Connections Between QBF Proof Complexity and Circuit Complexity
summary
Proves a tight correspondence: QBF refutations in Frege-like systems exactly mirror circuit lower bounds via strategy extraction that yields programs in C-decision lists (depth increases by two). Constructs Q_f (exists x forall z z ≠ f(x)) so any universal winning strategy must compute f, so small Q-Frege proofs imply small C-circuits and conversely. Gives iff statements (e.g. Q-Frege superpolynomial lower bounds ↔ PSPACE ⊄ non-uniform NC1), produces new QBF separations (AC0 mod p, Sipser gaps), and characterizes Q-Res by unified decision lists with n^d blowup for d alternations.
AI-written summary · about
chapters (23)
- 0:00Hard problems everywhere?
- 2:20The proof complexity theme song
- 2:34Proofs vs circuits: what's the question exactly?
- 5:07Game semantics of QBFs
- 6:03A core QBF system: QU-Resolution
- 8:00From propositional proof systems to QBF
- 9:46Key example: Frege systems
- 10:27Important propositional proof systems
- 11:16Genuine QBF lower bounds.
- 12:12QBF proof systems with NP oracles
- 13:27Reasons for QBF hardness
- 13:52Precise characterisations in QBF
- 15:37Strategy extraction by decision lists
- 18:02From decision lists to circuits
- 20:16From circuit lower bounds to proof size lower bounds
- 21:29Strong separations
- 22:59The current frontier: propositional vs QBF
- 23:20What happens for resolution?
- 24:11Unified decision lists
- 25:04The characterisation for Q-Res
- 26:30Comparison to QBF Frege
- 27:25Proof ingredients - Part 1
- 28:10Conclusion
related talks
18mOn the Verification of a Subgraph Construction Algorithm
37mJourney with Eric Allender: From Turing Machines to Circuits, and Back
55mLower Bounds for Subgraph Isomorphism and Consequences in First-Order Logic
01h 22mTutorial 5: The GCT Chasm II
32mA Finite Alternation Result for Reversible Boolean Circuits
53mNew Lower Bounds and Derandomization for ACC, and a...
01h 08mThe Simple, Little and Slow Things Count: On Parameterized Counting Complexity
01h 17mInapproximability of Constraint Satisfaction Problems V
