Post

New paper! arxiv.org/abs/2607.26156 ⚛️🧪 We study disorder-free localisation (DFL) - an intriguing phenomenon where dynamical degrees of freedom become localised and frozen, in the absence of the strong disorder (such as a random on-site magnetic field) normally needed to induce this effect. 🧵

This was recently the subject of a Google Quantum AI experiment, which simulated DFL on a quantum processor (in the context of a lattice gauge theory). The lack of disorder means far fewer experimental shots are needed to run your simulation, meaning DFL is ideal for studying localisation phenomena

The model in the Google experiment maps to an effective ensemble of disordered models, specifically a mixed-field Ising model with two-level discrete bond disorder. While they observe a localised regime at accessible timescales, we wanted to study how this changes with n-level disorder (for n > 2!)

We find, surprisingly, that the model mapping to two-level disorder wasn't enough to produce a true localised phase – in fact, we thermalise fairly quickly following a quench! Whereas for n = 4, 8 we see long-lived localisation and other signatures of a localised phase (like "area law" entanglement)