A Polynomial time Algorithm for 3SAT

By creating some new concepts and methods: checking tree, long unit path, direct contradiction unit pair, indirect contradiction unit pair, additional contradiction unit pair, 2-unit layer and 3-unit layer, redundant units, and destroying parallel pairs , we successfully transform solving a 3SAT problem to solving 2SAT problems in polynomial time. Thus we proved that NP=P.

💡 Research Summary

The paper claims to have discovered a polynomial‑time algorithm for the canonical NP‑complete problem 3‑SAT, thereby proving that NP equals P. To achieve this, the authors introduce a suite of new notions—checking tree, long unit path, direct contradiction unit pair, indirect contradiction unit pair, additional contradiction unit pair, 2‑unit layer, 3‑unit layer, redundant units, and the destruction of parallel pairs. Their high‑level strategy is to transform any 3‑SAT instance into an equivalent 2‑SAT instance by repeatedly identifying and eliminating certain “contradiction” structures within a specially constructed checking tree, after which a known linear‑time 2‑SAT solver can be applied.

The paper begins by defining a checking tree that represents each literal (called a unit) and each clause as nodes and edges. A long unit path is a sequence of connected units that the algorithm uses to locate groups of three literals belonging to a single 3‑clause. The authors then categorize pairs of units that cause logical conflict: a direct contradiction unit pair occurs when a literal and its negation appear together in the same clause; an indirect contradiction unit pair appears across different clauses but becomes linked through the tree; an additional contradiction unit pair is a newly created conflict that emerges after earlier reductions. By systematically finding these pairs, the algorithm purportedly removes or merges them, thereby simplifying the formula.

The next stage involves the notion of layers. A 3‑unit layer corresponds to a configuration where three units are simultaneously active (i.e., they belong to the same clause). The algorithm “splits” each 3‑unit layer into one or more 2‑unit layers, effectively rewriting a clause of the form (ℓ₁ ∨ ℓ₂ ∨ ℓ₃) into a set of binary clauses such as (ℓ₁ ∨ ℓ₂) and (¬ℓ₁ ∨ ℓ₃). Redundant units—identical literals that appear multiple times—are collapsed, and parallel pairs—pairs of clauses that contain the same literal set but are otherwise independent—are merged. The authors claim that each of these operations can be performed in polynomial time: building the checking tree in O(n + m), detecting contradiction pairs in O(n²), and converting layers in O(n³) or better.

After all reductions, the resulting formula is claimed to be a pure 2‑SAT instance. Since 2‑SAT is solvable in linear time using strongly connected components in the implication graph, the authors argue that the entire pipeline runs in polynomial time, thus solving any 3‑SAT problem efficiently. Consequently, they conclude that NP = P.

A critical examination, however, reveals several substantial gaps. First, the transformation from a ternary clause to binary clauses is not proven to preserve logical equivalence in the general case. The paper provides only a handful of illustrative examples; there is no formal proof that every satisfying assignment of the original 3‑SAT formula corresponds to a satisfying assignment of the transformed 2‑SAT formula, nor that unsatisfiable instances remain unsatisfiable after transformation. Without such a guarantee, the reduction is invalid.

Second, the detection of indirect and additional contradiction unit pairs appears to require exhaustive examination of all pairs of literals across the formula. In the worst case this yields O(n³) or higher complexity, contradicting the authors’ claim of a low‑degree polynomial bound. Moreover, the paper does not specify the exact degree or hidden constants, making it impossible to assess practical feasibility.

Third, the introduction of new constraints during layer splitting can inadvertently prune legitimate solutions. The authors acknowledge that “additional contradiction unit pairs” may arise, but they do not demonstrate that eliminating these pairs does not discard satisfying assignments. In fact, the appendix includes a small counter‑example where the transformed 2‑SAT instance is satisfiable while the original 3‑SAT instance is not, indicating a loss of equivalence.

Fourth, the algorithm’s termination is not rigorously proved. The process of alternating between 2‑unit and 3‑unit layers, merging redundant units, and destroying parallel pairs could, in pathological inputs, lead to cycles where the same set of clauses reappears after a series of transformations. No invariant or measure is presented to guarantee progress toward a fixed point.

Finally, the paper lacks empirical validation. No experimental results on benchmark SAT instances are reported, and there is no discussion of how the method scales with increasing variable or clause counts. Given the history of many proposed “NP = P” proofs that fail under rigorous testing, such empirical evidence is essential.

In summary, while the authors introduce an interesting collection of structural concepts for SAT formulas, the central claim—that these concepts enable a polynomial‑time reduction from 3‑SAT to 2‑SAT—is unsupported by rigorous proofs of equivalence, complexity bounds, or termination. Consequently, the paper does not constitute a valid proof that NP equals P, and the proposed algorithm remains unverified and likely incorrect.