arXiv:2609.39345
Constant-Overhead Injection into Quantum Codes
Louis Golowich, Venkatesan Guruswami
Abstract
We construct the first known fault-tolerant scheme for injecting states into quantum error-correcting codes with constant space and time overhead. That is, we construct a family of constant-rate quantum error-correcting codes for which a set of bare physical qubits can be injected, i.e. fault-tolerantly encoded, into a code block. Similarly, a code state can be ejected, i.e. fault-tolerantly decoded, back into bare physical qubits. We show that these injection and ejection procedures succeed under circuit-level locally stochastic noise, while incurring just a small constant probability of corrupting each qubit, which is unavoidable for bare physical qubits. We also show how to perform fault-tolerant error correction and code-state preparation under locally stochastic noise. All of our gadgets can be implemented with constant quantum circuit depth (i.e. are single-shot), and with a number of physical qubits growing linearly with the number of logical qubits, assuming the ability to run polynomial-sized noiseless classical circuits on the side. We construct our quantum codes by taking a high-dimensional hypergraph product of classical LDPC codes, which in turn are a simplified version of Spielman's linear-time encodable codes (STOC'95). As our resulting product codes are only resilient to physical errors occurring with non-uniform probabilities across qubits, we then show how to concatenate with inner codes of various sizes to obtain fault-tolerance against uniform noise.
AI Summary
The paper asks whether bare quantum data can be fault-tolerantly encoded into, and decoded from, quantum error-correcting codes with constant space and quantum-time overhead. It constructs a code family with parameters $[[n=\Theta(k),k,d=\mathrm{poly}(k)]]$ supporting injection, ejection, logical $|0\rangle^{\otimes k}$ and $|+\rangle^{\otimes k}$ preparation, transversal blockwise CNOT, and error correction. Each gadget uses $O(k)$ physical qubits and $O(1)$ quantum depth under circuit-level locally stochastic noise, assuming polynomial-sized noiseless classical computations can run within each timestep. Operations other than injection and ejection have failure probability $2^{-\mathrm{poly}(k)}$. The construction takes high-dimensional hypergraph products of hierarchical classical LDPC codes adapted from linear-time-encodable expander codes. Four-dimensional products supply syndrome redundancy for single-shot state preparation. Decoding combines pushing errors into selected sectors with a high-dimensional small-set-flip decoder. Ejection uses physical Pauli measurements and classical corrections; injection teleports through encoded–physical Bell pairs. The initial analysis assumes level-dependent noise rates $p^{2^{\Omega(\ell)}}$. Concatenation with differently sized inner codes converts uniform physical noise into this profile while retaining constant overhead. Injection and ejection necessarily introduce errors on bare data. The guarantees allow correlated errors: except with small constant probability, only a small fraction of data qubits are corrupted, yielding low marginal error probabilities on a large fraction of positions once the noise distribution is fixed. Locally stochastic output errors are not established. A complete universal fault-tolerance scheme and practical finite-size performance remain open.
ChatGPT 6 Astra was used to find references, make figures, and check over the final draft.
Central claim may not hold: The abstract's small corruption probability for every data qubit is not established by the body's weaker, explicitly acknowledged guarantee of low marginal probabilities only for a large fraction of positions identified after fixing the fault distribution. The written lossless-expansion definitions lack a restriction on the size of the expanded set, making the stated expander existence lemma false when applied to the entire left vertex set.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.