A Framework for Spatial Quantum Sensing

Analytical and algebraic geometry are valuable tools for dealing with problems involving analytical functions and polynomials. In what we connote as spatial quantum sensing the goal is, given an underlying field and a set of quantum sensors interrogating the field in a set of positions, to find an estimator for some property the field. This property can have multiple forms, be it distinguishing the source of a target signal, or evaluating the field (or a derivative thereof) in an arbitrary position. In this work we also link this problem to networks of quantum sensors, and the role and usefulness of entangling these sensors. We find that the estimators that come out as a solution to the problem are such that a non-local entangled strategy provides maximum precision. We start by working under the assumption of polynomial fields, which relates to the interpolation problem, and then generalize for any signal that is modeled via analytical functions, giving rise to any general least-squares estimator. We discuss the effects of the placement of the sensors in the estimation, namely, how to find well defined, construction error-free placements for the sensors. In the case of interpolation we provide concrete examples and proofs in a $m$-dimensional array of sensors, and discuss necessary and sufficient conditions for the more general cases. We provide clear examples of the possible use-cases and statements, and compare a non-local entangled strategy with the best local strategy for an interpolation problem, showing the benefit in terms of precision in a distributed sensing scenario. This is a key tool for a wide-range of problem in sensing problems, ranging from large-scale such as earth-sized experiments, to local-scale, such has biological experiments.

💡 Research Summary

**
The paper introduces a comprehensive theoretical framework for spatial quantum sensing, where a set of quantum sensors distributed across a spatial domain are used to estimate a property of an underlying field. The authors model the field as a linear combination of known analytical basis functions (f_j(x)) with unknown coefficients (\beta_j). Sensor locations ({x_i}) provide access to the field values through quantum channels that imprint the field’s value as a phase on a probe state, (\Lambda_{F(x_i)}(\rho)=e^{iF(x_i)H_i}\rho e^{-iF(x_i)H_i}).

The central mathematical problem is to reconstruct the coefficient vector (\beta) from the measurement vector (F = (F(x_1),\dots,F(x_p))^T). This leads to the linear system (X\beta = F), where the matrix (X) contains the basis functions evaluated at the sensor positions. The authors explore three increasingly general settings:

1. Interpolation (Polynomial case) – When the field is a polynomial of total degree (d) in (m) dimensions, the matrix (X) becomes a generalized Vandermonde matrix. The paper derives necessary and sufficient conditions for the existence of a unique solution: the sensor positions must be “generic”, i.e., they must make the Vandermonde matrix non‑singular. Under these conditions, the solution yields the exact Taylor coefficients of the field, allowing any higher‑order derivative or value at arbitrary points to be expressed as a linear functional (c\cdot F). The authors also provide explicit constructions for sensor lattices that guarantee non‑singular Vandermonde matrices.

2. Signal‑Isolation (General analytic functions) – The authors extend the analysis to arbitrary analytic basis sets (e.g., Fourier modes, orthogonal polynomials). Here the key is whether the desired property can be written as a linear functional of the field values. If such a functional exists, an entangled probe state that couples to all sensors simultaneously can achieve the optimal quantum Fisher information. The paper shows that when the coefficient vector (c) has uniform sign (or is proportional to the all‑ones vector), the precision scales as (\sqrt{np}) (Heisenberg scaling) rather than the classical (\sqrt{n}) limit, where (n) is the number of repetitions per sensor and (p) the number of entangled sensors.

3. Least‑Squares (Noisy measurements) – Realistic scenarios involve measurement noise. The authors introduce a weighted pseudo‑inverse solution (\hat\beta = (X^\top W X)^{-1} X^\top W F), where (W) encodes the signal‑to‑noise ratios of individual sensors. Regularization techniques are discussed to avoid over‑fitting when the number of sensors exceeds the number of basis functions. The quantum advantage persists: an entangled probe still yields a Fisher information matrix amplified by a factor of (p), leading to an overall precision improvement of order (1/N) with (N = np) total resources.

A major contribution is the analysis of sensor placement. The condition number of (X) determines numerical stability; poorly chosen locations can render the Vandermonde matrix nearly singular, destroying the quantum advantage. The authors define “error‑free subspaces” that are spanned by sensor‑dependent vectors and can be engineered by appropriate placement, guaranteeing that the linear estimator does not suffer from model‑indistinguishability errors.

The paper also discusses practical implementations with existing quantum sensor technologies such as atom interferometers, nitrogen‑vacancy (NV) centers, and trapped‑ion probes. It outlines how large‑scale networks (potentially hundreds of sensors) could be deployed for applications ranging from Earth‑scale gravimetry to cellular‑level magnetic field mapping in biology.

In conclusion, the work unifies interpolation, signal isolation, and least‑squares estimation under a single algebraic‑geometric framework, demonstrates rigorously when and how entanglement provides a genuine precision boost, and supplies concrete guidelines for sensor placement and estimator construction. This positions the framework as a versatile tool for the design of next‑generation distributed quantum sensing platforms.