Optimal measurement window, convergence rates, and beyond-Ehrenfest behavior for quantum trajectories of repeatedly measured quantum maps

For a semiclassical particle on the quantized torus $H_N$ (semiclassical parameter $h=(2\pi N)^{-1}$) that is repeatedly indirectly measured by Kraus operators $\mathrm{Op}_N(f_q)$ and evolved between measurements by a metaplectic operator or a quantized Hamiltonian, Borns-Weil and Oltman proved that the law $P^{(n)}_{N,\rho}$ of the quantum measurement trajectory converges in total variation to the classical trajectory law $P^{(n)}_\mu$, but only for measurement counts $n<\frac{\delta\log(N/C_0)}{\Gamma+\varepsilon}$ and at the algebraic rate $O(N^{-\min(1-2\delta,\theta)})$, where $\Gamma$ is the Lyapunov exponent of the classical dynamics and $(\delta,\theta)$ quantify the semiclassical concentration of the initial state. The authors state that these estimates are far from sharp, that the admissible $(\delta,\theta)$ ranges are not claimed optimal, and that the statistics of the measurement trajectories beyond the Ehrenfest window are uncontrolled. The problem is to establish the sharp quantitative form of this quantum-classical correspondence: the optimal growth of the maximal measurement count with $N$ (including a rigorous lower bound showing where correspondence provably fails), the optimal convergence rate together with the largest admissible $(\delta,\theta)$ ranges for coherent and torus-concentrated states, and the characterization of the observed-position statistics beyond the Ehrenfest window.

Background

A quantum trajectory is the record of outcomes of a quantum system subjected to repeated measurement. Mathematically, a quantum instrument is specified by Kraus operators (bounded operators $A_q$ indexed by outcomes $q\in\Omega$ with $\int A_q^*A_q\,d\nu(q)=\mathrm{Id}$), which describe indirect, weak measurement; the sequence $(q_0,\dots,q_n)$ of recorded values is a random variable whose distribution is computable from the evolved density operator.

Borns-Weil and Oltman (the source paper, published as Annales Henri Poincaré 25, 3699-3738, 2024) study this in the discrete semiclassical setting of the quantized torus. For $d$ degrees of freedom, the torus $\mathbb{T}^{2d}=\mathbb{R}^{2d}/\mathbb{Z}^{2d}$ is quantized on the $N^d$-dimensional Hilbert space $H_N$ with semiclassical parameter $h=(2\pi N)^{-1}$ and a discrete position basis; observables are obtained by $\mathrm{Op}_N$-quantization of smooth phase-space symbols. Between measurements the state is evolved by a unitary $U$: either the metaplectic quantization $\widetilde M_N$ of an integer symplectic matrix $M\in\mathrm{Sp}(2d,\mathbb{Z})$ (a 'quantum map', e.g. Arnold's cat map), or the Schrödinger propagator $e^{-2\pi iNP}$ of a quantized Hamiltonian $P$. Writing $\Phi^*_{N,q}(\rho)=U\,\mathrm{Op}_N(f_q)\rho\,\mathrm{Op}_N(f_q)^*U^{-1}$, the joint law of the measurement record is

$$P^{(n)}_{N,\rho}(F)=\int_F \mathrm{tr}_{H_N}\!\left(\Phi^*_{N,q_n}\circ\cdots\circ\Phi^*_{N,q_0}(\rho)\right)d\nu(q_0)\cdots d\nu(q_n).$$

The classical counterpart is built from the initial state's semiclassical defect measure (also called a Wigner measure): a probability measure $\mu$ on phase space such that $\mathrm{tr}(\rho_N\mathrm{Op}_N(a))\to\int a\,d\mu$ for all smooth symbols $a$. One interprets $\mu$ as the distribution of the initial classical phase point $\zeta$; the classical particle evolves by the flow $\varphi^t$ associated to $U$ ($M^t$ or $\exp(tH_p)$), and outcomes are drawn with density $|f_{q_k}(\varphi^k(\zeta))|^2$, giving a 'classical trajectory law' $P^{(n)}_\mu$. The classical measurement does not disturb the particle; the quantum one does, and the question is how the two laws on records compare as $N\to\infty$.

Two semiclassical mechanisms control the answer. First, Egorov's theorem (propagation of observables along the classical flow) holds only up to the Ehrenfest time $t_E\sim\frac{1}{\Gamma}\log(1/h)$, where $\Gamma$ is the Lyapunov exponent of the flow: beyond it, derivatives of symbols composed with the flow grow too fast. Second, to let the number of measurements $n$ grow with $N$ one must control how well the initial states concentrate, encoded in the notion of a uniform $(\delta,\theta)$-defect measure: $\mathrm{tr}(\rho_N\mathrm{Op}_N(a))=\int a\,d\mu+O(h^\theta)$ for all $h$-dependent symbols $a$ whose derivatives grow at most like $h^{-\delta}$ (symbol class $S_\delta$). Larger admissible $\delta,\theta$ mean stronger control. For generalized coherent states on $\mathbb{R}^d$, for their torus analogs, and for $\beta$-concentrated torus states (packets concentrated at scale $h^\beta$ in position), explicit admissible ranges are known: for torus coherent-type states $\delta<(d+2)^{-1}$, $\theta<1-(d+2)\delta$; for $\beta$-concentrated states $\delta<\min(\beta,(1-\beta)/(d+1))$, $\theta<\min(\beta-\delta,1-\beta-(d+1)\delta)$.

The paper's Theorem 1 then states: if $\mu$ is a $(\delta,\theta)$-defect measure, then for every $\varepsilon>0$ there are constants $C_0,C_1$ such that

$$n<\frac{\delta\log(N/C_0)}{\Gamma+\varepsilon}\quad\Longrightarrow\quad\bigl\lVert P^{(n)}_{N,\rho}-P^{(n)}_\mu\bigr\rVert_{\mathrm{TV}}\le C_1\,N^{-\min(1-2\delta,\theta)}.$$

This strengthens the fixed-time PDE result of Benoist-Fraas-Fröhlich (J. Math. Phys. 63, 062101, 2022) by allowing the time horizon to grow logarithmically with $N$ and by relating the window to the chaoticity ($\Gamma$) of the dynamics. Numerical simulations for the cat map ($\Gamma\approx 0.96$) confirm the qualitative transition: the observed positions track the classical prediction until $n$ near the Ehrenfest time and then decohere toward the uniform distribution; the convergence in $N$ is exponentially slow (the sufficient condition is roughly $N\gtrsim 50^n$ for the cat map, which the authors call 'far from sharp'); elliptic (zero-Lyapunov) maps agree for more steps.

The authors explicitly disclaim sharpness of the window, of the algebraic rate, and of the $(\delta,\theta)$ ranges, and leave the beyond-Ehrenfest regime uncontrolled; they also note that the $N^2$ simulation cost versus the logarithmic window makes larger-$N$ numerics infeasible. An adjacent but distinct line of work (Hernández-Ranard-Riedel; Galkowski-Zworski-Huang) proves beyond-Ehrenfest correspondence for observables of open systems governed by Lindblad evolution (the standard Markovian master equation for a system coupled to an environment) when the environment-induced diffusion strength satisfies $D\gg\hbar^{4/3}$, with the $\hbar^{4/3}$ threshold recently proven sharp in that setting. Those results concern continuously coupled systems at the observable level with noise scaled in $h$, not the discrete measurement-record law of the quantized-torus setting, where the measurement probes are fixed as $N\to\infty$; they frame but do not answer the source's questions.

Problem Statement

Work in the setting of the source paper: $H_N$ is the quantized $2d$-torus with semiclassical parameter $h=(2\pi N)^{-1}$; the particle is measured at each step by Kraus operators $\mathrm{Op}_N(f_q)$, $q\in\Omega$, whose smooth symbols satisfy $\int_\Omega\mathrm{Op}_N(f_q)^*\mathrm{Op}_N(f_q)\,d\nu(q)=\mathrm{Id}$ with uniformly bounded derivatives; between measurements the state is evolved by $U$, either the metaplectic quantization $\widetilde M_N$ of an integer symplectic matrix $M\in\mathrm{Sp}(2d,\mathbb{Z})$ or the Schrödinger propagator $e^{-2\pi iNP}$; the quantum trajectory law $P^{(n)}_{N,\rho}$ and the classical law $P^{(n)}_\mu$ (built from the defect measure $\mu$ of the initial states and the classical flow $\varphi^t$ with Lyapunov exponent $\Gamma$) are as defined above. Establish the sharp quantitative theory of the correspondence $P^{(n)}_{N,\rho}\approx P^{(n)}_\mu$, namely:

(1) Optimal window. Determine the true maximal growth $n_{\max}(N)$ of the number of measurements for which $\|P^{(n)}_{N,\rho}-P^{(n)}_\mu\|_{\mathrm{TV}}\to 0$ as $N\to\infty$, with the quantifiers of Theorem 1 (state sequences $\rho_N$ admitting a $(\delta,\theta)$-defect measure, a fixed symbol family $\{f_q\}$, and constants allowed to depend on the state, on $\varepsilon$, and on the symbol family), as a function of $N$, the Lyapunov exponent $\Gamma$, the state parameters $(\delta,\theta)$, and the measurement resolution of the symbols $\{f_q\}$ (the precision of the position measurement encoded by the probes: for the source's Gaussian Kraus symbols $f_q(x,\xi)=f(x-q)$ it is the width $\sigma$ of the Gaussian, smaller $\sigma$ meaning more precise; in Theorem 1 it enters implicitly through the constant $C_0$). Theorem 1 gives sufficiency of $n<\frac{\delta\log(N/C_0)}{\Gamma+\varepsilon}$; determine where correspondence provably breaks down - a rigorous lower bound showing that beyond some measurement count the total-variation distance fails to vanish would establish essential optimality of the logarithmic Ehrenfest-type window, while an improved upper bound extending the admissible range of $n$ would refute it. (The numerics support essential optimality only for the chaotic cat map, and only numerically.)

(2) Optimal rates and input classes. Determine the optimal rate of convergence in $N$ at admissible $n$-scaling - is $N^{-\min(1-2\delta,\theta)}$ sharp, or is a better algebraic (or non-algebraic) rate attainable? - together with the largest admissible $(\delta,\theta)$ ranges for torus coherent-type states (where $\delta<(d+2)^{-1}$, $\theta<1-(d+2)\delta$ is currently proven) and for $\beta$-concentrated torus states (where $\delta<\min(\beta,(1-\beta)/(d+1))$, $\theta<\min(\beta-\delta,1-\beta-(d+1)\delta)$), whose exponents directly cap the convergence rate; clarify the optimal dependence on the dimension $d$ and the concentration exponent $\beta$.

(3) Beyond the window. Characterize the statistics of the quantum measurement trajectories for measurement counts exceeding the Ehrenfest scale in this discrete setting, at least in the following respects: whether the observed-position marginals converge to the uniform distribution, at what time-scale and by what mechanism (measurement-induced decoherence versus chaotic spreading), and whether the transition from classical correspondence is sharp or gradual.

The common objective is the complete quantitative picture of how long, at what rate, and for which states a repeatedly measured quantum map behaves classically, and how that correspondence breaks down: (1) and (2) jointly fix the sharp boundary and rate (the $(\delta,\theta)$ ranges enter directly through the rate exponent and the window constant), and (3) is a required part, describing the complementary regime. Computational schemes or model reductions capable of probing larger $N$ (the present $N^2$ simulation cost versus the logarithmic window makes this infeasible with existing methods) are a legitimate auxiliary contribution for testing any claimed sharper bounds, not a substitute for the analytic core.

References

1. Y. Borns-Weil and I. Oltman, "Particle Trajectories for Quantum Maps", Annales Henri Poincaré 25, 3699-3738 (2024). DOI: 10.1007/s00023-023-01387-x, https://doi.org/10.1007/s00023-023-01387-x (preprint arXiv:2210.03224, https://arxiv.org/abs/2210.03224)
2. T. Benoist, M. Fraas, and J. Fröhlich, "The appearance of particle tracks in detectors. II. The semi-classical realm", Journal of Mathematical Physics 63, 062101 (2022). DOI: 10.1063/5.0088668, https://doi.org/10.1063/5.0088668 (preprint arXiv:2202.09558, https://arxiv.org/abs/2202.09558)
3. J. Galkowski, M. Zworski, and Z. Huang, "Classical-quantum correspondence in Lindblad evolution", Journal of Mathematical Physics 66, 091503 (2025). DOI: 10.1063/5.0224648, https://doi.org/10.1063/5.0224648 (preprint arXiv:2403.09345, https://arxiv.org/abs/2403.09345)
4. F. Hernández, D. Ranard, and C. J. Riedel, "Classical correspondence beyond the Ehrenfest time for open quantum systems with general Lindbladians", Communications in Mathematical Physics 406, 4 (2024). DOI: 10.1007/s00220-024-05146-9, https://doi.org/10.1007/s00220-024-05146-9 (preprint arXiv:2307.05326, https://arxiv.org/abs/2307.05326)
5. F. Hernández, D. Ranard, and C. J. Riedel, "Ehrenfest's theorem beyond the Ehrenfest time", arXiv:2306.13717 (2023), https://arxiv.org/abs/2306.13717
6. F. Hernández, D. Ranard, and C. J. Riedel, "The threshold for quantum-classical correspondence is $D\sim\hbar^{4/3}$", arXiv:2512.17623 (2025), https://arxiv.org/abs/2512.17623
7. I. Oltman, "An exotic calculus of Berezin-Toeplitz operators", Mathematika 71, e70016 (2025). DOI: 10.1112/mtk.70016, https://doi.org/10.1112/mtk.70016 (preprint arXiv:2207.09596, https://arxiv.org/abs/2207.09596)
8. J. Fröhlich and A. Pizzo, "The quantum mechanics of experiments", arXiv:2603.25335 (2026), https://arxiv.org/abs/2603.25335