On the Complexity of Hazard-Free Formulas

2024-11-13

arXiv.org

AI Summary

This paper investigates hazard-free formula complexity of Boolean functions. The central problems addressed are: (i) characterizing when hazard-free formula complexity equals monotone formula complexity, and (ii) establishing tight bounds for random and block-composed functions. To tackle these, the authors employ combinatorial analysis, semantic modeling of hazards, and adaptation of KRW-type communication games to the hazard-free setting. Key contributions include: (i) the first proof that only monotone functions satisfy the equality between hazard derivative and hazard-free formula complexity—thereby rigorously breaking the “monotone barrier” conjectured at ITCS’23; (ii) an improved upper bound of $2^{(1+o(1))n}$ on hazard-free formula complexity for random $n$-variable Boolean functions, substantially tightening the prior $O(3^n)$ bound; and (iii) exponential lower bounds on hazard-free formula depth for set-cover and multiplexer-based block-composed functions, resolving long-standing questions about structural hardness in the hazard-free regime.

Technology Category

• Knowledge Representation and Reasoning: Computational Complexity of Reasoning
• Reasoning under Uncertainty: Other Foundations of Reasoning under Uncertainty
• Constraint Satisfaction and Optimization: Satisfiability Modulo Theories

Application Category

• Security and Privacy: Large-scale security measurements
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs
• Responsible Web: Measurement, analysis, and circumvention of Web censorship

Abstract

This paper studies the hazard-free formula complexity of Boolean functions. As our main result, we prove that unate functions are the only Boolean functions for which the monotone formula complexity of the hazard-derivative equals the hazard-free formula complexity of the function itself. Consequently, every non-unate function breaks the so-called monotone barrier, as introduced and discussed by Ikenmeyer, Komarath, and Saurabh (ITCS 2023). Our second main result shows that the hazard-free formula complexity of random Boolean functions is at most $2^{(1+o(1))n}$. Prior to this, no better upper bound than $O(3^n)$ was known. Notably, unlike in the general case of Boolean circuits and formulas, where the typical complexity matches that of the multiplexer function, the hazard-free formula complexity is smaller than the optimal hazard-free formula for the multiplexer by an exponential factor in $n$. Additionally, we explore the hazard-free formula complexity of block composition of Boolean functions and obtain a result in the hazard-free setting that is analogous to a result of Karchmer, Raz, and Wigderson (Computational Complexity, 1995) in the monotone setting. We demonstrate that our result implies a lower bound on the hazard-free formula depth of the block composition of the set covering function with the multiplexer function, which breaks the monotone barrier.

Problem

• Analyzes hazard-free formula complexity of Boolean functions.
• Establishes upper bounds for random Boolean functions.
• Explores hazard-free complexity in block composition.

Innovation

• Unate functions optimize hazard-derivative approach
• Random Boolean functions have reduced complexity
• New converse method bounds hazard-free complexity