---
title: Quantum Cramér–Rao Bound
url: https://www.emergentmind.com/topics/quantum-cramer-rao-bound-f6e16788-c149-495a-ba85-d56a64cf172b
type: topic
---

# Quantum Cramér–Rao Bound

The quantum Cramér–Rao bound (QCRB) establishes the ultimate precision limit imposed by quantum mechanics on the estimation of unknown parameters encoded in quantum states. By optimizing over all positive operator-valued measures (POVMs), the QCRB generalizes the classical Cramér–Rao bound to the quantum regime, setting a lower bound on the mean-square error that any unbiased estimator can achieve. While the existence of a QCRB-saturating measurement is guaranteed for arbitrary single-parameter quantum models, the efficient implementation of such measurements—especially in multipartite or high-dimensional systems—can be challenging. Recent results demonstrate that local operations with one-way classical communication (LOCC) are sufficient to saturate the QCRB for broad classes of states, including all pure states and certain rank-two mixed states, while measurements restricted to local observables without communication generally fail to attain this quantum limit [1809.06017].

## 1. Formal Structure of the Quantum Cramér–Rao Bound

Consider a quantum state $\rho_\theta$ that depends smoothly on a real parameter $\theta$. Performing $n$ independent preparations and measurements, any unbiased estimator $\hat\theta$ satisfies the classical Cramér–Rao bound:
\[
\operatorname{Var}(\hat\theta) \geq \frac{1}{n F_{\mathrm{cl}}(\theta)},
\]
where $F_{\mathrm{cl}}$ is the Fisher information of the observed data. Optimizing over all possible POVMs yields the QCRB:
\[
\operatorname{Var}(\hat\theta) \geq \frac{1}{n J(\theta)},
\]
with the quantum Fisher information (QFI) defined by
\[
\frac{\partial \rho_\theta}{\partial\theta} = \frac{1}{2}\left(L_\theta \rho_\theta + \rho_\theta L_\theta\right),\qquad
J(\theta) = \operatorname{Tr}[\rho_\theta L_\theta^2].
\]
Here, $L_\theta$ is the symmetric logarithmic derivative (SLD) [1809.06017].

## 2. Criteria for QCRB Saturation by Measurements

Given a POVM $\{E_x\}$, the observed Fisher information $F_{\mathrm{cl}}$ satisfies
\[
F_{\mathrm{cl}}(\{p_x\}) \leq J(\theta),
\]
with equality if and only if, for every outcome $x$:
- $\operatorname{Re} \operatorname{Tr}[E_x L_\theta \rho_\theta] = \operatorname{Tr}[E_x \partial_\theta \rho_\theta]$,
- $\operatorname{Im} \operatorname{Tr}[E_x L_\theta \rho_\theta] = 0$,
- $E_x^{1/2} \rho_\theta^{1/2} \propto E_x^{1/2} L_\theta \rho_\theta^{1/2}$,

plus a regularity condition: if $\operatorname{Tr}[E_x \rho_\theta]=0$, then $E_x^{1/2} L_\theta \rho_\theta^{1/2}=0$. An equivalent spectral criterion involves constructing operators $M_{ij} = |\psi_i\rangle\langle\psi_j| L_\theta - L_\theta |\psi_i\rangle\langle\psi_j|$, and requiring $E_x^{1/2} M_{ij} E_x^{1/2} = 0$ for all $i, j$ and all $x$ [1809.06017].

## 3. LOCC Protocols for QCRB Saturation in Multipartite Settings

For arbitrary pure states and rank-two mixed states (with fixed eigenbasis and varying probabilities), a POVM based on LOCC can be constructed to saturate the QCRB:
- Decompose the parameter-dependent operator $M$ (pure: $M=|\psi\rangle\langle\psi^\perp| - |\psi^\perp\rangle\langle\psi|$; rank-two mixed: $M=|\psi_0\rangle\langle\psi_1|$ up to scalar).
- Sequentially, for each subsystem $s_k$, construct an orthonormal measurement basis such that the diagonal elements of the relevant reduced $M$ vanish. Basis selection at subsystem $k$ depends on measurement outcomes at subsystems $1,\dots,k-1$—an adaptive, one-way LOCC tree.
- The final global measurement is built from tensor products of these local projectors. This satisfies $E_{x_1\ldots x_n}^{1/2} M E_{x_1\ldots x_n}^{1/2} = 0$ for all composite outcomes, ensuring the classical Fisher information saturates $J(\theta)$ [1809.06017].

## 4. Non-Saturability of Product Local Measurements

POVMs restricted to purely local product form,
\[
E_{x_1\ldots x_n} = E^{(1)}_{x_1} \otimes \cdots \otimes E^{(n)}_{x_n},
\]
cannot generically saturate the QCRB in multipartite systems. Achieving $E M E=0$ for all outcome tuples imposes a large set of bilinear constraints that outstrip the local degrees of freedom as system size increases. Explicit counterexamples exist, including situations where discrimination of three non-LOCC-distinguishable Bell states is required, illustrating the impossibility for general rank-two states [1809.06017].

## 5. Illustrative Examples

### GHZ State
- $|\psi_\theta\rangle = (|0\rangle^{\otimes n} + e^{in\theta}|1\rangle^{\otimes n})/\sqrt{2}$,
- SLD: $L_\theta \propto i (|1\ldots1\rangle\langle0\ldots0| - |0\ldots0\rangle\langle1\ldots1|)$,
- Both global projections onto $L_\theta$'s eigenbasis and product LM in $\{|+\rangle,|-\rangle\}$ saturate the QCRB with $(\Delta\theta)^2 \sim 1/n^2$.

### Four-Qubit Open-Chain Example
- For a nearest-neighbor $XX$ interaction and Dicke-1 input, numerical analysis shows product LM in fixed bases fails, but a one-way LOCC protocol attains the QCRB, with measurement adaptivity implemented by feeding outcomes forward to subsequent sites [1809.06017].

## 6. Physical and Practical Implications

LOCC protocols allow ultimate precision in quantum parameter estimation without requiring global entangled measurements. The adaptive measurement tree is implementable on various platforms equipped with local control, fast readout, and feed-forward (e.g., trapped ions, Rydberg arrays, superconducting qubits). LOCC strategies naturally extend to decoherence-affected settings by combining with dynamical decoupling or logical encoding [1809.06017].

## 7. Summary Table: QCRB Saturation in Various Scenarios

| Scenario                        | QCRB-Saturating Measurement | Efficient (LOCC) Realizable?         |
|----------------------------------|----------------------------|--------------------------------------|
| Arbitrary single-parameter case  | Some POVM always exists    | Not always efficient                 |
| Pure/rank-two mixed, multipartite| Adaptive LOCC (one-way)    | Yes                                  |
| Product local measurements only  | Rarely (special symmetry)  | No, not generally                    |

Efficiently implementing the QCRB-saturating measurement in realistic systems is feasible for broad classes of states when an adaptive LOCC scheme is adopted, and infeasible when restricting to non-communicating local measurements [1809.06017].

Source: https://www.emergentmind.com/topics/quantum-cramer-rao-bound-f6e16788-c149-495a-ba85-d56a64cf172b