Worst-Case Upper Bound for (1, 2)-QSAT

The rigorous theoretical analysis of the algorithm for a subclass of QSAT, i.e. (1, 2)-QSAT, has been proposed in the literature. (1, 2)-QSAT, first introduced in SAT'08, can be seen as quantified extended 2-CNF formulas. Until now, within our knowledge, there exists no algorithm presenting the worst upper bound for (1, 2)-QSAT. Therefore in this paper, we present an exact algorithm to solve (1, 2)-QSAT. By analyzing the algorithms, we obtain a worst-case upper bound O(1.4142m), where m is the number of clauses.

💡 Research Summary

The paper addresses the problem of determining a worst‑case upper bound for the (1, 2)-QSAT problem, a restricted subclass of quantified Boolean formulas (QBF). A (1, 2)-QSAT instance has the form ∀ X ∃ Y φ where φ is a conjunction of 3‑clauses, each clause containing exactly one universally quantified literal and two existentially quantified literals. Although this subclass is co‑NP‑complete, it possesses a structural regularity that can be exploited for algorithmic analysis.

Main Contributions

1. The authors present a deterministic exact algorithm, named Sₘ, for solving (1, 2)-QSAT.
2. They introduce a set of preprocessing (reduction) rules—trivial falsity, unit clause, unit propagation, and monotone literal—that simplify the formula while preserving satisfiability.
3. By systematically branching on variables from the outermost quantifier block and converting remaining 3‑clauses into 2‑clauses, the algorithm reduces the problem to a polynomial‑time solvable QSAT‑2 instance.
4. Using a branching‑tree analysis, they prove that each branching step eliminates at least two clauses, leading to a branching factor that solves the recurrence T(m)=2·T(m−2). The positive root of the characteristic equation x⁻²+x⁻²=1 yields x≈1.4142, establishing a worst‑case time complexity of O(1.4142^m) where m is the number of clauses.

Technical Overview

Preprocessing Rules

• Trivial falsity rule: If a clause contains only universal literals, the whole formula is unsatisfiable.
• Unit clause rule: For an existential unit clause {ℓ}, assign ℓ = true, delete all clauses containing ℓ, and remove ¬ℓ from the remaining clauses.
• Unit propagation: Repeatedly apply the unit clause rule until no new unit clauses appear.
• Monotone literal rule: If a literal appears only positively (or only negatively) throughout the formula, its truth value can be fixed (false for universal, true for existential).

These rules are proved to preserve equi‑satisfiability (Theorem 1).

Algorithm Sₘ

1. Apply Reduceₘ (the preprocessing routine) to the input formula.
2. If the reduced formula is empty, return SAT; if it contains an empty clause, return UNSAT.
3. While 3‑clauses remain, select a variable from the outermost quantifier block.
   • If the selected variable is universal, recursively evaluate both assignments (true and false) because the universal quantifier requires both branches to be satisfiable.
   • If the selected variable is existential, only one satisfying branch is needed.
   • The algorithm distinguishes three specific patterns of clauses involving the chosen universal literal and analyzes the clause reduction in each case.
4. Once all clauses are 2‑CNF, invoke the known polynomial‑time QSAT‑2 solver.

Branching‑Tree Analysis
The authors model the recursion as a branching tree where each node’s branching tuple (r₁,…,r_k) records the decrease in clause count for each child. For all three patterns examined, the worst‑case reduction per branch is two clauses, yielding a tuple (2,2). The branching number β is defined as the unique positive root of Σ_i β^{−r_i}=1. Solving β^{−2}+β^{−2}=1 gives β≈1.4142. Consequently, the total running time satisfies T(m) ≤ β^m·poly(m), i.e., O(1.4142^m).

Significance
This work provides, to the best of the authors’ knowledge, the first explicit worst‑case upper bound for (1, 2)-QSAT expressed solely in terms of the number of clauses. The approach demonstrates how careful preprocessing combined with a disciplined branching strategy can dramatically reduce the exponential base compared to naïve exhaustive search. Moreover, the reduction to 2‑CNF QSAT leverages existing polynomial algorithms, showing a clean separation between the exponential core (handling the universal variables) and the tractable remainder.

Future Directions
Potential extensions include: (i) refining the reduction rules to achieve larger clause deletions per branch, thereby lowering the exponential base; (ii) empirical evaluation on random and structured (1, 2)-QSAT instances to assess practical performance; (iii) exploring whether similar techniques can be applied to other quantified formula subclasses, such as (k, ℓ)-QSAT for larger k and ℓ.

In summary, the paper delivers a rigorous theoretical analysis and a concrete algorithmic contribution, establishing an O(1.4142^m) worst‑case bound for solving (1, 2)-QSAT.