Abstract
We analyze the collective magnetoroton excitations of bosonic Laughlin $\nu=\frac{1}{2}$ fractional quantum Hall (FQH) states on the torus and of their analog on the lattice, the fractional Chern insulators (FCIs). We show that, by applying the appropriate mapping of momentum quantum numbers between the two systems, the magnetoroton mode can be identified in FCIs and that it contains the same number of states as in the FQH case. Further, we numerically test the single-mode approximation to the magnetoroton mode for both the FQH and FCI cases. This proves particularly challenging for the FCI because its eigenstates have a lower translational symmetry than the FQH states. In spite of this, we construct the FCI single-mode approximation such that it carries the same momenta as the FQH states, allowing for a direct comparison between the two systems. We show that the single-mode approximation captures well a dispersive subset of the magnetoroton excitations both for the FQH and the FCI cases. We find remarkable quantitative agreement between the two systems. For example, the many-body excitation gap extrapolates to almost the same value in the thermodynamic limit.
Introduction
In the study of topological phases of matter, research often focuses on universal properties such as ground state degeneracy or quasiparticle statistics. However, these properties do not fully characterize the dynamics—the energy costs and behaviors of excitations above the ground state. While the physics of the Fractional Quantum Hall (FQH) effect in the continuum is well-established, the discovery of Fractional Chern Insulators (FCIs) raised a critical question: how much of the continuum physics survives on a discrete lattice? Specifically, FQH systems exhibit a neutral collective excitation known as the magneto-roton mode. Because FCIs lack the continuous translational symmetry of the FQH effect, it was previously unclear if they would support a similar mode or if the lattice would fundamentally alter these dynamics.
Methodology
To bridge the gap between the continuum and the lattice, we employ three primary tools:
- Exact Diagonalization (ED): Numerical simulations on small systems (up to $N=13$ bosons) are used to find exact energy levels and eigenstates.
- Single-Mode Approximation (SMA): A variational method that constructs an excited state by acting on the ground state with a density operator, providing a way to model the collective excitation.
- Momentum Mapping: A technique developed to "fold" the momentum quantum numbers of the FQH system (on a torus) onto the Brillouin zone of the FCI lattice, enabling direct comparison.
A key challenge in the FCI case is that the lattice allows for a much larger manifold of variational states than are physically present in the magneto-roton mode. We introduced a selection criterion based on minimizing the momentum-space distance to the ground state to identify the most relevant excitations.
Key Insights and Results
Our analysis reveals that the magneto-roton mode is remarkably robust. Despite the lack of continuous magnetic translation symmetry in FCIs, the magneto-roton mode remains a well-defined, dispersive branch of excitations.
One of the most striking findings is the quantitative convergence between the two systems. When rescaled appropriately, the many-body excitation gap for both systems extrapolates to nearly the same value ($\approx 0.6 V_0$) in the thermodynamic limit. The SMA states showed a high overlap (up to $\approx 0.9$) with the exact eigenstates at low momenta, proving that the SMA is an excellent description of the physics in both regimes.
However, we also found that not all FCIs are created equal. While the ruby and kagome lattices showed clear, isolated magneto-roton modes, the Haldane model's excitations were harder to distinguish from the multi-particle continuum. This highlights that "good" FCI models need to minimize finite-size effects, such as ground-state splitting, to maintain clear collective dynamics.
Limitations and Future Work
The SMA works best at long wavelengths (small momentum). At short wavelengths (large momentum), the magneto-roton mode flattens out and merges with the continuum of multi-particle excitations, where the SMA loses its predictive power. Additionally, in FCIs, the density operator depends on the "embedding"—the specific physical locations of orbitals within the unit cell. Pathological embeddings can degrade the quality of the SMA.
Future work could extend this framework to more complex topological states, such as non-Abelian phases, to determine if the SMA remains a powerful tool for characterizing their collective dynamics across lattice and continuum models.
Summary updated