arXiv:2609.25699
Post-Selection-Based Stochastic Quantum Battery
Aman Verma, Roson Nongthombam, Amarendra K. Sarma
Abstract
Quantum batteries are quantum systems that store useful energy, which can be subsequently extracted to perform work. In this paper, we investigate the charging dynamics of the quantum battery under continuous measurements and post-selection. By post-selecting the trajectories with no quantum jump $|1\rangle \rightarrow |0\rangle$ in a three-level system, the dynamics is effectively confined to a two-level $(|2\rangle-|1\rangle)$ manifold. We analyze the resulting charging dynamics under both $σ_x$- and $σ_y$-driving protocols and investigate the roles of detuning, coherent driving, non-linearity, and measurement backaction. The generation and dynamics of quantum coherence, which plays a pivotal role in quantum battery, are also analyzed under post-selection. The advantage and comparison of our results over those of a generic two-level system and with the Lindblad dynamics are also highlighted. Our work explores the interplay between the measurement axis and the driving protocols that govern the charging rate, ergotropy, and quantum coherence of the non-Hermitian, highly tunable, and controllable quantum battery.
AI Summary
The paper studies how continuous monitoring and post-selection affect extractable work in a driven quantum battery. The model is a three-level system with decay channels $|2\rangle\to|1\rangle$ and $|1\rangle\to|0\rangle$. Homodyne monitoring of the upper transition and rejection of trajectories containing the lower transition leave a conditioned two-level manifold. Charging is characterized using ergotropy relative to the rotating-frame driven Hamiltonian, its time derivative, and computational-basis $l_1$ coherence. Kraus-operator Bayesian simulations compare $\sigma_x$ and $\sigma_y$ drives, trajectory-averaged ergotropy, and ergotropy of the averaged state. Representative parameters are $\Gamma_g=2\,\mathrm{MHz}$, $\Gamma_e=0.2\,\mathrm{MHz}$, driving strengths $\Omega=0.3$ or $2\,\mathrm{MHz}$, and detunings $\Delta=0$, $0.5$, or $1.5\,\mathrm{MHz}$. At resonance, mean trajectory ergotropy remains approximately constant, whereas mean-state ergotropy decays monotonically or with damped oscillations across the Liouvillian exceptional-point transition. Positive detuning produces oscillatory charging in the post-selected model; larger detuning gives faster saturation. A comparison with an otherwise monitored two-level system without the lower decay channel uses opposite detuning signs and reports higher ergotropy in the post-selected case. Analytical Bloch equations identify variance and covariance contributions to conditional charging power. The distinction between mean ergotropy and mean-state ergotropy follows the established convexity bound $\langle\mathcal E[\rho_c]\rangle\geq\mathcal E[\langle\rho_c\rangle]$. The study assumes efficient detection and restricts its numerical conclusions to this small-system model; inefficient detection is deferred to separate work. It also states that post-selected ensemble averages require the full Bayesian selection procedure rather than simply averaging the no-jump stochastic equations.
Central claim may not hold: The claimed performance advantage is only conditional on survival, and its headline comparison also gives the post-selected battery a higher initial ergotropy by reversing the detuning. The displayed Liouville matrix omits the population-loss entry −Γ_g and therefore does not represent the stated trace-decreasing no-jump generator. Correct survival weighting makes normalized trace-decreasing Liouvillian evolution reproduce the survivor ensemble and cancels the explicit covariance and variance contributions used in the paper's mean charging-power explanation. Under the stated efficient-measurement and pure-state assumptions, the correctly conditioned mean ergotropy is identical for the two drive phases, contrary to the claimed drive-dependent mean-ergotropy effects. The TLS steady-state ergotropy-change formula has 16Ω² where substitution of the paper's own expressions gives 12Ω².
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.