Four-Layer Approximation Scheme on Torus

• The paper introduces a four-layer projective construction for the bosonic Z4 Read-Rezayi state, recovering the full ground state and excitation spectrum on a torus.
• It employs twisted boundary conditions and layer symmetrization to combine four Laughlin states into a single effective state capturing non-Abelian features.
• Numerical validation indicates high variational accuracy with rapid convergence of energy gaps and overlaps, ensuring practical feasibility in FQH studies.

The four-layer approximation scheme on the torus constitutes a projective construction specialized for the $\mathbb{Z}_4$ Read-Rezayi series within the fractional quantum Hall (FQH) paradigm. This protocol, introduced and elucidated in Repellin et al. (Repellin et al., 2015), synthesizes multilayer Laughlin states by symmetrizing over layers, augmented with twisted boundary conditions. This enables recovery of the full ground state and excitation spectrum that standard multilayer constructions fail to capture on toroidal geometries. The construction is expressible in both second-quantized lowest Landau level (LLL) orbital terms and first-quantized conformal-block (theta function) language, and is tailored for rigorous numerical and theoretical investigations. The scheme is most directly pertinent for the bosonic $k=4$ Read-Rezayi state at filling $\nu=2$.

1. Pre-Symmetrized Four-Layer Wave Function Construction

Commencing with four independent copies of the bosonic Laughlin $1/2$ state, each layer is defined on a torus of size $4N_\phi$ flux quanta, where $N_\phi = L_x L_y / (2\pi \ell_B^2)$ represents the fundamental flux threading per torus, and $\ell_B$ is the magnetic length. The LLL single-particle orbitals on the rectangular torus, in the $x$-Landau gauge $\mathbf{A}=(0,-Bx)$, are

$$\phi_j(z) = \frac{1}{\sqrt{L_y \sqrt{\pi}}} \sum_{m \in \mathbb{Z}} \exp\left[i y\left(\frac{2\pi j}{L_y} + m L_x\right) - \frac{(x + \frac{2\pi j}{L_y} + m L_x)^2}{2}\right],$$

with $j = 0, \dots, N_\phi-1$. In second quantization, the four-layer "parent" state is simply the tensor product:

$$|\Psi_{\text{pre}}\rangle = |{\rm Laughlin}_{1}\rangle \otimes |{\rm Laughlin}_{2}\rangle \otimes |{\rm Laughlin}_{3}\rangle \otimes |{\rm Laughlin}_{4}\rangle,$$

and in first quantization:

$$\Psi_{\text{pre}}\left(\{z_i^{(\ell)}\}\right) = \prod_{\ell=1}^4 \exp\left[-\sum_i \frac{|z_i^{(\ell)}|^2}{4\ell_B^2}\right] \prod_{\ell=1}^4 \prod_{i<j} \theta_1\left(z_i^{(\ell)}-z_j^{(\ell)}|\tau\right)^2,$$

where $\tau=iL_y/L_x$ and $\theta_1$ is the odd Jacobi theta function. The conventional center-of-mass prefactor is omitted, as in the cited reference.

2. Twisted Boundary Conditions and Layer Coupling

The hallmark innovation of this scheme is the imposition of twisted boundary conditions, cyclically permuting layers as coordinates wind around the fundamental cycles of the torus:

• $x$-cycle: $\psi_\ell(x+L_x, y) = \psi_{\ell+1 \bmod 4}(x, y)$
• $y$-cycle: $\psi_\ell(x, y+L_y) = \psi_{\ell+1 \bmod 4}(x, y)$

In operator language this translates to:

$$v_\ell(\mathbf{r} + L_x \mathbf{e}_x) = v_{\ell+1}(\mathbf{r}), \quad v_\ell(\mathbf{r} + L_y \mathbf{e}_y) = v_{\ell+1}(\mathbf{r}),$$

for $\ell = 1, \ldots, 4$ modulo $4$. This twist identifies the four-layer torus as geometrically equivalent to a single torus of periods $(4L_x, L_y)$ (for $x$ twists) or $(L_x, 4L_y)$ (for $y$ twists), effectively folding the layers into an extended toroidal geometry. This step is indispensable for recovering all $\mathbb{Z}_4$ ground states on the torus.

3. Symmetrization Over the Layer Index

Projecting onto the $\mathbb{Z}_4$ Read-Rezayi subspace is accomplished by the symmetrizer $\mathcal{S}$, defined by summing over permutations $\sigma \in S_4$ that act on the layer indices:

$$\mathcal{S} = \frac{1}{4!} \sum_{\sigma \in S_4} U(\sigma),$$

where $U(\sigma)$ permutes layer labels on creation operators. In first quantization, the effect is:

$$(\mathcal{S} \Psi_{\text{pre}})(z_1, ..., z_N) = \frac{1}{4!} \sum_{\sigma \in S_4} \Psi_{\text{pre}}(z_1^{(\sigma(1))}, ..., z_N^{(\sigma(N))}),$$

with normalization ensuring the resulting state is properly normalized. This symmetrization is critical for accessing the non-Abelian parafermion structure inherent in the $\mathbb{Z}_4$ Read-Rezayi wave function.

4. Single-Layer $\mathbb{Z}_4$ Trial State and Ground-State Multiplet

The symmetrized state realizes a single-layer wave function at filling $\nu=2$ corresponding to the $k=4$ Read-Rezayi phase. In first-quantized theta-function language:

$$\Psi_{\mathbb{Z}_4}(\{z_i\}) = \mathcal{S}\left[\prod_{\ell=1}^4 \prod_{i<j} \theta_1(z_i^{(\ell)} - z_j^{(\ell)})^2\, e^{-\sum |z|^2 / 4\ell_B^2}\right]_{z_i^{(1)} = \ldots = z_i^{(4)}}.$$

Expanded, this wave function matches the parafermion conformal block of the $k=4$ series. For even $N_\phi$, the ground-state manifold on the torus exhibits a fivefold degeneracy (as $k+1=5$ for $k=4$). These states are distinguished by translation quantum numbers $(k_x,k_y)$ in the reduced Brillouin zone; for $N_\phi=2N/4$ they are: $(0,0)$, $(0,N_\phi/2)$, $(N_\phi/2,0)$, $(N_\phi/2,N_\phi/2)$, $(0,N_\phi/2)$. Generation of the full multiplet follows either from selecting distinct initial momenta for the parent Laughlin layers or by translation operators post-symmetrization.

5. Quasihole and Neutral Excitation Construction

Quasihole zero modes in the $\mathbb{Z}_4$ scheme are realized by increasing the parent system flux by four, producing one additional single-layer flux quantum after symmetrization. This produces the complete set of quasihole states (verified computationally for lower $k$ in (Repellin et al., 2015); for $g=4$ the same mechanism applies). Neutral excitation trial states are accessed by promoting one or more Laughlin parents to a collective neutral mode (e.g., magnetoroton), then symmetrizing. For instance, three layers can remain in their ground state while the fourth is excited prior to symmetrization, yielding a neutral excitation trial for $\mathbb{Z}_4$.

6. Variational Accuracy: Numerical Overlaps and Gaps

Repellin et al. (Repellin et al., 2015) provide numerical validation for $k=2$, exhibiting overlaps exceeding $0.98$ and energy gap estimates converging within $1\%$ for $N \sim 12$. For $g=4$, although explicit overlap tables are not published, the methodology implies similarly high variational accuracy, with expected overlaps in the range $0.90–0.99$ for small $N$ and rapid gap convergence scaling with system size. Calculation of such metrics for $g=4$ would entail full diagonalization of the $5$-body contact Hamiltonian and inner product evaluation with symmetrized trial states, ensuring completeness of the zero-mode manifold and consistency with the neutral excitation spectrum.

7. Practical Implementation Workflow

Concrete computational steps for the four-layer scheme entail:

1. Construct the LLL orbital basis $\phi_j(r)$ for the extended torus $(4L_x, L_y)$ or $(L_x, 4L_y)$.
2. Formulate four independent Laughlin many-body states at flux $4N_\phi$ via diagonalization of the $2$-body contact Hamiltonian, storing each as Fock vectors.
3. Realize the twist/extended-torus equivalence by projecting onto every fourth orbital subset, yielding four Laughlin copies.
4. Implement the symmetrizer $\mathcal{S}$ by summing over the $4!=24$ layer-index permutations, or equivalently by orbital shifts.
5. Apply $\mathcal{S}$ to the product state and normalize; project back to the single-layer Fock basis of $N_\phi$ orbitals.
6. Generate the fivefold ground-state multiplet by repeating steps for each initial momentum configuration or via translation operations.
7. For quasiholes, increment parent flux per q-hole and recompute; for neutral modes, substitute parent layers by their neutral-mode diagonalizations before symmetrization.
8. Finally, calculate overlaps and energies by taking inner products of symmetrized states with exact diagonalizations of the $5$-body contact Hamiltonian at flux $N_\phi$.

This sequence renders a comprehensive and information-rich application of the four-layer projective symmetrization on the torus (Repellin et al., 2015).