Calculating the quantum Fisher information via the truncated Wigner method

Abstract: In this work, we propose new methods of parameter estimation using stochastic sampling quantum phase-space simulations. We show that it is possible to compute the quantum Fisher information (QFI) from semiclassical stochastic samples using the Truncated Wigner Approximation (TWA). This method extends the class of quantum systems whose fundamental sensitivity limit can be computed efficiently to any system that can be modelled using the TWA, allowing the analysis of more meteorologically useful quantum states. We illustrate this approach with examples, including a system that evolves outside the spin-squeezing regime, where the method of moments fails.

• The paper introduces a trajectory-based protocol using the truncated Wigner approximation to estimate QFI without full quantum state reconstruction.
• It validates the method against analytic results in both undepleted and pump-depleted regimes, highlighting its precision in complex systems.
• The approach serves as an error diagnostic by identifying when semiclassical approximations break down due to strong non-Gaussian effects.

Calculating Quantum Fisher Information via the Truncated Wigner Method

Introduction

This paper presents an operational method for estimating the quantum Fisher information (QFI) in many-body quantum systems using classical stochastic sampling techniques based on the truncated Wigner approximation (TWA) (2603.29196). The principal contribution is a validated protocol for efficiently computing the QFI directly from sampled phase-space trajectories, circumventing the need to reconstruct the full quantum state or analytically differentiate the Wigner function. The approach applies broadly to any interacting quantum system amenable to TWA and addresses scenarios that go beyond the reach of spin-squeezing metrics and the restrictions of the method-of-moments (MoM) estimator.

Background: QFI in Phase-Space and the Truncated Wigner Approximation

The QFI quantifies the ultimate lower bound on the variance of unbiased estimators of a parameter encoded in a quantum system, setting the quantum Cramér–Rao lower bound for parameter estimation. For pure states, the QFI can be written as $$F_Q = 2 \operatorname{Tr}[(\partial_\omega \rho)^2]$$, where $\omega$ is the parameter of interest and $\rho$ the density operator. Evaluating $F_Q$ is tractable by analytic means only for a small subset of states and Hamiltonians, especially in the presence of non-Gaussianity or complex dynamics.

The TWA maps quantum dynamics onto the evolution of a (quasi)probability distribution in phase-space—the Wigner function. Quantum operators are replaced by stochastic averages over classical trajectories, greatly reducing computational complexity for large Hilbert spaces. While moments of observables are easily evaluated as trajectory averages, QFI requires computing infinitesimal changes in the state with respect to $\omega$, a task that historically reintroduces the exponential complexity TWA aims to avoid.

Trajectory-Based Quantum Fisher Information Estimation

The authors derive a protocol that leverages the TWA's phase-space structure to directly access the QFI. The main steps are:

1. Stochastic Sampling and Dynamics: The initial quantum state (typically Gaussian) is sampled to produce an ensemble of initial phase-space points. Each sample evolves via the classical equations of motion derived from the system's Hamiltonian, neglecting third- and higher-order derivatives in the Wigner function.
2. Parameter Differentiation via Finite Differences: Trajectories are evolved for slightly different values of the encoded parameter $\omega$. The derivative of evolved coordinates with respect to $\omega$ is estimated via finite differencing.
3. QFI Evaluation: For Gaussian initial states, the expression for $F_Q$ reduces to an explicit formula involving the covariance matrix of the initial distribution, the derivatives of the trajectory endpoints with respect to $\omega$, and simple averages over the stochastic ensemble. No explicit reconstruction of the Wigner function is necessary, even for systems with large mode occupation or spatial extension.

The authors rigorously demonstrate that for unitary parameter encodings $\hat{U} = \exp(-i\omega \hat{G})$, the QFI calculated from the variance is exactly reproduced by the trajectory method. For more general encodings—e.g., when the parameter couples via a non-commuting term in the Hamiltonian—the method computes the full QFI as required.

Illustrative Applications and Numerical Validation

Three representative quantum-optical/atomic physics models are discussed to benchmark the proposed method:

• Parametric Amplification (Undepleted Pump Approximation): The QFI is computed analytically and via TWA. Across a broad range of parameters, the trajectory-based TWA recovers the analytic QFI to numerical precision, reinforcing the method's validity when higher-order quantum corrections are negligible.
• Parametric Amplification with Pump Depletion: Including interactions and pump depletion (multi-mode case) removes analytic tractability. The method robustly reproduces the QFI accessible via TWA for observables, highlighting its value in complex interacting systems.
• Kerr Nonlinearity and Non-Gaussian Evolution: In a regime where the dynamics depart from Gaussianity, the TWA's neglect of third-order derivatives becomes significant. The method tracks the QFI accurately up to the breakdown point of the TWA; when the Wigner function develops strong non-Gaussian features (notably, negativity), both state fidelity and QFI estimation become unreliable as expected. This provides an in-situ diagnostic for the breakdown of semiclassical methods.

A crucial finding is that the method consistently identifies cases where the MoM estimator fails to register increased sensitivity due to state preparation (e.g., Kerr-induced non-classicality), whereas the QFI computation via TWA captures the correct scaling and metrological advantage.

Comparative Analysis with Alternative Semiclassical Techniques

An important point of comparison is the method introduced by RouhbakhshNabati et al., which evaluates QFI by propagating classical trajectories and using the action’s sensitivity to the parameter [cited as RouhbakhshNabati:2025]. The present approach does not require explicit knowledge of the classical action, rendering it suitable for scenarios featuring parameter quenches and time-local dynamics, common in quantum sensing applications. This broadens applicability to quantum fields and situations where the action is ill-defined or zero.

Implications and Prospective Developments

From a practical perspective, the availability of QFI estimators within the TWA framework greatly expands the toolkit for metrological analysis in many-body systems. Systems exhibiting complex quantum correlations, non-Gaussianity, or high spatial complexity (e.g., Bose-Einstein condensates with nonlinear dynamics or atom-light interfaces) may now have their metrological limits evaluated without brute-force Hilbert space simulation.

Theoretically, the method provides a bridge between semiclassical stochastic simulations and quantum estimation theory, sharpening the interface between classical and quantum descriptions in parameter estimation. It also offers an explicit diagnostic for the breakdown of semiclassical methods: when the QFI diverges from exact quantum calculations, higher-order quantum corrections become essential.

Possible extensions include the systematic correction for higher-order terms (beyond TWA), application to time-dependent encoding protocols, and hybridization with tensor-network representations for systems where the TWA's validity is marginal.

Conclusion

This paper establishes a robust, efficient method for computing quantum Fisher information within the truncated Wigner framework, delivering reliable metrological sensitivity bounds for a wide class of quantum many-body systems. The method's integration into existing TWA simulation pipelines enables practical evaluation of quantum-enhanced sensing protocols and provides a natural error diagnostic by pinpointing regimes where standard semiclassical treatments fail. The approach is poised to become an essential ingredient in the analysis and design of quantum sensors and metrological platforms grounded in complex quantum dynamics (2603.29196).