# Executed continuation: experiments beyond prerequisite tracking

Eight new calculation stages passed **293 scoped checks**. They implement ordinary nuclear calibration with reserved predictions, an empirically pole-matched charged continuum, joint isotope/chronology inference, interacting matter with leptons, coupled stability and boundary tests, reciprocal formation dynamics, explicit saved-shell release and transport, and thermal/gravitational response. [Seven figures with data and provenance](/research/metastable/full/experiments/index.html).

The main new positive result is a **localized, radially stable candidate within the specified cold cosine benchmark**, accessible from the tested imposed seeds. A density-normalized seed instead remains too small as its supporting cloud expands. The main material result is that **ordinary release/enrichment alternatives absorb much of the tested contrast**. Direct decay-calibration choices also shift the isotope inference. No result establishes an observed change of constants or a fully matched field-to-stellar-ejecta history. Numerical execution, conditional model results and physical acceptance remain distinct in the [program](research-program.md).

## P1: calibrate ordinary inputs, then reserve predictions

The journal version of the 2020 nuclear source retains the regulator expression already audited; it does not supply a missing dimensionful convention. We therefore implement an explicitly dimensionless short-range Yukawa coefficient as a separate calibrated convention. For both potential sets and three regulator cutoffs, it reproduces the declared −2.224 MeV ordinary binding. Scattering lengths, effective ranges and radii are then predictions, not additional fitting targets. The reserved singlet channel instead has a bound level between -0.728 and -0.475 MeV, failing the source’s unbound-singlet baseline. A deuteron-only refit therefore does not rescue the full interaction. Field-dependent contacts are varied independently; at theta=1 the models span -4.466 to -0.726 MeV in the lowest triplet box energy. [Primary journal source](https://doi.org/10.1103/PhysRevResearch.2.033392).

One E1 capture counterterm, 268830.329 MeV³, is calibrated at three energies. Seven other energies predict the author’s curve with a maximum fractional discrepancy of 0.525%, below the approximately 2.2% drawing-width comparison scale. The exact source coefficients and field-dependent threshold/contact terms are still separate questions; the source curve itself is not independent experimental data. This is a quantitative improvement over leaving the baseline mismatch entirely untested. [Capture source](https://arxiv.org/html/2601.11180v2).

## P2: an actual charged near-threshold model

A separate Coulomb-resummed halo EFT matches the empirical Be-8 pole at 91.84 keV and width 5.57 eV. Complex-pole matching, independent real-axis evaluation, elastic unitarity and the bound-state continuation agree. The calibrated inverse-length coefficient is 0.000502738 fm⁻¹. A contact shift of -0.000502738 fm⁻¹ reaches threshold. Which alpha displacement produces that contact shift is not fixed by the ordinary pole. The zero and opposite-sign derivative scenarios demonstrate the resulting ambiguity. Width/shape corners are sensitivity cases with no invented covariance. [Halo EFT](https://arxiv.org/abs/0802.3426v2), [pole reference](https://academic.oup.com/ptep/article/2021/11/113D02/6380952).

Greedy interpolation is checked on 64 disjoint direct EFT evaluations. Maximum reserved error falls from 2493.0 eV with three points to 17.0 eV with fifteen. This measures interpolation error in the defined EFT. It does not reproduce the unavailable paired NLEFT matrices, estimate their correlated theory error, or determine three-alpha/Hoyle physics.

An additional numerical audit found a maximum 0.547 keV difference between a real-axis resonance condition and the actual complex pole across the scan. The selected run solves the complex outgoing pole above 10 keV. Below that scale, Coulomb widths are negligible at the declared precision and are retained logarithmically. The previous real-axis scan is preserved as an earlier approximation, not silently promoted to exact pole data.

## P3: measured chronology now enters the likelihood

The published Re/Os analyses and individual corrected W observations enter a joint likelihood. Duplicate/sample offsets and unknown Re error correlations remain explicit. The four selected W observations give weighted mu182W=-308.315 ppm and a reference segregation delay of 3.796 Myr. The two W normalizations are not counted as independent analyses. Shared initial/chondritic W reference uncertainties are propagated; the chondritic precursor assumption remains conditional. [Re/Os measurements](https://pmc.ncbi.nlm.nih.gov/articles/PMC6457471/), [W/Os chronology](https://pmc.ncbi.nlm.nih.gov/articles/PMC6352993/).

The compressed 4.56730 Gyr Pb/Pb anchor is recalculated with **both** affected uranium decays. Its published ±0.00016 Gyr uncertainty is tested explicitly. The direct Re half-life prior is the **historical** 42.3±1.3 Gyr measurement at 95% confidence, converted to one sigma; it is not presented as current metrology. A separate W/Pt analysis gives 6.0±0.8 Myr and is tested as an alternative chronology, rather than multiplied into the W/Os fit as independent evidence. [Pb/Pb anchor](https://doi.org/10.1126/science.1226919), [historical direct Re calibration](https://doi.org/10.1016/0016-7037(89)90241-X), [W/Pt alternative](https://arxiv.org/abs/1802.10421).

The university-hosted dissertation supplies the study’s Appendix A2, including sample identifiers and duplicate-composition comparisons. Inspection did not identify a per-analysis Re/Os covariance matrix. These comparisons describe chemical heterogeneity and must not be invented into a new independent measurement-error covariance. [Dissertation](https://api.drum.lib.umd.edu/server/api/core/bitstreams/549ca320-2bbd-4bf5-8f13-110d56c78c70/content).

| Excursion center (Gyr ago) | Duration (Gyr) | Reference conditional 95% grid interval (ppm) |
|---:|---:|---:|
| 2 | 0.1 | [-63.00, 33.00] |
| 2 | 1 | [-9.00, 2.50] |
| 4.4 | 0.1 | [-63.00, 33.00] |

These intervals profile laboratory calibration, W chronology and Re sample scatter under each declared covariance/exchange scenario. They use the asymptotic one-parameter profile threshold and finite grids. The 240 null/injection trials give lab-dominated nominal-95% recovery fractions 0.925 and 0.933; they validate the stated simulation, not unknown real covariance or all geochemical misspecification. The 0.1 Gyr histories at 2 and 4.4 Gyr have the same reference interval: chronology has not magically supplied temporal resolution inside a common integration window. Heavy-nucleus alpha response is not relabeled as theta response.

Two later direct calorimetric measurements are now profiled separately: 41.2±0.2(stat)±1.1(syst) Gyr and 43.2±0.2(stat)±0.1(syst) Gyr. Treating each quoted error component as an independent Gaussian one-sigma term is an explicit conditional choice. They are not multiplied into a consensus calibration. The table below keeps the same 1 Gyr excursion, measured records and reference nuisance scenario. [Galeazzi et al.](https://doi.org/10.1103/PhysRevC.63.014302), [Arnaboldi et al.](https://arxiv.org/abs/hep-ex/0302006).

| Laboratory calibration | Best excursion (ppm) | Conditional 95% interval (ppm) | Twice delta NLL at zero |
|---|---:|---:|---:|
| Lindner-1989 | -3.528 | [-9.228, 2.678] | 1.290 |
| Galeazzi-2001 | 1.657 | [-8.335, 13.408] | 0.093 |
| Arnaboldi-2003 | -7.499 | [-9.450, -5.496] | 49.530 |

The inferred displacement depends materially on the laboratory calibration. A nonzero optimum under one calibration is therefore not an observed changing-constants signal. Laboratory systematics, chemical-environment effects and a justified shared calibration model must be addressed before promoting these intervals into a robust empirical constraint.

## P4: interacting static solutions, stability, formation and removal

The benchmark fits ordinary symmetric-matter saturation at n0=0.16 fm⁻³, binding −16 MeV, incompressibility 240 MeV, symmetry energy 32 MeV and slope 60 MeV. One energy functional supplies beta-equilibrated neutrons, protons, electrons and muons, chemical potentials, pressure and scalar source. The first electron-only audit exposed muon onset; the selected runs include that additional species and common lepton chemical potential. These anchors define a phenomenological functional, not a matched theta-dependent chiral Hamiltonian. [Nuclear-EOS context](https://arxiv.org/abs/2004.07805).

Local densities and fields vary independently at fixed baryon number, with f_a=10¹⁶ GeV and weak shell gravity. The QCD solution loses its field displacement and its 90%-mass radius grows from 9.602 to 14.397 km when the box grows from 10 to 15 km: it is not a localized object. The separate cosine candidate has radius 1.784 km, excess energy -67.412 MeV per baryon and central angle 3.124234. Its radius changes by 0.461% in the larger-box/same-spacing comparison. Its 160-to-240-cell energy difference is 0.00954 MeV per baryon. No negative mode is found in the stated fixed-number coupled **radial energy Hessian**. This is a static variational test, not a relativistic frequency spectrum or proof against all perturbations. [Finite-density model context](https://arxiv.org/abs/1904.00020), [density-induced instability](https://arxiv.org/abs/2105.13354).

The same cold functional now drives moving spherical mass packets and the field with reciprocal forces. Specified dense initial matter and a 0.001 seed produce half the full central scalar-mass contrast after 10.174 microseconds in the refined cosine run. Maximum speed is 0.080 c; the response-scaled energy error is 6.1e-06. A smaller seed delays growth, the QCD comparison stays close to its ordinary phase, and an exact zero seed remains at the symmetry point. This computes local access from those initial conditions. It does not predict an astrophysical supply of such clouds. The fixed smoothing kernel, cold closure, fixed ordinary inertial mass and missing shock heating delimit the model.

The seed amplitude is now tested against a declared reference all-axion density of 0.4 GeV/cm³. That energy density gives a coherent angle amplitude of 5.45e-19. Linear evolution on the moving ordinary-matter trajectory reaches only 4.32e-11 radians in the original window. Extending the calculation to 80.06 microseconds, with 480/720 cells and 32/48 matter packets in a 45 km box, gives a maximum 2.35e-09 radians in the finer calculation. Meanwhile the cloud expands to maximum density 0.0183 fm⁻³. The two gains differ by 2.46%. The angle remains well inside the linear regime, where matter feedback is quadratic and negligible. Thus **this density-normalized seed does not form the candidate within the tested evolving cloud**, despite strong initial amplification. A constant-density growth-time estimate cannot replace the calculated expansion. This is a bounded negative formation result for the specified cloud, not an exclusion of all progenitors.

The uniform coherent seed starts at rest, at its maximum field displacement. Replacing packet kinetic energy and drift by their special-relativistic forms changes the maximum linear gain by 1.15% and gives a peak angle of 2.37e-09. Its maximum packet speed is 0.210 c. This quantifies one kinematic correction while retaining the same instantaneous cold functional and weak gravity; it is not a covariant relativistic fluid solution.

After an explicitly imposed removal of supporting matter, the central **periodic mass contrast first crosses below half** at 4.251 microseconds. It subsequently oscillates and repeatedly returns near its full contrast. The 1.51% value at the chosen 20-microsecond endpoint is a phase sample, not a decay envelope or lifetime. Energy error in that unsupported evolution is 1e-05. An unwrapped angle near 2π is correctly treated as nearly ordinary matter coupling. This intervention is not a supernova, and this finite calculation establishes neither eventual erasure nor persistence during stellar burning. Matching the interaction’s field response, quantum corrections, a real formation history and applicable observations remains necessary for astrophysical viability.

## P5: release, transport, reserved observables and cloud response

The calculation now uses actual shell masses, radii, internal energies and compositions in eight saved MESA tracks. It executes 64 release cases: energy budgets from zero to 10⁵¹ erg, a 1.4-solar-mass retained floor, and zero/full availability of saved internal energy. Exterior material is removed contiguously and partial shells pay their proportional binding cost. Every species closes between released and retained reservoirs. These are **helium-exhaustion snapshot interventions**: explosive burning, shocks, fallback and final remnant identity are not invented.

Winds and released material feed 32 conservative, delayed two-reservoir histories. A halved timestep and a three-Myr release delay are tested. For six altered-history targets, three ordinary stellar families independently fit release energy and star-formation burst fraction to C/O and N/O. Selection uses those two channels only. Ne/O and Mg/O are then genuine reserved predictions.

The largest departure of a cohort’s birth metal fraction from the stellar calibration mixture is 52.2%. Positivity and mass conservation do not prove the accuracy of linear net-yield recycling over that change. Eighteen additional controls interpolate between frozen gross ejecta and the linear birth-composition response, with identical mass accounting. Their complete shifts are saved in `recycling-closure-sensitivity.csv`; this is a closure sensitivity test, not a replacement stellar grid or a measured uncertainty distribution.

| History; release energy | Reserved Ne/O residual (dex) | Reserved Mg/O residual (dex) | Maximum design significance² with stated systematic floor |
|---|---:|---:|---:|
| early; 1e+50 erg | -9.4967e-10 | -9.5427e-10 | 8.2389e-16 |
| late; 1e+50 erg | -9.8085e-10 | -9.8559e-10 | 8.7887e-16 |
| persistent-response; 1e+50 erg | -1.4498e-08 | -1.5306e-08 | 2.027e-13 |
| early; 1e+51 erg | -0.0027466 | -0.0041692 | 0.0134 |
| late; 1e+51 erg | -0.0028236 | -0.0042858 | 0.01416 |
| persistent-response; 1e+51 erg | -0.023314 | -0.05376 | 2.5088 |

The design covariance contains independent measurement/intrinsic scatter and a shared oxygen error, plus explicit systematic floors. None of these six cases reaches its stated 5-sigma target even with unlimited sample size under those floors. This is a conditional design result, not a real-survey exclusion. The floors and conventional-family set are assumptions, and a different material history requires a new test. It nevertheless shows why core sensitivity alone cannot establish a distinguishable delivered signal.

The transported compositions also drive 18 matched-metallicity cloud calculations. Atomic C II/O I cooling, thermal derivatives, contracting wavelengths, pressure and self-gravity evolve together. Runs stop at optical depth one or the linear-perturbation limit. Different terminal amplitudes can occur at different stop densities. No linear growth factor or Jeans mass is presented as a fragment count or stellar IMF. Molecular chemistry, dust, radiation transfer and nonlinear turbulent fragmentation are outside this experiment; the preceding reserved-pattern result does not justify claiming their outcome.

## Execution, acquisition and remaining scope

The new suite is executable rather than a list of prerequisites:

```sh
uv run --offline --frozen python run.py full experiment-suite
uv run --offline --frozen python run.py full experiment-report
uv run --offline --frozen python run.py smoke
```

The suite runs the eight new stages sequentially. The report command uses frozen results and also refreshes prior-batch reports. The advanced MESA continuation is a separate long calculation (`full E04-advanced`), not hidden in the suite. Raw sources, failed runs, output arrays, implementation archives and all original full-project acceptance criteria are preserved.

The advanced continuation `20260930T164728219271Z-E04-advanced` has lifecycle status **validated_within_scope** as of 2026-10-08T10:49:38.690940+00:00. Its latest readable history is model 8361. The [timestamped continuation snapshot](advanced-continuation-status.json) is kept separate from frozen completed evidence. The iron endpoint remains unavailable unless that run passes its endpoint check; it supplies no terminal ejecta.


The source packet remains unsent. New acquisitions include the 2020 journal PDF, 2008 halo-EFT source, 2021 alpha-cluster article, 2018 W/Pt paper, isotope dissertation, two later direct half-life sources and a primary nuclear-EOS source. The Galeazzi abstract supplies the stated calibration; its full PDF returned an authorization error and was not acquired. Two purported decay-evaluation PDFs were HTML responses and are not used as decay data; an acquired A=8 thermal-capture table does not supply the Be-8 pole. These distinctions are recorded in [acquisition audit](experiment-acquisition-audit.json).

The completed conditional experiments do not supply the missing paired lattice inputs, an exact missing source convention, unreported laboratory covariance, or quantum/astrophysical validation of the chosen action. The [twenty-package map](research-program.md) now displays executed conditional evidence even where its stronger physical gate remains open. The original 102-task and E15–E23 full-project statuses are not promoted on the strength of numerical checks alone.

| Stage | Frozen run | Checks |
|---|---|---:|
| P1-renormalized | [20260929T024206209567Z-P1-renormalized](runs/20260929T024206209567Z-P1-renormalized/manifest.json) | 25 |
| E18-halo | [20260929T023727021282Z-E18-halo](runs/20260929T023727021282Z-E18-halo/manifest.json) | 34 |
| E20-joint | [20260929T030548291050Z-E20-joint](runs/20260929T030548291050Z-E20-joint/manifest.json) | 14 |
| P4-interacting | [20260929T023344889478Z-P4-interacting](runs/20260929T023344889478Z-P4-interacting/manifest.json) | 30 |
| P4-region-audit | [20260929T024540159269Z-P4-region-audit](runs/20260929T024540159269Z-P4-region-audit/manifest.json) | 20 |
| P4-dynamics | [20260929T031718343243Z-P4-dynamics](runs/20260929T031718343243Z-P4-dynamics/manifest.json) | 29 |
| P5-material | [20260929T024518646066Z-P5-material](runs/20260929T024518646066Z-P5-material/manifest.json) | 121 |
| P5-clouds | [20260929T025855993746Z-P5-clouds](runs/20260929T025855993746Z-P5-clouds/manifest.json) | 20 |