Relationships between the number of inputs and other complexity measures of Boolean functions

Abstract

We generalize and extend the ideas in a recent paper of Chiarelli, Hatami and Saks to prove new bounds on the number of relevant variables for boolean functions in terms of a variety of complexity measures. Our approach unifies and refines all previously known bounds of this type. We also improve Nisan and Szegedy's well-known block sensitivity vs. degree inequality by a constant factor, thereby improving Huang's recent proof of the sensitivity conjecture by the same constant.