The correlation between parity and quadratic polynomials mod 3
Journal of Computer and System Sciences - Special issue on computational complexity 2002
Query-efficient algorithms for polynomial interpolation over composites
SODA '06 Proceedings of the seventeenth annual ACM-SIAM symposium on Discrete algorithm
LWPP and WPP are not uniformly gap-definable
Journal of Computer and System Sciences
Inverse conjecture for the gowers norm is false
STOC '08 Proceedings of the fortieth annual ACM symposium on Theory of computing
Hardness amplification proofs require majority
STOC '08 Proceedings of the fortieth annual ACM symposium on Theory of computing
Guest Column: correlation bounds for polynomials over {0 1}
ACM SIGACT News
On the Power of Small-Depth Computation
Foundations and Trends® in Theoretical Computer Science
Hardness Amplification Proofs Require Majority
SIAM Journal on Computing
Optimal Testing of Reed-Muller Code
Property testing
Optimal Testing of Reed-Muller Code
Property testing
Hard functions for low-degree polynomials over prime fields
MFCS'11 Proceedings of the 36th international conference on Mathematical foundations of computer science
On the correlation between parity and modular polynomials
MFCS'06 Proceedings of the 31st international conference on Mathematical Foundations of Computer Science
Hard Functions for Low-Degree Polynomials over Prime Fields
ACM Transactions on Computation Theory (TOCT)
Hi-index | 0.00 |
Abstract: We use a Ramsey-theoretic argument to obtain the first lower bounds for approximations over Z_m by nonlinear polynomials: 1) A degree-2 polynomial over Z_m (m odd) must differ from the parity function on at least a 1/2-1/2^{(log n)^\Omega(1)} fraction of all points in the Boolean n-cube. 2) A degree-O(1) polynomial over Z_m (m odd) must differ from the parity function on at least a 1/2 - o(1) fraction of all points in the Boolean n-cube. These nonapproximability results imply the first known lower bounds on the top fanin of MAJ \circ MOD_m \circ AND_{O(1)} circuits (i.e., circuits with a single majority-gate at the output node, MOD_m-gates at the middle level, and constant-fanin AND-gates at the input level) that compute parity: 1) MAJ \circ MOD_m \circ AND_2 circuits that compute parity must have top fanin 2^{(log n)^\Omega(1)}. 2) Parity cannot be computed by MAJ \circ MOD_m \circ AND_{O(1)} circuits with top fanin O(1). Similar results hold for the MOD_q function as well.