Standardised Synchronisation Benchmark Suite

May 18, 2026 ยท View on GitHub

This suite defines canonical coupled-oscillator benchmark instances and a stable result schema. It is a contract for future backends, not a new hardware claim.

Result schema

Required result fields:

  • benchmark_id
  • backend
  • backend_version
  • command
  • commit
  • dependency_lock
  • hardware_submission
  • wall_time_s
  • observables
  • claim_boundary

Observable row fields:

  • name
  • value
  • uncertainty
  • units
  • tolerance
  • passed

Canonical instances

BenchmarkFamilyNEvidence classRequired backendsClaim boundary
kuramoto_ring_n4_linear_omegakuramoto_xy_reference4analytic_referenceclassical_ode, exact_diagonalisationReference smoke instance for compiler and observable consistency; not a hardware-performance or advantage claim.
kuramoto_chain_n8_decay_omegakuramoto_xy_reference8simulator_referenceclassical_ode, exact_diagonalisationSimulator/reference benchmark for scaling behaviour; not a broad quantum-advantage claim.
phase1_dla_parity_n4_ibm_kingstonhardware_replay4hardware_replayhardware_replay, classical_parity_referenceReplays committed raw counts only; does not submit QPU jobs or promote broader scaling, mitigation, or advantage claims.
bkt_finite_size_grid_plannedphase_transition_reference16plannedclassical_xy, exact_or_tensor_referenceVisible roadmap row only; not an available benchmark result.

Hardware rule

hardware_submission must be false unless a preregistered manifest, QPU budget, raw-count target path, and explicit approval are recorded

Claim boundary

The registry defines benchmark contracts and replay rows. It does not create new hardware evidence or quantum-advantage claims.