AI Detects Quantum Phase Transitions in Driven Systems

Understanding how quantum materials behave when pushed out of equilibrium is crucial for developing advanced technologies like quantum computers and sensors. In a study of the periodically driven transverse field Ising model, researchers have shown that a concept from computer science—circuit complexity—can clearly identify non-equilibrium quantum phase transitions. This approach provides a geometric way to measure how difficult it is to transform one quantum state into another, acting like a ruler for quantum changes.

The key finding is that circuit complexity distinguishes between different quantum phases in the driven system. For example, in the paramagnetic phase, where spins are disordered, the complexity settles to a steady value over time. In contrast, in the ferromagnetic phase, where spins align, it oscillates indefinitely. At critical points where phases change, the time-averaged complexity shows sharp, non-analytic behavior, similar to how temperature causes sudden shifts in materials like ice melting to water.

Ology relies on Nielsen's geometric approach, which treats the transformation between quantum states as finding the shortest path in a mathematical space. The researchers applied this to a simplified model solved in the high-frequency driving limit, using techniques akin to averaging fast oscillations to reveal underlying patterns. They computed the complexity analytically by summing contributions from different momentum sectors, ensuring the analysis stays within the model's solvable regime.

From the paper, illustrated in Figure 2, show that complexity grows linearly at early times, independent of the constant driving field, up to a time scale inversely related to the field strength. Figure 3 highlights how the time-averaged complexity and its derivatives exhibit discontinuities and divergences at critical points, such as when the transverse field amplitude hits specific values like the roots of Bessel functions. For instance, at points where dynamic localization occurs, complexity drops to zero, indicating frozen quantum dynamics.

This matters because it offers a new lens to study quantum systems under periodic driving, which are experimentally realizable with ultracold atoms in optical lattices. By using complexity as a diagnostic tool, scientists can better control and design quantum materials for applications in computing and sensing, without needing to probe sensitive original data directly.

Limitations include the model's reliance on non-interacting particles and high-frequency approximations, leaving open how complexity behaves in more complex, interacting systems. The paper notes that future work could extend this to dynamical phase transitions and many-body localization, where universal properties remain poorly understood.