Community detection is a foundational problem in data science. Its natural extension to hypergraphs captures higher-order correlations beyond pairwise interactions. In this work, we develop a quantum algorithm for hypergraph community detection that achieves a quartic quantum speedup over the best known classical algorithm, along with superpolynomial savings in space. Our algorithm is based on the Kikuchi method, which we extend beyond previously considered problems such as Tensor PCA and $p$XORSAT to a broad family of generalized stochastic block models. To demonstrate (near) optimality of this method, we prove matching lower bounds (up to logarithmic factors) in the low-degree framework, showing that the algorithm saturates a smooth statistical-computational tradeoff. The quantum speedup arises from a quantized version of the Kikuchi method and is based on the efficient preparation of a guiding state correlated with the underlying community structure. Our work suggests that prior quantum speedups using the Kikuchi method are sufficiently robust to encompass a broader set of problems than previously believed; we conjecture that a quantity known as marginal order characterizes the existence of these quantum speedups.
翻译:社区检测是数据科学中的一个基础性问题。将其自然推广到超图可以捕捉超越成对相互作用的高阶相关性。在本工作中,我们提出了一种用于超图社区检测的量子算法,该算法相较于已知最佳经典算法实现了四次量子加速,同时在空间复杂度上实现了超多项式节省。我们的算法基于Kikuchi方法,我们将其应用范围从先前考虑的问题(如Tensor PCA和$p$XORSAT)扩展到了一个广泛的广义随机块模型家族。为了证明该方法的(近似)最优性,我们在低度框架下证明了匹配的下界(至多对数因子),表明该算法达到了一种平滑的统计-计算权衡。量子加速源于Kikuchi方法的量子化版本,其基础是高效制备一个与底层社区结构相关的引导态。我们的工作表明,先前使用Kikuchi方法实现的量子加速具有足够的鲁棒性,能够涵盖比先前认为更广泛的问题集;我们推测一个被称为边际阶的量可以刻画这些量子加速存在的条件。