Algorithms and Lower Bounds for Comparator Circuits from Shrinkage
summary
Proves an average-case lower bound: there is an explicit polynomial-time function f on n bits such that every comparator circuit with at most n^{1.5} gates agrees with f on at most 1/2 + 2^{-k} of inputs for k ≥ log n, matching prior worst-case bounds. Gives a nontrivial counting-SAT algorithm for size-n^{1.5} comparator circuits (runtime 2^n divided by a superpolynomial factor when k = ω(log n)), a pseudorandom generator that fools size-S circuits, an MCSP lower bound, and a new shrinkage-based wire-to-gate reduction plus a conversion data structure used in the algorithms.
AI-written summary · about
chapters (11)
- 0:00Intro
- 2:14Circuit complexity
- 4:42The power of comparator circuits
- 8:14Lower bounds for comparator circuits
- 9:20Our results
- 10:02Average-case lower bound
- 11:32Circuit analysis algorithms: #SAT
- 13:09Circuit analysis algorithms: PRGS
- 15:22Techniques
- 16:08Shrinkage for comparator circuits
- 17:40Open problems
related talks
16mOn Oracles and Algorithmic Methods for Proving Lower Bounds
01h 01mTop-Down Lower Bounds for Depth-Four Circuits
17mSatisfiability Algorithms Based on Concentrated Shrinkage
21mEfficiently Testable Circuits
47mHardness Amplification and the Approximate Degree of Constant-Depth Circuits
55mNon-commutative Computation
22mNew Lower Bounds and Derandomization for ACC, and a...
32mA Finite Alternation Result for Reversible Boolean Circuits
