arXiv:2610.12597
Error correction and quantum information spreading on sparse random graphs
Giovanni Cemin, Marin Bukov, Andrea Solfanelli
Abstract
Quantum error-correcting codes on graph geometries beyond regular euclidean lattices have emerged as a promising route toward robust quantum information processing. A major challenge in quantum error correction is the trade-off between favorable properties of codes constructed in high dimensionality and the need for locality in practical implementations on near-term hardware. Sparse intermediate geometries may combine the advantages of both, but how their structure governs quantum information dynamics and code performance remains unclear. Here we show that the spectral dimension $d_s$, defined from the low-energy spectrum of the graph Laplacian, is the universal parameter connecting these properties. Using Floquet random Clifford circuits on two distinct sparse graph architectures with tunable spectral dimension, we find that $d_s$ governs both quantum information spreading and the scaling of the resulting quantum error-correcting code distance. We identify a geometry-induced entanglement phase transition from localized area-law to ergodic volume-law, whose universality is determined by $d_s$ and exhibits long-range percolation criticality. In the ergodic phase, the circuit acts as a CSS code encoder whose code distance is set by the competition between two logical sectors --- the cost of severing the graph and the length of its shortest winding cycle --- which connectivity pushes in opposite directions. The spectral dimension organizes this trade-off, highlighting a maximum at intermediate $d_s$ retaining sparse connectivity and physical realizability.
AI Summary
The paper asks whether a geometric parameter can organize entanglement dynamics and error-correction properties on sparse, non-Euclidean interaction graphs. It studies periodically repeated Clifford circuits on two random graph families: algebraically distributed long-range bonds and bonds restricted to power-of-two separations. Both have tunable annealed spectral dimension $d_s$ and an average of $O(N)$ gates per period. An analytical cutting-probability calculation gives $P_{\rm cut}\sim\mathrm{const}>0$ for $d_s<2$, algebraic decay at $d_s=2$, and $P_{\rm cut}\sim\exp[-cN^{1-2/d_s}]$ for $d_s>2$, under stated spectral and eigenmode-resolution assumptions. The resulting picture distinguishes geometries localized by absolute cuts from those supporting a collective localization–delocalization transition. Stabilizer simulations and conditional-mutual-information finite-size scaling, primarily at $N=128$–$2048$, find compatible correlation-length exponents between the two graph families at matched $d_s$, broadly consistent with long-range percolation. The QEC analysis proposes a CSS interpretation with competing cut-like and winding-cycle logical sectors. It assigns scalings $d_X\sim N^{1-2/d_s}$ and, for long-range graphs, $d_Z\sim(\ln N)^\Delta$, where $\Delta^{-1}=\log_2(2/\alpha)$. Graph calculations up to $N\approx10^6$ show finite-size distance optima from this competition. Limitations include substantial finite-size corrections, critical-exponent estimates restricted to $d_s\le10/3$, assumptions connecting annealed spectra to random-graph behavior, and an approximate winding-cycle calculation. Extensions beyond CSS and Clifford circuits remain open. Experimental feasibility estimates neglect transport, idling, and single-qubit errors; the quoted $N\lesssim100$ scale concerns one Floquet period, not steady-state preparation.
Central claim may not hold: The claimed generic cutting-probability scaling requires embedding-dependent spectral-overlap assumptions beyond spectral dimension, with additional hypotheses at the marginal dimension. Numerical agreement between two related graph families at four matched annealed spectral dimensions does not establish the asserted universality for generic sparse interaction graphs. The specified CNOT-only encoder with |0> ancillas always has a weight-one logical Z and hence quantum distance one, invalidating the claimed diverging cut/cycle code distances. A unitary acting on one arbitrary input qubit and fixed ancillas always encodes a two-dimensional subspace, contrary to the claim that the logical qubit disappears above the percolation threshold. The asserted exact non-blocking-edge percolation mapping conflicts with the paper's own demonstration that information can propagate across an individual blocking CZ edge.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 31 pages, 8 figures