साक्षात्-तुलनाIndependent residual reproduction
sākṣāt-tulanā
Run the published bench against your own DE441 build. We publish the 1,000-date × 9-graha test set; you publish your residuals. If they reproduce ours within noise, we cite each other.
peer-review surface · cc by 4.0 · agpl · refutation invited
Bharat Ephemeris is a research artifact with a seeker-facing surface — not the reverse. Everything is reviewable: the substrate paper and reproducible proof scripts in the published Zenodo deposit, quarterly residual reports, machine-checked propositions. Scholars are invited to refute us.
Published: DOI 10.5281/zenodo.20024273 (v3 · 230 propositions · 83/83 machine-checked · 2026-05-04). Current internal catalog: 890 propositions · 805/805 machine-checked — APEX v5/v6, publication pending. The internal numbers are not yet covered by the cited DOI and are stated here only as work-in-progress.
Two separate claims, two separate proofs. The ephemeris accuracy claim — Sun 0.13′ p95 / 0.21′ max, Moon 0.51′ p95 / 1.0′ max, all nine grahas ≤ 3.3′ max across 1,000 samples spanning 2000–2050 — is an empirical measurement of Sūrya-Siddhānta + Kerala-school + dṛk computation against NASA/JPL Horizons, reproducible via scripts/audit/. The (R, g, k) substrate is a separate algebraic research program in number theory with its own DOI. Planetary longitudes do not derive from modular arithmetic, and this site does not claim they do.
| artifact | detail | tier | access |
|---|---|---|---|
| साधार-दत्तम्(R, g, k) algebraic substrate — published record | DOI 10.5281/zenodo.20024273 (v3 · 230 propositions · 83/83 machine-checked · 2026-05-04) | Tier-S | doi.org → |
| मूल-लिपिःProof scripts — bundled in the same Zenodo deposit | Reproducible proof scripts archived inside DOI 10.5281/zenodo.20024273 · CC BY 4.0 | Tier-S | doi.org → |
| तुलना-दर्शनम्Accuracy audit — Bharat vs NASA/JPL Horizons | 1,000-date × 9-graha bake-off · 50-year span · Tier-S/W per graha | Tier-S | /accuracy → |
| तुलनम्Compare table — 38 platforms · 4 ground-truth observatories | Structural rigour vs prediction-shaped · feature parity audit | Tier-W | /compare → |
sākṣāt-tulanā
Run the published bench against your own DE441 build. We publish the 1,000-date × 9-graha test set; you publish your residuals. If they reproduce ours within noise, we cite each other.
grantha-samīkṣā
If you publish on BPHS, Sārāvalī, Phaladīpikā, Jaimini-Sūtra and your edition meets the citation discipline (ISBN + translator + page), we link readers from /library to your publication; you cite the (R, g, k) substrate.
daśā-prastāva
We currently route 12 daśā systems. If your paramparā computes a 13th — a regional Vimśottarī variant, for instance — submit a PR or paper. We bench it, route it, and credit you on the new /dasha/[system] page.
megha-gaṇanam
Joint sprint to take Candra from the current ~1 arc-minute (p95 0.51′) vs NASA/JPL Horizons toward arc-second precision — by extending the indigenous Kerala-school series (additional Mādhava / Tantrasaṅgraha periodic terms: evection, variation, annual equation), 100% Bhāratīya, zero foreign runtime. Co-publish, named co-authorship, open-source release.
Send your CV or publication list and the track you want to enter. The verified-academic API tier is ₹0, permanently.
(R, g, k) = (ℤ/3³³⁷ℤ, 2, k=337) · published: DOI 10.5281/zenodo.20024273 (v3 · 230 propositions · 83/83 machine-checked) · internal catalog: 890 propositions · 805/805 — APEX v5/v6, publication pending