VED-008: Panchanga Shuddhi source research and admission proposal

Date: 8 October 2026. State: research complete for the decisions below. The user subsequently approved the named profiles together; the 9 October implementation standard and validation receipt record their engine/REST admission. Proposal language below preserves the research decision sequence, not current implementation status. Baseline: engine c647363a342a7756d85a5b012ff0e9004ba2ce80, version 6.9.9; generated wiki 2e74b0b6241c8db52a368d89274050a123b795da. Owner: Vedic remaining work register, VED-008. VED-009/010/011 retain general cancellation, strength and complete purpose-election scope. Website adoption is separate.

1. Decision

Proceed to a finite engine/public/REST package with separately named rules for:

1. Panchaka Rahita arithmetic and date-derived inputs, including the full thirty-tithi convention and the ascendant at the evaluated instant.
2. Tara cycle distinctions, with the Chintamani quarter reading and the different P. S. Sastri cycle rule exposed as alternatives.
3. The nine adverse Nitya Yogas, distinguishing whole-event prohibitions, initial ghatis and the first half of Parigha.
4. Karana-specific evidence from Brihat Samhita, and independently reported Bhadra occurrence, residence, day/night exceptions and mouth/tail windows.

There is substantive classical support. The obstacle is choosing and naming the exact reading and clock, not a general absence of texts. Preserve existing tara_bala, chandra_bala and weighted-score behavior; add an inspectable source-owned assessment. Do not silently reinterpret their existing booleans.

The recommended clocks for fractional intervals below are explicitly derived computational policies. Approval of this package would admit those named derivations, not establish that a verse directly specifies a modern algorithm. Conflicting readings and activity exceptions remain visible in results.

2. Sources actually examined

PDF page numbers here are one-based file pages, distinct from printed pages. Source PDFs and page renders remain in private research storage. Only this analysis, factual rule tables and small evidence receipts belong in the repo.

ID	Witness and identity	Reviewed locus	Evidential use
MC	Daivajna Rama, Muhurta Chintamani, Hindi commentary by Pandit Ramlal Avasthi, edited by Chandika Prasad Avasthi; Tejkumar Book Depot, Lucknow, tenth edition 2004. Scan.	Shubhashubha 34–35, 42–45: PDF 25–26, 29–30 / printed 16–17, 20–21. Gochara 11–13: PDF 79–80 / printed 70–71.	Printed Sanskrit, anvaya and commentary collated. Yoga, Bhadra and Tara authority; commentary discrepancy recorded below.
KP	Kalaprakasika, Sanskrit with English translation by N. P. Subramania Iyer; Asian Educational Services, New Delhi, 1982. Scan.	Panchakam in travel: PDF 188–189 / printed 151–152. Exceptions: PDF 230 / printed 193.	Independent printed Panchaka formulation and noon-exception witness. OCR incorrectly reads printed 151/152 as 161/162; images control.
PS	P. S. Sastri, Text Book of Scientific Hindu Astrology, Ranjan Publications; exact printing date not present in reviewed front matter. Local file Books by Authors/BV Raman/Scientific-hindu-astrology.pdf.	PDF 1 title, PDF 5 / printed 8 signed preface; PDF 268 / printed 273, “Constellations — Nature.”	A distinct cycle rule. Author is Sastri, not Raman: verified signed preface. Edition remains identified by scan hash; no date guessed from a bookseller.
BS	Varahamihira, Brihat Samhita, English translation/notes by V. Subrahmanya Sastri and M. Ramakrishna Bhat; V. B. Soobbiah & Sons, Bangalore, 1946. Local Classics books/brihatsamhita.pdf.	Title PDF 1; signed introduction PDF 11; chapter 100, verses 1–5, PDF 767–769 / printed 748–750.	Direct local scanned primary witness for all eleven Karana identities and activity distinctions. Text search fails because these pages are image-only; bookmarks and visual reading located the chapter.
R	B. V. Raman, Muhurtha, online reflowed transcription; printing/edition not established. File.	Chapter III, PDF 13–16.	Corroborates quarter reading and Panchaka; also preserves author-specific caution and an internal ghati/example discrepancy. Not treated as an edition-locked classical text.
MP	Darwin, NT Panchang, 2025; Hindu Council of Australia / myPanchang; calculations credited to Pandit Mahesh Shastri, editor Pandit Vishal Sharma. Published calendar.	PDF 1 acknowledgements; PDF 2, “Panchaka Rahita Vidhi for Muhurtha.”	Domain-primary operational authority explicitly specifying thirty tithis, sunrise-owned weekdays and a worked dark-fortnight case. This convention is attributed to this publication.

Digital corroboration:

• E-Bharatisampat, Chintamani supplies searchable Sanskrit. Its online text is a corroborating witness; the named printed MC scan owns the edition used here.
• Brihat Samhita, Iyer translation chapter 99 combines tithi and Karana material. Its verses 4–8 correspond to Karana 1–5 in the local 1946 chapter 100. Digital Sanskrit labels the Karana chapter 99. Do not cite a bare chapter number across editions.
• Muhurta Sadhana, Samjna 45–63, records Bhadra body divisions and alternative exceptions. Author/print edition is not identified there; it is a variant witness, not the default authority.
• Gochara 13–25 demonstrates that Janma and paksha-conditioned Chandra interpretations vary; it does not authorize changing the existing Chandra function in this package.

Local-corpus search limits

Searched the prior local extraction cache for Panchaka Rahita, paryaya and Tara Bala spellings; checked relevant matches in the Sastri textbook, Uttara Kalamrita and the modern Vedic textbook. The Uttara Kalamrita hit concerns birth-time verification, not this electional rule. The latter textbook's nine-tara table is a general transit baseline, not proof of cycle exceptions. Filename searches for Muhurta/electional/Panchanga titles did not locate a dedicated manual in this corpus; that is not an absence claim about rules embedded in other volumes. The local Raman Manual of Hindu Astrology had no relevant extracted-text match. Image-only pages cannot be cleared by text search: the local Brihat Samhita is the concrete counterexample.

Private audit directory: C:/dev/outputs/ved008-research-2026-10-08/. source_manifest.json records paths, hashes and reviewed pages; page PNGs permit visual rechecking. Supporting PDFs from the previous VED-007 research are reused without pretending they were found inside the local corpus.

ID	SHA-256 of reviewed PDF
MC	9e931eb7fdc1958516424590a168680be60753f4cfccea8f327cae973dfe4777
KP	d082d5b3b29cb6c054f3cfe8e5ff50dfa6a66220128b52afa485c5a91c10c074
PS	e82c5936775a5d3577433bb88271667ef18580bf449505b5bffa6fd12fa54bdf
BS	30345203af3be094776ef67f7a9ad2ae8da8392513c72ac7d24a791eb7b19302
R	8ab2c1fc0a4ca0e0f7145434b4c02f8efa97ce48d27fc03732cd632d6a3f9256
MP	81615141f0c729387304ba80bae4b3157d2b62c79030bec5a125f5ed3efbdccb

3. Panchaka Rahita: formula and counting policy

With one-based inputs, let S = tithi + weekday + nakshatra + lagna.

S % 9	Classification
0, 3, 5, 7	Rahita: this Panchaka restriction is absent
1	Mrityu
2	Agni
4	Raja
6	Chora
8	Roga

KP states five offset tests: add respectively 15, 12, 10, 8 or 4 to S; remainder 5 indicates respectively illness, fire, royalty, thieves or disaster. These are algebraically equivalent to the table, not five conflicting algorithms. Its rising sign is explicitly at departure, not frozen at sunrise. R gives the direct table and a worked 13 + 9 + 1 + 6 = 29, remainder 2 case.

MP resolves the dark-fortnight input explicitly: Shukla 1–15, Krishna 16–30; Sunday 1 through Saturday 7; Ashwini 1 through Revati 27; Aries 1 through Pisces 12. Its printed example is 17 + 7 + 19 + 11 = 54, remainder 0. The weekday continues through midnight until the next local sunrise.

Proposed profile: mypanchang_2025_panchaka_30_tithi.v1, with the KP remainder equivalence recorded as classical corroboration. Do not label the entire modern counting contract “universally classical.” A fortnight-reset convention is not admitted merely because it is easy to offer as a toggle. Keep the late-Dhanishtha-to-Revati usage of “Panchak” a separate product; it is not this four-input arithmetic.

Research-only enumeration checked 68,040 input combinations, with zero disagreements between the direct and offset formulas. Incorrectly resetting Krishna tithis to 1–15 changes the category in 30,240/34,020 dark-half combinations and the Rahita boolean in 22,680/34,020. These are combinatorial counts, not frequencies over real dates. Both printed examples were checked.

R also distinguishes purpose exceptions and explicitly calls some permissions its own inference. Preserve the five detected labels. Do not interpret Rahita as complete suitability, or automatically ignore Raja for a wedding.

4. Tara: keep three questions separate

For zero-based natal and target stars:

count = (target - natal) % 27 + 1

tara = (count - 1) % 9 + 1; cycle = (count - 1) // 9 + 1

This uses 27 equal nakshatras, inclusive counting and no inserted Abhijit. The base tara identity, the cycle-specific restriction, and a broader personal judgment are different outputs.

MC Gochara 13: proposed quarter profile

Cycle	Vipat (3)	Pratyari (5)	Vadha (7)
1: counts 1–9	Whole star restricted	Whole star restricted	Whole star restricted
2: counts 10–18	First quarter restricted	Fourth quarter restricted	Third quarter restricted
3: counts 19–27	Restriction lifted by this rule	Restriction lifted by this rule	Restriction lifted by this rule

Textual discrepancy: the printed Sanskrit/anvaya gives first, last and third portions; R explicitly interprets these as quarters 1, 4, 3. The MC Hindi paragraph instead describes three successive twenty-of-sixty portions. It does not match its own Sanskrit order for Pratyari. Record that reading as a commentarial discrepancy, not an interchangeable quarter implementation.

Recommended decision: admit mc_gochara_13_quarters.v1, citing the printed Sanskrit and R's corroborating interpretation; declare a quarter to be one 3°20′ sidereal lunar arc, with exact ingress roots for dated windows. This is an explicit interpretation of amshaka, not a claim that the Hindi paragraph says the same thing. Do not introduce the thirds profile without an internally reconciled witness and an identified time/arc basis.

The verse's cycle cancellation refers to the three named adverse taras. It does not make Janma unconditionally favorable. Keep existing Janma caution and report Janma purpose exceptions as unevaluated unless separately selected. R's personal advice to avoid Vipat/Vadha for important undertakings even in cycle 3 is a distinct conservative application, not grounds to erase the verse's cancellation. Its “fifty per cent” language is not a calibrated score.

PS printed 273: a real alternative

PS singles out Janma in cycle 1, Vipat in cycle 2, Pratyari in cycle 3, and Vadha in every cycle. This gives restricted inclusive counts {1, 7, 12, 16, 23, 25}, without quarter restrictions.

Admit as ps_sastri_scientific_273_navaka.v1, scan-owned and explicitly modern scholarly testimony. It disagrees with MC: count 25 is restricted here but lifted by MC's cycle rule; count 14, quarter 4 is restricted by MC but is not singled out by this PS rule. Neither is “the universal Tara exception.” Not singled out by a rule does not certify the whole election as favorable.

R's separate initial-ghati rule lists 7/3/8/6 for Janma/Vipat/Pratyak/Naidhana, then its Pushya-to-Magha example says seven for Vipat. The inspected reflow is internally inconsistent. Defer that specific initial-ghati profile until a printed edition or independent primary witness resolves 3 versus 7. This does not block either of the above complete cycle profiles.

5. Nitya Yoga: partial periods, not whole-day labels

MC Shubhashubha 34–35 and the Hindi explanation give:

Nitya Yoga	One-based index	Excluded portion
Vishkumbha	1	Initial 3 ghatis
Atiganda	6	Initial 6 ghatis
Shula	9	Initial 5 ghatis
Ganda	10	Initial 6 ghatis
Vyaghata	13	Initial 9 ghatis
Vajra	15	Initial 3 ghatis
Vyatipata	17	Entire Yoga occurrence
Parigha	19	First half of the Yoga occurrence
Vaidhriti	27	Entire Yoga occurrence

Proposed clock: fixed 24-minute ghatis for the explicitly numbered initial periods; actual solved start/end times for a whole occurrence; temporal midpoint of the solved occurrence for Parigha. Name this mc_34_35_fixed_ghati_temporal_half.v1. The midpoint conversion is an explicit derived choice. A 60%-long or 40%-long calendar display segment is not enough to establish the event start. Do not replace a whole occurrence by 24 hours, Parigha's half by a fixed 12 hours, or elapsed time by angular progress without declaring a different clock policy.

Intervals start at the true Yoga ingress, even if it preceded sunrise or the requested civil date. Initial-ghati intervals must be bounded by the actual Yoga end. Use [start, end) ownership and preserve root uncertainty.

The 27 longitudinal-sum Nitya Yogas are distinct from weekday/star Anandadi Yogas, named auspicious windows, upagrahas and declination-based Mahapata. KP printed 193 and MC commentary at 42 contain noon exceptions; those are separate cancellation rules. The base profile above does not silently apply an after-noon override or a strong-Lagna override.

6. Karana and Bhadra

BS chapter 100.1–5: finite Karana evidence

Propose a named, eleven-entry catalogue of supported activity tags, with verse references. It reports what this rule supports, not a complete election:

Karana	Source-supported character/activity, paraphrased
Bava	Auspicious, temporary or lasting acts; nourishment
Balava	Religious/meritorious acts; assistance to Brahmins
Kaulava	Affection and friendship
Taitila	Prosperity/popularity, shelter and household matters
Gara	Cultivation, sowing, house/shelter work
Vanija	Lasting undertakings and commerce
Vishti	Auspicious acts prohibited; hostile acts mentioned historically
Shakuni	Nourishment, medicines/roots and mantra work
Chatushpada	Cattle, Brahmins, ancestors and royal affairs
Naga	Lasting and fierce acts; coercive acts mentioned historically
Kimstughna	Meritorious, nourishing and auspicious acts

These are historical textual categories, not medical or practical advice. In particular, the four fixed Karanas cannot all receive one universal adverse flag under this source. VED-011 owns complete purpose-specific elections; these narrow tags and explicit prohibitions can be useful before that work.

MC 43–45: Bhadra occurrence, residence and exceptions

Vishti occupies these tithi halves: Shukla 4 latter, 8 former, 11 latter, 15 former; Krishna 3 latter, 7 former, 10 latter, 14 former. In the existing zero-based monthly Karana sequence these are {7,14,21,28,35,42,49,56}.

Residence is determined by the current sidereal Moon sign:

Residence	Signs
Earth	Cancer, Leo, Aquarius, Pisces
Heaven	Aries, Taurus, Gemini, Scorpio
Underworld	Virgo, Libra, Sagittarius, Capricorn

The text locates the effect in that residence. Return the residence and earth-applicability explicitly; do not delete the detected Vishti identity when the Moon is outside the Earth group.

MC 44 also permits latter-half-origin Bhadra during daytime and former-half-origin Bhadra during nighttime. The Sadhana witness expressly limits its corresponding exception to necessary work. Select a named source and preserve that condition; do not silently merge the qualifications. Day and night need actual local sunrise/sunset, not 06:00/18:00 constants.

MC 44: mouth/tail tables and a declared derived clock

Tithi	Mouth: beginning of eighth-part number	Tail: end of eighth-part number
Shukla 4	5	8
Shukla 8	2	1
Shukla 11	7	6
Shukla 15	4	3
Krishna 3	8	7
Krishna 7	3	2
Krishna 10	6	5
Krishna 14	1	4

The specified mouth is the initial five ghatis of its named yama; the tail is the final three ghatis of its named yama. They are not always the first and last portions of the Karana. MC calls the mouth adverse and tail favorable.

Recommended derived profile: mc_44_tithi_eighths_normalized.v1. For actual tithi duration D and start A, let one nominal tithi-ghati be D/60 and an eighth be D/8. Then:

• mouth: [A + (m-1)D/8, A + (m-1)D/8 + D/12);
• tail: [A + qD/8 - D/20, A + qD/8).

This normalization is a proposed computation derived from the table, not a directly verified universal time-conversion instruction. It must be named in policy/provenance. Intersect the resulting interval with the actual solved Vishti occurrence and retain both original and clipped bounds: equal time halves of a tithi do not exactly equal the two 6° Karana intervals when lunar speed changes. Validate unequal-duration cases, not only a 24-hour tithi. Do not quietly mix these normalized ghatis with the fixed Yoga clock.

Sadhana's sequential 30-part body scheme and its alternative snake/scorpion tail prohibitions are genuinely different readings. Retain them in the source registry as unselected; do not combine them with MC's tithi-position table. The named MC derived profile can be implemented without waiting for every body-anatomy variant to be admitted.

Precedence recommendation: first return detection, mouth/tail membership, residence and each applicable exception as separate source facts. Where a mouth restriction and another exception overlap, report the simultaneous claims; the inspected passage does not establish a universal precedence ordering. An optional conservative “mouth takes priority” selector would be a Moira application policy, not an invented classical consensus. No generic favorable boolean should hide this unresolved precedence.

7. Current Moira behavior and proposed contracts

Existing surface	Verified limitation	Proposed addition
tara_bala	Inclusive count and nine-tara polarity; no cycle/pada exception evaluator	Separate source-policy assessment retaining base identity/polarity
chandra_bala and personal score	Existing bounded rule/weight profile	Preserve compatibility; expose the new assessment separately
Panchanga Yoga classification	Adverse indices {5,8,9,16,26} zero-based; five entire Yoga labels	New nine-Yoga timing profile with start/end evidence; no silent replacement of shared constants
Karana classifier	Vishti adverse; every other Karana neutral	Source-owned eleven-entry evidence and Bhadra components
Daily Panchanga	Sunrise-owned day, phase transitions and uncertainty already exist	Reuse solver/reader/frame policy; obtain preceding and following event bounds where fractional rules require them
Muhurta REST models and router	Strict current personal score; no VED-008 source-policy result	New typed direct and dated assessment; catalogue/policy discovery; canonical serialization

Suggested new owning module: moira/panchanga_shuddhi.py. Its data structures should distinguish detected, restricted, exception_applies, not_applicable, unavailable and not_evaluated, together with stable source/rule IDs. A conflict is not success; unavailable astronomy is not a negative finding. Each result needs selected policy, applied policy, actual inputs, raw finding, exception evidence and any remaining restriction. Policy receipt fields must describe execution rather than echoing unimplemented options.

Proposed REST scope: typed policy/catalogue discovery, direct component assessment and date/location assessment. Exact route names are implementation design, but all three must arrive with engine exports/facade and tests. A bounded one-day interval result is preferable to silently widening generic sampled Muhurta search. Keep new rules out of weighted search until its composition/weight policy is separately selected.

Input requirements:

• Strict integer ranges and explicit zero/one-based adapters. Reject booleans, strings, unknown profiles, extra fields and contradictory redundant inputs.
• Natal star for Tara; lunar longitude/pada when the selected quarter rule needs it. Missing natal data must not suppress unrelated evaluations.
• A single coherent ayanamsa/frame for Moon, Sun, nakshatra and ascendant. Reuse the actual serving reader; do not compute an unrelated chart behind the response or substitute a different ayanamsa at one stage.
• Location/time zone and real solar anchors for dated weekday/day-night rules; current Lagna for Panchaka. Sunrise determines the weekday, not the Lagna. Handle polar absence and ascendant singularities explicitly.
• Actual predecessor/successor boundaries for Yoga, tithi and Karana. A clipped sunrise-day segment cannot stand in for the full event duration.
• Bound requests, solver horizons, transitions and response sizes. Preserve coincident-boundary reasons and uncertainty bands; do not fabricate precision.

The new nine-Yoga contract and BS Karana catalogue expose limitations in the legacy classifier. Keep that compatibility behavior identified and documented; a future migration can change defaults explicitly. Broad BPHS chapter-85 comments in the legacy module are not edition-collated authority for the new rules. Cite the precise MC/BS loci instead.

8. Admission, deferral and validation plan

Item	Research decision
Panchaka direct remainder rule; MP thirty-tithi/date convention	Admit for implementation as a named operational profile, with KP corroboration
MC Tara quarter rule; PS alternative cycle rule	Admit separately; keep Janma and broader purpose judgments distinct
MC nine-Yoga initial/whole/half table	Admit with explicitly selected fixed-ghati/temporal-half clock
BS eleven-Karana activity evidence	Admit catalogue and bounded rule evaluation; complete purpose elections remain VED-011
MC Bhadra residence, origin-half/day-night, mouth/tail table	Admit component facts; normalized mouth/tail clock is a named derived policy
One universal precedence among Bhadra exceptions	Exclude as unsupported; expose overlaps or choose a separately named application policy
MC Hindi thirds; R 7-versus-3 Vipat initial-ghati example	Defer these particular readings; exact discrepancy and resolution evidence identified above
Sadhana anatomical/scorpion variants	Document as alternatives; edition and complete clock/precedence required before selection
Noon, strong-Lagna, benefic-aspect and ritual-remedy cancellations	Retain references; finite separately selected VED-009/010/011 rules, not an automatic override

Implementation sequence should be one reviewable package with three internal milestones, not twelve mandatory phases:

1. Freeze named policy IDs, source tables and typed component results. Write independently transcribed source fixtures and compatibility checks first.
2. Implement pure evaluators and date/day composition with real phase/solar/ ascendant boundaries. Complete all selected profiles, including missing-data and conflicting-exception states.
3. Finish exports/facade and REST together, execute focused source/boundary/ resource/HTTP validation, reconcile standard/reference/register and generate documentation. Publication requires its own user instruction.

Minimum acceptance cases:

• Both printed Panchaka examples; every remainder; all 68,040 direct input combinations against KP's offset formulation. Krishna Pratipada/Amavasya, midnight-versus-sunrise, and Lagna ingress cases must detect counting errors.
• All 27×27 natal/target combinations, all four padas, wraparound and exact pada boundaries. Explicit disagreement fixtures: count 14/pada 4 and count 25 between MC and PS; count 12/pada 2 and count 16/pada 4. Janma remains visibly outside MC's three-tara cancellation scope.
• Every one of the nine Yoga rules, immediately before/at/after exclusion end, pre-sunrise starts, month/360° wrap, and non-24-hour Yoga occurrences. A 20-hour Parigha occurrence has a 10-hour excluded temporal half under the selected clock; this is a synthetic arithmetic fixture, not a published event.
• All 60 Karana positions, all eleven catalogue entries, all eight Vishti occurrences, all twelve residence signs and all eight mouth/tail rows. For a synthetic 24-hour Shukla Ashtami: mouth 03:00–05:00 from tithi start; tail 01:48–03:00. Recheck with non-24-hour duration and nonuniform lunar speed.
• Residence change during Vishti; sunrise/sunset splits; tail followed by mouth; exception/restriction overlap; clipping and exact endpoint ownership. Missing event start, natal star or solar anchor must yield the correct component state.
• Independent root/bracket checks, real serving-reader and startup lifecycles, ordinary/DST/polar dates, finite operational budgets, typed response validation and direct/HTTP equality. Test both configured and discovered kernel paths.
• Regression of existing Tara/Chandra, score, sampled search, daily Panchanga and named Muhurta contracts. No silent changes to their existing policy IDs.

No astronomical or predictive accuracy claim follows from agreeing with these rule tables. Source fidelity, arithmetic, event solving, transport and predictive validity are distinct evidence classes.

9. Research receipt

Executed with C:/dev/moira/.venv/Scripts/python.exe (Python 3.14.3): source extraction, bookmark lookup, rendering, SHA-256 manifests and the private check_research_arithmetic.py enumeration. Result: 68,040 formula agreements, zero disagreements, two source examples checked. No runtime implementation, kernel event validation or new REST behavior is claimed by that script.

Documentation verification: canonical consistency, relative-link resolution, in-memory rendering through the owning wiki generator and git diff --check. The generated checkout is left for the next authorized synchronization/publication; no source PDFs are copied into either repository. Runtime, tests, dependencies, version and existing source-publication commits remain unchanged.