arXiv:2610.32724
Optimal probes for estimating the dephasing of a quantum spin system
Leonardo Ratini, Stefano Carretta, Francesco Albarelli
Abstract
We investigate the optimal estimation of dephasing in finite-dimensional quantum spin systems. We first discuss some properties of the optimal probe states and present an efficient numerical algorithm to compute the optimal precision in dephasing estimation and identify the optimal probe states. For the dephasing channel arising from Gaussian phase diffusion, a spin cat state is known to be optimal for small dephasing. We analytically determine the range of dephasing strengths over which the cat state remains globally optimal for spins $J=1$ and $J=3/2$, and derive upper bounds on its optimality region for larger spins. We then compare numerically several probe states proposed in the literature, showing in particular that the simple class of probes obtained from uniform superpositions of all levels performs near-optimally for stronger dephasing. Finally, we apply our framework to a central-spin model coupled to a nuclear-spin bath, whose dynamics we simulate using the cluster-correlation expansion. Our results show that, while the cat state is optimal for weak Gaussian dephasing, pulse-controlled non-Markovian dynamics can favor states with amplitudes distributed over multiple spin levels, and in some cases the uniform state outperforms the other fixed probes considered. This shows how the choice of probe state depends both on the dephasing strength and on the dynamics.
AI Summary
The paper studies which finite-spin probe states maximize quantum Fisher information (QFI) for estimating parameters encoded in pure dephasing, emphasizing Gaussian phase diffusion. For general differentiable dephasing channels $\Lambda_\theta(\rho)=E_\theta\circ\rho$, it shows that ancillary entanglement provides no advantage, probe phases are irrelevant, and QFI is concave in probe populations. Concavity permits reflection-symmetric probes when the channel has that symmetry. Specializing the channel-QFI optimization gives a second-order cone program whose dual variables directly yield optimal populations. For Gaussian diffusion, $(E_\Delta)_{mn}=e^{-\Delta(m-n)^2}$, concavity turns an endpoint-derivative calculation into a global optimality test for extremal spin-cat probes. The exact cat-optimal regions are $0<\Delta\lesssim0.1423$ for $J=1$ and $0<\Delta\lesssim0.0650$ for $J=3/2$. Restricted-subspace and two-level comparisons give larger-spin upper bounds proportional to $J^{-2}$; for integer spins, $\Delta_{\mathrm{cat}}(J)\leq0.1423/J^2$. Numerically, uniform superpositions perform near-optimally at stronger dephasing, while an optimized one-parameter cat–uniform interpolation spans both regimes. With generator $J_z^2$, conventional cats carry no information, and populating increasingly separated levels can reduce precision. Cluster-correlation-expansion simulations of a spin-$3/2$ probe coupled to 50 nuclear spins compare free and prescribed pulse-controlled evolution. Under pulses, coherence revivals can favor multilevel probes, with the uniform state outperforming the other fixed probes considered. These results use an approximately dephasing channel. Optimization is local in the unknown parameter; QFI-rate comparisons assume independent bath reinitialization and neglect preparation/readout/reset costs. Experimentally constrained measurements and pulse-sequence optimization remain open.
Error found: The pulse sequence in Eq. (ECHO_time_evolution) has net permutation V2 even at zero duration, so its output cannot have the claimed population-preserving Schur-channel form without an unstated output-frame correction.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.