Many-body localization in two dimensions

Imagine atoms arranged in alternating crowded and sparse rows on a flat grid. Once they can move and interact, does that pattern eventually disappear? Random variations in the energy of grid sites can impede spreading. The question is whether a large isolated system can keep memory indefinitely.[4]

An isolated plane of interacting atoms can begin in a stripe pattern of occupied and less occupied sites. Once hopping is allowed, can random disorder in site energies keep a lasting density imbalance in arbitrarily large systems? This is one way to probe many-body localization, a proposed failure of interacting quantum matter to thermalize.

The competing avalanche argument starts with a rare patch of unusually weak disorder that can relax internally. It may exchange energy with neighboring localized degrees of freedom. If those neighbors join the thermal region, the enlarged region may then influence more distant neighbors. That growth is a theoretical mechanism, not a conclusion from a persistent stripe alone.[1] [2] [3] [4]

A pattern that survives an experiment may still fade later. A rare patch that relaxes internally might cause neighboring regions to relax too, so finite observations do not settle the large-system, long-time question.[1] [2] [4]

The conference-linked preprint reports different finite-size trends for random and quasiperiodic potentials, with finite observation times. Such scaling can support an avalanche scenario without proving the asymptotic result. Avalanche arguments depend on assumptions about the growing region and its couplings. The 2023 confinement theory changes localization profiles with a trap; its proposed stabilization does not settle the homogeneous random-disorder target. Quasiperiodic potentials likewise lack the same random rare-region statistics.[1] [2] [3] [4]

It tests how quantum systems retain information about their starting state.[3] [4]

The question connects microscopic quantum dynamics with the equilibrium descriptions used for macroscopic matter. Retaining a local trace of preparation would limit when those descriptions apply. Resolving the distinction between a lasting phase and extremely slow relaxation also gives a principled interpretation of experiments whose observation window ends before any eventual loss of memory.[3] [4]