On Rotationally Symmetric Norms
For two-dimensional norms that generate rotationally-symmetric norms in all higher dimensions, necessary and sufficient conditions are established for them to be subspaces of $L_1$. \
math.FA↗
arXiv subjects
Yossi Lonke
Publications and source records attributed to Yossi Lonke.
A note on spherical discrepancy
A non-algorithmic, generalized version of a recent result, asserting that a natural relaxation of the Komlós conjecture from boolean discrepancy to spherical discrepancy is true, is proved by a very short argument using convex geometry.
math.MG↗
The real part of the complementary error function
It is shown that the real part of the complementary error function is bounded below by 1 in the subset of the complex plane where the principal argument is between $3π/4$ and $5π/4$. This improves a previous result asserting that the complementary error function has no zeros in the same set, whose proof was essentially based on calculus of two variables. The present proof uses an almost $60$ years old, relatively unknown estimate for the error function, combined with some elementary complex analysis, thus serving as a humble illustration to a quote attributed to Jacques Hadamard: "The shortest path between two truths in the real domain passes through the complex domain".
math.FA↗
On a question of Pietch
The main result is that a finite dimensional normed space embeds isometrically in $\ell_p$ if and only if it has a discrete Levy $p$-representation. This provides an alternative answer to a question raised by Pietch, and as a corollary, a simple proof of the fact that unless $p$ is an even integer, the two-dimensional Hilbert space $\ell_2^2$ is not isometric to a subspace of $\ell_p$. The situation for $\ell_q^2$ with $q\neq 2$ turns out to be much more restrictive. The main result combined with a result of Dor provides a proof of the fact that if $q\neq 2$ then $\ell_q^2$ is not isometric to a subspace of $\ell_p$ unless $q=p$. Further applications concerning restrictions on the degree of smoothness of finite dimensional subspaces of $\ell_p$ are included as well.
math.FA↗
A characterisation of zonoids
Let $K$ be a unit ball of some norm in $R^n$. For an arbitrary direction $u\in R^n$, there is associated a unit-ball $K_u$, which is rotationally invariant with respect to rotations keeping $u$ fixed, called the $u$-spin of $K_u$. It is proved that $K$ is a zonoid if and only if all of its spins are zonoids.
math.MG↗
Shadows of Cube Vertices
Suppose that a finite-dimensional cube is orthogonally projected onto a central section of itself by a subspace of one dimension less. Up to dimension $9$, at least one vertex is projected onto the section, but for dimension $10$ or larger, there are orthogonal projections for which all the vertices are projected outside the section. In fact, this is the case for "most" orthogonal projections, as the dimension tends to infinity.
math.FA↗
Derivatives of the L^p cosine transform
The $L^p$-cosine transform of an even, continuous function $f\in C_e(\Sn)$ is defined by: $$H(x)=\int_{\Sn}|\ip{x}ξ|^pf(ξ) dξ,\quad x\in {\R}^n.$$ It is shown that if $p$ is not an even integer then all partial derivatives of even order of $H(x)$ up to order $p+1$ (including $p+1$ if $p$ is an odd integer) exist and are continuous everywhere in ${\R}^n\backslash\{0\}$. As a result of the corresponding differentiation formula, we show that if $f$ is a positive bounded function and $p>1$ then $H^{1/p}$ is a support function of a convex body whose boundary has everywhere positive Gauss-Kronekcer curvature.
math.MG↗
On random sections of the cube
Let $f(j,k,n)$ denote the expected number of $j$-faces of a random $k$-section of the $n$-cube. A formula for $f(0,k,n)$ is presented, and for $j\geq 1$, a lower bound for $f(j,k,n)$ is derived, which implies a precise asymptotic formula for $f(n-m,n-l,n)$ when $1\leq l<m$ are fixed integers and $n\to\8$.
math.PR↗
On zonoids whose polars are zonoids
Zonoids whose polars are zonoids cannot have proper faces of dimension other than $n-1$ or zero ($n\geq 3$). However, there exist non smooth zonoids whose polars are zonoids. Examples in $R^3$ and $R^4$ are given.
math.MG↗