DC-M4 — Empirical Constituent Calibration and Full-State Body Integration adds independent physical optical-trap relaxation shape evidence and a new full-state moving-volume finite-bath body with actual native24→48 remesh/continuation. This preserves the qualification below. Full Digital Cell chemical calibration and whole-organism integration remain unqualified; production values remain null. Draft PR55.
PARTIAL — DESIGNED REFERENCE RECOVERY AND BOUNDED ACTUAL NATIVE MECHANICS QUALIFIED. Empirical four-parameter calibration, constant epsilon_A and whole-organism integration remain unqualified.
Owner requests parameter-calibration and native-integration qualification. The sealed protocol tests blind K/lambda/theta recovery with eight prepared training probes and six held-out probes. Values are designed computational inputs; production K/lambda/theta/epsilon_A remain null.
The opt-in feature retains the full six-species fixed-volume source and calls actual native six-vertex mesh mechanics, accounting for every native stretch receiver. Native cubic biology, transport, growth, remesh and fission remain outside this fixture.
An exact coarse-state counterexample shows that identical mapped a/b/A/W/P/length can have different rates and chemical free energy. Exact full-domain four-parameter closure remains inadmissible; a separately budgeted local approximation has not been ruled out or qualified.
Draft PR54 follows DC-M4 — Chemical/Kinetic Computational Reference: Built and Measured. Hosted run39134962997 passed thirteen native and eleven Python tests and the sealed four-run campaign. Parent reran the downloaded hash-verified native executable, obtained byte-identical raw observations and independently reconstructed its storage, rates, forces, source updates, native vertex motion, work and ledgers. Distinct semantic review found no remaining scientific/implementation blocker; final bound review/exact-head CI and custody readbacks follow. Entry bc44dbac88b0754ae53f030d0852f462468ce3af. Thirteen prior local checkouts, PR44–53, 2208 scoped Atlas files and original dirty source are protected. No merge, deployment or automatic successor.
Eight training and six held-out probes recover designed K=2, lambda=.75 and theta=1 to floating-point precision. Held-out force error8.33e-17 and rate log-ratio error7.22e-16. This is ideal designed-model recovery, with no independent empirical parameter data or uncertainty estimate.
Actual native mesh mechanics accepted5000 joint steps across coarse/fine coupled runs and zero-lambda/zero-chemistry controls. At horizon2, independent RK4 endpoint errors are6.19015e-5/.0000309699 for dt.002/.001; signed balance defects .00140354/.000700588 decrease with refinement and remain below the unchanged .01 budget. Respective cumulative absolute defects equal those signed values in this campaign; no universal continuum/absolute-error guarantee is claimed.
Fine native receiver peak energy.000339318 and net coupling work.000671227; both controls stay below6.17e-33 receiver energy. Source drift remains below2.23e-15; instantaneous identity error below8.89e-16; exact next-step restart replay and whole-state rejected-trial rollback pass. No calorimetric heat or biological effect is inferred.
The first hosted attempt39134579699 failed four native tests; its logs/artifact remain preserved. Fixes enable feature-scoped exact float restart parsing and correct test-fixture assumptions, preserving sealed equations, runtime guards and numerical budgets.
The explicit two-state coarse-closure witness gives a source energy gap.427362191532 and active-rate gap.045 at identical mapped native state; any energy function of that state alone incurs at least.213681095766 maximum error over those states. An exact broad constant-epsilon_A/coarse cubic mapping is inadmissible there. Retain full-state source ownership or separately define a bounded local approximation before wider integration. Native organism transport, dilution, reaction replacement, growth/remesh/fission, physical scales and independent empirical calibration remain unresolved. Four production values remain null. No merge or automatic successor.
Local evidence: /home/sketch/Projects/digital_cell-calibration-native-evidence/. Additive Atlas group: evidence/calibration-native-qualification/. Preservation recheck confirms13 prior local checkouts, PR44–53,2208 prior archive files and original dirty source unchanged. Historical global governance diagnostics remain exactly2625 lines with no additions; scoped records are checked separately.
Final commit 59edb9ca2b1730d870f7a7f238f05718b97347ac; exact-head CI39135582161 PASS, draft PR54 remains open/unmerged. Artifact12139483753 ZIP SHA256 63b036a294467e57c1861a7460ad77b895a7fa3ebf3fb594e638ce54718d7403 was downloaded and independently checked against3431 exact repository bindings,37 evidence bindings and27 bundled source/record files. Native executable and raw-observation hashes are verified; final observations are byte-identical to the parent rerun. Final workflow enforces all14 independent review bindings, exact opt-in feature/export changes, frozen historical science, append-only ledgers, scoped structured records and stored-versus-regenerated measurements. Parent independently requalified the final artifact.