QALBench: A Tiered Audit Benchmark and Reference Implementation for Quantum Artificial-Life Claims

Quantum artificial-life (QAL) claims combine two demanding vocabularies. Artificial-life labels such as inheritance, mutation, interaction, turnover, and selection require population records and lineage controls. Quantum-resource labels such as coherence, entanglement, Bell nonlocality, and channel resources require resource diagnostics and resource-destroying controls. QALBench makes these claims independently auditable. It is a tiered benchmark suite and reference implementation that separates artificial-life adequacy, quantum-resource evidence, and computational nonclassicality.

The reference results are deliberately diagnostic. Under matched seeds, basis-selection demographic, phenotype, and lineage summaries are identical for the quantum, dephased, and classical controls, excluding resource diagnostics. A software positive-control assay inserts a stored negativity-derived birth-event score into the reproduction rule and shows that resource-coupled reproduction is removed by dephasing or Markov controls. A no-inheritance resource control remains resource-positive but transmission-negative, establishing that resource-coupled reproduction and inherited resource-mediated artificial life are distinct claims. QALBench also supplies finite-shot certification and simulator-challenge submission workflows for external evidence. Its included reference packages exercise those workflows without asserting quantum advantage, loophole-free hardware certification, open-ended evolution, or persistent population-level entanglement.