Substratum

A curated, schema-validated catalog of open tensions and anomalies in contemporary physics, spanning particle physics, cosmology, gravitation, and condensed matter.

The repository has two parallel layers:

• data/ — a record of where the Standard Model and ΛCDM are currently under empirical stress. Each entry tracks the discrepancy's history, current significance, primary references, and status (open, contested, resolved, theoretical).
• benchmarks/ — the non-negotiable constraint set that any candidate replacement framework must satisfy before it is worth taking seriously. Reductions to known theories, foundational-principle tests, forbidden phenomena, and a meta-entry pointing back into data/ for the open tensions any new framework has to explain.

Together: tensions tell you where the current models break; benchmarks tell you what the next model has to look like.

Why

Tension trackers exist inside individual subfields (PDG for particle physics, review articles for cosmology), but no single, structured, cross-domain dataset does this in a way that survives time. Significances drift. The 2022 W-mass anomaly was 7σ at CDF and is now consistent with the SM after ATLAS and CMS measurements. The muon g-2 tension was ~4σ in 2021 and has shrunk substantially as lattice QCD evaluations of the hadronic vacuum polarization have matured. A useful atlas tracks that motion, not a single snapshot.

What this repo is not

• Not a unified theory.
• Not a "translation layer" that maps unrelated phenomena into shared jargon. Such mappings, without a derivation, are numerology.
• Not a claim that the Standard Model or General Relativity are wrong. They are extraordinarily well-tested effective theories. Tensions are where the next physics might live, or where systematics are still being understood.

Scope of an entry

Each anomaly is a JSON file under data/ validated against schema/anomaly.schema.json. Required fields:

• observable — what is measured (e.g. a_mu = (g-2)/2)
• characteristic_scale — approximate length and energy scale
• standard_prediction — value, uncertainty, model, references
• measurements — array of independent experimental results
• tension_history — σ over time, so the record reflects how the picture has actually evolved
• status — open, contested, resolved, or theoretical
• references — primary literature, DOIs where available

Optional structured hooks (filled per domain when meaningful):

• trend — growing / stable / shrinking / resolved / theoretical. Orthogonal to status.
• principles_stressed — which foundational principles (unitarity, causality, Lorentz, CPT, equivalence, lepton-flavor universality, …) the anomaly would stress if it survives systematics. Most entries explicitly record principle: none, status: none — that the anomaly does not violate a foundational principle is itself information.
• smeft_context — for particle entries, the Warsaw-basis SMEFT operator(s) most directly probed plus references to standard global fits (Fitmaker, SMEFiT, HEPfit). No invented dictionaries; we use the BSM community's working framework.
• cosmology_context — for cosmology / astrophysics entries, the extension classes that move the needle (early dark energy, interacting dark sectors, modified recombination, …) and the cost elsewhere of each (which other tension or constraint they worsen).

See schema/anomaly.schema.json for the full schema and CONTRIBUTING.md for how to add or update entries.

Current entries

data/ — open tensions

ID	Domain	Status	Trend	Latest σ
muon_g_minus_2	particle	contested	shrinking	~1–4 (depends on SM input)
hubble_tension	cosmology	open	stable	~5
w_boson_mass	particle	contested	shrinking	CDF outlier; world avg consistent
s8_tension	cosmology	open	stable	~2–3
cosmological_constant	theoretical	theoretical	—	naturalness, not σ

benchmarks/ — what any new framework must satisfy

Each entry includes a procedural block: the canonical parameterization (PPN, SMEFT, SME, ΛCDM extensions, Yukawa…), the predicted observable, the bound restated in those parameters, what a candidate framework must expose to be evaluated, and an evaluator_status flag indicating how hard the evaluation is (trivial / tractable / requires_code / requires_simulation / requires_lattice / research_problem).

53 benchmark entries across 12 domains. Every entry is backed by primary literature (arXiv / DOI). See ROADMAP.md for the full populated list and remaining additions.

Domain coverage:

Domain	Entries	Canonical parameterizations
Gravity (PPN, GW, EP)	12	PPN, PPK, TIGER, Yukawa, SME-grav
Particle precision (QED)	4	SMEFT (Warsaw), bound-state QED
Flavor and CP violation	10	WET, CKM, SMEFT CP-odd
Weak / EW precision	3	SMEFT, Peskin-Takeuchi S/T/U
Lorentz / CPT	4	SME (photon, meson, lepton sectors)
Cosmology	7	ΛCDM + extensions, Boltzmann codes
Dark matter	5	DM-EFT, SIDM, axion couplings
QCD	3	Lattice QCD, χPT
Astrophysics	3	TOV, helioseismic inversion
Constants / clocks	1	dilaton d_e, ULDM
Meta	1	links to data/

The procedural block makes each constraint executable in principle. For trivial and tractable cases (graviton speed, EP, m_W, S/T/U, PPN parameters, axion-photon coupling), evaluation against a candidate framework is a one-liner; for requires_simulation (CMB, BBN, LSS, Lyman-α, helioseismology), the framework needs to be implemented as a module in CLASS / CAMB / PArthENoPE / GARSTEC.

Roadmap

1. Data layer (this milestone). Schema, ~5–10 high-quality seed entries, validation tooling.
2. Coverage. Extend to: KOTO/NA62 rare kaon decays, B-anomalies (R(K), R(D*)), atomic parity violation, neutron lifetime, ANITA anomalous events, lithium problem, CMB cold spot, axis of evil, superconductor tensions.
3. Visualization. Plot σ-vs-time per entry; cross-domain landscape on (length_scale, energy_scale).
4. Optional, clearly speculative. A separate, explicitly-flagged notes/ area where reformulation ideas (geometric algebra, tensor networks, modified gravity) can be discussed without being mixed into the rigorous data layer.

Validation

python scripts/validate.py

Validates every file in data/ against the schema and reports problems.

Visualization site

A static visualization (σ-vs-time per tension, length–energy landscape, and a filterable catalog of every entry) is published to GitHub Pages from the docs/ directory. The data layer for the site is built from data/ and benchmarks/ by:

python scripts/build_site.py    # writes docs/data.json

To preview locally:

python scripts/build_site.py
python -m http.server -d docs 8000
# open http://localhost:8000

docs/data.json is generated and excluded from the repo. The .github/workflows/pages.yml workflow rebuilds and redeploys the site on every push to main.