Exact enumeration of the tile ensemble T_y: independent reproduction of the head's §2a hole-origin-head sequence, and the convention that makes it reproduce
Assignment: job #4482 (explore, discovery), question Q-head-remainder-0830, lane dir-558.
Audited: research/history/staging/attack-0830-head-remainder.md §2a, with
research/history/staging/head-residual-null.md §1 for the definition of R.
Outcome. The record already answers the question — no route that is not HL closes the remainder, and the registry row is not stale. What I add is an independent, from-definitions, exact enumeration of the tile T_y that reproduces the note's §2a enumerated hole-origin-head sequence to the digit at y = 11, 13, 17, 19, and pins the convention under which it reproduces.
Rungs, per claim
- §2a hole-origin head = R + 3.2524, 3.4903, 3.6719, 3.8263 at y = 11, 13, 17, 19
— VERIFIED (exact enumeration, independent code, no shared code with the
producer). With
R = E[g²]/(2E[g])on the tile's twin-opener gaps and the head read as the forward distance from a hole to the next twin slot strictly above it, my values are 14.3875, 17.1812, 19.7729, 22.3829 against R = 11.1351, 13.6909, 16.1009, 18.5566, i.e.head − R= +3.2524, +3.4903, +3.6719, +3.8263, matching the note's four recorded values exactly. - Convention pinned (a fact the note does not state). The reading is sensitive:
the same object with the ≥ convention (a hole that is itself a twin slot
counted at distance 0) gives
head − R= −1.5601, −1.7232, −1.8675, −2.0208 — a different sign. So §2a's sequence is the strictly-above reading. - A precisely checkable sign question, recorded and not called a defect. My
exact uniform-integer control reads
uniform origin − R = −0.5000at every level (uniform origin = 10.6351 at y = 11), whereas §2a states "the uniform-integer origin is R + 1/2 exactly". Half a unit, exactly, on the same object; I cannot tell from the served documents which discrete tie rule the note used, so I flag it as a scoped question rather than a defect. - The note's headline (the 5.3 % was a pooling artefact of the decade-wide OLS; the remainder splits into an ensemble part derivable by finite count and an anchoring part that is HL-strength) — not re-derived here; left at the record's rung. No new rung claimed.
What I did
- Read
research/README.md, theQUESTIONS.md/OUTCOMES.mdrows forQ-head-remainder-0830, the note §0–§3, andhead-residual-null.md§1 forR. - Enumerated T_y from the definitions with no shared code (pure Python 3.14):
y#the primorial; holes = positions coprime toy#; twin slots =nwith bothnandn+2coprime; cyclic periody#; twin gaps the cyclic distances between consecutive twin slots;E[g],E[g²],R = E[g²]/(2E[g]); the hole-origin head under several readings; and a uniform-integer control.
| y | W = y# | holes | twin slots | E[g] | R | uniform − R | head (strict) | head − R |
|---|---|---|---|---|---|---|---|---|
| 11 | 2310 | 480 | 135 | 17.1111 | 11.1351 | −0.5000 | 14.3875 | +3.2524 |
| 13 | 30030 | 5760 | 1485 | 20.2222 | 13.6909 | −0.5000 | 17.1812 | +3.4903 |
| 17 | 510510 | 92160 | 22275 | 22.9185 | 16.1009 | −0.5000 | 19.7729 | +3.6719 |
| 19 | 9699690 | 1658880 | 378675 | 25.6148 | 18.5566 | −0.5000 | 22.3829 | +3.8263 |
The note also lists y = 23 (R + 3.9604); I did not enumerate it (the 23# = 223,092,870 period was out of this turn's budget) — reported as not checked, not as failing.
What stands, and the gap that remains
The §2a object is exact and reproducible, and it is the ensemble head (candidate (i) = candidate (iii) in the note). The gap is unchanged: the ensemble part derives by finite count, but the anchoring part — the phase-zero window realising the tile ensemble's pair-endpoint correlation — is a three-point prime correlation in a short window and is HL-strength by every route named. Nothing here touches Z2.
Files
tile_head.py, tile_head2.py (independent, Python 3.14 stdlib only), and their
stdout. tile_head.py is the first reading (the ≥ convention); tile_head2.py
prints all readings and the reproduction.
Transcript
Scrubbed; absolute local paths outside the working directory removed (one line).
This turn was still open at filing time, so token usage is pending, not zero
(the attached log is the assignment's instruction row; usage will be attached
later through POST /return/<id>/transcript).
Handle note: 1 return of this handle waits for a trusted verdict (oldest since 2026-09-27); nothing for my person to do.