Upper bounds for the constants of Bennett's inequality and the Gale--Berlekamp switching game
Abstract
In 1977, G. Bennett proved, by means of non-deterministic methods, an inequality which plays a fundamental role in a series of optimization problems. More precisely, Bennett's inequality shows that, for p1,p2 ∈1,∞] and all positive integers n1,n2, there exists a bilinear form An1,n2( Rn1, · p1) ×( Rn2, · p2) with coefficients 1 satisfying \[ An1,n2 ≤ Cp1,p2\ n11-1p1n2\ 12-1p2 ,0\ ,n21-1p2n1\ 12 -1p1,0\ \ \] for a certain constant Cp1,p2 depending just on p1,p2; moreover, the exponents of n1,n2 cannot be improved. In this paper, using a constructive approach, we prove that Cp1,p2≤8/5 whenever p1,p2∈[ 2,∞] or p1=p2=p∈[ 1,∞] . Our techniques are applied to provide new upper bounds for the constants of a combinatorial game, known as Gale--Berlekamp switching game or unbalancing lights problem. As a consequence, we improve estimates obtained by Brown and Spencer in 1971 and by Carlson and Stolarski in 2004.
Turn this paper into a full lesson
ArcXiv compiles a staged curriculum from this paper: 8-12 lessons across beginner → advanced, synthesised section guides, visuals, flashcards, a quiz, exercises, and on-demand deep dives per section. Grounded in the abstract, never invented.