A preprint posted to arXiv in April 2026 (not yet peer-reviewed) introduces a time-integrated spread complexity as a diagnostic tool for quantum ergodicity. Lead author Mehmet A. Süzen and collaborators examine how maximally entangled states scramble under unitary evolution generated by different Hamiltonians. Rather than relying on instantaneous snapshots like out-of-time-order correlators (OTOCs), they integrate a complexity measure over time, then use numerical bootstrapping on Rosenzweig-Porter random-matrix ensembles to tune continuously from integrable to fully chaotic regimes. The method reveals finer resolution of ergodic transitions than many earlier probes.

Methodology note: The study is purely numerical, generating ensembles of Hamiltonians via bootstrapped sampling from the Rosenzweig-Porter distribution, which interpolates between Poissonian level statistics (integrable) and Gaussian orthogonal ensemble (chaotic) statistics. No explicit system sizes are given in the abstract, but such computations are typically limited to Hilbert-space dimensions of a few thousand because the unitary evolution and entanglement calculations scale exponentially. Limitations include finite-size effects that may not fully capture thermodynamic-limit behavior, possible ensemble-specific biases, and the absence of analytic proofs for general interacting many-body systems.

This work goes beyond simply proposing a new metric. It illuminates fundamental connections between entanglement scrambling, integrability breaking, quantum chaos, and operator complexity that previous literature has treated separately. Classic results on the eigenstate thermalization hypothesis (ETH) by Deutsch (1991) and Srednicki (1994) established that chaotic eigenstates look thermal, but left open the dynamical question of how systems actually reach that equilibrium. The 2016 paper by Shenker and Stanford (arXiv:1512.07631) on OTOCs linked fast scrambling to black-hole horizons in AdS/CFT, yet OTOCs often saturate rapidly and lose sensitivity at late times. Cotler et al. (arXiv:1608.06950) showed random-matrix universality in chaotic spectral form factors. The new integrated spread complexity synthesizes these threads: by averaging over unitary paths and tracking how entanglement spreads in Krylov space (echoing Dymarsky et al., arXiv:2202.05701 on Krylov complexity saturation), it captures the gradual breakdown of conserved quantities that integrable systems possess.

What the original preprint under-emphasizes is the bridge to many-body localization (MBL) transitions and experimental realizability. While Rosenzweig-Porter ensembles model the ergodic-to-localized crossover, the paper stops short of linking its measure to realistic Hamiltonians such as disordered spin chains where MBL has been observed in trapped-ion and superconducting-qubit experiments (e.g., Google's 2021 quantum processor results). It also misses explicit discussion of how this complexity diagnostic could benchmark near-term quantum simulators attempting to emulate quantum thermalization.

The deeper insight revealed here is that quantum thermalization is neither purely spectral nor purely entangling; it is a complexity phenomenon. Integrable systems keep information localized in a few conserved charges, suppressing operator growth. Chaotic systems allow exponential spreading in Hilbert space until the complexity saturates at a volume-law entanglement scale, mirroring black-hole information scrambling. The time-integrated nature of the new measure smooths out early-time transients and late-time plateaus, exposing intermediate dynamical regimes where partial thermalization occurs. This has broad implications: it suggests complexity-based diagnostics could unify our understanding of quantum chaos, holographic duality, and quantum algorithm design, where controlling scrambling rates may improve error mitigation.

By placing the Süzen preprint in conversation with ETH, OTOC scrambling literature, and Krylov-complexity studies, a clearer picture emerges: thermalization mechanisms are governed by how fast and how completely entanglement explores the available operator space. The transition is not binary but graded, and the right integrated complexity observable makes those grades quantitatively accessible for the first time.