A new upper bound on the query complexity for testing generalized Reed-Muller codes

Noga Ron-Zewi, Madhu Sudan

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

4 Scopus citations

Abstract

Over a finite field double-struck F q the (n,d,q)-Reed-Muller code is the code given by evaluations of n-variate polynomials of total degree at most d on all points (of double-struck F q n). The task of testing if a function f : double-struck F q n → double-struck F q is close to a codeword of an (n,d,q)-Reed-Muller code has been of central interest in complexity theory and property testing. The query complexity of this task is the minimal number of queries that a tester can make (minimum over all testers of the maximum number of queries over all random choices) while accepting all Reed-Muller codewords and rejecting words that are δ-far from the code with probability Ω(δ). (In this work we allow the constant in the Ω to depend on d.) For codes over a prime field double-struck F q the optimal query complexity is well-known and known to be Θ(q ⌈(d+1)/(q-1)⌉), and the test consists of testing if f is a degree d polynomial on a randomly chosen (⌉(d + 1)/(q - 1)⌈)-dimensional affine subspace of double-struck F q n. If q is not a prime, then the above quantity remains a lower bound, whereas the previously known upper bound grows to O( q⌈(d+1)/(q-q/p)⌉) where p is the characteristic of the field double-struck F q. In this work we give a new upper bound of (c q) (d+1)/q on the query complexity, where c is a universal constant. Thus for every p and sufficiently large q this bound improves over the previously known bound by a polynomial factor. In the process we also give new upper bounds on the "spanning weight" of the dual of the Reed-Muller code (which is also a Reed-Muller code). The spanning weight of a code is the smallest integer w such that codewords of Hamming weight at most w span the code. The main technical contribution of this work is the design of tests that test a function by not querying its value on an entire subspace of the space, but rather on a carefully chosen (algebraically nice) subset of the points from low-dimensional subspaces.

Original languageEnglish
Title of host publicationApproximation, Randomization, and Combinatorial Optimization
Subtitle of host publicationAlgorithms and Techniques - 15th International Workshop, APPROX 2012, and 16th International Workshop, RANDOM 2012, Proceedings
Pages639-650
Number of pages12
DOIs
StatePublished - 28 Aug 2012
Externally publishedYes
Event15th International Workshop on Approximation Algorithms for Combinatorial Optimization Problems, APPROX 2012 and the 16th International Workshop on Randomization and Computation, RANDOM 2012 - Cambridge, MA, United States
Duration: 15 Aug 201217 Aug 2012

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume7408 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference15th International Workshop on Approximation Algorithms for Combinatorial Optimization Problems, APPROX 2012 and the 16th International Workshop on Randomization and Computation, RANDOM 2012
Country/TerritoryUnited States
CityCambridge, MA
Period15/08/1217/08/12

Fingerprint

Dive into the research topics of 'A new upper bound on the query complexity for testing generalized Reed-Muller codes'. Together they form a unique fingerprint.

Cite this