Skip to content

TN/MPS Crossover Stage-1 Gate

This QWC-5.1 artifact admits the larger-than-16-node N=30-40 tensor-network crossover row format before any owner-gated compute run.

Boundary

Stage-1 schema and admission evidence for N=30-40 tensor-network crossover rows only; no N=30-40 solver row, hardware advantage, tensor-network hardness, or broad quantum-advantage claim is established.

Row Schema

  • protocol: qwc_5_1_tn_mps_crossover_stage1
  • target sizes: 30, 32, 36, 40
  • required baselines: classical_ode, mps_tensor_network, aer_statevector_or_skip
required field
protocol_id
n_qubits
baseline
status
wall_time_ms
memory_bytes
max_bond
discarded_weight
entropy_proxy
truncation_policy
omitted_coupling_mass
command
machine
dependencies
git_commit
host_load
claim_boundary
notes

Stage-1 Gates

gate passed evidence
target_sizes_exceed_sixteen True target_sizes=(30, 32, 36, 40)
row_schema_pinned True 18 required row fields pinned
stage2_compute_owner_gated True stage-2 N=30-40 execution remains owner-gated
claim_boundary_closed True design keeps advantage_claim_allowed=false and benchmark_execution_performed=false

Blocked Claims

  • broad quantum advantage
  • tensor-network hardness
  • hardware scaling win
  • GPU tensor-network comparison
  • Julia/ITensor parity

Owner-Gated Follow-ups

  • execute N=30-40 CPU-first quimb rows with isolated host metadata
  • promote or skip GPU tensor-network rows only through the owner-gated GPU lane
  • add maintained Julia/ITensor parity only after Julia toolchain ownership exists

Regeneration

scpn-bench s2-tn-crossover-stage1