arXiv:2607.16278
Number Fluctuations and Entanglement-Spectrum Participation in Monitored Free Fermions
Enso O. Torres Alegre
Abstract
Monitored free-fermion chains show a crossover in entanglement scaling as the measurement rate is increased, from a subextensive weak-monitoring regime to an area law at strong monitoring. I test, using trajectory-resolved correlation-matrix simulations of chains up to $L=96$, whether two diagnostics built from the single-particle entanglement spectrum add practical value: the bipartite particle-number fluctuation $F_A=\sum_kν_k(1-ν_k)$ and a participation-style effective number of entangling modes $\mathcal{M}_A=\exp[-\sum_k w_k\ln w_k]$, with $w_k\proptoν_k(1-ν_k)$. Both quantities track the crossover, decreasing monotonically with measurement probability and losing system-size dependence in the strongly monitored regime. However, at the level of trajectory- and time-averaged steady-state values, $\mathcal{M}$ is almost a deterministic function of $F$: a pooled curve captures 99.7\% of its variance across all measurement rates, so $\mathcal{M}$ carries little independent information in this model. Its main advantage is statistical: the trajectory-to-trajectory coefficient of variation of $\mathcal{M}$ is up to a factor of about 2 smaller than that of the entropy or $F$ at strong monitoring. A model comparison of subsystem scaling at $L=48$--$96$ further shows that, at weak monitoring, a pure logarithmic law is not statistically preferred at these sizes; a small residual quasi-extensive component and a drifting effective logarithmic coefficient remain. I conclude that $F$ is the natural experimentally motivated companion to the entropy, while $\mathcal{M}$ is best viewed as a variance-reduced numerical summary of largely the same information, not an independent diagnostic.