arXiv:2610.03194
Emergent Dissipation from Fluctuating Quantum-Network Topology
Alireza Nourmandipour, Stefano Mancini
Abstract
We investigate how dissipation can emerge in a quantum network from temporal fluctuations of the interaction topology. For a network whose graph is redrawn at short time intervals, the averaged dynamics yields a Markovian master equation: the mean network fixes the coherent part and the covariance fixes the dissipative structure. The covariance eigenvalues set the noise strengths, while its eigenvectors determine the collective operators on which those strengths act. Two ensembles with identical mean adjacency and covariance spectra but different eigenvectors produce distinct dynamics despite identical purity. A physically realizable model yields an analytically predicted dissipative gap, giving quasi-protected dynamics. These results establish fluctuating connectivity as a resource for engineering quantum dissipation and protection.
AI Summary
The paper asks how temporal fluctuations of network connectivity determine ensemble-averaged decoherence and protected sectors. It considers $N$ sites restricted to the single-excitation manifold, with binary edges independently redrawn between intervals of duration $\Delta t$, allowing correlations between edges within each draw. Expanding the exact random-unitary channel and its logarithm gives $\mathcal L\rho=-i[\bar H,\rho]-(J^2\Delta t/2)\sum_{ef}C_{ef}[V_e,[V_f,\rho]]+O(\Delta t^2)$. The mean adjacency determines $\bar H$; covariance eigenvalues $c_\alpha$ give Lindblad strengths $J^2\Delta t\,c_\alpha$, while eigenvectors specify collective Hermitian jump operators. Standard common-eigenspace and Hamiltonian-invariance conditions identify decoherence-free subspaces. Two correlated graph ensembles have identical mean adjacency and covariance spectra but different collective noise operators. Closed-form finite-step dynamics show different return fidelities and sector coherences, despite identical purity for an initially localized excitation. A latent-variable ensemble interpolates between independent edges and community-correlated switching. An explicit Liouvillian eigenmode has decay rate $\gamma_{\rm slow}=J^2\Delta t\,\xi(1-\xi)(N-2)(1-\kappa^2)$. Full diagonalization identifies this rate with the gap for the tested parameters: $N=10$, $\xi=1/2$, $J=1$, $\Delta t=0.01$, giving $0.02(1-\kappa^2)$. At full correlation, there are 66 stationary and 16 undamped oscillatory operator modes. The generator requires $\|H_{\mathcal G}\|\Delta t\ll1$; dissipation vanishes as switching becomes arbitrarily fast. The gap identification is numerical, not a general analytical bound. Finite-step benchmarks use 5,000 graph samples and carry sampling error. Temporally correlated switching is left for future work.
Error found: The abstract omits that its Markovian master equation is only a leading rapid-switching approximation to the exact discrete random-unitary dynamics. The Discussion incorrectly proposes the |+⟩–|−⟩ coherence as a signal decaying at γ_slow: the derived slow mode is a population difference, and that coherence remains damped even when γ_slow vanishes.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 3 figures