arXiv:2609.36008
Entanglement spectrum and magic in higher-dimensional free fermionic systems
Beatrice Magni, Daniele Iannotti, Riccardo Travaglino
Abstract
We study fermionic non-local magic in higher-dimensional free-fermion systems through the lens of the entanglement spectrum. For translationally invariant Gaussian states in strip geometries, dimensional reduction maps the problem to independent lower-dimensional momentum sectors, leading in two dimensions to a multiplicative logarithmic violation of the boundary law analogous to Gioev-Klich-Widom scaling. We further show that the second Rényi fermionic non-local magic is bounded by the capacity of entanglement and introduce an entanglement-temperature deformation that probes the low-energy structure of the entanglement Hamiltonian. Applied to the matter-Majorana sector of the Kitaev model, this response is exponentially suppressed in the Abelian phase and algebraic in the non-Abelian phase, where its ratio with the capacity approaches a universal constant. Out of equilibrium, we extend the quasi-particle description to fermionic non-local magic in higher-dimensional quenches and show that random Gaussian circuits exhibit diffusive spreading in two spatial dimensions.
AI Summary
The paper asks how fermionic non-local magic (FNL)—stabilizer Rényi entropy minimized over bipartition-local Gaussian unitaries—is controlled by the entanglement spectrum in higher-dimensional free-fermion systems. It uses the existing additive covariance-spectrum formula to study equilibrium scaling, entanglement-temperature response, and dynamics. For translation-invariant strips, transverse Fourier transformation gives an exact decomposition into one-dimensional momentum sectors. A finite density of critical sectors produces $M_\alpha^{\mathrm{FNL}}\sim L_y\log\ell$ in two dimensions. Fisher–Hartwig asymptotics yield explicit coefficients and subleading terms for square-lattice hopping, checked numerically. A pointwise comparison of spectral kernels establishes $M_2^{\mathrm{FNL}}\leq C_E$, where $C_E$ is the capacity of entanglement. Introducing $\rho_A(T_E)\propto\rho_A^{1/T_E}$ resolves low-energy entanglement modes. In the Kitaev model’s Gaussian matter-Majorana sector, FNL scales as $L_y\sqrt{T_E}e^{-\Delta_E/T_E}$ for a gapped band with a quadratic minimum, versus $L_yT_E$ for linear gapless branches. For the latter, $M_2^{\mathrm{FNL}}/C_E\to(6/\pi^2)[\pi^2/4-\log^2(2+\sqrt3)]$. These results follow from spectral-kernel asymptotics and numerical spectra. For quadratic quenches, a quasiparticle formula reproduces numerical FNL dynamics. Random Gaussian circuits show $L_y\sqrt t$ growth numerically, consistent with an analytical diffusion equation for averaged Majorana weight; this does not establish the FNL growth law analytically. The analysis largely assumes strip geometries and pure Gaussian states. The Kitaev quantity is not generally the physical spin-state non-local magic; claims about unrestricted local-unitary minimization remain conjectural. Low-temperature continuum scaling also requires sufficiently large boundary length.
Error found: The random-circuit analytical argument does not establish M_2^{FNL}∝L_y√t, although the paper reports numerical evidence for that growth law. The statement that the capacity-of-entanglement upper bound is uniformly tighter than the existing entropy bounds is false, as a single covariance singular value λ=0.99 demonstrates. The critical-coefficient formulas mix natural-log and base-two conventions, producing values differing by a factor of 1/ln 2 without consistently distinguishing their use.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 10 figures, comments are welcome