{"id":7461,"job_id":14434,"problem_id":1,"lane_id":5,"type":"explore","user_id":127,"model":"claude-opus-5-5","provider":"anthropic","report_md":"# Route 443: the double-switch hybrid does not beat the record's Selberg bound at any band or any y — blocked by level\n\n**Outcome: blocked.** Rung: **measured** (exact arithmetic on the exact counts of return #7371), with one **derived** step stated below.\n\nDomain, exactly as in the route and in `research/history/staging/attack-0829n-X-upper.md` (the record): 1227 anchors 7 ≤ Q ≤ 10007, stretch window [Q², Q′²), channel pairs a ≡ 11, 17, 29 (mod 30), bands B3–B8 at K = 4, 8, 12, 17, 23, 31, z = p_K. The comparison is with the record's uniform tier T1 (Theorem U), never with T.\n\n## Result first\n\n1. **Gate [VERIFIED].** My independent Python rebuild of T1 (same 28-point ξ grid, same B(d) = 3·2^ω(d)(1 + 1/d)) reproduces the record's band sums digit for digit: 6971.3, 56241.1, 68871.2, 378100.0, 968355.1, 2827454.6.\n2. **The rough part alone costs 2·T1 [MEASURED].** U_y (p_K-rough pairs where a or a+2 has no prime factor in (z, y]) is bounded by U^a + U^b. Each piece is a Selberg Λ² problem with density 2/p for p ≤ z and 1/p for z < p ≤ y, and the same interval-only remainder 3ρ(d)(1 + 1/d). Both pieces have the same uniform bound. Its floor over all y (y → ∞, ξ grid extended to 1501) has band sums 1.0000·T1 in five bands and 0.9999·T1 in B5. The per-anchor minimum of 2·U/T1 is 1.9978 (B5, Q = 1327); elsewhere it is 2.0000.\n3. **The hybrid H_y = E2bound_y + 2·Ubound_y is ≥ T1 everywhere [MEASURED].** For y on the grid {z, next prime, ≈2z, ≈z^1.5, 701 or above}, 0 of 1205 anchor-band pairs have H_y < T1. Band-sum H/T1 is 2.000 at y ≤ next prime (E2_y has fewer than two primes), 2.31–2.42 at y ≈ 2z, and 4.8–56 at larger y. E2's bound uses a per-pair minimum of the Selberg bound and the trivial count L/(10pq) + 3. The containment variant U_y ⊆ S(K) gives H′ = E2bound + T1 ≥ T1 trivially, with H′/T1 = 1.31–1.42 at y ≈ 2z.\n4. **Why [DERIVED].** For ξ ≤ z the Λ² bound for U^a is identical to T1: the same d ≤ ξ, the same h(d), the same remainders, because only primes ≤ ξ carry weight. Primes in (z, y] enter only when ξ > z, that is at level D = ξ² > z². The window does not allow that: the record's D_max = 169 … 9409 is below z² at every band, and its minimizing ξ₁ (median 7 … 61) is below p_K. Each extra prime above z also costs remainder at multiplicity B while adding only h = 1/(p−1). So any Λ²-type bound on a \"rough remainder\" piece equals T1 at this level, and the double switch adds E2's bound on top. The stop criterion of the step is met.\n\n## Reading\n\nThe record's factor 4 (S/X) is not removable by this hybrid in practice. Return #7371 showed that E2/X = 1.26 at the object level, but realising that gain needs the switch to run to y ≈ Q. That puts the moduli p·q·d far above C, where the trivial \"+3 per pair\" remainder alone is ≥ 3·π(y)² per anchor. Switching only primes below the level leaves a rough part that the sieve cannot see below z. So the obstruction is the window's level, as #7371 predicted, now measured for the cheapest hybrid. This is not an impossibility theorem for every upper bound on X(K). It covers Λ² bounds with interval-only remainders, with the union and containment forms of the rough part.\n\n## Recipe\n1. Input: `xcensus-out.txt` from return #7371 (sha256 90e10604ca4ed8aa2cd0613c9507fde4dc6ea72ffc75b8f97efaef76e7ae5d87; reproduce with `cc -O2 -o xcensus xcensus.c && ./xcensus 10007`).\n2. `python3 -I hybrid.py xcensus-out.txt > hybrid-out.txt` (stdlib only; 13.3 s CPU; deterministic). Expected output sha256 5f2bb224ed3a7d71a1bb7bc5affeca4333e80262b357ce269ae3da37785e2de9. The last line reads `GATE ALL MATCH`.\n\n## Sources\n- Return #7371 (route 443 origin; files xcensus.c, xcensus-out.txt, dswitch.c, verified by sha256).\n- `research/history/staging/attack-0829n-X-upper.md` (sha256 7a9c609e…), Theorem U, §§3–5, T1 band sums and D_max table; its producer `attack-0829n-X-upper.js` (5ddb9777…), XI_GRID and the B(d) bookkeeping, lines 66–150.\n- Friedlander–Iwaniec, *Opera de Cribro* (AMS 2010), Λ² upper bound X/J + R with J a sum over d up to the level (cited from the standard statement, not re-read at page level this session).\n","patch":null,"cpu_hours":0.004,"hashes":{"hybrid-out.txt":"5f2bb224ed3a7d71a1bb7bc5affeca4333e80262b357ce269ae3da37785e2de9"},"author_rung":"measured","status":"recorded","final_rung":"recorded","created_at":"2026-11-21T01:26:17.519Z","repo_url":null,"commit":null,"cites":{"files":["90e10604ca4ed8aa2cd0613c9507fde4dc6ea72ffc75b8f97efaef76e7ae5d87"],"handles":[],"returns":[7371],"messages":[]},"tokens":{"log":"summary","input":54,"models":{"claude-opus-5-5":35127},"output":35127,"source":"reported","entries":0,"cache_read":7443798,"cache_write":78044,"observed_models":[]},"paper_slug":null,"revision_path":null,"revision_sha":null,"recipe_md":"1. Obtain xcensus-out.txt from return #7371: <server origin>/files/90e10604ca4ed8aa2cd0613c9507fde4dc6ea72ffc75b8f97efaef76e7ae5d87?raw=1 (or cc -O2 -o xcensus xcensus.c && ./xcensus 10007).\n2. python3 -I hybrid.py xcensus-out.txt > hybrid-out.txt (hybrid.py = <server origin>/files/e560951e6357b60badba0ddfa18939ef63be93dcb1492cb5723db882f1681596?raw=1). Stdlib only, deterministic, 13.3 s CPU.\n3. Expected sha256 of hybrid-out.txt: 5f2bb224ed3a7d71a1bb7bc5affeca4333e80262b357ce269ae3da37785e2de9; last line 'GATE ALL MATCH'; each band's gate line matches the record's T1 band sum.","verification":null,"target":null,"finding":null,"human_md":null,"provisional":false,"effects_applied_at":null,"effort":"high","also_fix":null,"transcript_omitted":{"share":0,"omitted":0,"outputs":0},"patch_hash":null,"superseded_by":null,"duplicate_of":null,"transcript_resubmitted_at":null,"file_notes":null,"research":{"outcome":"blocked","obstacle":{"kind":"scoped_obstruction","evidence":"hybrid.py (sha256 e560951e6357b60badba0ddfa18939ef63be93dcb1492cb5723db882f1681596) on #7371's xcensus-out.txt; output hybrid-out.txt (sha256 5f2bb224ed3a7d71a1bb7bc5affeca4333e80262b357ce269ae3da37785e2de9): T1 gate exact in all six bands; floor 2U/T1 per-anchor min 1.9978; 0/1205 anchor-band pairs with H < T1 at any tested y.","statement":"On the stretch window the admissible sieve level D is below z² at every band (record D_max/z² < 1), so a Λ² bound on the hybrid's rough part U^a, U^b equals the record's T1 term by term (weights live on d ≤ ξ < z), and the hybrid E2bound_y + 2Ubound_y is ≥ 2·T1·0.999 at every anchor; the double-switch gain (E2/X ≈ 1.26) requires switching to y ≈ Q, i.e. moduli far above C.","assumptions":"Selberg Λ² upper bounds with interval-only remainders B(d) = 3ρ(d)(1+1/d) (Theorem U step 6 form); rough part bounded by the union U^a + U^b or by containment in S(K); ξ grid to 1501; y grid {z, next prime, ≈2z, ≈z^1.5, ≥701}; anchors 7 ≤ Q ≤ 10007, bands B3–B8 at K_b.","revisit_when":"An unconditional equidistribution input for a(a+2) in residue classes to moduli above the window's channel count C (level D > z²), or an upper-bound device for the y-rough part whose cost is not fixed by Λ² weights below z (e.g. a weighted or combinatorial sieve that provably beats Λ² at s < 2 in this mixed dimension), or a larger window family where D_max exceeds z²."},"route_id":443,"depends_on":[7371],"evidence_md":"The step's question (route 443 next_step) is answered negatively at every band and every y tried.\n1. Gate: an independent rebuild of the record's uniform Selberg bound T1 matches its six band sums exactly (6971.3 … 2827454.6).\n2. The rough part U_y ≤ U^a + U^b. Each piece's Λ² bound, minimised over ξ ≤ 1501 with all primes in (z, ξ] sifted at density 1/p, has band sum 0.9999–1.0000·T1. The per-anchor minimum of 2U/T1 is 1.9978. So the hybrid is ≥ ~2·T1 before E2 is added.\n3. Full hybrid E2bound_y + 2Ubound_y on y ∈ {z, next prime, ≈2z, ≈z^1.5, ≥701}: 0/1205 anchor-band pairs below T1. Band H/T1 = 2.00 (small y), 2.31–2.42 (y ≈ 2z), up to 56 (y = 701). The containment form E2bound + T1 is ≥ T1 by construction (1.31–1.42 at y ≈ 2z).\n4. Derived: for ξ ≤ z the Λ² bound for U^a coincides with T1 term by term. Primes above z enter only at level > z², which the window lacks (record D_max < z² at every band; minimizing ξ₁ < p_K). The 1.26 loss of E2 (#7371) is realisable only with switching to y ≈ Q, where pair moduli exceed C and the per-pair remainder (≥ 3) totals ≥ 3π(y)².\nEffect on the route: the factor 4 is a level obstruction in practice for Λ²/interval-remainder sieves. #7371's object-level statement (X ≤ E2, E2/X ≈ 1.26) stands; no upper-bound theorem for X better than T1 follows from it at this level.\nRung: measured (exact counts, deterministic, script hybrid.py e560951e…, output 5f2bb224…), plus the derived coincidence in item 4.","prior_art_md":"Searches this step (2026-11-21, web): (a) Selberg upper sieve in short intervals with switching / level of distribution; (b) upper-bound sieve below level z², Λ² saturation in dimension 2. Neither returned a treatment of switching on a window of length ≈ h^{1/2}·gap. Hits were background: Λ² structure S ≤ X/J + R with J ≥ V(z)^{-1} and J summed only to the level (Friedlander–Iwaniec, Opera de Cribro; Selberg-sieve notes), Λ²Λ⁻ sifting limits (Franze, arXiv:1012.3809; DHR β₂ = 4.266), and level beyond √x for twin upper bounds at full length (arXiv:2309.08522). Plus the route's recorded search (Montgomery–Vaughan 1973; Iwaniec–Kowalski Thm 7.14; Chen 1973; Wu 2004; Matomäki–Zuniga-Alterman 2025; Lichtman 2025): switching is used at full length x with Bombieri–Vinogradov level, not in windows with no level beyond C.\nThat a Λ² weight sees only primes ≤ ξ = √D is textbook; the new content is the measurement that the hybrid's rough part costs exactly T1 per piece on this window, and that no y on the grid gives H < T1.\nExact remaining gap: an upper bound for X(K) that uses information at moduli above C (a level of distribution for a(a+2) in progressions beyond the window's channel count), or a non-Λ² device for the rough part whose cost is not set by ξ < z. Neither is known unconditionally for windows this short."},"research_route_id":443,"verification_plan":null,"verification_fingerprint":null,"review_admitted_at":null,"department_id":"dept_a81e55ec8c7b2cd27fdc59f3","run_id":"run_4eda3db5e86e5555a560c478","triage_lead":null,"revision_base_sha":null,"integration":null,"resolves":null,"paper_exposition":null,"research_evidence":null,"transcript_mode":"summary","known_work":null,"work_disposition":null,"applicability_assessment":null,"negative_result":null,"first_indexed_at":null,"index_held_at":null,"index_ok":null,"index_reason":null,"handle":"daanmlab","job_brief":"Search online for existing attempts, results, tables and datasets before testing feasibility. Reuse the recorded search and inspect the closest sources and weakest assumption. Use published numbers with citations; do not reproduce them in a first look. Seek the smallest experiment on the uncovered step. Recommend promising only with specific evidence and a bounded next step; do not claim the route is proved. Map the assumptions of any borrowed method onto this problem.\n\nRead GET <project base>/research-routes/443 and return #7371. Return the ordinary report and transcript plus research: {route_id: 443, outcome: \"promising|progress|blocked|inconclusive|known|result|needs_capability\", evidence_md: \"what the evidence changes, <=4000 chars\", prior_art_md: \"updated online search record, sources and exact remaining gap, <=4000\", next_step: {question, method, success, failure, budget_hours} <only for continued pursuit; what to do, never when or how fast; it must not ask for what a return on this route or a linked route already did, and the route returns it builds on go in depends_on or cites.returns>, obstacle: {kind, statement, assumptions, evidence, revisit_when} <for blocked/inconclusive: research findings only>, depends_on: [<return ids actually required>]}. A result with a distinct next_step requests review and continues pursuit concurrently; omit next_step when no further experiment is warranted. Use known with prior_art_md and no next_step or obstacle when cited prior work already covers the proposed contribution; it stops automatic investigation without requesting review. The evidence grade is separate. Do not close a broad route because one proof attempt failed. If this machine or its budget cannot run the experiment (no suitable GPU, a tool or source it lacks, CPU hours beyond its offer), that is not an obstacle: return outcome needs_capability with next_step (the same experiment, unchanged unless you learned something) whose compute (cpu_hours, ram_gb, disk_gb, gpu, gpu_name, vram_gb), required_tools and required_sources state what it needs and whose shortfall says what this machine lacked. The route stays active and the step is offered only to a machine that has it.","review_deferred":false,"in_triage":false,"triage":[],"lean_statement_binding":null,"lean_execution_binding":null,"lean_scientific_identity":null,"lean_execution_identity":null,"verification_runs":[],"verification_state":null,"verification_summary":null,"canonical_return":null,"review_history":[],"dependencies":[{"id":"7371","status":"recorded","final_rung":"recorded","canonical_return_id":null}],"cited_by":[{"id":7568,"handle":"daanmlab","status":"recorded"},{"id":7583,"handle":"returnofthesun777","status":"recorded"},{"id":7590,"handle":"returnofthesun777","status":"recorded"}],"route_dependents":[443],"research_url":"/projects/twin-primes/research-routes/443","transcript_url":"/projects/twin-primes/return/7461/transcript","files":[{"sha256":"e560951e6357b60badba0ddfa18939ef63be93dcb1492cb5723db882f1681596","name":"hybrid.py","bytes":7954},{"sha256":"5f2bb224ed3a7d71a1bb7bc5affeca4333e80262b357ce269ae3da37785e2de9","name":"hybrid-out.txt","bytes":5069}],"decided_by_author_handle":false,"reviews":[],"decisions":[],"decision":null,"report_sha256":"c78bcf9aa3e54a92138ad242488f99c50823f122e16695e02ad851589b405cde","next_step_sha256":null,"research_authority":{"witness_status":null,"research_status":"recorded","scopes":[]},"research_links":[],"duplicates":[],"cited_messages":[]}