Where do homogeneous polynomials on _1ⁿ attain their norm?

From MaRDI portal
Publication:1826871





In 2001, Zalduendo conjectured that almost all homogeneous polynomials on \(\ell_\infty^n\) will attain their norms at vertices of the unit ball when \(n\) tends to infinity. In this article, the authors show that this conjecture fails for \(\ell_1^n\). Namely, using the normalised Lebesgue measure \(\mu_n\) on the unit ball of the space of symmetric bilinear forms on \(\ell_1^n\), it is shown that for \(C=\{P \in {\mathcal P}(^2 \ell_1^n)\) such that \(\| A\|=1\) and \(P\) does not attain its norm at a vertex\(\}\), we have \(\mu_n(C) \geq 1/6\) (here \(A\) is the symmetric bilinear form associated with \(P\) by the polarisation formula). However, in the main result of the article, it is shown that most of the polynomials \(P \in {\mathcal P}(^2 \ell_1^n)\) attain their norm, in the low dimensional faces of the unit ball of \(\ell_1^n\). If \(S_n^m=\{P \in {\mathcal P}(^2 \ell_1^n)\) such that \(\|A\|=1\) and \(P\) attains its norm in the interior of a \((m-1)\)-face\(\}\), then \(\lim_{n \rightarrow \infty} \mu_n (\bigcup_{m>16\sqrt n} S_n^m) = 0\).











This page was built for publication: Where do homogeneous polynomials on \(\ell_{1}^{n}\) attain their norm?

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1826871)