Title :
Degree complexity of Boolean functions and its applications to relativized separations
Author_Institution :
Dept. of Comput. Sci., Rochester Univ., NY, USA
fDate :
30 Jun-3 Jul 1991
Abstract :
It is shown that a simple function in AC0, OR of √n disjoint ANDs, cannot be computed by decision trees of depth logO(1)n where each node asks whether or not p(x1, . . .,xn)=0 for some polynomial p of degree logO(1)n. This is in contrast to recent results that every function in AC0 can be computed probabilistically by just one such query and can be deterministically computed by such decision trees if each node asks whether or not p(x1, . . .,xn )>0. The proofs are based on simple algebraic arguments that also provide alternative proofs for some known results
Keywords :
Boolean functions; computational complexity; Boolean functions; decision trees; degree complexity; relativized separations; Application software; Artificial intelligence; Boolean functions; Circuits; Computer science; Decision trees; Polynomials;
Conference_Titel :
Structure in Complexity Theory Conference, 1991., Proceedings of the Sixth Annual
Conference_Location :
Chicago, IL
Print_ISBN :
0-8186-2255-5
DOI :
10.1109/SCT.1991.160282