Observational Data
The datasets consumed by the computational program of the model, with their status in the repository's data layer and the contrast in which each one is used.
The data layer of the repository records each dataset with its provenance, its pinned version, and its SHA-256 checksum, and declares a status: available (ingested and verifiable) or not available (the contrast that needs it stays frozen, with no verdict). The program's contract is explicit: ingested data is not the same as a validated likelihood, and the latter is not the same as a validated model.
Source: data/sources.yaml and the manifests in data/manifests/ of the repository (30 September 2026).
Data in Use
| Dataset | What it contains | Where it is used |
|---|---|---|
| Cosmic chronometers | 31 H(z) measurements by the differential age method, compiled with their references | Production fits v1 and v2 |
| BAO BOSS/eBOSS | 6 measurements of baryon acoustic oscillations | Production fits v1 and v2 |
| Pantheon+ | 1371 Type Ia supernovae with full covariance; the absolute magnitude MB is marginalized over, with no Cepheid anchor (Scolnic et al. 2022; Brout et al. 2022) | Production fits v1 and v2 |
| Planck 2018, compressed geometry | Three numbers (R, lA, ωb) following Chen, Huang, and Wang 2019; declared caveats: Hu–Sugiyama and diagonal matrix | Production fit v2 |
| fσ8 | 11 measurements of redshift-space distortions | Fit v2 and out-of-sample prediction of the linear sector |
| DESI DR2 BAO | 13 components (DH/rd, DM/rd, DV/rd) of the public likelihood of CobayaSampler, pinned commit (DESI Collaboration 2025) | Background contrast (front 6A) |
| DES-SN5YR (Dovekie) | Hubble diagram of 1820 Type Ia supernovae and the official ΛCDM chain as a reference | Validation of the pipeline by mocks and first application to real data (indeterminate) |
| Union3 / UNITY | Compilation of supernovae and their μ matrices (Rubin et al. 2023) | Available in the data layer |
| Primordial abundances | YP (EMPRESS XV; control Aver et al. 2021), D/H (Cooke et al. 2018), ωb prior (Planck 2018), and τn (PDG 2023): five published values, transcribed by hand | Residues contrast via nucleosynthesis (preregistered) |
| Solar neighborhood | Local dynamical and baryonic densities and Σ(<1.1 kpc) (Holmberg and Flynn 2000; McKee et al. 2015; Bovy and Rix 2013), transcribed | Oort–Kz bound on the Cronos amplitude (front 5) |
| PPN bounds | Preferred-frame bounds (Müller et al. 2008; Shao and Wex 2012; Nordtvedt 1987; Shao et al. 2013), transcribed | Bounding αa of the Atlas sector (front 3) |
Values transcribed by hand carry a provenance notice: their official bytes have not been ingested, and the contrasts that use them are declared provisional for that reason.
Data Not Available
| Dataset | What it would be needed for | Status of the contrast |
|---|---|---|
| SPARC (Lelli, McGaugh, and Schombert 2016) | Rotation curves for the 5E cross-falsification with the Cronos amplitude frozen | Ingestion blocked from the execution environment; prediction frozen, no verdict |
| Walker, Mateo, and Olszewski 2009 | Individual stellar data for the velocity dispersion profile of Sculptor | Frozen preregistration, not run |
| Official DESI DR2 chains | Compare the in-house ΛCDM with the official fit | Audit pending |
| DES photometric mocks | Validate the DES stages prior to the Hubble diagram | Outside the current scope of the pipeline |
What the Current Contrasts Do Not Include
No fit of the program yet uses the full CMB likelihood (Cℓ spectra from Planck, ACT, or SPT) or weak gravitational lensing. The H0 and σ8 figures of fit v2—68.19 ± 0.40 and 0.799 ± 0.029—are obtained with the compressed geometry and the 11 fσ8 measurements, with the caveats declared in the fit itself. The current result, with ΛCDM favored, is in MCMC versus ΛCDM.