arXiv:2610.37765
Long-Range Toric Code: Finite-Temperature Topological Order in Two Dimensions
Jacopo Gliozzi, Taylor L. Hughes, Kyung-Su Kim
Abstract
We introduce two-dimensional toric code models with additional long-range loop interactions. Our models preserve the topologically ordered ground states of the ordinary toric code, while confining anyons at zero temperature. For sufficiently long-ranged interactions, we argue that this confinement persists at finite temperatures, strongly constraining anyon mobility and giving rise to finite-temperature topological order. This topological phase acts as a self-correcting quantum memory and exhibits long-range entanglement. For shorter-ranged interactions, signatures of topological order survive only up to an exponentially large length scale at finite temperature, yielding a partially self-correcting quantum memory.
AI Summary
Can long-range interactions stabilize finite-temperature topological order and passive quantum memory in two dimensions without a superextensive energy spectrum? This paper supplements the toric code with commuting square-loop stabilizers weighted by perimeter as $\ell^{-\alpha}$. The ground-state manifold is unchanged, while the bandwidth remains extensive for $\alpha>1$; interactions have unbounded support and are not $k$-local. Loop counting gives isolated anyon-pair energies growing as $d^{3-\alpha}$ for $2<\alpha<3$, logarithmically for $\alpha=3$, and remaining bounded for $\alpha>3$. For $1<\alpha\leq2$, even nearest-neighbor dipoles have energies diverging with system size; compact quadrupoles instead have finite energy, producing emergent row/column parity constraints. A dilute-excitation analysis predicts that thermal dipoles screen confinement for $2<\alpha\leq3$ beyond $L_{\mathrm{conf}}\sim e^{\beta\Delta_2}$, yielding partial self-correction. For $1<\alpha\leq2$, quadrupoles are argued not to screen confinement, supporting finite-temperature topological order and full self-correction. A quasi-local renormalization-group decoder supplies a related argument for long-range entanglement of low-temperature Gibbs states. Classical Monte Carlo simulations examine specific heat, Wilson loops, decoding errors, and memory lifetimes. Rigorous correlation inequalities establish asymptotic Wilson-loop decay $\langle W_R\rangle\sim\beta/R^\alpha$ for $\alpha>2$. The central $1<\alpha\leq2$ Wilson-loop plateau remains conjectural: only an upper bound, dilute-quadrupole reasoning, and finite-size evidence are supplied. Diverging memory lifetime is likewise not proved; collective logical-error paths and entropic effects remain unresolved. The transition and crossover are not precisely characterized, and numerical sizes cannot establish the asymptotic distinction between full and partial self-correction.
Central claim may not hold: The proposed finite-temperature ordered phase relies on a nonzero asymptotic Wilson-loop plateau that is supported heuristically and numerically but remains explicitly conjectured. The claimed fully self-correcting memory has no established diverging logical lifetime because its Arrhenius argument does not control entropy or collective logical-error paths. The Gibbs-state long-range-entanglement argument assumes unestablished cluster statistics and decoder behavior, while its numerical RG test concerns the partially confining regime. The stated simulation normalization contradicts its claimed toric-code energy matching and, when inserted into the paper's confinement-length estimate, gives approximately 2.1 x 10^14 rather than the quoted 10^4.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 12+6 pages, 9+3 figures