TL;DR
This paper introduces the "displacement" of the radial probability distribution in Fock space as a novel, experimentally relevant tool to characterize the critical behavior of many-body localization transitions (MBLTs). They demonstrate that this measure, combined with finite-size scaling analysis, can distinguish between different universality classes of MBLTs, particularly favoring a BKT-type scaling for random systems over the more commonly assumed power-law scaling.
What Problem Does This Paper Address?
Understanding the transition from an "ergodic" phase to a "many-body localized" (MBL) phase in quantum systems is a major challenge in condensed matter physics. This transition, known as the MBL transition (MBLT), is characterized by a loss of ergodicity and the emergence of local integrals of motion. While various methods exist to identify this transition, precisely characterizing its critical properties, such as the location of the transition point and its universality class, remains difficult. Traditional methods, like analyzing level-spacing ratios, sometimes yield results that violate theoretical bounds (e.g., the Harris-Chayes bound for 1D systems). This has motivated exploring alternative approaches, including those based on the dynamics of local observables and, more recently, the structure of quantum states in Fock space. However, a comprehensive finite-size scaling analysis using Fock-space properties has been lacking.
Methods
The authors investigate three distinct models exhibiting MBLTs:
- A 1D spin-1/2 Heisenberg model with quasiperiodic fields.
- A 1D spin-1/2 Heisenberg model with random fields.
- A random-circuit Floquet model.
For these models, they simulate the time evolution of a quantum state starting from a Néel state. The key observable they analyze is the displacement, defined as the width of the radial probability distribution in Fock space. This distribution quantifies how a wave packet spreads in the vast Hilbert space (Fock space).
To characterize the MBLT, they employ finite-size scaling analysis. This involves studying how the displacement changes with system size ($L$) and disorder strength ($W$). They specifically compare two theoretical frameworks for describing the critical behavior:
- Power-law scaling: Assumes a divergent correlation length with a power-law behavior.
- BKT (Berezinskii-Kosterlitz-Thouless) scaling: A different type of scaling often associated with phase transitions in 2D systems, but here applied to characterize MBLTs.
They perform data collapses of their simulated displacement data onto scaling functions corresponding to these two ansatzes to determine the critical exponents and transition points.
Datasets
Not discussed in the paper. The study relies on numerical simulations of the described quantum models.
Key Insights and Ideas
- Displacement as a MBLT Probe: The "displacement" of the radial probability distribution in Fock space is shown to be a sensitive and experimentally relevant indicator of MBLTs. Its behavior, particularly its peak around the critical point, mirrors that of other established MBLT indicators.
- Distinguishing Universality Classes: The study highlights that different types of disorder (quasiperiodic vs. random) can lead to distinct MBLT universality classes.
- BKT Scaling for Random Systems: Crucially, the paper finds that BKT-type scaling provides a more accurate description of MBLTs in random-field and Floquet models compared to the commonly assumed power-law scaling. This is evidenced by better data collapses near the critical point.
- Larger Critical Points with BKT: The BKT scaling analysis, when applied to the displacement, yields larger estimates for the critical disorder strength ($W_c$) compared to power-law scaling, and these are found to be consistent with other theoretical analyses (e.g., many-body resonances).
- Experimental Relevance: The authors emphasize that the Fock-space dynamics, and thus the displacement, can be measured in current analog quantum simulation devices, opening avenues for experimental verification.
Key Results
- Quasiperiodic Model: The MBLT in the quasiperiodic model is well-described by power-law scaling, with a scaling exponent $\nu \approx 1.0$, which satisfies the Harris-Luck bound.
- Random Model & Floquet Model: For both the random-field Heisenberg model and the random-circuit Floquet model, BKT-type scaling provides a better data collapse than power-law scaling.
- For the random model, BKT scaling yields a critical point $W_c \approx 6.8$.
- For the Floquet model, BKT scaling yields a critical point $W_c \approx 8.2$.
- Power-law scaling for random/Floquet models: While power-law scaling can be applied, it results in less convincing data collapses and larger uncertainties in the critical parameters ($\Delta W$) compared to the quasiperiodic model. The extracted scaling exponents ($\nu \approx 2.9$ for random, $\nu \approx 3.0$ for Floquet) are significantly larger than for the quasiperiodic case.
- Consistency with other methods: The BKT critical points obtained for the Floquet model are consistent with estimates from many-body resonance analysis.
Limitations and Future Work
- Power-law scaling in quasiperiodic model: While power-law scaling appears most compelling for the quasiperiodic model with accessible system sizes, the authors acknowledge that BKT-type scaling cannot be entirely ruled out and suggest exploring alternative functional forms for $W^*(L)$ in the BKT ansatz.
- Numerical limitations: The study is based on numerical simulations, and finite-size effects are inherent. The authors suggest that larger system sizes and improved computational resources could further refine the analysis.
- Experimental verification: A key future direction is the experimental realization of this analysis using analog quantum simulators, which could overcome numerical limitations and provide direct validation.
- Broader applicability: The authors propose extending this Fock-space perspective to more complex systems, including 2D systems, systems with long-range interactions, and those with specific symmetries or potentials.
Summary updated