NABL & ISO/IEC 17025:2017 Calibration Reference Portal
Official Lab Services: multitek.in
Solutions for: Quality Heads & Auditors Shop Floor Inspectors Tool Room Technicians Engineering Students

Quantum Metrology Explained: Fisher Information and Measurement Limits


Quantum Metrology: Fisher Information, Cramér-Rao Bound and the Physical Limits of Measurement

Last reviewed: October 5, 2026

Every measurement has a limit. Sometimes that limit comes from the instrument, calibration, environment or data-processing method. At much higher levels of precision, another question appears: even with an ideal experiment, how much information can the measurement outcomes contain about the quantity we want to determine?

Quantum metrology studies that question. It uses parameter-estimation theory and quantum mechanics to determine the best precision that a physical measurement strategy can, in principle, achieve. Three concepts sit at the center of this subject: Fisher information, the Cramér-Rao bound, and quantum Fisher information.

The important metrology point is that these are not substitutes for calibration, traceability or a measurement-uncertainty budget. They describe a different layer of the measurement problem: how precisely a parameter can be inferred from measurement statistics.

What is quantum metrology?

Quantum metrology is the study of precision measurement when the probe, measurement process or noise must be described using quantum mechanics. The unknown quantity might be a phase shift, frequency, displacement, magnetic field, optical path difference or another parameter that changes the state of a physical system.

A useful way to view the complete measurement chain is:

unknown parameter → physical interaction → state change → measurement outcomes → probability distribution → estimator → parameter estimate

Quantum metrology asks how much information is available at the measurement stage and whether nonclassical resources such as squeezing or entanglement can increase it.

Start with parameter estimation

Suppose the unknown quantity is represented by θ. A measurement produces an outcome x. Because measurements can fluctuate, we describe the possible outcomes by a conditional probability distribution:

p(x|θ)

This means the probability of obtaining outcome x when the true parameter has value θ.

If a small change in θ produces a large and measurable change in the distribution of x, the experiment contains substantial information about θ. If the distributions for nearby values of θ are almost indistinguishable, the parameter is difficult to estimate precisely.

What is Fisher information?

Fisher information quantifies how strongly the probability distribution of the measurement outcomes depends on the parameter being estimated.

For discrete outcomes, the Fisher information can be written as:

F(θ) = Σx p(x|θ)[∂ ln p(x|θ)/∂θ]2

The equation is easier to understand conceptually than it may first appear. The derivative asks how rapidly the likelihood of the observed result changes when θ changes. The squared term ensures that changes in either direction contribute positively. The probability weighting accounts for how often each outcome is expected to occur.

Therefore:

  • High Fisher information means nearby parameter values produce measurement statistics that are relatively easy to distinguish.
  • Low Fisher information means the distributions overlap strongly, so it is harder to tell which parameter value produced the observations.

Fisher information is not the same thing as calibration accuracy or total measurement uncertainty. It is a statistical property of a measurement model and estimator problem.

The Cramér-Rao bound turns information into a precision limit

Once Fisher information is known, the Cramér-Rao bound provides a lower bound on the variance of an unbiased estimator under the required regularity conditions.

For ν independent repetitions:

Var(θ̂) ≥ 1/[νF(θ)]

Here:

  • θ̂ is the estimate of the unknown parameter.
  • ν is the number of independent measurement repetitions.
  • F(θ) is the Fisher information for one repetition under the chosen measurement model.

Expressed as a standard deviation:

Δθ ≥ 1/√[νF(θ)]

The result has an important interpretation. More Fisher information allows a smaller statistical uncertainty in the estimate. Repeating an independent experiment also improves precision, with the familiar square-root dependence on the number of trials.

The bound does not say that every estimator will achieve this precision. Saturation depends on the estimator, statistical model, amount of data and assumptions behind the bound.

Fisher information as statistical distinguishability

Another way to understand Fisher information is to compare two nearby models:

p(x|θ) and p(x|θ + dθ).

If they are almost identical, many observations may be required to distinguish them. If the distributions separate rapidly as θ changes, fewer measurements are needed.

This is why Fisher information is closely connected to statistical distance. In quantum theory that geometric idea becomes especially useful because the parameter changes a quantum state before a measurement converts the state into classical outcomes.

From classical probabilities to quantum states

In a quantum experiment, the parameter may be encoded in a state written as ρ(θ), or |ψ(θ)⟩ for a pure state. A particular quantum measurement then generates the ordinary probabilities p(x|θ).

This introduces an additional optimization problem. The same quantum state can be measured in different ways, and different measurements can extract different amounts of Fisher information.

That leads to a central question in quantum metrology:

What is the greatest Fisher information obtainable from this parameter-dependent quantum state when the measurement is optimized?

The answer is described by quantum Fisher information.

What is quantum Fisher information?

Quantum Fisher information, usually written FQ, characterizes the maximum classical Fisher information that can be obtained from a quantum state for the parameter of interest when the measurement is optimized over the allowed quantum measurements.

For a single parameter, the quantum Cramér-Rao relation is commonly written as:

Var(θ̂) ≥ 1/[νFQ(θ)]

For a pure parameterized state, one useful expression is:

FQ = 4[⟨∂θψ|∂θψ⟩ − |⟨ψ|∂θψ⟩|2]

This equation measures how quickly the state moves through quantum-state space as the parameter changes, after removing changes that do not make the physical state more distinguishable.

For certain pure-state unitary parameter encodings generated by an observable G, the result reduces to:

FQ = 4 Var(G)

This compact relation helps explain why quantum fluctuations can become a resource for sensing. A probe whose state changes more distinguishably under the parameter-generating interaction can potentially support more precise estimation.

Standard quantum limit and Heisenberg scaling

A common benchmark is the standard quantum limit, often associated with independent probes or shot-noise-limited measurements. Under a suitable definition of the resource N, the precision commonly scales as:

Δθ ∝ 1/√N

Ideal quantum strategies can, in some settings, exhibit Heisenberg-type scaling:

Δθ ∝ 1/N

The difference is substantial. Increasing the resource by a factor of 100 would improve a 1/√N-limited uncertainty by a factor of 10, while ideal 1/N scaling would improve it by a factor of 100.

However, the resource must be defined carefully. It might mean photon number, atoms, channel uses, interrogation time, total energy or another experimentally relevant quantity. Claims of "Heisenberg-limited" performance are not meaningful unless the resource accounting and assumptions are stated.

Why quantum advantage is difficult in real instruments

The ideal scaling picture is not the complete story. Real experiments contain loss, decoherence, phase diffusion, detector inefficiency, imperfect state preparation, mode mismatch and technical noise. These effects can strongly reduce quantum enhancement.

For optical metrology, photon loss is particularly important. A fragile nonclassical state may contain high theoretical quantum Fisher information but lose much of its advantage before detection. General studies of noisy quantum metrology show that uncorrelated noise can remove ideal asymptotic Heisenberg scaling in many important cases, even though a useful constant-factor advantage over a classical strategy may remain.

That is why a credible quantum-metrology comparison should specify:

  • the classical reference strategy,
  • the quantum probe or measurement,
  • the resource being counted,
  • the loss and noise model,
  • the detector assumptions, and
  • the actual performance metric.

Optical interferometry is a natural quantum-metrology problem

An interferometer converts a change in optical phase into a change in an observable output. This makes it a natural platform for parameter estimation.

A simplified measurement chain is:

light source → state preparation → interferometer → phase shift → interference → photodetection → estimator

The unknown phase may arise from a displacement, refractive-index change, path-length difference or another physical effect. The phase changes the probability distribution of the detector outcomes. Fisher information tells us how much those outcomes reveal about the phase.

Quantum optical techniques try to reshape the input fluctuations or correlations so that the measurement carries more usable information about that phase.

How squeezed light improves interferometric measurement

Squeezed light redistributes quantum uncertainty between conjugate field quadratures. The uncertainty of one quadrature can be reduced below the ordinary vacuum level while the conjugate quadrature increases.

If the reduced-noise quadrature is aligned with the quantity that carries the desired signal, an interferometer can achieve lower quantum measurement noise than the corresponding unsqueezed strategy.

This does not remove quantum uncertainty. It redistributes it in a controlled way.

It also does not remove classical noise sources. Laser noise, thermal effects, mechanical disturbances, optical loss, alignment errors and environmental noise can still limit the measurement.

LIGO provides a real example of quantum-enhanced interferometry

Gravitational-wave detectors show why the distinction between theoretical quantum limits and practical instrumentation matters.

A gravitational wave produces an extremely small differential change in interferometer arm lengths. That changes the relative optical phase and therefore the detector output. At high sensitivity, quantum fluctuations of the light contribute significantly to the detector noise.

In work published by the LIGO O4 Detector Collaboration in 2023, frequency-dependent squeezing reduced both shot noise and quantum radiation-pressure noise across a broad frequency range. Near 1 kHz, the reported noise-amplitude reductions were about 4.0 dB at Hanford and 5.8 dB at Livingston. The authors also reported a 15% to 18% improvement in detector range relative to operation without squeezing for the low-frequency improvement discussed in that study.

This is an important practical example because the quantum technique did not make the interferometer perfect. Instead, it reduced a specific physical noise contribution enough to improve the overall measurement capability.

Quantum metrology is not the same as measurement uncertainty

This distinction is essential for calibration and industrial metrology.

Fisher information and the Cramér-Rao bound answer a question such as:

Given this statistical measurement model, how precisely can the parameter be estimated?

A conventional measurement-uncertainty evaluation asks a broader practical question:

What uncertainty should be associated with the reported measurement result, considering the relevant input quantities, effects, calibration information and measurement model?

A measurement uncertainty budget can include contributions from reference standards, calibration, repeatability, resolution, drift, alignment, environment, corrections and correlations. These are not automatically contained in a Fisher-information calculation.

The JCGM Guide to the Expression of Uncertainty in Measurement applies general rules for evaluating and expressing uncertainty across fields ranging from industrial measurement to fundamental research. The current JCGM publication set also includes the 2026 Amendment 1 to JCGM 100 on nonlinearity in measurement models and GUM-5:2026 containing examples.

Quantum estimation theory therefore complements conventional metrology. It does not replace it.

Quantum Fisher information does not establish traceability

Metrological traceability is a property of a measurement result. It requires a documented unbroken chain of calibrations, with each calibration contributing to the measurement uncertainty.

A sensor could have exceptional Fisher information and still lack a demonstrated traceability chain. Likewise, a well-calibrated instrument can be fully traceable while operating far from any fundamental quantum precision limit.

These concepts belong to different layers of measurement science:

Question Main framework
How much information do the observed data contain about a parameter? Fisher information
What statistical variance is achievable by an estimator? Cramér-Rao bound
What information could an optimized quantum measurement extract? Quantum Fisher information
What uncertainty should accompany the reported measurement result? GUM measurement-uncertainty framework
How is the result related to a stated reference or the SI? Metrological traceability

Where quantum metrology is moving

Quantum sensing is increasingly moving from laboratory demonstrations toward practical instruments. A 2025 NASA review describes active development across atom interferometry, squeezed-light sensing, superconducting detectors, optical clocks and correlated-photon metrology, while emphasizing engineering constraints such as size, mass, power and environmental robustness.

This transition is important. A theoretical increase in quantum Fisher information is only the first step. A practical metrology system must also preserve the quantum state, perform a suitable measurement, control systematic effects, quantify uncertainty and connect the final result to an appropriate reference framework.

A practical way to read quantum-metrology claims

When a paper or product claims quantum-enhanced sensitivity, check six things:

  1. What is the measurand or parameter? Phase, frequency, field, displacement or something else?
  2. What is the classical benchmark? A coherent state, independent particles or another reference strategy?
  3. What resource is being counted? Photons, atoms, energy, time or repetitions?
  4. What metric improved? Variance, standard deviation, Fisher information, signal-to-noise ratio or bandwidth?
  5. Were loss and technical noise included? Ideal theory and an operating instrument are not the same problem.
  6. How is the final measurement result validated? Quantum enhancement does not eliminate the need for uncertainty evaluation and traceability.

Key takeaway

Quantum metrology is fundamentally an information-extraction problem. A physical parameter changes a quantum state. A measurement converts that change into outcome probabilities. Fisher information measures how informative those outcomes are. The Cramér-Rao bound converts that information into a lower bound on estimator variance. Quantum Fisher information asks how much information the quantum state could provide when the measurement is optimized.

Quantum resources can improve that chain, but only under defined assumptions and resource constraints. Loss, decoherence and technical imperfections can reduce or even eliminate the ideal scaling advantage.

For practical metrology, one final distinction must remain clear: the fundamental statistical precision limit is not the same thing as the uncertainty of a traceable measurement result. A complete measurement system has to address both.

Frequently asked questions

What does Fisher information mean in metrology?

Fisher information measures how strongly the probability distribution of measurement outcomes changes with the parameter being estimated. Greater Fisher information generally permits a lower statistical variance for an estimator, subject to the assumptions of the Cramér-Rao bound.

What is the Cramér-Rao bound?

The Cramér-Rao bound gives a lower bound on the variance of an unbiased estimator under appropriate regularity conditions. For independent repetitions, it is commonly expressed as Var(θ̂) ≥ 1/[νF(θ)].

What is quantum Fisher information?

Quantum Fisher information is the maximum Fisher information obtainable from a parameter-dependent quantum state when the quantum measurement is optimized. It is used to define the quantum Cramér-Rao precision bound.

Is the Heisenberg limit always achievable?

No. Heisenberg-type scaling is an ideal result that depends on the sensing model, resource definition and noise assumptions. Loss and decoherence can substantially change the achievable scaling and may leave only a constant-factor quantum advantage.

Does squeezed light remove measurement uncertainty?

No. Squeezing can reduce quantum fluctuations in a selected quadrature and therefore improve a specific measurement, but other quantum and classical noise sources remain. Calibration, uncertainty evaluation and traceability are still required when applicable.

Is quantum Fisher information part of a calibration uncertainty budget?

Not normally as a direct replacement for uncertainty components. Quantum Fisher information describes an estimation limit associated with a quantum state and measurement problem. A calibration uncertainty budget evaluates the uncertainty associated with the reported measurement result using the relevant measurement model and input quantities.

References

  • Marco Barbieri, "Optical Quantum Metrology," PRX Quantum 3, 010202 (2022).
  • S. L. Braunstein and C. M. Caves, "Statistical Distance and the Geometry of Quantum States," Physical Review Letters 72, 3439 (1994).
  • M. Szczykulska, T. Baumgratz and A. Datta, "Multi-parameter quantum metrology," Advances in Physics: X 1, 621-639 (2016).
  • D. Ganapathy et al., LIGO O4 Detector Collaboration, "Broadband Quantum Enhancement of the LIGO Detectors with Frequency-Dependent Squeezing," Physical Review X 13, 041021 (2023).
  • JCGM 100:2008, "Evaluation of measurement data - Guide to the expression of uncertainty in measurement," including Amendment 1:2026.
  • JCGM GUM-5:2026, "Guide to the expression of uncertainty in measurement - Part 5: Examples."
  • NIST, "Metrological Traceability: Frequently Asked Questions and NIST Policy."
  • C. R. Mercer et al., "Quantum sensing for NASA science missions," EPJ Quantum Technology 12, 56 (2025).