On Minimal Unsatisfiability and Time-Space Trade-offs for k-DNF Resolution
Abstract
In the context of proving lower bounds on proof space in k-DNF resolution, [Ben-Sasson and Nordstrom 2009] introduced the concept of minimally unsatisfiable sets of k-DNF formulas and proved that a minimally unsatisfiable k-DNF set with m formulas can have at most O((mk)(k+1)) variables. They also gave an example of such sets with Omega(mk2) variables. In this paper we significantly improve the lower bound to Omega(m)k, which almost matches the upper bound above. Furthermore, we show that this implies that the analysis of their technique for proving time-space separations and trade-offs for k-DNF resolution is almost tight. This means that although it is possible, or even plausible, that stronger results than in [Ben-Sasson and Nordstrom 2009] should hold, a fundamentally different approach would be needed to obtain such results.
Turn this paper into a lesson
ArcXiv compiles a structured reading guide from this paper's metadata: plain-English importance, contributions, prerequisite concepts, which sections to read first, flashcards, and a quiz. Grounded in the abstract, never invented.