Error-Correction Transitions in Finite-Depth Quantum Channels

Arman Sauliere, Guglielmo Lami, Pedro Ribeiro, Andrea De Luca, Jacopo De Nardis · 2026

Read Error-Correction Transitions in Finite-Depth Quantum Channels by Arman Sauliere, Guglielmo Lami, Pedro Ribeiro, Andrea De Luca, Jacopo De Nardis (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-20369.txt.

Introduction

Sauliere and co-authors study how error correction in quantum channels fails as noise grows. A channel encodes logical information into a larger quantum space. They model channels as one-dimensional random noisy circuits with local gates. Noise can strike after encoding or during the encoding circuit itself. In the deep limit, random matrix theory predicts a sharp transition. Past a critical noise rate, encoded information is lost for good. Finite depth behavior differs by case, approaching perfect encoding at distinct rates.

Catalog record

CREATORArman Sauliere, Guglielmo Lami, Pedro Ribeiro, Andrea De Luca, Jacopo De Nardis
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE27919
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.20369

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Error-Correction Transitions in Finite-Depth Quantum Channels Arman Sauliere, Guglielmo Lami, Pedro Ribeiro, Andrea De Luca, Jacopo De Nardis Abstract We study error-correction–type protocols in which a quantum channel encodes logical information into an enlarged Hilbert space. Specifically, we consider channels realized by one-dimensional random noisy quantum circuits with spatially local interaction gates. We analyze both noise acting after the encoding and noise affecting the encoding circuit itself. Using the coherent information as a metric, we show that in both cases the infinite-depth limit is governed by random-matrix theory, which predicts a universal phase transition at a critical noise rate. This critical point separates an error-correcting phase, in which encoded information is preserved, from a phase in which it is irretrievably lost. Going beyond the infinite-depth limit, we characterize the systematic finite-depth deviations from random-matrix universality. In particular, we show that these deviations behave parametrically differently depending on whether the noise acts after the encoding or also affects the encoding itself. For noiseless encoders, the approach is exponential in circuit depth, although boundary effects can delay perfect encoding relative to the circuit design time. For noisy encoders, we find that the circuit fidelity effectively replaces the Hashing bound, and perfect encoding is approached polynomially with depth. I Introduction . Quantum computers are expected to unlock computational capabilities beyond those of classical devices, from quantum many-body simulation to algorithmic speedups in specific tasks Feynman 1982 ; Lloyd 1996 ; Shor 1997 ; Grover 1997 ; Bernstein and Vazirani 1997 ; Aaronson and Arkhipov 2011 ; Zhong et al. 2020 ; Kretschmer et al. 2025 . Realizing this promise, however, requires going beyond the noisy intermediate-scale quantum (NISQ) regime and entering the fault-tolerant one, where logical information can be processed reliably despite decoherence and imperfect control Preskill 2018a ; Daguerre et al. 2025 ; Aasen et al. 2025 ; Peham et al. 2025 . The standard route is quantum error correction (QEC), which encodes logical degrees of freedom into a larger Hilbert space using structured codes, often based on stabilizers and Clifford operations Shor 1995 ; Gottesman 1997 ; Gottesman 1998 ; Nielsen and Chuang 2010 ; Bluvstein et al. 2023 ; Gidney 2021 . This remains one of the most active directions in quantum information research Yuan et al. 2026 ; Gu et al. 2026 ; Wills et al. 2026 ; Perlin et al. 2026 ; Hetényi et al. 2026 ; Lo et al. 2025 ; Jing et al. 2025 ; Sala and Verresen 2025 ; Knill et al. 2000 ; Wilde 2013 ; Preskill 2018b . Alongside this structured paradigm, it is useful to study generic encoding channels generated by random circuits Hayden and Preskill 2007 ; Choi et al. 2020 ; Fan et al. 2020 ; Turkeshi and Sierant 2024 ; Sommers et al. 2025 ; Nelson et al. 2025 ; Dias et al. 2022 ; Yi et al. 2025 ; Kroll and Helsen 2026 ; Li et al. 2025 . These serve both as theoretical laboratories and as effective models of hardware-efficient scramblers. Here we study channels that encode k=rN logical qudits into N physical qudits through a generic one-dimensional brickwall circuit of depth t (Fig. 1). As a diagnostic of information transmission we use the coherent information I_{\mathrm{\rm c}} , the key quantity underlying quantum capacity theorems Devetak and Shor 2005 ; Hayden et al. 2008 . Our framework unifies and extends several recent works Nelson et al. 2025 ; Choi et al. 2020 ; Zhuang et al. 2023 , with particular emphasis on finite-depth random encoders and the threshold phenomena they develop as functions of depth and noise strength. We consider two scenarios (Fig. 1 ). In setup I, the encoding circuit is unitary and noise acts only after encoding on the physical qudits. In setup II the encoding itself is noisy: a noise channel is applied after each unitary gate. Figure 1: We study the coherent information of a random encoding circuit at finite depth t and investigate its ability to protect quantum information against noise, whether the noise acts after the encoder or within the encoding circuit itself. In setup I, the large-depth limit is governed by random-matrix (RM) universality: the encoder generates an effectively Haar-random code subspace, and the capacity problem reduces to an RM calculation closely related to decoupling Hayden et al. 2008 . One then finds a sharp transition at a critical noise rate fixed by the Hashing bound Wilde 2013 , H=1-r , separating an error-correcting phase from one in which information is lost. In the protected phase, recovery becomes essentially perfect, with I_{\mathrm{\rm c}}=k up to corrections exponentially small in N . These connections between capacity, scrambling, and open-system information dynamics have been widely explored Gullans and Huse 2020 ; Choi et al. 2020 ; Zhuang et al. 2023 ; Sommers et al. 2025 . A main goal of this work is to go beyond this asymptotic RM picture and characterize finite-depth corrections. In setup I, within the protected phase, the approach to perfect encoding is exponentially fast in depth, as noticed also in recent works, but with corrections of the form Ne^{-2t/\tau}+O(e^{-t/\tau}) , analogous to those associated with the formation of a quantum state design, governed by Ne^{-2t/\tau} Lami et al. 2025 ; Dowling et al. 2025 ; Harrow and Low 2009 ; Brandão et al. 2016 ; Liu et al. 2026 , where \tau is the purity-decay time-scale Nahum et al. 2017a ; Nahum et al. 2018a . In setup II, recently considered also in Ref. Nelson et al. 2025 , the encoder does not generically reach the same ideal RM universality at large depth. Nevertheless, we show that the coherent information retains the same universal functional form when expressed in terms of the circuit fidelity F , whose logarithm replaces the Hashing bound as the relevant measure of noise. For circuits with fixed fidelity, the convergence to perfect encoding is much slower than in the previous case: corrections are inversely proportional to the circuit depth. Our results connect both settings to the statistical-mechanics description of Haar-averaged random circuits in replica space, where the averages map to a gas of domain walls between permutations. II Background . A quantum error-correcting code maps k logical qudits into N physical qudits through an encoding channel. The k logical qudits, each of local dimension d , are supplemented by N-k ancillas initialized in |0\rangle , and a quantum channel \Lambda describing the encoding circuit and possible noise acts on the full system. We denote the resulting physical output by B . Equivalently, the logical input can be taken maximally entangled with a reference system R of k qudits via Bell pairs \ket{\phi_{+}}=d^{-1/2}\sum_{i=0}^{d-1}\ket{i}\ket{i} , while R remains untouched by \Lambda . The final state is then \rho_{RB}=(\mathbbm{1}_{R}\otimes\Lambda)(\ket{\boldsymbol{\phi_{+}}}\bra{\boldsymbol{\phi_{+}}}\otimes\ket{\boldsymbol{0}}\bra{\boldsymbol{0}}), where \ket{\boldsymbol{\phi_{+}}}=\ket{\phi_{+}}^{\otimes k} and \ket{\boldsymbol{0}}=\ket{0}^{\otimes N-k} . The coherent information is I_{\mathrm{\rm c}}(\Lambda)=S(\rho_{B})-S(\rho_{RB})\,, (1) where S(\rho) denotes the Von Neumann entropy. The coherent information optimized over the input state for single-qudit noise \mathcal{N} is directly related to the quantum channel capacity through \mathcal{Q}(\mathcal{N})=\lim_{N\to\infty}1/N\,\max_{\rho}I_{\mathrm{\rm c}}(\rho,\mathcal{N}^{\otimes N}) Lloyd 1997 ; Devetak and Shor 2004 . In our setting, the input state is chosen to be maximally mixed, i.e. \rho=\mathbbm{1}/d^{k} . A value I_{\mathrm{c}}(\Lambda)/k=1 corresponds to perfect protection

Read the complete text