Catalytic Quantum Error Correction: Theory, Efficient Catalyst Preparation, and Numerical Benchmarks
Hikaru Wakaura · 2026
Read Catalytic Quantum Error Correction: Theory, Efficient Catalyst Preparation, and Numerical Benchmarks by Hikaru Wakaura (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-2603-25774.txt.
Introduction
Wakaura and Tanimae present catalytic quantum error correction, a new way to recover quantum states from noise. Standard correction fails once noise passes about one percent and costs many extra qubits. Their method uses catalytic coherence amplification, an effect from the resource theory of coherence. It restores a known target state from noisy copies with no noise threshold, as long as key modes survive. In an effective model, fidelity topped 0.99 across 200 noise settings. A small exact test under dephasing also showed real recovery, though shallow circuits drain the catalyst.
Catalog record
| CREATOR | Hikaru Wakaura |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 66921 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.25774
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Catalytic Quantum Error Correction: Theory, Efficient Catalyst Preparation, and Numerical Benchmarks Hikaru Wakaura Abstract We introduce Catalytic Quantum Error Correction (CQEC), a state recovery protocol exploiting catalytic covariant transformations 32 . CQEC recovers a known target state from noisy copies without an error magnitude threshold: recovery succeeds whenever the coherent modes satisfy 𝒞 ( ρ 0 ) ⊆ 𝒞 ( ρ noisy ) \mathcal{C}(\rho_{0})\subseteq\mathcal{C}(\rho_{\mathrm{noisy}}) , regardless of noise strength. The main practical bottleneck—catalyst preparation requiring n ∗ ∼ d 4 e 2 γ n^{*}\sim d^{4}e^{2\gamma} copies—is resolved by a three-stage pipeline combining CPMG dynamical decoupling, Clifford twirling, and the recursive swap test, achieving F cat > 0.96 F_{\mathrm{cat}}>0.96 with only 8 copies ( 10 9 10^{9} -fold reduction). Numerical validation across four quantum algorithms ( d = 4 d=4 – 64 64 ), a cryptographic protocol, and three noise models confirms F > 0.999 F>0.999 in the asymptotic limit across 200 configurations. I Introduction Quantum computers promise exponential speedups for problems in cryptography 30 ; 34 , quantum simulation 6 ; 13 , and machine learning 19 ; 4 , but their practical realization is hindered by the fragility of quantum coherence under environmental decoherence 46 ; 29 . Quantum error correction (QEC) addresses this challenge by encoding logical information redundantly across physical qubits, enabling the detection and correction of errors below a code-specific threshold 33 ; 15 . However, the qubit overhead of conventional QEC is substantial: the Steane [\![7,1,3]\!] code requires 7 physical qubits per logical qubit 36 , while surface codes at distance d require O(d^{2}) qubits 21 ; 14 , and correction fails entirely when error rates exceed the threshold p_{\mathrm{th}}\approx 1\% 22 . A distinct approach to quantum state protection arises from the resource theory of quantum coherence 2 ; 37 . Coherence—the presence of off-diagonal elements in a state’s density matrix with respect to a preferred basis—is a quantifiable resource under the framework of covariant (energy-conserving) operations 26 ; 24 . Shiraishi and Takagi 32 recently established that the transformation rate between coherent states can diverge arbitrarily: given n copies of a weakly coherent state \rho , one can produce m\gg n copies of a target state \rho^{\prime} via catalytic covariant operations, provided that the coherent modes of \rho^{\prime} are contained within those of \rho . This result reveals that the output-to-input copy ratio m/n can be made arbitrarily large (the asymptotic rate diverges), creating an infinitely sharp boundary between recoverable and irrecoverable quantum states—a feature with no analog in conventional QEC. The role of coherence in quantum thermodynamics and resource theories has been extensively studied 24 ; 37 ; 26 , with catalytic transformations appearing in the context of thermodynamic state conversion 2 and entanglement manipulation. The interplay between symmetry constraints and quantum error correction has been studied in the context of covariant codes 12 and approximate QEC 17 . Covariant codes 12 encode information in a way that respects a symmetry group, with an inherent trade-off between code accuracy and the symmetry charge variance (the Eastin–Knill theorem for exact codes 11 ). CQEC differs fundamentally: it does not encode information redundantly but instead uses catalytic transformations to amplify coherence in the original Hilbert space. However, the specific application of catalytic coherence amplification to quantum error correction has not been explored. In particular, it remains unclear whether the infinite amplification rate translates into effective protection for realistic quantum algorithms operating under decoherence, what the finite-copy resource costs are, and how these costs compare with conventional QEC. Moreover, even assuming the theoretical validity of catalytic recovery, the practical bottleneck is catalyst preparation. The constructive proof of Ref. 32 requires n^{*}\sim d^{4}e^{2\gamma} noisy copies to prepare a catalyst with fidelity F_{\mathrm{cat}}\geq 0.99 , where d is the Hilbert space dimension and \gamma is the dephasing strength. For a 6-qubit system ( d=64 ) under moderate dephasing ( \gamma=2 ), this translates to n^{*}\approx 7.3\times 10^{10} copies—far beyond current experimental capabilities. Recent advances in quantum state purification 8 ; 44 ; 45 and dynamical decoupling 39 ; 27 offer paths to dramatically better scaling. In this paper, we bridge these gaps by (i) introducing the Catalytic Quantum Error Correction (CQEC) protocol with explicit success conditions, (ii) developing a quantum circuit architecture based on energy-conserving gates, (iii) proposing four catalyst preparation strategies culminating in a DD+Twirl+Swap Test pipeline that achieves a 10^{9} -fold copy-count reduction, and (iv) providing comprehensive numerical validation. Our central contributions are: 1. Protocol and circuit. We formulate the CQEC protocol with explicit success/failure conditions based on mode inclusion, implement it via energy-conserving gates in a minimal 4-qubit circuit, and provide finite-copy fidelity bounds. 2. Efficient catalyst preparation. We develop four catalyst preparation strategies culminating in the DD+Twirl+Swap Test pipeline, which achieves a 10^{9} -fold reduction in copy count over distillation under strong dephasing. 3. Comprehensive benchmarks. We validate CQEC across four quantum algorithms ( d=4 – 64 ), a cryptographic protocol, and three noise models (200 configurations), including comparison with conventional QEC codes, threshold verification across 10 orders of magnitude, and 100-cycle catalyst reuse. While Theorems 1 and 2 are due to Shiraishi and Takagi 32 , the application to quantum error correction, the circuit architecture, the catalyst preparation pipeline, and the comprehensive numerical validation are original to this work. II Theoretical Framework II.1 Unspeakable coherence and covariant operations We work within the resource theory of unspeakable coherence 32 ; 25 ; 26 ; 24 , where the free operations are covariant channels—completely positive trace-preserving maps \Lambda satisfying \Lambda\circ\mathcal{U}_{t}=\mathcal{U}_{t}\circ\Lambda\quad\forall\,t\in\mathbb{R}, (1) where \mathcal{U}_{t}(\rho)=e^{-iHt}\rho\,e^{iHt} is the time evolution generated by the system Hamiltonian H=\sum_{i}E_{i}\ket{i}\!\bra{i} . Condition ( 1 ) enforces energy conservation: [U,H_{\mathrm{total}}]=0 for any unitary U implementing \Lambda 26 . This is physically motivated by the Wigner–Araki–Yanase theorem, which constrains measurements and operations that do not commute with conserved quantities 42 . For a d -level system, the \ell_{1} -norm of coherence quantifies the total off-diagonal content: C_{\ell_{1}}(\rho)=\sum_{i\neq j}|\rho_{ij}|. (2) The quantum Fisher information \mathcal{F}(\rho,H)=2\sum_{i,j}\frac{(\lambda_{i}-\lambda_{j})^{2}}{\lambda_{i}+\lambda_{j}}|\!\braket{i|H|j}\!|^{2} 5 provides a complementary coherence monotone under covariant operations 32 . Unlike C_{\ell_{1}} , QFI is operationally meaningful for metrology and satisfies \mathcal{F}(\rho_{\mathrm{noisy}})\leq e^{-2\gamma}\mathcal{F}(\rho_{0}) under dephasing. We use QFI in Sec. VI.2 to quantify the coherence available for recovery. The two measures are complementary: C_{\ell_{1}} counts total off-diagonal weight (relevant for mode coverage), while QFI weights contributions by their metrological utility. Under dephasing, both decay exponentially but at different rates: C_{\ell_{1}}\sim e^{-\gamma} (slowest mode), \mathcal{F}\sim e^{-2\gamma} (quadratic suppression), making QFI a more sensitive indicator of coherence loss. II.2 Coherent modes and mode inclusion The modes of asymmetry 24 ; 25 of a state \rho are defined as \mathcal{D}(\rho)=\bigl\{\Delta_{ij}=E_{i}-E_{j}\mid\rho_{ij}\neq 0\bigr\}. (3) The resonant coherent modes \mathcal{C}(\rho) are the integer-linear span of \mathcal{D}(\rho) : \mathcal{C}(\rho)=\biggl\{\sum_{(i,j)}n_{ij}\,\Delta_{ij}\;\bigg|\;n_{ij}\in\mathbb{Z},\;\Delta_{ij}\in\mathcal{D}(\rho)\biggr\}. (4) \mathcal{C}(\rho) forms a subgroup of (\mathbb{R},+) and determines the transformability of states under covariant operations 32 . Theorem 1 (Shiraishi–Takagi 32 ). If \mathcal{C}(\rho^{\prime})\subseteq\mathcal{C}(\rho) and \rho is full rank, the asymptotic marginal transformation rate R(\rho\to\rho^{\prime})=\sup\biggl\{\frac{m}{n}\;\bigg|\;\rho^{\otimes n}\xrightarrow{\mathrm{cov}}\sigma,\;\mathrm{Tr}_{[m]\setminus\{k\}}[\sigma]\approx\rho^{\prime}\;\forall k\in[m]\biggr\} diverges: R(\rho\to\rho^{\prime})=\infty . Theorem 2 (Shiraishi–Takagi 32 ). For any \rho,\rho^{\prime} with \mathcal{C}(\rho^{\prime})\subseteq\mathcal{C}(\rho) , there exists a catalyst c and a covariant operation \Lambda such that \mathrm{Tr}_{C}\bigl[\Lambda(\rho\otimes c)\bigr]=\rho^{\prime},\quad\mathrm{Tr}_{S}\bigl[\Lambda(\rho\otimes c)\bigr]=c. (5) That is, \rho\to\rho^{\prime} is achievable by correlated-catalytic covariant transformation. Crucially, the mode inclusion condition