matter-embryogenesis / results /V2_RESULTS.md
PureOne's picture
Release Matter Embryogenesis v3.0.0: theory, code, data and audit
52fc221 verified
|
Raw
History Blame
3.1 kB

Results — Matter Embryogenesis v2

All results below are synthetic reduced-model results. The v1 data are preserved separately and are not a directly matched baseline.

Dimension Method Completed functional / 32 Mean G/G0 Added material Scheduled time
2 open_loop 0/32 0.652926 0.000000 245.472286
2 blueprint 0/32 0.797943 4.263125 251.582841
2 blind_reserve 1/32 0.841290 7.056000 250.385086
2 sensor_only 0/32 0.652890 0.000000 278.472286
2 early_seal 0/32 0.745264 3.203390 280.047146
2 contract 32/32 0.967874 10.707725 263.776243
2 conventional_joint 32/32 0.967874 10.707725 263.776243
2 biased_reference 6/32 0.922817 8.986592 276.035343
3 open_loop 0/32 0.588141 0.000000 709.048333
3 blueprint 0/32 0.708413 12.686771 729.007943
3 blind_reserve 1/32 0.773905 20.448000 723.583333
3 sensor_only 0/32 0.588008 0.000000 744.048333
3 early_seal 0/32 0.703078 11.563083 760.393162
3 contract 32/32 0.966938 38.773435 764.032782
3 conventional_joint 32/32 0.966938 38.773435 764.032782
3 biased_reference 0/32 0.915541 33.134518 777.418398

There are 512 default runs plus 32 seal-stress runs. Nominal functional material is 39.2 units in 2-D and 113.6 in 3-D. Scaffold material is additional. Reserve capacity is present in all designs; added material measures actual use. Blueprint repair is given ideal pre-conversion defect diagnosis and is charged replacement-bond material, discarded bonds, and delivery time.

The conventional_joint controller uses the same capabilities and policy and ties exactly. This is a required negative control on claims of algorithmic uniqueness.

Biased reference: 12 completed 2-D objects and four completed 3-D objects have invalid local certificates. Only six of those 2-D objects, and none of the 3-D objects, meet the global 8% conductance criterion.

Seal stress: all 32 objects have some invalid local certificates; all 32 nevertheless pass the one global scalar response. Mean local-contract fraction is 0.458705. Local contracts are sufficient, not necessary. This does not validate an incorrectly assumed 1% final-seal bound.

Theorem checks: 300 compositions, 250 energy bounds, 400 quadratic checks. Sixteen tests pass, including 300 bounded-noise repair cases in one test. Numerical checks are not independent proof certification.

No reported time is in seconds. No molecular species count, physical energy, nucleotide sequence, or verified chemical actuator is claimed. Complete parameters and seeds are in results/runs.json.

Canonical capsules are 126 bytes (1,008 bits) each; the fixed grammar interpreter is approximately 3 kB. The 300-module packed three-bit role list would be 900 bits. These small demonstrations are not evidence of net compression; the family result concerns asymptotic reuse of a paid interpreter. Molecular encoding and runtime remain unimplemented.