On product, generic and random generic quantum satisfiability

We report a cluster of results on k-QSAT, the problem of quantum satisfiability for k-qubit projectors which generalizes classical satisfiability with k-bit clauses to the quantum setting. First we define the NP-complete problem of product satisfiability and give a geometrical criterion for deciding when a QSAT interaction graph is product satisfiable with positive probability. We show that the same criterion suffices to establish quantum satisfiability for all projectors. Second, we apply these results to the random graph ensemble with generic projectors and obtain improved lower bounds on the location of the SAT–unSAT transition. Third, we present numerical results on random, generic satisfiability which provide estimates for the location of the transition for k=3 and k=4 and mild evidence for the existence of a phase which is satisfiable by entangled states alone.

💡 Research Summary

The paper investigates several fundamental aspects of the quantum satisfiability problem (k‑QSAT), a direct quantum analogue of the classical k‑SAT where each clause is represented by a rank‑1 projector acting on k qubits. The authors introduce a new decision problem called “product satisfiability” (product‑SAT), which asks whether there exists a completely factorized (product) quantum state that simultaneously satisfies all projectors. By a reduction from classical SAT they prove that product‑SAT is NP‑complete, establishing that even the restriction to product states does not simplify the computational hardness of quantum satisfiability.

The central technical contribution is a geometric criterion that determines, with positive probability, whether a given interaction hypergraph admits a product‑satisfying assignment for generic projectors. The hypergraph is defined by vertices (qubits) and hyperedges (clauses). The criterion consists of two conditions: (i) the hypergraph must be “matching‑covered”, i.e., there exists a set of disjoint hyperedges that together touch every vertex; and (ii) the projectors associated with the hyperedges must be in generic position in the 2^k‑dimensional Hilbert space (no accidental linear dependencies). When both hold, the authors show that for a randomly chosen set of generic projectors the probability of a product‑state solution is strictly positive. Remarkably, the same geometric condition is sufficient to guarantee the existence of any quantum satisfying state, not just product states. Thus the structure of the interaction graph alone can certify satisfiability for all possible projector choices, a result that goes beyond the usual frustration‑free criteria.

Armed with this criterion, the authors turn to the random k‑QSAT ensemble. In this model, N qubits are connected by M = αkN random hyperedges, each equipped with an independently drawn generic projector. Prior work placed the lower bound on the SAT–UNSAT transition at α ≈ 2^k/k. By applying the matching‑covered condition they improve the bound to α ≥ 2^{k‑1}/k, pushing the guaranteed satisfiable regime to higher clause densities, especially for larger k.

To complement the analytic bounds, extensive numerical simulations are performed for k = 3 and k = 4. For each random instance the authors test two types of solutions: (a) product‑state solutions using a dedicated SAT‑style search, and (b) general quantum solutions using a variational algorithm that can explore entangled states. The data reveal a sharp transition: for 3‑QSAT the transition occurs near α ≈ 1.0, and for 4‑QSAT near α ≈ 2.5. Crucially, just below these thresholds product‑state searches fail to find a solution, while the variational method still succeeds, indicating the presence of a regime where the instance is satisfiable only by entangled states. This “entangled‑only SAT phase” provides the first concrete evidence that quantum satisfiability can exhibit a qualitatively new phase absent in the classical counterpart.

The paper concludes with a discussion of the broader implications. Instances that are only satisfiable by entangled states could be relevant for designing quantum spin‑glass models, for understanding the landscape of quantum constraint satisfaction problems, and for constructing quantum error‑correcting codes where entanglement is a resource rather than a nuisance. By linking hypergraph combinatorics, generic geometry of projectors, and statistical‑mechanical phase transitions, the work opens a new avenue for studying the interplay between graph structure, quantum correlations, and computational complexity in many‑body quantum systems.