arXiv:2609.34559
Microsecond-Programmable Synthetic Quantum Matter: Circumventing the Adiabatic Bottleneck for Heisenberg-Limited Metrology
Soi-Chan Lei
Abstract
Generating macroscopic quantum entanglement is central to quantum metrology. While traversing a quantum critical point yields Heisenberg-limited precision, adiabatic schemes are restricted by slow dynamics and decoherence. Here, we propose a non-adiabatic Floquet protocol in a single-cavity Bose-Einstein condensate to circumvent these limitations. By applying rapid rotations and phase jumps, we dynamically synthesize effective Two-Axis Twisting (TAT) interactions from native longitudinal couplings. To evaluate this beyond mean-field limits, we develop an exact finite-state automaton Matrix Product Operator (MPO) encoding that strictly bounds the virtual dimension to $\mathcal{O}(1)$, enabling exact DMRG and TDVP simulations for large systems. Tracking non-equilibrium Floquet TAT dynamics and environmental decoherence for $N=10^3$ ensembles, we demonstrate that the quench rapidly generates Heisenberg-limited spin squeezing ($ξ_R^2 \sim 1/N$). The microsecond timescale of this protocol decouples the generation process from typical millisecond decoherence, with phenomenological models revealing an optimal extraction time near $t \approx 3.26~μs$. This work establishes a feasible route for realizing Heisenberg-limited metrology in current cavity-QED platforms.
AI Summary
The paper studies whether nonadiabatic control of cavity-mediated collective spins can generate metrologically useful entanglement faster than adiabatic preparation near a quantum critical point. It proposes independently tunable longitudinal and transverse all-to-all interactions in a single-cavity Bose–Einstein condensate. Within the symmetric spin manifold, the static Hamiltonian reduces to one-axis twisting (OAT), with nonlinear coefficient $(4\chi-2\lambda)/N$. A Floquet sequence combines $\pi/2$ rotations with a proposed phase-controlled interaction-sign inversion to synthesize $H_{\mathrm{TAT}}\simeq(\chi_{\mathrm{eff}}/N)(J_z^2-J_y^2)$, where $\chi_{\mathrm{eff}}=(4\chi-2\lambda)/2$. The intended two-axis-twisting dynamics yield ideal squeezing scaling as $\xi_R^2\sim1/N$, compared with the $N^{-2/3}$ OAT limit. Numerical methods use an exact constant-bond-dimension finite-state-automaton MPO, DMRG for static calculations at $N=10^4$, and TDVP dynamics at $N=10^3$. The static calculations report a finite-size critical-point shift and enhanced bipartite entanglement. Symmetric-subspace exact diagonalization additionally treats ideal dynamics at $N=10^4$. Decoherence is represented by the additive phenomenological model $\xi_{R,\mathrm{real}}^2\approx\xi_{R,\mathrm{ideal}}^2+\Gamma_{\mathrm{eff}}t$, rather than an open-system evolution. With an assumed interaction scale near $1\,\mathrm{MHz}$, reported optimal extraction times are approximately $3.26\,\mu\mathrm{s}$ for $N=10^3$ and $3.34\,\mu\mathrm{s}$ for $N=10^4$; the latter ideal minimum occurs near $4.5\,\mu\mathrm{s}$. The effective Floquet Hamiltonian assumes fast pulses and a leading-order Magnus approximation. TDVP trajectories are limited by an MPS bond-dimension cap of 128; exact MPO encoding does not remove this state-representation constraint. Full Lindblad dynamics and spatially inhomogeneous drives are left for future work.
Specifically, the AI was employed to assist with language refinement, the generation of Python scripts for data visualization (Matplotlib), and the exploratory derivation of intermediate mathematical equations.
Central claim may not hold: The stated global pi phase pulse does not reverse the sign of Jy^2, so the proposed Floquet sequence has not been shown to generate the TAT Hamiltonian assumed in the simulations. An exact Hamiltonian MPO neither makes truncated TDVP evolution exact nor establishes classical insimulability, particularly when the simulated dynamics admit the paper's own N+1-dimensional Dicke-subspace treatment. Decoherence is represented only by adding Gamma_eff t to unitary squeezing, which neither simulates a dissipative state nor establishes robust Heisenberg scaling as N increases. The stated symmetric static Hamiltonian has nondegenerate Dicke ground states with Wineland squeezing at least one, contradicting the reported sub-SQL static ground-state squeezing. The microsecond feasibility claim assumes an approximately 1 MHz effective interaction without a quantitative cavity and pulse specification establishing that rate, the Floquet approximation, and the associated noise.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.