scripts

Analysis Scripts — Investigation & Verification

Python (and a few JavaScript) scripts for investigating, verifying, and reproducing the results of the Expanding Solar System Resonance Theory (ESSRT): the climate formula and its tests, the ΔT/LOD stack, the eclipse data pipeline, the deep-time cross-check and the browser-modal exports. Every script imports the model through the §2f bridge below, and npm run test:py-smoke imports each one so the collection cannot rot unseen.

§2f — analysis only, and it is enforced. Python here may read the model: import from tools/lib/python/constants_scripts.py, which loads public/input/*.json. It must never define model physics. With JS and Python side by side there is no compiler keeping the two in step, so the boundary is checked rather than trusted: tools/check-python-physics.mjs runs in npm run check and in CI.

A frozen analysis may pin the value it ran at — that is provenance, not a fork — but the pin goes on the gate's ledger, and the gate then asserts the pin still matches the live model. Nine scripts currently pin the anchor; the day it is recalibrated they all go red at once, which is the intended behaviour. New work should import from constants_scripts rather than add a ledger entry.

Shared library (constants_scripts.py) lives in tools/lib/python/. All scripts here load it via sys.path at startup.

Finished investigations leave this folder for scripts/archive/ (gitignored; git history carries the files) together with their result files (data/archive/). A script stays here when a gate runs it, a governed artifact names it as generator, a live document cites it, or another script imports it.

---

Quick Start

# The canonical climate formula fit — the governed generator of
# data/milankovitch-climate-formula.json
python3 milankovitch_climate_formula.py

---

Governed generators

These write artifacts that npm run check re-checks against their recorded input hashes (check:artifacts) or renders into the docs (docs:tables).

Script	Produces
milankovitch_climate_formula.py	data/milankovitch-climate-formula.json — the canonical L1 + L2 + L3 climate formula, fit per regime with sequential ridge regression; the Earth Climate Analysis modal's coefficients
l1_physical_lines.py	data/l1-physical-lines.json — the one home of the climate formula's orbital line list (the beats of the engine's secular modes and its precession rates, plus the 405.6-kyr family). Two rates, two families: the climatic-precession lines p + gᵢ carry the general precession in longitude at J2000; the obliquity lines p + sᵢ carry the pole rate's mean over the LR04 window on the orbital forcing alone (v23.0 — a record sees the window mean, not the J2000 instant; the T1 exporter tools/explore/t1-export-engine-frequencies.mjs writes both)
t7_fixed_phase_l1.py	data/t7-fixed-phase-l1.json, data/t7-model-orbital-histories.json — pre-registered test T7: the fixed-phase L1
generate_doc97_tables.py	the generated table blocks of doc 92 (npm run docs:tables; --check in the gate chain)

---

The climate formula — tests and diagnostics (doc 92, doc 94)

Script	Description
fit_methodology_diagnostics.py	Diagnostics on the L1 design matrix of the shipped lines: conditioning, Rayleigh groups, VIF, forward-projection growth, the OLS / ridge / thinned ladder (doc 92 §9.5; data/fit-methodology-diagnostics.json)
milankovitch_insolation_extension.py	Does adding Berger insolation features to the formula buy cross-window-stable gain? (doc 94)
milankovitch_insolation_laskar_check.py	The same test hardened on La2004/La2010 orbital features (doc 94)
milankovitch_insolation_stability.py	Cross-window stability of the insolation extension (doc 94)
extract_insolation_features.js	Extracts the Berger insolation features at the LR04 sample times (data/insolation-features.csv)
extract_insolation_features_deep.js	The deep-source feature set — hybrid obliquity, deep orbital history, the physical climatic precession of date (data/insolation-features-deep.csv)
milankovitch_l1_line_audit.py	Every shipped line on the same footing, per regime (doc 94 §9.1)
milankovitch_ecc_period_scan.py	Which period the climate record prefers in the ~100-kyr band, against the shipped lines (doc 94 §9.2)
climate_formula_mwp_check.py	Does the formula coincidentally place a peak at the Medieval Warm Period?

The comb-era research scripts are gone. The variance-budget decomposition that led to the physical line set, the Laskar-eigenmode and invariant-manifold comparisons, the cross-domain and paleo comb tests and the pre-registered comb tests T1 and T5 were deleted with their result files at plan 09 step 7e — git history carries them, and doc 109 §22 records each script's headline numbers. What stays: nbody_50myr_backward.py (the backward N-body ground truth; needs rebound, see data/PROVENANCE.md) and test_evolving_clock_climate_formula.py (the p-carrying lines on the composed clock of date, with the warp-scale, sign and frequency-offset controls — doc 99).

---

Climate sensitivity (ECS) — record

Charney-ECS decomposition across paleoclimate eras on the L1 lines, cross-validated on LR04, EPICA, Snyder GAST and boron-isotope CO₂ reconstructions. The write-up is archived; the datasets and their rows stay in data/PROVENANCE.md.

climate_ecs_tight.py (the entry point: frequency-dependent ice fraction, regime-conditional kernels), climate_ecs_cross_proxy.py, climate_ecs_boron.py, climate_ecs_full_forcing.py, climate_ecs_monte_carlo.py, climate_ecs_per_regime.py, climate_ecs_snyder.py, climate_ecs_phase_lag.py.

---

The ΔT / LOD stack (doc 102, doc 104, doc 105)

Script	Description
stephenson_observation_density.py	Observation-density analysis across the Stephenson 2016 tables — which centuries are well observed
parse_stephenson_deltaT_polynomial.py	Parses the Stephenson 2016 piecewise ΔT polynomial → public/input/stephenson-2016-deltaT-polynomial.json

---

Eclipse data pipeline (doc 102, doc 103)

Script	Description
fetch_nasa_lunar_canon.py	Scrapes NASA's 5-Millennium Canon of Lunar Eclipses (12,064 events) → public/input/lunar-eclipses-nasa.json
fetch_nasa_historical_lunar.py	Parses NASA's "Lunar Eclipses of Historical Interest"
parse_stephenson_lunar.py	Parses the Stephenson, Morrison & Hohenkerk 2016 timed-lunar tables → public/input/lunar-eclipses-stephenson-2016.json
parse_stephenson_solar.py	Parses the Stephenson 2016 timed-solar tables → public/input/solar-eclipses-stephenson-2016.json

---

Browser-modal data exports

One-shot utilities that prepare data for the in-app modals. Outputs are committed under public/input/; no re-runs expected unless source data updates.

Script	Description
export_climate_formula_browser.py	Climate-formula coefficients → public/input/climate-formula-data.json (the Earth Climate Analysis modal)
export_cenogrid_browser.py	Westerhold 2020 CENOGRID (δ¹⁸O + δ¹³C, 67 Myr) → public/input/cenogrid-data.json
export_cenco2pip_browser.py	CenCO2PIP atmospheric CO₂ proxy → public/input/cenco2pip-data.json
export_epica_browser.py	EPICA Dome C CO₂ (Bereiter 2015) → public/input/epica-co2-data.json
process_climate_proxy.js	The LR04 climate proxy → public/input/climate-proxy.json

---

Utilities

(test_phase0_inline.js, the deep-time chain's cumulative-integral validator against its own mirrored copies, left with the era device's counters at plan 09 step 5c.)

---

Data Files

File	Description
../data/lr04-stack.txt	LR04 benthic δ¹⁸O stack (Lisiecki & Raymo 2005, Paleoceanography 20, PA1003) — 5.3 Myr orbitally-tuned marine climate record.
../data/cheng2016-speleothem.txt	Cheng 2016 U-Th-dated Asian Monsoon speleothem record (Science 352, 343) — 640-kyr non-tuned chronology bias control.
../data/epica-co2-bereiter2015.txt	Bereiter et al. 2015 (GRL 42, 542) EPICA Dome C composite atmospheric CO₂ record — 0–800 kyr BP, Antarctic ice cores.
../data/westerhold2020-cenogrid.tab	Westerhold et al. 2020 (Science 369, 1383) CENOGRID — 67-Myr astronomically tuned benthic δ¹⁸O+δ¹³C reference splice.

Every dataset's source, citation and licence is in data/PROVENANCE.md; two datasets are not ours to redistribute and must be fetched separately (the download steps are there).

---

Dependencies

• Python 3.8+
• numpy, scipy — numerical computations
• pandas — data manipulation
• openpyxl — Excel file reading
• astropy — where a script converts calendar dates

pip install -r requirements.txt in the repository root installs them.

---

Related Resources

• Interactive 3D Simulation
• Model Documentation
• Fitting Pipeline

---

License

These scripts are part of the Interactive 3D Solar System Simulation project and are released under the GNU Affero General Public License v3.0.