Escrever uma avaliação PREreview

We test whether the local occupation variance $F_i=\langle \hat n_i^2\rangle-\langle \hat n_i\rangle^2$ identifies causally effective intervention sites in a one-dimensional open Bose-Hubbard chain under Lindblad dephasing. Additional dephasing at the top-$k$ highest-$F_i$ sites is compared against matched-budget random targeting across 100 independent trials per condition, using exact Lindblad evolution in the fixed-particle-number sector ($L\in{6,7,8}$, $\tau\in{1,2,3}$). The primary sweep reveals three regimes: at $J/U=0.12$ targeted intervention is harmful; near $J/U\simeq0.20$ there is a crossover sensitive to system size and horizon; and at $J/U\ge0.30$ targeted intervention is reliably beneficial, with 95% confidence intervals above zero at every tested $(L,\tau)$ combination. Controls using disorder, shell-matched permutations, deterministic tilt chains, and extra-dephasing-rate scans show that the effect is not reducible to geometric centrality, shell position, mean occupation, disorder amplitude, or a tuned intervention strength. Because the systems are small enough for exhaustive enumeration, we rank the $F_i$-selected subset against all $\binom{L}{k}$ possible intervention subsets; in the positive-pocket regime it lies at the top of the subset-response distribution and the conclusion survives signed, absolute, and redistribution-based target definitions. Finite-difference susceptibility analysis shows that $F_i$ is not a local-depletion susceptibility but a redistribution susceptibility: high-$F_i$ sites are those where small extra dephasing most strongly perturbs the global occupation pattern. Thus, within the tested finite chains, local number variance functions as a regime-dependent redistribution handle for targeted dephasing.