The local Hamiltonian problem is the canonical $\mathsf{QMA}$-complete problem, and $O(2^n)$ time classical algorithms and $O(2^{n/2})$ time quantum algorithms are known to solve the problem in the worst case. It is not clear how to improve these brute force strategies for a broad class of the problem because ground states are highly entangled in general, and we cannot directly apply known strategies for classical CSPs.
In this work, we present exponentially faster classical and quantum algorithms under two mild assumptions: (1) the Hamiltonian is frustration-free on YES instances, and (2) it is approximately regular, meaning that every qubit is acted upon by approximately the same number of constraints.
We complement these upper bounds by showing that, assuming (Q)SETH, quantum 5-SAT admits no non-trivial worst-case speedup. Our lower bound further demonstrates that the dependence of our algorithms on regularity is in some sense nearly optimal. Specifically, quantum 5-SAT remains (Q)SETH-hard even for Hamiltonians in which all but $O(\sqrt{n})$ qubits participate in only constantly many constraints, while the remaining $O(\sqrt{n})$ qubits each participate in $O(\sqrt{n})$ constraints. By contrast, if either the size of this high-degree subset or the degrees of its qubits is reduced by a factor of $n^δ$, for any $δ>0$, our algorithm solves the problem in time $O(2^{(1- \varepsilon)n})$ for some $\varepsilon>0$.
Together, our upper and lower bounds establish a fine-grained complexity dichotomy for quantum satisfiability.
The local Hamiltonian problem is the canonical QMA-complete problem, and O(2n) time classical algorithms and O(2n/2) time quantum algorithms are known to solve the problem in the worst case. It is not clear how to improve these brute force strategies for a broad class of the problem because ground states are highly entangled in general, and we cannot directly apply known strategies for classical CSPs.
In this work, we present exponentially faster classical and quantum algorithms under two mild assumptions: (1) the Hamiltonian is frustration-free on YES instances, and (2) it is approximately regular, meaning that every qubit is acted upon by approximately the same number of constraints.
We complement these upper bounds by showing that, assuming (Q)SETH, quantum 5-SAT admits no non-trivial worst-case speedup. Our lower bound further demonstrates that the dependence of our algorithms on regularity is in some sense nearly optimal. Specifically, quantum 5-SAT remains (Q)SETH-hard even for Hamiltonians in which all but O(n) qubits participate in only constantly many constraints, while the remaining O(n) qubits each participate in O(n) constraints. By contrast, if either the size of this high-degree subset or the degrees of its qubits is reduced by a factor of nδ, for any δ>0, our algorithm solves the problem in time O(2(1−ε)n) for some ε>0.
Together, our upper and lower bounds establish a fine-grained complexity dichotomy for quantum satisfiability.
The local Hamiltonian (LH) problem is the canonical QMA-complete problem introduced by Kitaev. In this paper, we show its hardness in a very strong sense: we show that the 3-local Hamiltonian problem on n qubits cannot be solved classically in time O(2(1−ε)n) for any ε>0 under the Strong Exponential-Time Hypothesis (SETH), and cannot be solved quantumly in time O(2(1−ε)n/2) for any ε>0 under the Quantum Strong Exponential-Time Hypothesis (QSETH). These lower bounds give evidence that the currently known classical and quantum algorithms for LH cannot be significantly improved.
Furthermore, we are able to demonstrate fine-grained complexity lower bounds for approximating the quantum partition function (QPF) with an arbitrary constant relative error. Approximating QPF with relative error is known to be equivalent to approximately counting the dimension of the solution subspace of QMA problems. We show the SETH and QSETH hardness to estimate QPF with constant relative error. We then provide a quantum algorithm that runs in O(2n) time for an arbitrary 1/poly(n) relative error, matching our lower bounds and improving the state-of-the-art algorithm by Bravyi, Chowdhury, Gosset, and Wocjan (Nature Physics 2022) in the low-temperature regime.
To prove our fine-grained lower bounds, we introduce the first size-preserving circuit-to-Hamiltonian construction that encodes the computation of a T-time quantum circuit acting on N qubits into a (d+1)-local Hamiltonian acting on N+O(T1/d) qubits. This improves the standard construction based on the unary clock, which uses N+O(T) qubits.
3-Local Hamiltonian Problem and Constant Relative Error Quantum Partition Function Approximation: O(22n) Algorithm Is Nearly Optimal under QSETH
We investigate the computational complexity of the Local Hamiltonian (LH) problem and the approximation of the Quantum Partition Function (QPF), two central problems in quantum many-body physics and quantum complexity theory. Both problems are known to be QMA-hard, and under the widely believed assumption that BQP=QMA, no efficient quantum algorithm exits. The best known quantum algorithm for LH runs in O(22n(1−o(1))) time, while for QPF, the state-of-the-art algorithm achieves relative error δ in O∗(δ1Z2n) time, where Z denotes the value of the partition function. A nature open question is whether more efficient algorithms exist for both problems. In this work, we establish tight conditional lower bounds showing that these algorithms are nearly optimal. Under the plausible Quantum Strong Exponential Time Hypothesis (QSETH), we prove that no quantum algorithm can solve either LH or approximate QPF significantly faster than O(2n/2), even for 3-local Hamiltonians. In particular, we show: 1) 3-local LH cannot be solved in time O(22n(1−ε)) for any ε>0 under QSETH; 2) 3-local QPF cannot be approximated up to any constant relative error in O(22n(1−ε)) time for any ε>0 under QSETH; and 3) we present a quantum algorithm that approximates QPF up to relative error 1/2+1/poly(n) in O∗(2n/2) time, matching our conditional lower bound. Notably, our results provide the first fine-grained lower bounds for both LH and QPF with fixed locality. This stands in sharp contrast to QSETH and the trivial fine-grained lower bounds for LH, where the locality of the SAT instance and the Hamiltonian depends on the parameter ε in the O(22n(1−ε)) running time.
Previously, all known variants of the Quantum Satisfiability (QSAT) problem - consisting of determining whether a k-local (k-body) Hamiltonian is frustration-free - could be classified as being either in 𝖯; or complete for NP, MA, or QMA₁. Here, we present new qubit variants of this problem that are complete for BQP₁, coRP, QCMA, PI(coRP,NP), PI(BQP₁,NP), PI(BQP₁,MA), SoPU(coRP,NP), SoPU(BQP₁,NP), and SoPU(BQP₁,MA). Our result implies that a complete classification of quantum constraint satisfaction problems (QCSPs), analogous to Schaefer’s dichotomy theorem for classical CSPs, must either include these 13 classes, or otherwise show that some are equal. Additionally, our result showcases two new types of QSAT problems that can be decided efficiently, as well as the first nontrivial BQP₁-complete problem. We first construct QSAT problems on qudits that are complete for BQP₁, coRP, and QCMA. These are made by restricting the finite set of Hamiltonians to consist of elements similar to H_{init}, H_{prop}, and H_{out}, seen in the circuit-to-Hamiltonian transformation. Usually, these are used to demonstrate hardness of QSAT and Local Hamiltonian problems, and so our proofs of hardness are simple. The difficulty lies in ensuring that all Hamiltonians generated with these three elements can be decided in their respective classes. For this, we build our Hamiltonian terms with high-dimensional data and clock qudits, ternary logic, and either monogamy of entanglement or specific clock encodings. We then show how to express these problems in terms of qubits, by proving that any QCSP can be reduced to a qubit problem while maintaining the same complexity - something not believed possible classically. The remaining six problems are obtained by considering "sums" and "products" of some of the QSAT problems mentioned here. Before this work, the QSAT problems generated in this way resulted in complete problems for PI and SoPU classes that were trivially equal to NP, MA, or QMA₁. We thus commence the study of these new and seemingly nontrivial classes. While [Meiburg, 2021] first sought to prove completeness for coRP, BQP₁, and QCMA, we note that those constructions are flawed. Here, we rework them, provide correct proofs, and obtain improvements on the required qudit dimensionality.
Abstract Quantum k-SAT (the problem of determining whether a k-local Hamiltonian is frustration-free) is known to be QMA 1 1 -complete for k≥3 k ≥ 3 , and hence likely hard for quantum computers to solve. Building on a classical result of Alon and Shapira, we show that quantum k-SAT can be solved in randomised polynomial time given the ‘property testing’ promise that the instance is either satisfiable (by any state) or far from satisfiable by a product state; by ‘far from satisfiable by a product state’ we mean that ϵnk ϵ n k constraints must be removed before a product state solution exists, for some fixed ϵ>0 ϵ > 0 . The proof has two steps: we first show that for a satisfiable instance of quantum k-SAT, most subproblems on a constant number of qubits are satisfiable by a product state. We then show that for an instance of quantum k-SAT which is far from satisfiable by a product state, most subproblems are unsatisfiable by a product state. Given the promise, quantum k-SAT may therefore be solved by checking satisfiability by a product state on randomly chosen subsystems of constant size.
The Quantum k-SAT problem is the quantum generalization of the k-SAT problem. It is the problem whether a given local Hamiltonian is frustration-free. Frustration-free means that the ground state of the k-local Hamiltonian minimizes the energy of every local interaction term simultaneously. This is a central question in quantum physics and a canonical QMA_1-complete problem. The Quantum k-SAT problem is not as well studied as the classical k-SAT problem in terms of special tractable cases, approximation algorithms and parameterized complexity. In this paper, we will give a graph-theoretic study of the Quantum k-SAT problem with the structures core and radius. These hypergraph structures are important to solve the Quantum k-SAT problem. We can solve a Quantum k-SAT instance in polynomial time if the derived hypergraph has a core of size n-m+a, where a is a constant, and the radius is at most logarithmic. If it exists, we can find a core of size n-m+a with the best possible radius in polynomial time, whereas finding a general minimum core with minimal radius is NP-hard.
We show a linear-size reduction from gap Max-2-Lin(2) (a generalization of the approximate Maximum Cut, or gap Max-Cut, problem) to γ-CVP_p for γ = O(1) and finite p ≥ 1, as well as a no-go theorem against poly-sized non-adaptive quantum reductions from k-SAT to CVP₂. This implies three headline results: (i) Faster algorithms for γ-CVP_p are also faster algorithms for Max-2-Lin(2) and Max-Cut. Depending on the approximation regime, even a 2^{0.78n}-time or 2^{0.3n}-time algorithm would improve upon state-of-the-art algorithms such as Williams' 2004 algorithm [TCS 2005] or Arora, Barak, and Steurer’s 2010 algorithm [JACM 2015]. This provides evidence that γ-CVP_p for γ = O(1) requires exponential time, improving upon the previous exponential lower-bound for γ-CVP₂ with γ < 3 by Bennett, Golovnev, and Stephens-Davidowitz [FOCS 2017]. (ii) A new almost 2^{(1/2 + ε/4ς + o(1)) n}-time classical algorithm and a new almost 2^{(1/3 + ε/6ς + o(1)) n}-time quantum algorithm for (1-ε, 1-ς)-gap Max-Cut. This algorithm is faster than the algorithm of Arora, Barak and Steurer [JACM 2015], as well as the algorithm of Williams [TCS 2005], and the algorithm of Manurangsi and Trevisan [APPROX 2018] when c₀ ε < ς < c₁ ε for constants c₀, c₁. (iii) If the Quantum Strong Exponential Time Hypothesis (QSETH) can be used to show a 2^{δ n}-time lower-bound for Max-Cut, Max-2-Lin(2), or CVP₂ for any constant δ > 0, it must be via an adaptive quantum reduction unless NP ⊆ pr-QSZK. This illuminates some difficulties in characterizing the hardness of approximate constraint satisfaction problems and shows that the post-quantum security of lattice-based cryptography likely cannot be supported by QSETH. This result complements the no-go results of Aggarwal and Kumar [FOCS 2023], who showed that the classical security of lattice-based cryptography likely cannot be supported by the classical Strong Exponential Time Hypothesis (SETH).