2013 IEEE 54th Annual Symposium on Foundations of Computer Science (1991)

San Juan, Puerto Rico

Oct. 1, 1991 to Oct. 4, 1991

ISBN: 0-8186-2445-0

pp: 424-430

J. Matousek , Dept. of Appl. Math., Charles Univ., Praha, Czechoslovakia

ABSTRACT

Let (X, R) be a set system on an n-point set X. For a two-coloring on X, its discrepancy is defined as the maximum number by which the occurrences of the two colors differ in any set in R. It is shown that if for any m-point subset Y contained in X the number of distinct subsets induced by R on Y is bounded by O(m/sup d/) for a fixed integer d is a coloring with discrepancy bounded by O(n/sup 1/2-1/2d/ (log n)/sup 1+1/2d/). Also, if any subcollection of m sets of R partitions the points into at most O(m/sup d/) classes, then there is a coloring with discrepancy at most O(n/sup 1/2-1/2d/ n). These bounds imply improved upper bounds on the size of in -approximations for (X, R). All of the bounds are tight up to polylogarithmic factors in the worst case. The results allow the generalization of several results of J. Beck (1984) bounding the discrepancy in certain geometric settings to the case when the discrepancy is taken relative to an arbitrary measure.

INDEX TERMS

geometric settings, discrepancy, in -approximations, bounded VC-dimension, two-coloring, m-point subset, upper bounds, polylogarithmic factors

CITATION

L. Wernisch,
E. Welzl,
J. Matousek,
"Discrepancy and in -approximations for bounded VC-dimension",

*2013 IEEE 54th Annual Symposium on Foundations of Computer Science*, vol. 00, no. , pp. 424-430, 1991, doi:10.1109/SFCS.1991.185401