Subgroup Rank-1 Lattice for Practical High-dimensional Black-box Integral Approximation

2026-09-28

AI Summary

This study addresses the bottleneck in high-dimensional black-box integration where standard quasi-Monte Carlo point sets incur an O(nd) computational complexity for feature map evaluation. We propose subgroup Rank-1 lattice rules, which leverage finite field splitting to transform nonlinear mappings into FFT-solvable short cyclic correlations, enabling efficient gradient-free integration approximation. Departing from classical component-by-component construction, convergence is established directly via cyclotomic polynomials. By integrating Korobov generators with algebraic number theory techniques, the proposed method significantly reduces both computational and memory overhead at a fixed order. Experiments demonstrate that our approach outperforms random features and Sobol/Halton sequences across most benchmarks, constructing ultra-large-scale high-dimensional samples within milliseconds.

Abstract

Estimating integrals of black-box, high-dimensional functions, from expectations and kernel mean embeddings to the softmax kernel in self-attention, is a basic subroutine in machine learning. Rank-1 lattice rules suit this setting: they query the integrand only at a fixed point set and need no gradients. When the $n$ points serve as a design matrix $X\in\mathbb{R}^{n\times d}$ for a feature map, however, computing $\Psi(X)^\top v$ or $\Psi(X)w$ for an elementwise nonlinearity $\Psi$ costs $O(nd)$ time and memory for any standard quasi-Monte Carlo point set. We study subgroup rank-1 lattices, whose Korobov generator $(1,t,\dots,t^{d-1})$ uses a scalar $t$ of fixed multiplicative order $m$. Splitting $\mathbb{F}_n^\times$ into cosets of $\langle t\rangle$ reduces both maps to short cyclic correlations evaluated by FFT, giving exact results for arbitrary $\Psi$ in $O(n\log m)$ time and $O(n)$ memory, without forming $X$. Since fixing $m$ falls outside classical component-by-component theory, we prove convergence directly: via resultants with the cyclotomic polynomial $\Phi_m$, the squared worst-case error in the Korobov space decays as $O(n^{-(\alpha-1)/(m-1)})$ for prime $m\ge d+1$, and this threshold is exact. Using the splitting of $n$ in $\mathbb{Q}(\zeta_m)$, averaging over the $m-1$ admissible generators improves the constant by a factor $\Theta(m-1)$. Empirically, the subgroup lattice beats Gaussian and orthogonal random features and scrambled Sobol'and Halton points in 49 of 54 synthetic kernel-estimation settings and all 45 softmax-attention settings on nine real datasets, and builds a sample set with $d=2048$, $n\approx4.1\times10^7$ in 2.3 ms.

Problem

• high-dimensional integration
• black-box functions
• quasi-Monte Carlo
• rank-1 lattice
• machine learning

Innovation

• Subgroup Rank-1 Lattice
• Quasi-Monte Carlo
• Fast Fourier Transform
• Black-box Integration
• Korobov Space