Applications of Grothendieck's inequality to linear symplectic geometry (Q2022920): Difference between revisions
From MaRDI portal
Set profile property. |
ReferenceBot (talk | contribs) Changed an Item |
||
(2 intermediate revisions by 2 users not shown) | |||
Property / OpenAlex ID | |||
Property / OpenAlex ID: W3080295537 / rank | |||
Normal rank | |||
Property / arXiv ID | |||
Property / arXiv ID: 2005.08596 / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Poisson brackets of partitions of unity on surfaces / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Q3239535 / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Grothendieck’s Theorem, past and present / rank | |||
Normal rank |
Latest revision as of 15:23, 25 July 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Applications of Grothendieck's inequality to linear symplectic geometry |
scientific article |
Statements
Applications of Grothendieck's inequality to linear symplectic geometry (English)
0 references
30 April 2021
0 references
\noindent Consider the bilinear anti-symmetric form \(J_{2n}\) defined on the Hilbert space \(\ell_2^{2n}=(\mathbb{R}^{2n}, \|\cdot\|_2)\) by \((u,v) \mapsto \langle u, J_{2n}v \rangle\), where \[ J_{2n} = \left( \begin{matrix} 0 & -\text{id}_{\mathbb{R}^n} \\ \text{id}_{\mathbb{R}^n} & 0 \end{matrix} \right) \] is the \(2n \times 2n\)-dimensional matrix corresponding to the multiplication by \(\pmb{ i}\) under the identification \(\mathbb{R}^{2n} \cong \mathbb{C}^n\). This is the standard example of a symplectic bilinear form, a basic notion in symplectic geometry. The main result shows that for all choices of finitely many vectors \(v_1, \ldots, v_N \in \mathbb{R}^{2n}\) \[ \sum_{i,j=1} |\langle v_i,J_{2n}v_j\rangle| \leq 3K_G \sqrt{2n} \sup_{|t_i|,|s_j|\leq 1} \Big\langle \sum_{i=1}^{N} t_iv_i,\sum_{j=1}^{N} s_jv_j \Big\rangle\,, \] where \(K_G>0\) denotes Grothendieck's constant (i.e., the best constant in Grothen\-dieck's famous inequality). This is a considerable improvement of a result of [\textit{L. Buhovsky} et al., Comment. Math. Helv. 95, No. 2, 247--278 (2020; Zbl 1454.53063)], who proved that the best constant \(c(n)>0\) in the preceding inequality grows at most exponentially in \(n\). An example shows that the growth \(c(n) = O (\sqrt{n})\) is even sharp. A second interesting result concerns the orbit of arbitrary finite collections \(v_1, \ldots, v_N \in \mathbb{R}^{2n}\) under the action of all symplectic matrices \(S\) on \(\mathbb{R}^{2n}\), i.e., all matrices \(S\) on \(\mathbb{R}^{2n}\) such that \(S^TJ_{2n}S = J_{2n}\). Grothendieck's inequality provides a sharp upper bound for \(\inf_S \sum_{j=1}^N \|v_j\|_{\ell_2^{2n}}\) in terms of the rank of the matrix \((\langle v_i, J_{2n}v_j \rangle)_{i,j=1}^N\) and its norm as an operator from \(\ell_\infty^{N}\) into~\(\ell_1^{N}\).
0 references
symplectic matrices
0 references
Grothendieck's inequality
0 references
spaces of matrices
0 references