Eigenstate Thermalization for Local versus Translationally Invariant Observables
AI Breakdown
Get a structured breakdown of this paper — what it's about, the core idea, and key takeaways for the field.
Abstract
Local observables and their translationally invariant counterparts are generally thought as providing the same predictions for experimental measurements. This is used in the context of their expectation values, which are indeed the same in clean systems (up to finite-size effects), but also in the context of their correlation functions, which need not be the same. We examine this intuition from the perspective of the eigenstate thermalization hypothesis. Specifically, we explore the diagonal matrix elements and the spectral functions of local and translationally invariant observables in the spin-1 tilted field Ising chain with periodic and open boundary conditions. We discuss in which ways those observables are different and in which contexts they can be thought as being the same. Furthermore, we unveil a novel form of off-diagonal eigenstate thermalization in translationally invariant systems that applies to pairs of energy eigenstates with different quasimomenta.