Some applications of duality relations

From MaRDI portal





[For the entire collection see Zbl 0755.00015.] The main object of investigation in the paper is multidimensional (mostly symmetric) convex bodies, or, more precisely, affine classes of such bodies. (Two bodies are equivalent if there is an affine transform which maps one of them onto another.) It is well known that in many senses (e.g. with respect to Banach-Mazur distance) for large dimensions there are affine classes which are very far from each other. It turns out (and it is more surprising) that in some natural senses (usually of volumetric nature) the difference between affine classes of different bodies has a common bound for all dimensions. The series of such results (most of them are due to the author) was obtained in the period 1984-1985. A nice source to get acquainted with these results is the book: \textit{G. Pisier}, `The volume of convex bodies and Banach space geometry' (1989; Zbl 0698.46008). In the present paper some more results of this kind are proved. We formulate one of them: There is a numerical constant \(C\) such that for arbitrary \(n\in \mathbb{N}\), and an arbitrary symmetric convex body \(K\) in \(\mathbb{R}^ n\) there are two linear mappings \(u_ 1,u_ 2\in \text{SL}_ n\) (i.e. \(\text{det } u_ 1= \text{det } u_ 2= 1\)) satisfying the following conditions. Let \(K_ 1= K+ u_ 1 K\) and let \(K_ 2= K^ 0_ 1+ u_ 2(K^ 0_ 1)\) (where \(K^ 0_ 1\) is the polar of \(K_ 1\)). Then for some ellipsoid \(E\subset \mathbb{R}^ n\), \[ E\subset K_ 2\subset CE. \] The proofs of the results rely heavily on previous results. The main purpose of the paper is to derive consequences which may be of interest to mathematicians working outside the asymptotic theory of normed spaces. At the end of the paper the author describes some relations of the results of the asymptotic theory to a study of the growth of polynomials of many variables and to an estimation of a number of integer points in convex regions. The paper contains corrected versions of some statements of the paper: \textit{V. D. Milman}, Isr. Math. Conf. Proc. 3, 151-161 (1990; Zbl 0705.46005). Misprint: In the formulation of Kashin's result on p. 19 \(\ell_ 1\)-norm should be divided by \(\sqrt n\).











This page was built for publication: Some applications of duality relations

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4037237)