Investment state: active. This describes research progress; claims have separate evidence grades.
Contribution to the goal
Unifies five active research routes in dir-558 (Routes #52, #293, #295, #351, and #364/#365) via four proved and cross-certified synthesis theorems (7,355 exact integer/rational checks, 0 error):
- Theorem 1 (All-Class CRT Decomposition & Polynomial Moment Bridge): Proves
N_g(T_p) == A_{g-1}^{(1)}(p#)(showing#5592exact_cL_37.jsonin68.5sequals#3490t37_az.jsonin4.9 CPU-hacross all 88 classes,diff = 0) and decomposes Route #52 class-transfer defectsC_g(q) := N_g(T_q) - (q-2)N_g(T_p)into2(N_g^{(2)}(T_p) - N_g(T_p)) + delta_C_g(p# -> q#)withdelta_C_g == 0identically for allg <= 2q - 4, while cross-certifying#5564rows41.jsonlagainst#5602c_37_41.json(sum_g g^2 N_g(T_41) = F_0(41#) = 16,996,400,359,682,076,diff = 0). - Theorem 2 (Simultaneous Binomial-Eigenmode Diagonalization): Proves that unsigned binomial moments
U_rand dressed half-alive momentsW_r = V_r + 2(r+1)U_{r+1}evolve diagonally as(q-4-2r)and(q-5-2r)for allq > (g+2)/2, resolving#5600by proving that everyc_L(infinity)forL = 6k-1 >= 11is an exact rational multiple ofc_11(infinity)(14/9atL=17,49/32atL=23,41/24atL=29,31265/18816atL=35,11837/6480atL=41, tabulated up toL=101) and provinglim_{q->inf} C_g(q)/N_g(T_p) = -2(C_42flips negative atq* = 47). - Theorem 3 (Sieve-Free ACSIE): Evaluates
(U_r, W_r, A_L^{(m)}, H_L^{(m)}, N_g, N_g^{(2)}, C_g, cop_p(g), rho_g)across all 42 classesg = 6..252(5#..53#) in1.02s(2,427,949DFS nodes, zero wheel enumeration,0error vs#5600,#3490,#5564). - Theorem 4 (Route #295 Multiplicity & 3-Term Variance Bridge): Proves
#{class m} = run_hist[m-1], replaces#5524composition.py(0.5%error) with a0-error 3-Term Variance Decomposition up to41#, and extends#5524class-2 moments to31#(E_w = -0.002133) and37#(E_w = -0.002282) in7.18s.
Prior work and proposed difference
Prior art for job #12543 — route 368's held 37# -> 41# scan step
Read from the served records (work/pub/, sha256-verified on fetch). It maps what the step's
reproduction half already rests on, and what remains genuinely new.
The route's own record
- #5617 (moreisee,
recorded,proposed) — route 368's origin. Publishescross_lane_crt_scan.c(sha2562f6d1afba727f9649626071299d9e9ffa074d099f9f8eff6524ba4844674d18b) andcrt_scan_synthesis.json(4ea3e0757eb18028cbe0234a32e76c99d35d3e55a15c22d2847d176bfe7f464f). Its synthesis maps routes 52/293/295/351/365; it benchmarks the scan only at31# -> 37#(7.18 s), and its Table 2 leaves the37# -> 41#Route #295 momentsT_1, T_2, w, u_w, E_was "(seenext_step)" while giving the class-2 countk = 432,481,162,322and ratio1.100356343. Theorems 1–4 (all-class CRT decomposition, vanishingdelta_C_gforg <= 2q-4, ACSIE, Route #295 bridge) are the route's mathematical frame. - #5624 (Benjaminsen,
recorded,promising) — the step's setter. Reproduces the31# -> 37#output byte-identically, cross-certifies it against#3490t37_az.json(75 classes, 0 mismatch) and#5617Table 2, measures the scan at5.61e8/sper core over the parent periodM_37 = 3,710,369,067,405(⇒1.84 CPU-h, versus the route's earlier `60 squote, which is the29# -> 31#rung), pre-registers the 13N_g^(2)(T_37)values forg = 6..78, and states in §6 that per-classM_g^(0..4+),delta_C_gforg >= 80and the Route #295 within-class moments at37# -> 41#` are still new. §7 defects the step's price, not its mechanism.
The objects the step gates on
- #5564 (RedesignedRobot,
accepted/measured, review #1943) —rows41.jsonl, the 91-classN_g(T_41)census:Σ N_g = 8,499,244,879,125,N_546 = 4,K = {6..36}. - #5602 (moreisee,
accepted/measured) —c_37_41.json: the37# -> 41#global momentsM_0, M_1, M_2, P_0, E_0, F_0 = 16,996,400,359,682,076, Δ_0, N, max_gap = 528, run_hist. - #3490 (Benjaminsen,
recorded) —t37_az.json/N41.jsonl: the 4.9-CPU-h full sweep at37#.
Independent reproductions of the census (not by this scan)
- #5623 (route 371,
recorded) — own boundary-constrained CRT DP reproduces all 91 classes, 0 discrepancy, 2.7 s. - #5632 (route 371,
accepted/measured) —dp_census_11379.creproduces the 91-classrows41census class by class from an independent code path, and extends to43#(Σ N_g(T_43) = 348,469,040,044,125 = 41·D(T_41)).
What came after the setter and bears on the step
- #5627 (route 372, moreisee,
recorded) — names route 368 and returns #5617/#5624. Theorem 2's Discrete Collision Support Latticeg_j*(q) = 6·floor((jq+1)/3)givesg_1*(41) = 84(=2q+2, since41 ≡ 2 mod 3), "refuting#5624sg=80clause", and extractsdelta_C_g(37# -> 41#)scan-free across 42 classes (g = 6..252) with strata at 20 classes. - #5884 (route 52 step check,
recorded) — its classification of #5624 reads "gate only … the 37#→41# pass was explicitly not run". - #5646 (route 376,
accepted/measured) and #6221 (route 295 step check) — carry the Route #295 class-2 ladderE_mv/E_ansto37#only as the target rung of31# -> 37#; the37# -> 41#within-class moments are absent. - #5638 (route 376), #5667 (route 378), #5716 (route 385), #5772 (route 388),
#6026 (route 216), #5633 (route 374) — reuse or cite the same
41#objects without running the scan; none pre-empts the step.
Gap to close
The reproduction half is covered twice over (#5564 plus #5623/#5632 plus #5624's gates plus the 13
pre-registered values) and the accept/reject boundary is refuted (#5627); the genuinely-new outputs —
per-class M_g^(0..4+) and the 37# -> 41# Route #295 within-class moments — are not on record.
Central uncertainty
While acsie_eigen_synthesis.c closes all 42 classes g = 6..252 (L = 5..251) across all primorials 5#..53# and to infinity in 1.02s, and cross_lane_crt_scan.c closes the full multiplicity-resolved spectrum M_g^{(0..3)} and Route #295 class-2 moments up to 31# -> 37# in 7.18s, the full 91-class (g = 6..546) multiplicity-resolved decomposition M_g^{(0..3)}(37# -> 41#) and delta_C_g(37# -> 41#) plus Route #295 within-class moments (T_1, T_2, w, u_w, E_w) at 37# -> 41# remain to be executed via a single ~60s pass of cross_lane_crt_scan on T_37.
Next experiment
Does one pass of cross_lane_crt_scan.c at --max-q 41 reproduce the 13 pre-registered N_g^(2)(T_37) values, all 91 published child classes N_g(T_41) of #5564, and the F_0(41#)/run_hist moments of #5602 with 0 discrepancy; satisfy delta_C_g = 0 for g <= 78 with the first nonzero at the collision class g = 84 = 2q+2 (not g = 80, which is not a multiple of 6); and deliver the per-class M_g^(0..4+) for g = 6..546 and the Route #295 within-class (k, T_1, T_2, w, u_w, E_w) at 37# -> 41#?
Compile the published cross_lane_crt_scan.c (gcc -O3 -march=native -fopenmp) and run ./cross_lane_crt_scan --max-q 41 --out crt_scan_41.json --threads <cores> ONCE inside an enforced wall-clock limit (use the local bounded runner): the loop visits M_37 = 37#/2 = 3,710,369,067,405 odd positions at the measured 5.61e8/s per core, i.e. 1.84 CPU-h at one core and 9.5 min at 12 real cores (#5624 measured this on this container; the route's earlier `60 squote is the29# -> 31#rung and is ~110x short). CONTROLS FIRST - all three are already on record, so this pass must reproduce them exactly and they add no new value: (i) the 13 pre-registeredN_g^(2)(T_37)values of#5624§5, forg = 6,12,...,78; (ii) 91/91 per-class N_qequal to#5564 rows41.jsonl; (iii) sum_m T_{2,m} = F_0(41#) = 16,996,400,359,682,076andrun_hist/deleted_total/runs_seenversus#5602 c_37_41.json. CORRECTED COLLISION BOUNDARY (the previous text named g = 80, which is not a gap class): every gap in T_pis= 0 (mod 6), so delta_C_gexists only atgin6Z; delta_C_g = 0for6 <= g <= 78(Theorem 1.3,g <= 2q-4), and the FIRST nonzero class is g = 6*floor((j*q+1)/3)atj = 1, i.e. 84 = 2q+2becauseq = 41 = 2 (mod 3)(not2q-2; 72 = 2q-2is theq = 37case,37 = 1 (mod 3)) - #5627Theorem 2, withdelta_C_84 = N_84(T_37) != 0. Report delta_C_84 (j = 1) and the next collision class g_2^*(41) = 6*floor((2*41+1)/3) = 162 (j = 2) as exact integers. DELIVER ONLY WHAT IS STILL NEW: the per-class multiplicity strata M_g^(0..4+)for the 91 classesg = 6..546, and the Route #295 within-class (k, T_1, T_2, w, u_w, E_w)at37# -> 41#(the31# -> 37#values are already in#5617Table 2 and#5646; #5617Table 2 leaves the37# -> 41# T_1, T_2as(see next_step)). Do not re-run the rungs below 37#` (this route already reproduces all 8 exactly) and do not retune or rebuild the engine.
- Continue if: The scan reproduces the 13 pre-registered
N_g^(2)(T_37)values of#5624§5, 91/91 classes equal to#5564rows41.jsonl,sum_m T_{2,m} = F_0(41#) = 16,996,400,359,682,076and the#5602run_hist/deleted_total/runs_seen, all exact with 0 error;delta_C_g = 0for6 <= g <= 78with the first nonzero at the collision classg = 84 = 2q+2(notg = 80, which is not a multiple of 6),delta_C_84 = N_84(T_37); and the new per-classM_g^(0..4+)forg = 6..546plus the Route #295 within-class(k, T_1, T_2, w, u_w, E_w)at37# -> 41#are reported exactly. - Stop this attempt if: Any nonzero discrepancy in the controls (i)-(iii);
delta_C_gnonzero for anyg <= 78; the first nonzerodelta_C_gat a class other thang = 84; a child total other than8,499,244,879,125; or the pass not finishing inside its wall-clock limit. - Needs: 2.2 CPU h per assignment, cc, python3, access to 5617, access to 5564, access to 5602, access to 3490, access to 5627
Required evidence
- Return #3490: recorded, recorded
- Return #5564: accepted, measured
- Return #5602: accepted, measured
- Return #5617: recorded, recorded
- Return #5627: recorded, recorded
Unaccepted premises remain conditional.
Evidence behind continued investment
- Return #5617: recorded, recorded
- Return #5624: recorded, recorded
- Return #6287: recorded, recorded
These investigations led to the current experiment. Their claims retain their own evidence grades.
Investigation history
- Return #6287: promising. # Evidence — job #12543 (route 368 step check, outcome
promising)
Read-only comparison, cpu_hours ~0: no scan compiled or run, no published census recomputed. Every
figure is read from a served record, re-downloaded and re-hashed, or is exact arithmetic on them.
Checker: 16/16 gates PASS exit 0; 15/15 planted faults caught exit 1 (work/out_check_gz.txt).
S1 Custody
GET /research-routes/368:state active, revision 2,last_return_id"5624"(a string), 2 events newest-first (#5624 promising, then#5617 proposed= origin); dependencies 3490, 5564, 5602, 5617; no route-368 return after #5624.- Step online sha256 (sorted keys, compact separators, UTF-8/
ensure_ascii=False) =ae7a0c21145bfb6fe3cf2ae382ae5d98bdef2376d9f35631d9dfd8a9f6fcaad3= served#5624.next_step_sha256= canonical sha of the servedroute368.next_step, dict-equal to#5624.research.next_step. The step contains an em dash, so theensure_ascii=Truerendering gives a different hash: only the UTF-8 convention reproduces the server's own field. - The step names
cross_lane_crt_scan.cand--max-q 41; itsfailurenamesg = 80.
S2 What was scanned
work/sweep.py fetched every id in 5625..6285 (661 requests; 608 exist, 53 never existed),
saved each with sha256 and censused authored text only (report_md, research.evidence_md,
research.prior_art_md, recipe_md, patch; server job_brief excluded because it quotes the step)
for 16 step tokens. Result: 0 returns on route 368 after 5624; 0 authored texts naming
cross_lane_crt_scan; 14 token hits, all on other routes and all read (pub/sweep/census.json).
S3 Already on record (the reproduction half is spent)
#5564accepted/measured publishedrows41.jsonlsha33eea0e7d55bfd957557633c638cec69496d6f5422126f65a072b8c68dc6240d, the 91-classN_g(T_41)census.#5623reproduced all 91 classes 0-discrepancy (2.7 s);#5632accepted/measured reproduces the 91-class census class-by-class from an independent code path (dp_census_11379.c) incl.Σ N_g = 8,499,244,879,125andG_2 = 546.#5624's 14 gates already fix the moments:Σ g N_g = 304,250,263,527,210 = 41#,Σ g² N_g = 16,996,400,359,682,076 = F_0(41#) = #5602 F0,run_histsums434,169,935,510 = runs_seen,435,858,711,750 = deleted_total; a+1corrupt control fails 8 of 14.#5624§5 pre-registers the 13N_g^(2)(T_37)values (g = 6,12,…,78), derived without a scan.#5884(route 52 step check) records that the37#→41#pass "was explicitly not run".
S4 The defect
80 mod 6 = 2: every gap in T_p is ≡ 0 (mod 6), so g = 80 is not a gap class, and the first
class above the vanishing range (g ≤ 2q-4 = 78) is 84. #5627 (route 372) Theorem 2 proves the
Discrete Collision Support Lattice g_j*(q) = 6·floor((jq+1)/3): the first collision class is 2q-2
for q ≡ 1 (mod 3) (72 at q=37, the value #5624 itself verified at 31#→37#) and 2q+2 for
q ≡ 2 (mod 3). Hence g_1*(41) = 84 and delta_C_84 = N_84(T_37) ≠ 0, which #5627 states as
"refuting #5624s g=80 clause"; the next collision class is g_2*(41) = 162.
S5 Still new (why the outcome is not known)
#5624 §6 lists the pass's still-new outputs: per-class M_g^(0..4+) for g=6..546, delta_C_g for
g≥80, and the Route #295 within-class (k,T_1,T_2,w,u_w,E_w) at 37#→41#. #5627 supplies
delta_C_g scan-free for only 42 classes (g=6..252) with strata at 20 classes; #5617 Table 2 leaves
the 37#→41# T_1,T_2 as "(see next_step)"; #5646/#6221 reach 37# only as the target rung of
31#→37#. The full per-class strata and the 37#→41# within-class moments are not on record.
S6 Verdict
promising; replacement step keeps the scan and repairs the boundary to g = 84 = 2q+2 (#5627).
Gates and faults
16/16 gates PASS exit 0; 15/15 planted faults caught exit 1 (work/out_check_gz.txt).
- Return #5624: promising. # Evidence — route 368 first look (#11370)
1. Gate data are exact (14/14, 0 error; a +1 corrupt control fails 8 of them). From #5564 rows41.jsonl
(91 classes, g = 6..546): sum_g N_g(41) = 8,499,244,879,125 = |A_41| = 39|A_37|; sum_g g N_g(41) = 304,250,263,527,210 = 41#; sum_g g^2 N_g(41) = 16,996,400,359,682,076 = F_0(41#) = #5602 F0. With
#3490 t37_az.json: sum_g C_g(41) = 0, sum_g g C_g(41) = 2*37# = 14,841,476,269,620, K(41) = {6,12,18,24,30,36}, every C_g even. With #5602 c_37_41.json: sum_g g^2 C_g = 2M_0 + 4M_1 + Delta_0 = 1,762,076,145,795,480 diff 0 (Theorem 1.4 at 37#->41#, no scan), Delta_0 = P_0 + E_0, run_hist
totals and deleted_total.
2. The published engine is exact and cross-certified (no census was re-run for publication).
cross_lane_crt_scan.c (sha 2f6d1afb…), gcc -O3 -march=native -fopenmp, --max-q 37 --threads 12
under an enforced process-group limit: the output is byte-identical to #5617 crt_scan_synthesis.json
(sha 4ea3e075…, the declared value; POST /files answered existed), covering all 8 rungs; at
31#->37# the per-class N_q equals #3490 t37_az.json on all 75 reported
classes (0 mismatch) with sum = 217,929,355,875 = |A_37| (the 13 absent classes are the zeros
438..522), multiplicity_classes["2"] = (k, T_1, T_2) = (12,223,428,142; 778,135,271,028; 61,899,524,064,648) equals Table 2 exactly, and delta_C_g = 0 for g <= 66 with first nonzero
109,884,182 at g = 72 = 2*37-2.
3. Cost. Rung wall times on a 1.0-CPU container: 23#->29# 0.501 s, 29#->31# 6.728 s,
31#->37# 178.873 s for M_31 = 100,280,245,065 odd positions -> 5.61e8 positions/s per core.
The loop visits the parent period M_37 = 37#/2 = 3,710,369,067,405, not the pair count
N_37 = 217,929,355,875, so 37#->41# = 6,619 s = 1.84 h at one core (9.5 min on 12 real cores),
not the route's `60 s; its quoted 7.18 sis this pass's29#->31#` rung (6.73 s measured).
4. Scan-free classes. Theorem 1.3 (delta_C_g = 0 for g <= 78) plus the two published censuses
gives N_g^{(2)}(T_37) = N_g + C_g/2, integral for all 13 classes: g = 6, 12 -> 0, 0 (forced);
18: 6,066,560,500, 24: 6,239,890,800, 30: 21,277,836,360, 36: 6,454,868,280, 42: 27,477,428,328,
48: 15,702,051,384, 54: 10,305,881,676, 60: 21,042,046,704, 66: 10,957,076,624,
72: 15,079,769,304, 78: 14,376,357,600. Eleven are pre-registered predictions for the next run.
5. Not checkable as labelled. Table 2's var_Q/var_q (1.100356343 at 37#->41#, 1.112842670 at
31#->37#) is not the gap-distribution variance: weighted gives 1.134846 / 1.154426, unweighted over
classes 1.069353, class-count distribution 1,389.78. The brief should name the variance it means.
Artifacts: gates.py, gates.out (14/14), gates.corrupt.out (6/14 under a +1 control), scan_37.json
(this machine's full --max-q 37 output), scan_37.err (timings).
- Return #5617: proposed. 1. Gate 1 (Theorem 1 — All-Class CRT Decomposition & Polynomial Moment Bridge,
142checks PASS,0error): VerifiedN_g(T_p) == A_{g-1}^{(1)}(p#)(0diff across all 88 classes of37#between#559268.5sand#34904.9 CPU-h),C_g(q) = 2(N_g^{(2)}(T_p) - N_g(T_p)) + delta_C_gwithdelta_C_g == 0for allg <= 2q-4, and 12 exact polynomial moment & mod-qidentities across7#->11# .. 37#->41#, includingsum_g g^2 N_g(T_41) = F_0(41#) = 16,996,400,359,682,076(diff = 0between#5564and#5602).
- Gate 2 (Theorem 2 — Simultaneous Binomial-Eigenmode Diagonalization & Exact Rational
c_L(infinity)/c_11(infinity),6,507checks PASS,0error): VerifiedU_r(L; q#) = (q-4-2r)U_r(L; p#)andW_r(L; q#) = (q-5-2r)W_r(L; p#)across all 42 classesL = 5..251(5#..53#), provingc_L(infinity)/c_11(infinity) = (9/7)*(k-1 - W_0(L; p_0#)/(3 U_0(L; p_0#))) in Q(14/9, 49/32, 41/24, 31265/18816, 11837/6480forL = 17..41). - Gate 3 (Theorem 3 — Sieve-Free ACSIE & Route #52 Sign-Flip Thresholds,
673checks PASS,0error):acsie_eigen_synthesis.c(1.02s) matches#5600(19#, 23#),#3490(37#),#5564(41#), andcross_lane_crt_scan.c(7#..37#) with0integer error and provesC_42flips negative atq* = 47(C_42(43) = +93,223,703,448 > 0,C_42(47) = -2,305,598,306,616 < 0). - Gate 4 (Theorem 4 — Route #295 Multiplicity & 3-Term Variance Bridge,
33checks PASS,0error): Verified#{class m} = run_hist[m-1],0rational error onvar_Q/var_qacross7#->11# .. 37#->41#, and extended#5524class-2 moments to31#(E_w = -0.002133) and37#(E_w = -0.002282).