Marco A. Rodríguez-García,Ruynet L. de Matos Filho,Pablo Barberis-Blostein
Abstract:We investigate strategies for reaching the ultimate limit on the precision of frequency estimation when the number of probes used in each run of the experiment is fixed. That limit is set by the quantum Cramér-Rao bound (QCRB), which predicts that the use of maximally entangled probes enhances the estimation precision, when compared with the use of independent probes. However, the bound is only achievable if the statistical model used in the estimation remains identifiable throughout the procedure. This in turn sets different limits on the maximal sensing time used in each run of the estimation procedure, when entangled and independent probes are used. When those constraints are taken into account, one can show that, when the total number of probes and the total duration of the estimation process are counted as fixed resources, the use of entangled probes is, in fact, disadvantageous when compared with the use of independent probes. In order to counteract the limitations imposed on the sensing time by the requirement of identifiability of the statistical model, we propose a time-adaptive strategy, in which the sensing time is adequately increased at each step of the estimation process, calculate an attainable error bound for the strategy and discuss how to optimally choose its parameters in order to minimize that bound. We show that the proposed strategy leads to much better scaling of the estimation uncertainty with the total number of probes and the total sensing time than the traditional fixed-sensing-time strategy. We also show that, when the total number of probes and the total sensing time are counted as resources, independent probes and maximally entangled ones have now the same performance, in contrast to the non-adaptive strategy, where the use of independent is more advantageous than the use of maximally entangled ones.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is how to reach the limit of precision in frequency estimation when the number of probes used in the experiment is fixed. Specifically, the paper explores the performance differences between using quantum - entangled probes and independent probes in frequency estimation, and proposes a time - adaptive strategy to overcome the non - identifiability problem of the statistical model, thereby improving the precision of frequency estimation.
### Background and Problem of the Paper
Frequency estimation is an important part of theoretical and experimental tests in physics, and is also the basis for many practical applications such as magnetometers, gravimeters and atomic clocks. When estimating an unknown frequency in a quantum system, one needs to measure the observables of the system and use these data as the input of the estimation function. The output of the estimation function represents the estimated value of the frequency. Parameter estimation theory aims to find measurement strategies and estimators that can achieve small estimation errors. Quantum mechanics not only sets the lower limit of the estimation error, but also determines the resources required to reach these lower limits. The quantum Cramér - Rao bound (QCRB) gives the ultimate quantum limit of the estimation error.
### Main Research Contents
1. **Comparison between Quantum - Entangled and Independent Probes**:
- **Independent Probes**: When using independent probes, the form of the quantum Cramér - Rao bound (QCRB) is:
\[
\text{Var}_\omega(\hat{\omega}(X_1, \ldots, X_N)) \geq \frac{1}{N t^2_{\text{prod}}}
\]
where \(X_1, \ldots, X_N\) are the results of \(N\) independent measurements, and \(t_{\text{prod}}\) is the evolution (sensing) time of the independent probes.
- **Quantum - Entangled Probes**: When using the maximally entangled state (such as the GHZ state), the form of the QCRB is:
\[
\text{Var}_\omega(\hat{\omega}(\{X^{(i)}_1, \ldots, X^{(i)}_N\}_{i = 1}^\nu)) \geq \frac{1}{\nu N^2 t^2_{\text{GHZ}}}
\]
where \(t_{\text{GHZ}}\) is the sensing time of the GHZ state. Theoretically, quantum - entangled probes can provide better estimation precision because its variance bound decreases with the square of the number of probes, that is, the Heisenberg limit.
2. **Non - identifiability Problem**:
- Frequency information is usually encoded in the relative phase \(\phi = (\omega - \omega_0)t\) of the probes. Since the relative phase is \(2\pi\) - periodic, when \(\phi\) exceeds \(\pi\), the measurement results will repeat, leading to the non - identifiability problem of the statistical model.
- For independent probes, in order to maintain the identifiability of the statistical model, the sensing time \(t\) must be limited to \(t\leq\frac{\pi}{2\Delta\Omega}\), where \(\Delta\Omega\) is the frequency interval.
- For quantum - entangled probes, since the evolution speed of the relative phase \(\phi_{\text{GHZ}} = N_{\text{GHZ}}(\omega - \omega_0)t\) is faster, the sensing time must be more strictly limited to \(t\leq\frac{\pi}{2N_{\text{GHZ}}\Delta\Omega}\).
3. **Time - Adaptive Strategy**:
- In order to overcome the non - identifiability problem, the paper proposes a time - adaptive strategy to improve the estimation precision by gradually increasing the sensing time. Specifically, in the initial stage, a shorter sensing time is used for measurement. After obtaining a preliminary estimate, the sensing time is gradually increased until the required estimation error is reached.
- This strategy can significantly improve the scaling relationship of the estimation uncertainty with the total number of probes and the total sensing time, making independent probes and maximally entangled probes have the same performance under the same resources.
### Conclusion
When considering the total number of probes and the total sensing time as fixed resources