Precision limits for time-dependent quantum metrology under Markovian noise
Luca Previdi, Francesco Albarelli · 2026
Read Precision limits for time-dependent quantum metrology under Markovian noise by Luca Previdi, Francesco Albarelli (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-18392.txt.
Introduction
Luca Previdi and Francesco Albarelli derive ultimate precision limits for estimating parameters in time-dependent quantum systems under general Markovian noise. Their 2026 work extends a minimization-over-purifications approach to time-varying channels, giving a differential bound on quantum Fisher information that semidefinite programming can compute. They prove a universal long-time scaling law: in one regime noisy scaling matches the noise-free scaling, while in the other it falls one power of time short. Their error-correction constructions meet the bounds.
Catalog record
| CREATOR | Luca Previdi, Francesco Albarelli |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 27271 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.18392
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Precision limits for time-dependent quantum metrology under Markovian noise Luca Previdi, Francesco Albarelli Abstract We derive ultimate precision bounds for estimating parameters encoded in time-dependent Hamiltonians in the presence of general Markovian noise, allowing for arbitrary adaptive protocols with fast controls and noiseless ancillas. Extending the minimization-over-purifications framework to time-varying continuous channels, we obtain a differential upper bound on the achievable quantum Fisher information (QFI) that can be evaluated at all times via semidefinite programming. For parameter-independent noise, we prove a universal long-time scaling law: if the coherent (noiseless) dynamics yields Q coh ( T ) ∼ T 2 k Q_{\mathrm{coh}}(T)\sim T^{2k} , then under Markovian noise the QFI scales at most as Q ( T ) ∼ T 2 k Q(T)\sim T^{2k} in the DHNLS regime, whereas in the DHLS regime it is fundamentally limited to Q ( T ) ∼ T 2 k − 1 Q(T)\sim T^{2k-1} . We illustrate these behaviors on paradigmatic driven-qubit sensors, exhibiting T 4 T^{4} and T 3 T^{3} scalings under dephasing and spontaneous emission, respectively. Finally, we provide explicit continuous exact and approximate quantum error correction constructions—supplemented by spin-squeezed probes—that asymptotically saturate the bounds, establishing their tightness. Introduction. —Quantum metrology exploits non-classical resources, such as entanglement and superposition, to push the boundaries of measurement precision beyond classical limitations 23 ; 45 ; 4 ; 3 ; 36 ; 9 . The extreme sensitivity of quantum systems to external perturbations, often a hurdle for quantum control, may become a powerful asset for estimating macroscopic parameters. In standard parameter estimation schemes relying on N uncorrelated probes or an interrogation time T , classical protocols are fundamentally bounded by the standard quantum limit (SQL), where precision scales as 1/\sqrt{N} or 1/\sqrt{T} . In contrast, fully quantum strategies can asymptotically achieve the Heisenberg limit (HL), improving the scaling to 1/N or 1/T 24 ; 17 ; 18 ; 19 , with far-reaching applications ranging from gravitational wave detection 6 and atomic clocks 47 to magnetometry 27 ; 2 . Despite these theoretical promises, realistic quantum sensors inevitably suffer from decoherence. Under Markovian noise, governed by the Gorini-Kossakowski-Sudarshan-Lindblad (GKSL) master equation 21 ; 31 , the performance can be severely degraded, often forcing the ultimate precision back to the SQL 13 ; 10 ; 11 . Recent systematic analyses of time-independent signals have shown that the achievable precision is fundamentally dictated by the algebraic relationship between the signal Hamiltonian and the noise jump operators 41 ; 8 ; 50 ; 44 ; 30 ; 32 ; 20 . Specifically, if a time-independent Hamiltonian lies within the span of the Lindblad operators, known as the Hamiltonian-in-Lindblad-span (HLS) condition, the ultimate precision is bounded by the SQL, which can be saturated via approximate quantum error correction (AQEC) 48 . Conversely, if the Hamiltonian-not-in-Lindblad-span (HNLS) condition holds, the HL can be perfectly recovered using exact quantum error correction (QEC) 50 . However, the metrological limits of time-dependent signals under Markovian noise remain largely unexplored. In noiseless scenarios, dynamically modulating the parameter-encoding Hamiltonian, often supplemented by adaptive control, can yield exceptional precision scalings that surpass even the standard HL in time, achieving variances that scale as 1/T^{k} with k>1 35 ; 46 ; 15 ; 34 ; 16 . Whether, and under what conditions, these exceptional scalings survive in the presence of realistic open-system dynamics represents a fundamental open question in quantum parameter estimation. This is a timely question, since time-dependent and oscillatory signals arise naturally in AC magnetometry, spectroscopy, and driven many-body sensing settings 40 ; 33 ; 22 . Further, recent theoretical studies have highlighted how oscillatory signals lead to metrological tradeoffs and optimality questions beyond the time-independent setting, especially in the broadband setting 37 ; 1 ; 12 ; 38 . In this Letter, we establish the ultimate precision bounds for time-dependent quantum metrology under general Markovian noise and provide the explicit error-correction strategies required to saturate them. By extending the minimization-over-purifications (MOP) framework 14 ; 13 ; 10 ; 8 ; 30 to time-varying continuous channels, we derive rigorous upper bounds on the quantum Fisher information (QFI) that are valid for arbitrary adaptive control strategies, which can be efficiently evaluated at all times via a semidefinite program (SDP) 7 . Crucially, we demonstrate that when the derivative of a time-dependent Hamiltonian with respect to the parameter of interest does not lie in the Lindblad span, the exceptional QFI scaling of the noiseless dynamics is asymptotically preserved. Conversely, if the derivative of such Hamiltonian lies in the Lindblad span, the noise penalty reduces the exponent of the optimal time scaling by exactly one, a bound that may still surpass standard coherent limits. Finally, extending standard recovery maps to the time-dependent regime, we show that these generalized bounds are not merely mathematical constructs, but are asymptotically saturable utilizing continuous QEC, AQEC, and spin-squeezed states 43 . Precision bounds. —We consider a time-dependent Markovian quantum channel, which depends on a parameter \omega , generated by a GKSL master equation. Denoting by \rho_{\omega}(t) the state at time t , we assume \displaystyle\dot{\rho}_{\omega}(t)= \displaystyle-i\bigl[H(\omega,t),\rho_{\omega}(t)\bigr] \displaystyle+\sum_{j=1}^{J}\!\big(L_{j}(\omega,t)\,\rho_{\omega}(t)\,L_{j}^{\dagger}(\omega,t) \displaystyle-\frac{1}{2}\bigl\{L_{j}^{\dagger}(\omega,t)L_{j}(\omega,t),\rho_{\omega}(t)\bigr\}\big), (1) where \dot{\rho}_{\omega}(t)\equiv\partial_{t}\rho_{\omega}(t) and \{\cdot,\cdot\} denotes the anticommutator. To estimate \omega we allow for a very general adaptive strategy (Fig. 1 ): the probe undergoes the evolution ( 1 ), while we can intersperse the dynamics with an arbitrary number of unitary controls acting jointly on the probe and any number of noiseless ancillas. Figure 1: Pictorial representation of the most general adaptive strategy considered in this Letter. A probe evolves with a channel \Lambda^{i}_{\omega} and then a unitary control is C_{i} applied, coupling the probe to any number of noiseless ancillas. This step can be repeated an arbitrary number of times N . Notice that, differently from 30 , the channels could be different each step, \Lambda^{i}_{\omega}\neq\Lambda^{j}_{\omega} , which is equivalent to consider a time-dependent channel. In close analogy with the time-independent case 11 ; 8 ; 30 ; 41 ; 51 ; 28 , one can upper bound the instantaneous growth of the quantum Fisher information (QFI) Q(t) achievable by such protocols as \frac{dQ(t)}{dt}\leq 4\,\min_{\kappa(t),\,\eta(t),\,\gamma(t)}\left(\bigl\|\alpha(t)\bigr\|+\bigl\|\beta(t)\bigr\|\,\sqrt{Q(t)}\right). (2) Here \|\cdot\| denotes the operator norm, and the minimization is over the auxiliary quantities \kappa(t) , \eta(t) and \gamma(t) introduced in the derivation. We use the compact notation \mathbf{L}(\omega,t)\equiv[L_{1}(\omega,t),\ldots,L_{J}(\omega,t)]^{T} and the prime (\cdot)^{\prime} denotes the derivative with respect to \omega . Moreover, we defined \alpha(t)=[\eta(t)\,\mathbb{I}+\gamma(t)\,\mathbf{L}(\omega,t)+i\,\mathbf{L}^{\prime}(\omega,t)]^{\!\dagger}[\eta(t)\,\mathbb{I}+\gamma(t)\,\mathbf{L}(\omega,t)+i\,\mathbf{L}^{\prime}(\omega,t)] and -i\,\beta(t)=H^{\prime}(\omega,t)-\frac{i}{2}(\mathbf{L}^{\prime\dagger}(\omega,t)\,\mathbf{L}(\omega,t)-\mathbf{L}^{\dagger}(\omega,t)\,\mathbf{L}^{\prime}(\omega,t))+\kappa(t)\,\mathbb{I}+\mathbf{L}^{\dagger}(\omega,t)\,\eta(t)+\eta^{\dagger}(t)\,\mathbf{L}(t)+\mathbf{L}^{\dagger}(\omega,t)\,\gamma(t)\,\mathbf{L}(t) similarly to 44 ; 30 . A proof of Eq. ( 2 ) is presented in the Supplemental Material (SM) 42 , but can alternatively be obtained from the results of 44 . We now specialize to the case of \omega -independent noise, i.e. \mathbf{L}^{\prime}(t)=0 . Moreover, we assume that for t\to\infty the Hamiltonian derivative admits the asymptotic expansion H^{\prime}(\omega,t)=t^{n}\,H_{0}(t)+O(t^{n-1}), (3) with H_{0}(t) bounded for all t . This assumption is discussed in more detail in the SM 42 and is valid for a wide class of Hamiltonians. In this regime it is convenient to introduce \alpha_{0}(t)=[\eta(t)\,\mathbb{I}+\gamma(t)\,\mathbf{L}(t)]^{\!\dagger}[\eta(t)\,\mathbb{I}+\gamma(t)\,\mathbf{L}(t)],-i\,\beta_{0}(t)=t^{n}\,H_{0}(t)+\kappa(t)\,\mathbb{I}+\mathbf{L}^{\dagger}(\omega,t)\,\eta(t)+\eta^{\dagger}(t)\,\mathbf{L}(t)+\mathbf{L}^{\dagger}(\omega,t)\,\gamma(t)\,\mathbf{L}(t). We say that the system satisfies the derivative of the Hamiltonian in Lindblad span (DHLS) condition if there exist T_{0} and a measurable set I\subset(T_{0},\infty) whose complement has zero (Lebesgue) measure, such that for every t\in I one