Complex Field Formulation of the Quantum Estimation Theory

1. Problem Statement

Quantum estimation theory (QET) is a fundamental tool for determining the precision limits of estimating parameters encoded in quantum states. Traditionally, QET is formulated over the field of real numbers. Even when the physical parameters of interest are complex (e.g., the amplitude and phase of coherent states, or parameters in quantum neural networks), standard approaches require decomposing these complex parameters into their real and imaginary parts ($\theta = \alpha + i\beta$) and treating them as a vector of real variables.

The authors argue that this real-valued adaptation is "unnatural" because:

• Quantum mechanics is fundamentally formulated in the field of complex numbers.
• Many quantum states (coherent, squeezed) and processes are naturally described by complex coefficients.
• Decomposing complex parameters into real components can obscure the intrinsic structure of the estimation problem and complicate optimization.
• Recent studies suggest that native complex optimization can outperform real-valued counterparts in specific tasks.

The paper aims to develop a native complex-field formulation of quantum estimation theory that works directly with complex statistics and complex parameters without resorting to real-valued representations.

2. Methodology

The authors employ Wirtinger calculus (complex differentiation) and a specific complex map to bridge the gap between real and complex formulations.

• Wirtinger Calculus: Instead of restricting derivatives to holomorphic functions, the authors use Wirtinger derivatives ($\partial_z$ and $\partial_{z^*}$) which treat a complex variable $z$ and its conjugate $z^*$ as independent variables. This allows for the differentiation of non-holomorphic functions common in physics (e.g., $|z|^2$).
• Conjugate Extension: A complex parameter $\theta$ is extended to a vector $\hat{\theta} = [\theta, \theta^*]^T$. Similarly, statistics and density operators are extended to include their conjugates.
• Complex Map $\langle \cdot \rangle_C$: The authors utilize a linear map $\langle \cdot \rangle_C$ that transforms real block matrices (representing real and imaginary parts) into complex block matrices. This map preserves algebraic properties such as inversion, trace, and positive semi-definiteness, allowing the translation of real-valued QET results into the complex domain.
• Derivation of Complex Logarithmic Derivatives:
  • Complex Symmetric Logarithmic Derivative (CSLD): Defined via $\partial_{\theta^*} \rho = \frac{1}{2}(\rho L^S_{\theta^*} + L^S_{\theta^*} \rho)$.
  • Complex Right Logarithmic Derivative (CRLD): Defined via $\partial_{\theta^*} \rho = \rho L^R_{\theta^*}$.
  • These derivatives are shown to be related to their real counterparts (SLD and RLD) via the complex map.

3. Key Contributions

The paper establishes a comprehensive framework for complex-parameter quantum estimation, providing the following key theoretical results:

• Complex Quantum Fisher Information Matrices (QFIM):
  • Defined the CSLD-QFIM ($J^S_{\hat{\theta}}$) and CRLD-QFIM ($J^R_{\hat{\theta}}$) as block matrices containing terms for both the parameter and its conjugate.
  • Proved that these complex QFIMs are related to the standard real QFIMs by the relation $J^X_{\hat{\theta}} = \frac{1}{4} \langle J^X_{\bar{\theta}} \rangle_C$.
• Complex Cramér-Rao Inequalities (QCRI):
  • Derived the complex version of the Cramér-Rao bound: $\text{Cov}_{\hat{\theta}}(\hat{t}) \geq D_{\hat{\mu}} (J^X_{\hat{\theta}})^{-1} D_{\hat{\mu}}^\dagger$.
  • Provided explicit block-matrix forms for the covariance bounds of the complex statistic $t$ (separating the bounds for $\theta$ and $\theta^*$), which are more informative than the bound for the conjugate extension alone.
• Weighted Mean Square Error (WMSE) Bounds:
  • Defined complex weighting matrices and derived lower bounds for the WMSE ($w^S_{\hat{\theta}}$ and $w^R_{\hat{\theta}}$) in the complex domain.
  • Showed that the complex bounds are equivalent to the real bounds but offer a more compact and natural representation.
• Pure State Specialization:
  • Derived explicit formulas for the CSLD-QFIM and CRLD-QFIM for pure states $|\psi_\theta\rangle$.
  • Established the necessary and sufficient condition for saturating the SLD bound in the complex domain: $[\hat{L}^S_{\theta_j}, \hat{L}^S_{\theta_k}] = 0$ (commutativity of CSLDs).
  • Extended the Holevo bound logic to the complex case, showing how the CRLD bound relates to the SLD bound for pure states via a commutator matrix $K_{\hat{\theta}}$.

4. Results and Application

The authors applied their theory to the problem of estimating a complex parameter $z$ encoded in a coherent state $|\alpha_z\rangle$, a scenario relevant to quantum communication.

• Single-Mode Coherent State:
  • The encoding was defined as $\alpha_z = \epsilon z + \eta z^*$.
  • The authors found that the condition for the CSLD-QCRB to be attainable ($|\eta|^2 = |\epsilon|^2$) leads to a singular Fisher Information Matrix (determinant is zero). This implies that for a single mode, no measurement can simultaneously estimate the real and imaginary parts of $z$ with the precision predicted by the symmetric bound.
  • However, using the CRLD-QCRB, they showed that the bound for estimating $z$ specifically (ignoring $z^*$) is attainable. The CRLD bound provides a rank-1 matrix, indicating that while the joint estimation of $(z, z^*)$ is limited, the estimation of $z$ alone can reach the fundamental limit using independent homodyne measurements of quadratures $X$ and $Y$.
• Two-Mode Coherent State:
  • By extending the protocol to a two-mode coherent state, the authors demonstrated that it is possible to find encoding keys $(\epsilon_1, \eta_1, \epsilon_2, \eta_2)$ that satisfy the optimality condition ($|\eta_1|^2 + |\eta_2|^2 = |\epsilon_1|^2 + |\epsilon_2|^2$) and result in a non-singular QFIM.
  • In this two-mode scenario, the symmetric Cramér-Rao bound is attainable. This is achieved by performing a two-mode homodyne measurement where one mode measures the real part and the other measures the imaginary part of $z$.

5. Significance

• Theoretical Consistency: The formulation provides a mathematically consistent framework for QET that respects the complex nature of quantum mechanics, avoiding the artificial decomposition of parameters into real and imaginary parts.
• Computational Efficiency: Working natively in the complex field can simplify optimization problems in quantum metrology and quantum machine learning (e.g., Variational Quantum Eigensolvers, Quantum Neural Networks) where parameters are inherently complex.
• Practical Insight: The application to coherent states clarifies the limitations of single-mode estimation and provides a clear path (two-mode encoding) to saturate the fundamental precision limits.
• Future Extensions: The framework is designed to be extensible to other advanced topics in QET, such as the Holevo-Cramér-Rao bound for multiparameter estimation, potentially offering computational advantages over real-valued formalisms in complex optimization landscapes.

In conclusion, this paper successfully generalizes quantum estimation theory to the complex domain, offering new tools for analyzing and optimizing precision in quantum systems where complex parameters are intrinsic.