LYZ ellipsoid and Petty projection body for log-concave functions (Q1621476): Difference between revisions

The authors introduce functional versions of some geometric results for the class of log-concave functions. They first define LYZ ellipsoid for log-concave functions, which is an important tool in convex geometry and introduced to the literature first by E. Lutwak, D. Yang and G. Zhang [\textit{E. Lutwak} et al., Duke Math. J. 104, No. 3, 375--390 (2000; Zbl 0974.52008)]. This article shows that the LYZ ellipsoid for log-concave functions is a unique solution to an optimization problem, which is about minimizing the first variation of the total mass of a log-concave function along with a Gaussian function (mixed volume for log-concave functions). This optimization problem is originated in convex geometry for convex bodies and solved by E. Lutwak, D. Yang and G. Zhang. The paper under review also defines Petty projection bodies for log-concave functions, which is another important tool in convex geometry and introduced by Minkowski in connection with Cauchy's surface area formula. As a consequence, the authors prove a functional version of the Loomis-Whitney inequality, by using a result by \textit{G. Zhang} [J. Differ. Geom. 53, No. 1, 183--202 (1999; Zbl 1040.53089)]. The authors also introduce a functional version of the famous Petty projection inequality (an inequality stronger than the classical isoperimetric inequality). In the proof, they use a variety of ideas, such as Fubini's theorem, Jensen's inequality, and a result by \textit{A. Colesanti} and \textit{I. Fragalà} [Adv. Math. 244, 708--749 (2013; Zbl 1305.26033)]. It is also noted that the results, which they found for log-concave functions, are the generalizations of the results for convex bodies. They do that simply by using the characteristic function of convex bodies, which is a log-concave function. Finally, they discuss the valuation property of log-concave functions. They use some earlier results by A. Colesanti and I. Fragalá (the above cited paper) and by \textit{A. Colesanti} et al. [``Valuations on convex functions'', Int. Math. Res. Not. 2017, rnx189, 1--27 (2017; \url{doi:10.1093/imrn/rnx189})]. The paper under review shows that the operator which is used to introduce Petty projection body for log-concave functions is a continuous, SL(n) contravariant valuation. A classification of valuations for log-concave functions on a subset where the convex functions are sublinear is also given.