{
 "key": "v3a",
 "shas": {
  "BS": "a9830b417dd43806c144e8b4a217cda1317850bb52578b18a3237307cf8dfa16",
  "VAR": "357d9b5ada279be132765650dab59638e254386ffc749139e423a84e4bd526e7",
  "NP2": "51f1c1a9d4999ee460eb8bbc66880da8074b173399a4f5d71f96671482e18ac0"
 },
 "n": 0,
 "complete": false,
 "blocked": {
  "step": "check_np2_0_999",
  "error": "NP2 on seeds 0..999 gave mean edge 526.115573224169, expected 526.1155722459862 (difference 9.782e-07). Per the brief a mismatch blocks the item, so V3/BS/NP2 were not run on seeds 3208000..3209999. All three file SHA-256s matched (BS a9830b41..., VAR 357d9b5a..., NP2 51f1c1a9...). The simulator was built from the pinned commit's source tarball on macOS aarch64 (rustc 1.99.1, Python 3.12) because this machine is not linux x86_64. The gap is about 2e-9 relative, which points to platform floating-point (libm) differences in the Rust simulator, not a config or bytecode error. Suggestion: run this block on linux x86_64, or accept a tolerance (e.g. |diff|<1e-5) for the check."
 },
 "check_np2_0_999": 526.115573224169,
 "pairs": {},
 "worst5_var_bs": [],
 "seconds_per_simulation": 0.03142583322525024
}