PIE·INDUSTRIAL
claim integrity

The claims manifest.

compiled 42/42 · verdict ALL_TRACED

This site says every solver result it publishes traces to a hashed artifact. This page is the proof, not the promise. Before the site is built, each number below is resolved against the JSON artifact it came from, compared to the value actually stored there, and checked against a declared tolerance. If any claim fails to resolve, the build stops.

last compile
ALL_TRACED·42/42

Reproduce it with python claims/claim_compiler.py claims/site_claims.json. The bare command without that argument runs a self-test and proves nothing — a trap worth naming, because it looks green.

Every claim, and what it resolved to

stated is the number as it appears on the site. evidence is the number read out of the artifact. Δ is the relative difference between them, and it must sit under the tolerance the claim declared before it was compiled. The checksum is the SHA-256 of the artifact the value was read from, truncated to twelve characters.

Every claim published on this site, its stated value, the value read from the evidence artifact, the relative difference, the declared tolerance, the traced verdict and the artifact checksum.
Claim Stated Evidence Δ Tol. Traced Artifact SHA-256
combustion-chamber sector coupon
sector_axial_modulus_gpa_136_72 136.72 136.71942290768752 4.2e-6 0.001 yes 161b65d29d29…
sector_first_yield_mpa_165_01 165.01 165.01363345510612 2.2e-5 0.001 yes 161b65d29d29…
sector_lbm_iterations_1400 1,400 1,400 0 0 yes 2b58162fe0ac…
sector_lbm_permeability_um2_410_6 410.6 410.60007543688215 1.8e-7 0.001 yes 2b58162fe0ac…
sector_mesh_nodes_98691 98,691 98,691 0 0 yes 161b65d29d29…
sector_mesh_tets_406889 406,889 406,889 0 0 yes 161b65d29d29…
sector_mesh_triangles_134608 134,608 134,608 0 0 yes 13b683fb7441…
sector_thermal_agreement_1_002 1.002 1.0019385951456807 6.1e-5 0.001 yes 562de87f6653…
sector_thermal_conductivity_7_970 7.97 7.970388776552465 4.9e-5 0.001 yes 562de87f6653…
engine-support campaign
ch7_v8_winner_mass_kg_4_539 4.538786399995498 4.538786399995498 0 0.0001 yes d848e3e70eb8…
ch7_v8_winner_sf_1_730 1.7296608893243002 1.7296608893243002 0 0.0001 yes d848e3e70eb8…
ch7_v9_first_pass_mass_kg_5_485 5.485183919998773 5.485183919998773 0 0.0001 yes 10d053795349…
ch7_v9_first_pass_sf_1_833 1.8332975282907649 1.8332975282907649 0 0.0001 yes 10d053795349…
ch7_v9_vs_baseline_mass_ratio_2_515 2.5151277454091576 2.5151277454091576 0 0.0001 yes b1bcc0149a49…
ch7_v9_vs_ownfirstpass_mass_ratio_0_437 0.4370183452304088 0.4370183452304088 0 0.0001 yes b1bcc0149a49…
ch7_v9_vs_v8_mass_ratio_0_528 0.5281425008246011 0.5281425008246011 0 0.0001 yes b1bcc0149a49…
ch7_v9_winner_mass_kg_2_397 2.3971260000023107 2.3971260000023107 0 0.0001 yes 11b97eb83f54…
ch7_v9_winner_passes_true 1 1 0 0.001 yes 11b97eb83f54…
ch7_v9_winner_sf_1_759 1.7591978282795169 1.7591978282795169 0 0.0001 yes 11b97eb83f54…
engine-support campaign — declared baseline
ch7_baseline_mass_kg_0_953 0.9530831999995887 0.9530831999995887 0 0.0001 yes 6ac5cffa37ca…
ch7_baseline_passes_false 0 0 0 0.001 yes 6ac5cffa37ca…
ch7_baseline_sf_0_748 0.747862864544351 0.747862864544351 0 0.0001 yes 6ac5cffa37ca…
ground-truth ledger
ground_truth_ledger_record_count_402 402 402 0 0.001 yes 0b6818dd0a16…
hydraulic network benchmark
hanoi_design_rel_vs_best_known_0_269pct 0.002693 0.002693 0 0.001 yes d98ef7173683…
hanoi_solver_vs_epanet_max_head_diff_0_0143m 0.0143 0.0143 0 0.001 yes 15b535c5fd02…
math shelf
math_shelf_theorem_count_968761 968,761 968,761 0 0 yes 41b5e5956d89…
payload bracket demonstrator
bracket_winner_first_mode_hz_171_58 171.58 171.5802 1.2e-6 0.001 yes c75906121b6a…
bracket_winner_governing_sf_2_055 2.055 2.0554252760692346 2.1e-4 0.001 yes c75906121b6a…
bracket_winner_mass_kg_1_165 1.165 1.1646967199989717 2.6e-4 0.001 yes c75906121b6a…
bracket_winner_mesh_nodes_31487 31,487 31,487 0 0 yes c75906121b6a…
bracket_winner_mesh_tets_126665 126,665 126,665 0 0 yes c75906121b6a…
bracket_winner_vertical_3_5g_sf_7_40 7.4004 7.400416426964802 2.2e-6 0.001 yes c75906121b6a…
payload bracket — conventional baseline
baseline_first_mode_hz_54_4 54.4 54.39108 1.6e-4 0.001 yes 1236eb31fb0a…
baseline_foreaft_sf_1_64 1.64 1.64388426440355 2.4e-3 0.005 yes 1236eb31fb0a…
baseline_governing_sf_0_427 0.427 0.42677245065880937 5.3e-4 0.001 yes 1236eb31fb0a…
baseline_mass_kg_0_946 0.946 0.9459062399996104 9.9e-5 0.001 yes 1236eb31fb0a…
payload bracket — proxy-driven deep run
deep_champion_governing_sf_1_082 1.0821 1.0820500166857923 4.6e-5 0.001 yes 80409ab09c73…
deep_champion_mass_kg_0_898 0.898 0.8980384799991354 4.3e-5 0.001 yes 80409ab09c73…
deep_run_sf_range_max_1_28 1.2758 1.275759368379655 3.2e-5 0.001 yes 1067b769d161…
deep_run_sf_range_min_0_97 0.9713 0.9712666478734413 3.4e-5 0.001 yes 3378ed54c206…
structural core benchmark
cantilever_core_vs_simp_ratio_1_66pct 1.0166 1.0166372816156435 3.7e-5 0.001 yes b29c5df556fa…
mbb_beam_compliance_ratio_0_9893 0.9893 0.9893 0 0.001 yes c84a25c4b497…

What is withheld, and why

The manifest holds one more field per claim that this page does not show: the path to the evidence artifact. Those paths are directory names from real campaigns, and they carry a customer's name and an engine designation that is not ours to publish. So they are cut at the build step rather than paraphrased, and the build fails if any of them survive into the output.

Be clear about what that costs you. A checksum you cannot resolve against a file you do not have is not verification — it is a commitment. What it buys is that the value cannot be quietly changed later without the checksum changing too, and that the same manifest is what gets compiled under a work agreement, where the artifacts themselves come with it. That is what collaboration includes.

What is not in this manifest

Five numbers appear on this site and are not in the table above. All five come from the proxy-drift oracles described in the honest-failure case study: the proxy measured 40.1% optimistic, its reported accuracy understated ~81×, its confidence band ~48σ too narrow, and the surrogate's reported error of 0.0056 against an effective error of 0.454.

Those oracles wrote their results into an engineering journal entry rather than into a JSON artifact, so there is nothing for the compiler to resolve them against. They are not claim-grade and they are not counted in the 42. They stay on the site because deleting the measurement of our own blind spot would be the worse dishonesty — but they are marked here, by name, so that no reader has to discover the gap themselves.

The general rule: a number that cannot be compiled is either removed or declared. It is never quietly left to look like the others.