arXiv:2610.31443
ConQEC: Concolic Testing for Quantum Error-Correcting Code Implementations
Zihao Li, Ji Guan, Mingsheng Ying
Abstract
Quantum error-correcting code (QECC) implementations are hybrid software-control systems: measured syndromes drive decoder dispatch, recovery actions, and logical checks. Existing QECC verification techniques can prove formal obligations, and general quantum testing can expose final failures, but developers still lack a scalable testing workflow that generates concrete faults and explains where a QECC repair path first diverges. This paper presents ConQEC, to our knowledge the first concolic testing framework for QECC implementations. By combining concrete (Pauli) fault replay with symbolic repair-path constraints, ConQEC generates tests that expose not only final logical failures but also fault--syndrome--decoder--recovery (FSDR) diagnostics: the injected fault, observed syndrome, decoder decision, recovery action, oracle result, and earliest divergent stage. Its symbolic encoding of Pauli faults, syndrome predicates, decoder and recovery relations, and oracle checks enables this diagnostic workflow to scale beyond exponentially increasing amplitude-level quantum-state reasoning. We implemented ConQEC and evaluated it on generated source-level QECC families and public QECC artifacts with up to thousands of qubits. ConQEC detects the seeded bug in every experimental case and reports a replay-valid, fault-localizing FSDR witness for each.
AI Summary
ConQEC addresses test generation and fault localization for quantum error-correcting-code software, where syndrome extraction, decoder dispatch, and recovery can diverge from a reference implementation. It combines concrete Pauli-fault replay with symbolic constraints over affine syndrome/detector parities and available decoder/recovery relations, rather than quantum-state amplitudes. Solver-generated faults are replayed on implementation and reference; reports record the fault, syndrome, decoder decision, recovery, oracle outcome, and earliest observable divergence. Generation targets semantic coverage or implementation–reference disagreement. Implemented over a symbolic-stabilizer backend and artifact adapters, ConQEC localizes all 195 seeded mutants across 13 generated code configurations, including 1,500-qubit subjects. Counting the clean seed and generation attempts, it finds all witnesses within four attempts; branch-order search finds none and random testing averages 15.7/195 at that budget. On a 1,000-qubit repetition code with weight-two faults, 500 tests cover 49.9% of adjacent-pair activation obligations, versus 0.10% for random testing. Completed source-level campaigns reach 3,000-qubit repetition and 3,362-qubit surface/toric subjects; repeated SMT solving, not replay, limits larger campaigns. Public-artifact tests record 36 logical-prediction disagreements between two decoders and reproduce a previously fixed PyMatching regression. The evidence is bounded by declared fault models, exposed predicates, and reference contracts. Seeded mutations do not establish real-world bug prevalence; coverage comparisons use repetition codes and weight-two faults. Public decoder disagreements are not established bugs. New interfaces require manual mappings, and broader support for structured repetition blocks and temporal fault histories remains future work.
Error found: The abstract's statement that ConQEC detects the seeded bug 'in every experimental case' should be restricted to the 195 seeded-mutant cases, because the public-artifact evaluation also includes decoder disagreements that the paper explicitly does not establish as bugs.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 12 pages. Accepted at the 49th IEEE/ACM International Conference on Software Engineering (ICSE 2027)