More-efficient Quantum Multivariate Mean Value Estimator from Generalized Grover Operator
Letian Tang
TL;DR
The paper develops quantum algorithms for multivariate mean value estimation by leveraging a generalized Grover operator and multidimensional phase estimation. It presents two estimators—a memory-efficient simple estimator and a meticulously optimized version—that achieve near-optimal sample complexity under unknown covariance $\Sigma$, reducing polylogarithmic overhead and approaching the known lower bounds. The work provides a detailed univariate backbone, a directional multivariate reduction via random projections, and a complete classical-reduction pipeline to handle unknown $\operatorname{tr}\Sigma$, with concrete complexity bounds and memory considerations. These results advance practical quantum mean-value estimation and suggest paths toward broader tasks involving multiple observables. The methods hold potential impact for quantum Monte Carlo, quantum sensing, and related quantum statistical tasks.
Abstract
In this work, we present an efficient algorithm for multivariate mean value estimation. Our algorithm outperforms previous work by polylog factors and nearly saturates the known lower bound. More formally, given a random vector $\vec{X}$ of dimension $d$, we find an algorithm that uses $O\left(n \log \frac{d}δ\right)$ samples to find a mean estimate that $\vec{\tildeμ}$ that differs from the true mean $\vecμ$ by $\frac{\sqrt{\text{tr } Σ}}{n}$ in $\ell^\infty$ norm and hence $\frac{\sqrt{d \text{ tr } Σ}}{n}$ in $\ell^2$ norm, where $Σ$ is the covariance matrix of the components of the random vector. We also presented another algorithm that uses smaller memory but costs an extra $d^\frac{1}{4}$ in complexity. Consider the Grover operator, the unitary operator used in Grover's algorithm. It contains an oracle that uses a $\pm 1$ phase for each candidate for the search space. Previous work has demonstrated that when we substitute the oracle in Grover operator with generic phases, it ended up being a good mean value estimator in some mathematical notion. We used this idea to build our algorithm. Our result remains not exactly optimal due to a $\log \frac{d}δ$ term in our complexity, as opposed to something nicer such as $\log \frac{1}δ$; This comes from the phase estimation primitive in our algorithm. So far, this primitive is the only major known method to tackle the problem, and moving beyond this idea seems hard. Our results demonstrates that the methodology with generalized Grover operator can be used develop the optimal algorithm without polylog overhead for different tasks relating to mean value estimation.