Weighted version of multivariate Ostrowski type inequalities (Q5948864): Difference between revisions

This paper gives extensions of the Ostrowski inequality to several variables. The idea is as follows. Suppose that \(Q\) is a compact convex set in \(\mathbb{R}^m\) on which there is given a probability Radon measure \(\mu\). We want to estimate the deviation of a function \(f\in C^n(Q)\) from its mean value of order \(n\), i.e., the difference \[ {\mathcal O}_n(x,f,w)= \int_Q f(y) w(y) dy- \sum^n_{j=0} \sum_{|\alpha|=j} {D^\alpha f(x)\over \alpha!} E_\alpha(x;\mu), \] where \(E_\alpha(x;\mu)= \int_Q(y- x)^\alpha d\mu(y)\) represents the \(x\)-centered moment of order \(\alpha\) (of \(\mu\)). This can be done by integrating in the Taylor's formula, which yields \[ |{\mathcal O}_k(x,f;\mu)|\leq {1\over n!}\cdot \sup_{|\alpha|= n} \|D^\alpha f(\cdot)- D^\alpha f(x)\|_\infty\cdot \int_Q\|y- x\|^n_1 d\mu(y) \] and then estimating the last integral under convenient circumstances.