On convex perturbations with a bounded isotropic constant (Q866883)

A convex body \(K\subset\mathbb{R}^n\) is in isotropic position if it has volume \(1\), its barycentre is at the origin and \([\int_K x_ix_j dx]_{i,j}=L_K^2 I_n\) for some \(L_K>0\), called the isotropic constant of \(K\) (\(I_n\) the identity matrix). It is known that for any convex body \(K\) there is an affine map \(T\) on \(\mathbb{R}^n\) such that \(T(K)\) is in isotropic position, and we define the isotropic constant of \(K\) to be \(L_K=L_{T(K)}\). The hyperplane conjecture (also known as the slicing problem) asserts that any convex body of volume \(1\) has a hyperplane slice \(K\cap H\) of volume bounded below by a universal constant \(c>0\). This is known to be equivalent to the conjecture that the isotropic constants of all convex bodies are bounded above by a universal constant. The previously best known upper bound for the isotropic constant was \(L_K < cn^{1/4}\log n\), due to \textit{J. Bourgain} [Lect.\ Notes Math.\ 1469, 127--137 (1991; Zbl 0773.46013)]. In the paper under review the upper bound is improved to \(L_K < cn^{1/4}\). This improvement follows from a theorem of \textit{G. Paouris} [Geom.\ Funct.\ Anal.\ 16, 1021--1049 (2006; Zbl 1114.52004)] that if the Banach-Mazur distance of two convex bodies is less than \(1+1/\sqrt{n}\) then their isotropic constants have the same order, in combination with the main result of the paper: For each convex body \(K\subset\mathbb{R}^n\) and \(\varepsilon>0\) there exists a convex body \(\tilde{K}\subset\mathbb{R}^n\) at Banach-Mazur distance less than \(1+\varepsilon\) to \(K\) and with isotropic constant almost universally bounded in the sense that \(L_{\tilde{K}}<c/\sqrt{\varepsilon}\). The proof uses a certain measure transportation.