq 0\) and if \(\xi\) is determined by the mean value theorem \(f (x) = f (a) + f' (\xi)(x - a)\), then \(\lim_{x\to a} {(\xi-a)}/{(x-a)}=1/2\). If, in this result, one sets \(h = x - a\), then the first difference, of \(f\), at \(a, a + h\) is given by \(f (a) = f (a + h) - f (a) = f' (\xi)h\), and one has \(\lim {(\xi-a)}/{h}=1/2\). The authors' generalization of the aforementioned result uses a relationship between the \(m\)th difference of \(f\) at \(x_0 , x_1 ,\dots , x_m\) and the \(m\)th derivative of \(f\); viz., \(\Delta^m f (x_0) = h^m f^{(m)} (\eta)\), where each \(x_k\) lies in \([a, b]\), \(x_k = a + kh\), \(\forall h\), and \(\eta\in (a, b)\). Theorem. If \(f\in C^{m+1}([a, b])\); \(f ^{(m+1)} (a) \neq 0\), \(h = (b - a)/m\), \(x_k = a + kh\), \(k = 0,\dots , m\), and \(\eta\) is taken from the above difference relation, then \(\lim_{h\to 0^+}{(\eta-a)}/{h}=m/2\).











This page was built for publication: A problem about the difference of functions with equidistant knots

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q865295)