A measurement concept defining how outcomes are modeled and how state descriptions are updated after an outcome is recorded. It governs outcome probabilities, information extraction, and the disturbance introduced by the measurement interaction. It does not yield reliable inference without adequate calibration, sufficient data, and appropriate estimation procedures. It supports reconstruction and validation of state and process descriptions from experimental statistics. The concept is generally stable, though practical implementations and estimation methods evolve over time.

Principle

For a fixed measurement, Fisher information measures the inverse lower bound on the variance of unbiased estimators of a parameter via the classical Cramér–Rao inequality; it depends on both the quantum state family and the specific measurement mapping states to outcome probabilities.

Demonstration

Estimate a small phase φ encoded on a qubit prepared in (|0>+|1>)/√2 and measured in the {|+>,|->} basis. The outcome probabilities p±(φ)=½[1±cos(φ)] yield a Fisher information I(φ)= [∂φ ln p+]^2 p+ + [∂φ ln p-]^2 p- = cos^2(φ)/[1−cos^2(φ)] evaluated at the operating point; comparing different measurement bases shows how measurement choice changes the Fisher information.

Misapplication

Treating Fisher information computed for one detector/POVM as if it were a measurement-independent, intrinsic property of the quantum state, or using it without accounting for measurement imperfections, correlated noise, or finite sample bias in estimators.

Consequence

Correct use allows experimental optimization of measurement settings and data analysis to approach the classical Cramér–Rao bound for that measurement, guiding estimator choice and required sample sizes.

Reversal

If one inverts the concept, one obtains a measure of insensitivity: parameters that produce identical outcome distributions for all POVM elements have zero Fisher information under that measurement.

Boundary

Applies only to the specified measurement mapping quantum states to classical probabilities and to statistical estimation of parameters; it does not by itself capture the maximum achievable information over all POVMs (the quantum Fisher information), nor nonstatistical error sources such as calibration bias or model misspecification.

Semantic Tension

Often confused with quantum Fisher information; the tension is that classical Fisher information is measurement-dependent and operational, while the quantum version is an upper bound over measurements and is a property of the state family and generators.

Synthesis

Fisher information (quantum) is the measurement-conditioned sensitivity metric: given a quantum state family and a chosen POVM, it quantifies how much classical data from that measurement can, in principle, tell you about a parameter through the classical Cramér–Rao framework.