The Gasca-Maeztu conjecture for \(n=5\) (Q398759): Difference between revisions

Let \(\Pi_n\) denote the real linear space of bivariate polynomials with total degree \(\leq n.\) The dimension of \(\Pi_n\) is \(N=\binom{n+2}{2}\). Let \(\mathcal X_s=(X_1,X_2,\dots,X_s)\) be a set of distinct points (interpolation nodes) in \(\mathbb R^2\). The set \(\mathcal X_s\) is called \(\Pi_n\)-poised (or \(n\)-poised) if for any data \((c_1,c_2,\dots,c_s)\) there is a unique \(\overline p\in\Pi_n\) such that \(\overline p(X_j)=c_j,\, j=1,2,\dots,s.\) A necessary condition for \(n\)-poisedness is \(s=\#\mathcal X_s = \)\,dim\,\(\Pi_n=N.\) Let \(\mathcal X\) be a set of nodes. A polynomial \(p\in\Pi_n\) is called fundamental for \(A\in\mathcal X\), if \(p(A)=1\) and \(p|_{\mathcal X\smallsetminus \{A\}}=0.\) Such a polynomial is denoted by \(p^*_A.\) The authors are interested in \(n\)-poised sets with the property that all fundamental polynomials are product of linear factors. \textit{M. Gasca} and \textit{J. I. Maeztu} [Numer. Math. 39, 1--14 (1982; Zbl 0457.65004)], conjectured that every such set must contain \(n+1\) colinear nodes. This conjecture had been confirmed for \(n\leq 4.\) Using some new methods, which, in their opinion, might be useful in deciding the still open case \(n\geq 6,\) the authors prove in this paper the validity of the conjecture for \(n=5.\)