On non-existence of certain error formulas for ideal interpolation (Q979050)

Let \({\Bbbk }\) be either the real field \({\mathbb R}\) or the complex field \({\mathbb C}\) and let \({\Bbbk}[{\mathbf x}]:={\Bbbk}[x_1,\ldots,x_d]\) be the ring of polynomials in \(d\) variables with coefficients in \({\Bbbk}\). A linear idempotent operator \(P\) on \({\Bbbk}[{\mathbf x}]\) is called an ideal projector if \(\ker P\) is an ideal in \({\Bbbk}[{\mathbf x}]\). Recently, \textit{C. de Boor} [Approximation Theory XI: Gatlinburg 2004, Nashboro Press, pp.\ 59--91 (2005; Zbl 1126.41003)] asked whether an arbitrary ideal projector \(P\) onto the subspace of polynomials of degree at most \(n\) has a ``good'' error formula, in the sense that the error \(q-Pq\) admits a certain representation in terms of an ideal basis of \(\ker P\), for all \(q\in {\Bbbk}[{\mathbf x}]\). The present paper proves that the answer is negative for the ideal projector from \(k[x,y]\) onto the space of linear polynomials in two variables whose kernel is the ideal generated by the polynomials \(x^2-y\), \(xy\), and \(y^2\).