Core Concepts
Symmetric exponential time (S2E) requires near-maximum circuit complexity, and this holds in every relativized world.
Abstract
The content discusses the following key points:
Proving circuit lower bounds has been a fundamental problem in complexity theory, with connections to questions like P vs NP and derandomization.
While almost all n-bit boolean functions require near-maximum (2^n/n)-sized circuits, limited progress has been made in finding small complexity classes with exponential circuit lower bounds.
In a recent breakthrough, Chen, Hirahara and Ren (CHR24) proved that S2E/1 (symmetric exponential time with 1 bit of advice) requires near-maximum circuit complexity.
This result is a corollary of their single-valued algorithm for the Range Avoidance (Avoid) problem, which is closely connected to finding hard truth tables and proving circuit lower bounds.
Building on CHR24's work, the author presents a simple single-valued FS2P algorithm for Avoid that works for all input sizes.
As a result, the author obtains the following:
Almost-everywhere near-maximum circuit lower bound for S2E, ZPENP, and Σ2E ∩ Π2E.
Pseudodeterministic FZPPNP constructions for various combinatorial objects like Ramsey graphs, rigid matrices, pseudorandom generators, etc.
The key technical ingredients are:
Modifying Korten's reduction to obtain a succinct description of the computational history.
Leveraging the post-order traversal structure to show the computational history has small circuit complexity.
Designing a selector algorithm to choose the correct computational history when multiple candidates are given.