Fabian Arevalo

Chowla's conjecture · analytic number theory

The width of the parity barrier

Exact descent identities for Liouville pair correlations · a certified census to x = 1012 · every sign pattern of length ≤ 33 occurs, independently audited by a clean-room reimplementation · the measured gap between truth and theorem.
July 29 – August 1, 2026 · two research sessions · code, data and certificates in the Matematica repository (conjectures/chowla).

Start here · no maths required

Take any number and break it into prime factors. Count the pieces. 12 = 2·2·3 is three pieces (odd). 15 = 3·5 is two (even). Write down H for even, T for odd, and do this for 1, 2, 3, 4, … forever. You get an endless string of coin flips: T T T H T H T T H H…

Every test anyone has ever run says this string behaves exactly like a fair coin with no memory: knowing today's flip tells you nothing about tomorrow's. That innocent-sounding claim is Chowla's conjecture, and it is one of the deepest unsolved problems in mathematics. Proving even the weakest version, that a flip and its immediate neighbour agree only half the time in the long run, has resisted everyone, because every known technique loses exactly this kind of "parity" information. The obstruction has a name: the parity barrier. It is the same wall that stands between us and the twin prime conjecture.

The figure below is the coin's memory, measured further than it has been measured before. Each curve tracks how much the string agrees with a shifted copy of itself, in units where a fair coin would wander between roughly −2 and +2. Nearly five orders of magnitude in x, no drift.

Shifts
hover the plot to read exact values
Measurement, not model. The curves are Sh(x)/√x for shifts h = 1…8: certified integer sums Sh(x) = Σn≤x λ(n)λ(n+h), plotted from the actual census grid (decimated for display; every underlying value is exact). The dashed band at ±2 is the only model element: where a memoryless ±1 coin would spend ~95% of its time. The conjecture says each curve, divided by a growing √x, must eventually hug zero; what is actually provable says almost nothing at these scales (§7).

What this note adds

The unaveraged conjecture is open for every shift; the modern results (Tao 2016, Helfgott–Radziwiłł 2021, Pilatte 2023/25) control only a logarithmically averaged version. This note contributes, from one session:

  1. An exact self-similarity. The correlation at an even shift literally contains, inside it, the half-scale correlation at half the shift: an identity, not an estimate. You can verify it live in §2's figure, and watch it collapse for a shuffled coin. It reduces the whole conjecture to its odd-shift core, and it forces a cross-scale echo of exactly 1/√2 = 0.707…, which the data reproduces as +0.7085 ± 0.0046 (§6).
  2. A certified census. Correlations, triple and quadruple correlations, and all 256 sign-pattern counts on a grid to x = 1012, every row certified by exact integer identities, and the pipeline reproducing every published number it can be tested against, including the famous first failure of Pólya's conjecture at n = 906,150,257 and the values of L(10k) (§3).
  3. Every pattern shows up on schedule. All 233 = 8,589,934,592 heads/tails patterns of length 33 occur, and the time for the last one to arrive tracks the fair-coin "coupon collector" formula across 33 doublings, with scatter as wide as that formula's own randomness predicts. The sequence of completion times is new to the OEIS (§4). For context: mathematics can currently prove such things only for patterns of length 3.
  4. The barrier, measured. At x = 1012 the true correlation is about 1.6·10−7 of its trivial maximum; the strongest theorem ever proved about it guarantees only that it is below ~0.9. A factor of five million of cancellation that nobody can prove (§7).
  5. An audit. A 2023 preprint claims the natural-density equidistribution unconditionally; §8 locates the invalid step, an average-transfer argument with an explicit counterexample, so the record stays straight: the conjecture is open.
  6. A second, independent look. A follow-up session rebuilt the counting engine from zero — different code, different machine, no shared libraries — and it reproduced all thirty of the first session's checkpoints exactly, then pushed three further: every pattern up to length 33 now shows up on schedule. It also ran a sharper test, comparing not just the slowest pattern but all sixteen million patterns of length 24 against a fair coin's exact prediction for when each should first appear. Same verdict: a coin (§10).

Everything below is the technical version. The fastest route through it: the identity figure in §2, then §7.

AbstractSummary of results

Let λ be the Liouville function and Sh(x) = Σn≤x λ(n)λ(n+h). Chowla's conjecture asserts Sh(x) = o(x) for every h ≥ 1; it is open for every h. We prove elementary but organizing exact identities: a 2-adic descent Sh(x) = Uh(x) + Sh/2(⌊x/2⌋) (h even, Uh the odd-n restriction), which exhibits the pair-correlation field as an exact fixed point of a renormalization map with odd-restricted correlations as innovations, and a Walsh–Hadamard duality between the k-window sign-pattern census and multipoint correlations (PROVED). We compute a certified integer census of Sh (h ≤ 32), Uh, the triples {0,1,2}, {0,2,4}, {0,1,3}, the quadruple {0,1,2,3}, the full 256-cell 8-window census, and harmonic sums ℓh(x) = Σ λ(n)λ(n+h)/n, on a grid of spacing 2·107 to x = 1012, every row passing the descent and duality certificates exactly, with external anchors reproduced (L(10k) per OEIS A090410 for k = 9…12; the Pólya island 906,150,257…906,488,079 with max L = 829 at 906,316,571; BFM's 252 zeros and 133 sign changes of L below 109) (CERTIFIED). Every ±1 pattern of length ≤ 27 occurs; completion points Nk track the coupon-collector model across 27 doublings (mean ratio 1.007 for k ≥ 10, scatter matching the model's own Gumbel width) and are new to the OEIS (CERTIFIED by exhibition). Six predictions registered before the large computation ran resolve to 14 scored clauses: 10 hits and 4 misses, all reported, including a parameter-free cross-scale increment correlation 1/√2 forced by descent plus innovation-independence, measured +0.7085 ± 0.0046 across 1012 (NUMERICAL). At x = 1012, |S1|/x ≈ 1.6·10−7 while the best proven almost-all-scales bound (Pilatte 2025, Remark 2.8) is numerically ≥ 0.9: the parity barrier is six to seven orders of magnitude wide at real scales. Finally we audit arXiv:2211.09736, whose claimed unconditional equidistribution rests on transferring logarithmic-average control to natural averages, an implication refuted by f(n) = niα. A second pre-registered instrument set went 10-for-10: a certified 4-point census to 1011 (first data at an order where even averaged Chowla is open; all twelve quadruples square-root-sized, quad echo +0.721 ± 0.013), the full covariance web of the descent structure matching the fair-coin model in every entry, and coverage extended to every pattern of length ≤ 30 (N30 = 22,249,147,014). A second, independent session (§10) then reimplemented the coverage engine from scratch — different language, different machine, no shared code — and reproduced all thirty Nk values and last-completing pattern codes exactly, together with L(10j) for j = 1..8; coverage was then extended to every pattern of length ≤ 33 (N31, N32, N33 = 43,901,697,682, 99,494,377,311, 196,202,853,829, each CERTIFIED by exhibition and independently re-verified by trial division). The same session introduced the first-occurrence spectrum, scoring every one of the 224 sign patterns of length 24 against Conway's exact fair-coin waiting time: mean R = 1.000123 against 32 decorrelated control streams at 0.999953 ± 0.000319 (+0.53 control-σ), indistinguishable from a fair coin at a resolution of about 3·10−4 (NUMERICAL). The conjecture survives, its open core isolated in the odd-restricted correlations.

Section 1The conjecture and where it stands

λ(n) = (−1)Ω(n), where Ω counts prime factors with multiplicity; λ is completely multiplicative with λ(p) = −1 at every prime.

Conjecture · Chowla, two-point case · open

For every fixed h ≥ 1:  Sh(x) = Σn≤x λ(n)λ(n+h) = o(x).

Status of the art (each statement checked against the cited paper's text, not a summary of it):

Section 2Exact structure (PROVED)

Lemma 1 · dilation descent

Let d ≥ 1 and h = d·h′. Then Σn≤x, d|n λ(n)λ(n+h) = Sh′(⌊x/d⌋). In particular, for even h, with Uh(x) := Σn≤x, n odd λ(n)λ(n+h):

Sh(x) = Uh(x) + Sh/2(⌊x/2⌋).

Proof. For n = dm, complete multiplicativity gives λ(dm)λ(dm+dh′) = λ(d)²·λ(m)λ(m+h′) = λ(m)λ(m+h′), and n ↦ m bijects {n ≤ x : d|n} with {m ≤ ⌊x/d⌋}. ∎

The identity is checkable in the strongest possible sense: exactly, integer for integer, at any scale. The figure computes λ live and does so; the shuffle button destroys multiplicativity while keeping the exact same multiset of signs, and the identity collapses. That contrast is the content of the lemma: the self-similarity is carried by multiplicative structure, not by statistics.

…
shift h (even) 2
x 40,000
Lemma 1, executed live. λ(n) is computed in your browser by an exact sieve up to n = 60,000+16; the three sums are then formed directly from the definition; nothing is looked up. For real λ the identity holds exactly at every x and every even h (the residual is 0). Shuffling the λ values (deterministic seed, same multiset of ±1) breaks complete multiplicativity and the residual becomes a random-walk-sized integer.
Proposition 2 · 2-adic reduction

Write h = 2ah₀ with h₀ odd. Then Sh(x) = Σj<a Uh/2^j(⌊x/2j⌋) + Sh₀(⌊x/2a⌋), and if |Ug(y)| ≤ δy for all even g and all y then |Sh(x)| ≤ 2δx + x/2a. Hence the two-point conjecture is equivalent to: Sh = o(x) for odd h, and Uh = o(x) for even h. The odd-restricted correlations are the irreducible content.

Proof. Iterate Lemma 1; bound the tail trivially; sum the geometric series. ∎

Lemma 3 · census–correlation duality

For ε ∈ {±1}k let Nε(x) count n ≤ x whose sign window matches ε, and let RS(x) = Σn≤x Πi∈S λ(n+i). Then Nε(x) = 2−k ΣS⊆[k] (Πi∈S εi) RS(x), and RS(x) = Σε (Πi∈S εi) Nε(x).

Proof. Expand Πi(1+εiλ(n+i))/2, the indicator of the window; invert by Walsh–Hadamard orthogonality. ∎

Proposition 4 · a maximal near-miss

Let f be completely multiplicative with f(3) = −1 and f(p) = χ(p) for p ≠ 3, χ the quadratic character mod 3. Then Σn≤x f(n)f(n+1) = −x/3 + O(log x) and Σn≤x f(n)f(n+3) = (5/9)x + O(log x).

Proof. Exactly one of n, n+1 is divisible by 3. If neither would be (n ≡ 1 mod 3), f(n)f(n+1) = χ(1)χ(2) = −1: density 1/3, contribution −x/3. If 3 | n, write n = 3au: the factor (−1)aχ(u) summed over u ≤ x/3a is O(1) per level by the bounded character sum, O(log x) in all; symmetrically for 3 | n+1. For h = 3: the coprime class contributes χ(n)² = 1 at density 2/3, and the class 3|n reduces by Lemma 1's mechanism (with d = 3) to the h = 1 sum at scale x/3: (2/3) − (1/3)(1/3) = 5/9. ∎

Measured at 107: −0.3333334 and +0.5555556. A completely multiplicative ±1 function can fail Chowla maximally; the conjecture is precisely the assertion that λ is far from every such character-like ("pretentious") function. This is the known structural obstruction (Frantzikinakis–Host; MRT), rediscovered here from the computation before the literature was consulted; recorded as validation, not contribution. Note the failure constants themselves obey the descent recursion s₃ = 2/3 + s₁/3.

Section 3The certified census (CERTIFIED)

Engine: an exact segmented sieve for λ: for each n the parity of small-prime-power divisors plus an exact integer product of the small-prime part, which detects the single possible prime factor > √x. No floating point decides any certified value: correlation sums are float64 BLAS dot products of ±1 vectors in which every intermediate is an integer < 253, so every IEEE-754 addition is exact. Validated against brute-force trial division (in full on [1, 2·105]; windows near 109, 3·1010, 1012, 240) and against an algorithmically independent smallest-prime-factor sieve (all 310 columns equal at 106); two independent kernels agree on all rows ≤ 109.

Certificates, checked at every grid row:

External anchors, the published numbers the pipeline must and does reproduce:

anchorpublishedthis census
L(109) … L(1012) (OEIS A090410) −25216, −116026, −342224, −522626equal, exactly
first n with L(n) > 0 (Tanaka 1980)906,150,257906,150,257
max L below 109 (Tanaka)829 at 906,316,571829 at 906,316,571
zeros / sign changes of L below 109 (BFM 2008)252 / 133252 / 133
Turán's Σλ(k)/k > 0 up to 7.2·1013 (BFM)positivepositive at every grid row

BFM's L-computation to 2·1014 took ~2.5 CPU-years in 2008; this census covers 0.5% of that range but records 300+ statistics per grid row that were never computed at any scale: the pair-correlation and pattern-census columns are new measurements.

One more thing the certificates did: across the two long runs they caught and localized ten transient memory-corruption events (single zeroed pages, statistically invisible; one detectable only by the descent identity C1). Each was repaired from a deterministic recompute, raw pre-repair files are kept in the repository, and the two independently corrupted, independently repaired runs agree bit for bit on all 5000 overlapping rows, while L(109)…L(1012) match the published values exactly. Details: VERDICTS.md.

Section 4Results

4.1 Size of the correlations

Certified values at decades (h = 1…8; each cell an exact integer, shown with Sh/√x underneath):

xS₁S₂S₃S₄S₅S₆S₇S₈
109−46,68243,79038,86634,01616,03224,6604,132−8,508
1010−71,450−118,58618,508153,92283,0421,176132,516137,364
1011−219,878153,142494,21065,362−62,982479,064142,722−203,530
1012−163,27457,386−604,438600,878−407,7941,672,602−996,980−299,864

At 1012, maxh≤32 |Sh|/√x = 2.149 (h = 13) and the log-log fit of RMSh Sh against x over [108, 1012] gives exponent α = 0.467, against 0.436 on the range ending at 1011: the estimate moves toward ½ as range is added, exactly as the single-path-noise reading predicts, though it stays marginally below the registered band (§5). The fluctuation level is square-root-sized across nearly five decades of x. NUMERICAL interpretation of CERTIFIED integers; the fit describes this range only and no asymptotic guarantee follows.

4.2 Sign-pattern coverage

Every ±1 pattern of length k ≤ 33 occurs (k ≤ 30 from this session; k = 31, 32, 33 from the independent session-2 audit, §10). Completion points Nk (least x by which all 2k patterns have appeared; exact first occurrences on file):

hover a point
Points are measurements, the curve is a zero-parameter model. Nk for k = 1…33 (certified by exhaustive scan; every completing position for k ≥ 25 independently recomputed, k ≤ 30 in this session and k = 31–33 by the clean-room reimplementation of §10) against the uniform coupon-collector expectation 2k(k ln 2 + γ). No fitting anywhere: the curve is drawn from the closed form. Over k ≥ 10 the measured/model ratio has mean 1.004 and sd 0.11, scatter matching the model's own Gumbel width (hover to read each ratio; §10.2 reads the same data as a Gumbel z-score instead). Occurrence of all patterns is proved only for k ≤ 3 (MRT 2015); these are, to our knowledge, the first certified occurrence bounds at these lengths, and the sequence Nk is not in the OEIS (checked 2026-07-29; not re-checked in session 2, whose sandbox had no network access — see §10.4).

At 1012 the 256 window counts deviate from x/256 by at most 2.98 standard deviations (4.21 at the 1011 cap; Gaussian-max expectation ≈ 3.5): no pattern is favoured beyond fair-coin fluctuation.

4.3 Harmonic pair sums

ℓh(x) = Σn≤x λ(n)λ(n+h)/n is what the modern theorems actually bound (trivial bound: log x ≈ 27.6 at 1012). Measured values are O(1) and frozen by small n: between 109 and 1012 no ℓh (h ≤ 32) moved by more than 9.63·10−5 (expected maximum over 32 shifts under the model: 9.2·10−5), against a fair-coin tail σ of 3.2·10−5. ℓ1(1012) = −0.842488. Convergence of the series is itself open (strictly stronger than logarithmic Chowla, strictly weaker than power-saving Chowla), and we propose it as a benchmark problem (§7). These columns are float64 with a rigorous < 10−8 rounding bound (every other column is exact).

4.4 A known drift, reproduced

The log-weighted mean of L(x)/√x over the fine grid is −0.671, against the conjectured density-mean 1/ζ(1/2) ≈ −0.6848 from the pole of ζ(2s)/ζ(s): agreement to 2%. Illustration only (NUMERICAL).

Section 5Out-of-sample predictions

Six predictions were written to disk (PREDICTIONS.md, timestamped) when the census had only reached 109, each naming its generating model; the large run then ran through the targets. Verdicts:

prediction (model)registeredmeasuredverdict
RMSh Sh/√x at 1010, 1011, 1012 (fair coin)0.90 ± 0.35 / ± 0.401.155 / 0.794 / 0.898hit / hit / hit
exponent α on [108, xcap][0.47, 0.53]0.436miss
cross-scale corr, pooled (descent + independence)[0.68, 0.73]+0.688 ± 0.017hit
per-h corridor ±0.05 of 0.70716 of 6 inside4 of 6 insidemiss
census max|z| (Gumbel over 256 cells)[2.7, 4.3]4.21hit
L(1010)…L(1012) (A090410); L<0; hL>0exact / all rowsexact / all rowshit
hL(1012) band (from BFM local minima)[5·10−8, 6·10−7]+8.23·10−7miss (band calibrated low)
N25 (coupon collector, 2σ)(6.0 ± 0.9)·1087.228·108 (ratio 1.203)miss (~2.6% tail)
N26, N27(1.25 ± 0.17)·109 / (2.6 ± 0.34)·1091.313·109 / 2.795·109hit / hit
harmonic freeze 109 → 1012 (±2·10−4, model σ 3.2·10−5)max drift in band9.63·10−5 (expected max-of-32: 9.2·10−5)hit

Scoring happened twice: at the pre-registered 1011 cap, and again when the full 1012 run resolved the remaining horizon clauses. Final ledger: 14 scored clauses, 10 hits, 4 misses; misses reported exactly as registered (full detail: VERDICTS.md in the repository).

The point of the protocol is not the hit rate: it is that the models named in advance (fair-coin scaling, coupon collector, descent + independence, published anchors) are exactly the models the data keeps electing.

Section 6The cross-scale echo (NUMERICAL)

Lemma 1 forces block increments to satisfy ΔS2h[2y, 2y+2Δ) = ΔU2h[2y, 2y+2Δ) + ΔSh[y, y+Δ): every fluctuation of Sh at scale y is replayed, exactly, inside S2h at scale 2y. If the innovation ΔU is uncorrelated with ΔSh and carries half the variance (the fair-coin hypothesis on the odd subsequence only), the correlation of the coupled increments is exactly 1/√2 = 0.7071, with no free parameter.

predicted / measured
corr( ΔS2h at (2y, 2Δ), ΔSh at (y, Δ) ) → 1/√2
measured, pooled over h ∈ {1,2,3,4,6,8}, disjoint blocks: +0.7085 ± 0.0046 (1247 block pairs per h at 1012; +0.688 ± 0.017 at the pre-registered 1011 scale)

The identity contributes the coupling; the data contributes the 1/√2. A conspiracy in the odd innovations would move this number; with ten times the data it instead converged to within one standard error of the parameter-free value. To our knowledge this 2-adic self-similarity of the Liouville correlation field has not been measured before.

Section 7The measured width of the parity barrier

One measured curve, two theoretical shapes. Solid: measured RMSh Sh(x)/x from the census (certified integers; decimated for display). Dashed: the fair-coin model x−1/2, a model drawn from its formula and visibly the same line. Dotted: the shape of the strongest proven bound, (log x)−c/2, drawn at c = 0.06 and, absurdly generously, at c = 1; Pilatte's c is inexplicit and structurally small. The vertical gap at the right edge is the parity barrier: six to seven orders of magnitude between what is true and what is provable.

At x = 1012: |S1(x)|/x = 1.6·10−7. For the proven shape to reach that value at this x, its exponent would need c ≈ 9.4; the method that produced the theorem caps c at a small constant, and its author states the power-of-log saving is the limit of current techniques. Nothing in this census can close that gap, but it fixes its width, at real scales, for the first time we know of.

Three open problems, in increasing strength, each strictly between what is proved and what is believed:

  1. Effectivity. Exhibit any explicit c > 0 in the (log x)1−c theorem.
  2. Harmonic convergence. Prove Σ λ(n)λ(n+h)/n converges for a single h.
  3. Odd core. Prove Uh(x) = o(x) for a single even h: by Proposition 2 this is the irreducible obstruction at even shifts.

Section 8An audit: the claimed equidistribution of arXiv:2211.09736

A 2022/23 preprint (Carella, math.GM) claims unconditionally that the four double sign patterns of λ have natural density 1/4, equivalent to unaveraged two-point Chowla, which Pilatte (Dec 2025) lists as open. Reading the preprint in full: the claim reduces (its eq. 3.9) to Σn≤x λ(n)λ(n+t) = O(x/(log log x)1/2−ε) (its Thm 6.4), obtained from Helfgott–Radziwiłł's logarithmically averaged bound via its Lemmas 6.1–6.2, which assert: if Σn≤x f(n)/n = o(log x) then Σn≤x f(n) = o(x).

That implication is false for bounded sequences. With f(n) = niα (α ≠ 0): Σ f(n)/n = O(1), yet |Σn≤x f(n)|/x → 1/|1+iα| ≠ 0. The proof's contradiction step needs its assumed lower bound B(z) ≥ z(log log z)−1/2+ε pointwise across the whole integration range, while the negation of the conclusion supplies it only along a sequence of scales, with uncontrolled sign; ∫B(z)/z² dz can cancel between scales. The preprint's Theorems 1.1, 1.2, 6.4 and 6.6 are therefore unproven. (Our census shows the claim itself is empirically on track: the flaw is in the proof, not the belief. The audit keeps the record straight: the conjecture remains open.)

Section 9Reproducibility

All code in conjectures/chowla/ of the Matematica repository; scripts run from that directory; no randomness in any certified path. Machine: Apple M3 Pro (6P+6E cores, 36 GB), Python 3.12, NumPy 2.3.5. Runtimes: fine run 231 s; harmonic baseline 267 s; first main run 2752 s wall to the 1011 cap; full 1012 run 23,362 s wall on 10 workers; coverage extension 978 s.

artifactcontents
liouville.py + test_liouville.pyexact sieve; brute-force validation
census_run.py / certify.pythe census; exact certificate checks C1–C3
coverage_ext.py / signchange.pypattern coverage k ≤ 27; L sign-change counts
PREDICTIONS.md / VERDICTS.mdthe out-of-sample protocol, timestamped
data/*_grid.csv, *_firstocc.npz, *_meta.jsoncertified grids; first occurrences; runtimes

Section 10Session 2 (2026-08-01): independent audit, and the first-occurrence spectrum

A second, unrelated session returned to this conjecture three days later, in a sandbox with four cores, 15 GB of RAM, and no outbound network. It also had no NumPy, SciPy or networkx — every script in §§1–9 imports NumPy, so none of the results above could be rerun on that machine at all. That is a reproducibility defect worth stating plainly rather than working around silently: a certified computation whose reproduction depends on a package set a clean machine may not have is less reproducible than it looks. The response was to rebuild the coverage half of the pipeline from scratch in C, sharing no code with the original — which turns the obstacle into the more useful thing, an independent audit.

10.1 Clean-room reproduction (CERTIFIED)

lambda_coverage.c is an exact segmented sieve for λ, written independently of the NumPy pipeline: Ω(n) mod 2 is tracked by flipping a parity bit on the multiples of every prime power pe ≤ x, with a per-segment residue that carries the one possible prime factor above √x, divided out exactly by multiplication with p−1 mod 264 rather than by hardware division. No floating point anywhere in the critical path.

Pointed at the previous session's certified output, it reproduces all thirty values of Nk and all thirty last-completing pattern codes, k = 1…30, across three separate files (fineA_coverage.csv, coverage_ext.csv, coverage_2830.csv), plus L(10j) for j = 1…8 (0, −2, −14, −94, −288, −530, −842, −3884) — exactly, integer for integer. A third, independent implementation (verify_coverage.py, pure Python standard library, trial division) confirms Nk for k = 1…14. Reproduction log: data/repro_k25_30.csv.

10.2 Coverage extended to k = 33 (CERTIFIED by exhibition)

One run, 4 cores, 7387 s wall, peak RSS ≈ 2.5 GB, tracking k = 31, 32, 33 simultaneously against a single λ stream:

kNklast-completing codeGumbel z
3143,901,697,6821,784,492,180−1.26
3299,494,377,3112,930,773,200+0.32
33196,202,853,8293,712,643,644−0.48

Each of the three completing windows was independently re-verified by pure-Python trial division, along with 10, 12 and 12 further endgame windows; data/endgame_3133.txt ships the last 256 completions per k, every one an independently checkable exhibition. As in §4.2, "certified" means exhibited and re-verified, not proved minimal — see §10.4.

hover a bar
The same data as §4.2's figure, read as a standardized residual instead of a ratio. zk = (Nk/2k − k ln 2 − γ)/(π/√6), the coupon-collector model's own z-score (model mean 0, sd 1, dashed band at ±1). k = 16…33 (below k = 16 the asymptotic model is not meaningful). Mean z = +0.24, least-squares slope −0.02 per unit k: no record is an outlier and there is no drift toward or away from the model as k grows — the newest records (k = 31…33) least of all.

The unseen-pattern count tracked the model 2ke−x/2^k to three significant figures the whole way down: at x = 1011 the k = 33 tracker had 75,925 patterns left against a predicted 75,600. Overlapping windows are not literally uniform coupon draws — periodic patterns wait longer, by Conway's formula (§10.3) — but the excess probability that a low-period pattern is what's unseen at the collector scale is below 0.3% at k = 32 and 33 (coverage_model.py), so the plain coupon-collector model is the right one to read Nk against to well under a percent.

10.3 The first-occurrence spectrum (NUMERICAL)

Nk is one number: the last pattern to appear. The --firstocc mode instead records the first occurrence of every one of the 2k patterns — 16,777,216 data points at k = 24 instead of one — and each has an exact, parameter-free prediction under the fair-coin null. Conway's leading number gives the expected waiting time to the first occurrence of a word w as A(w) = Σj δj(w)·2j, where δj = 1 iff w's length-j prefix equals its length-j suffix. With s(w) the first-occurrence start index, the statistic

model
R(w) = (s(w) + k − 1) / A(w)     E[R(w)] = 1 for every w

has model mean exactly 1 for every individual pattern, not just on average. A(w) itself ranges over a factor of two — 2k for a pattern with no self-overlap, up to 2k+1−2 for a constant pattern — and that clumping is a property of the word, not of λ; dividing it out is what leaves room to see anything else. Measured directly at k = 24 (all 16,777,216 patterns): mean R = 1.000123, sd R = 1.000251 — matching the Exp(1) shape an i.i.d. fair coin predicts, to four digits.

The 2k values of s(w) come from a single stream and are not independent, so σ/√2k is the wrong error bar, by a large factor. Every ± below is instead the dispersion of 32 independent i.i.d. fair-coin control streams, pushed through the identical window/bitmap/first-occurrence code path.

hover a mark
32 controls define the ruler; λ is one more point on it. Mean R over all 16,777,216 patterns of length 24, one mark per stream (data/firstocc_k24_controls.csv); λ is the filled marker. Axis zoomed to [0.9990, 1.0010]. λ = 1.000123; controls 0.999953 ± 0.000319 — λ sits at +0.53 control-σ, rank 23 of 33, inside the pack.
hover a point
Band is the control ensemble's own spread, not a fit. Mean R resolved by popcount (number of −1's among the 24 entries), popcount 4…20 only — outside that range a class holds a few hundred patterns or fewer and is pure noise. Shaded band: control mean ± sd over the 32 streams. λ's slope across this range is −2.64·10−4 per unit popcount against a control band of −1.10·10−4 ± 3.53·10−4 (−0.44σ): nothing anomalous.

That popcount slope is not free-floating structure. Across the 32 controls it correlates at r = +0.80 with the stream's own realised one-point bias L(107)/107 — a density excess has to tilt a popcount-resolved waiting time, and to first order the tilt is linear in that bias. Regressing λ's slope through the control-calibrated relation leaves a residual of −0.56 residual-σ. (Fitted coefficient 0.92 ± 0.12 against the crude first-order value 2 — expected, since the relevant density is a weighted average over the whole range of first-occurrence times, not its value at one reference scale; this regression is a calibration, not a second prediction.) The overlap-class breakdown (how many nontrivial self-overlaps w has) is flat at 1.000 for both λ and the controls, which is the check that the Conway normalisation does what it claims; no figure for this — the table in data/firstocc_k24_analysis.txt is the whole content, and there is also none for the endgame logs of §10.2, for the same reason: no sentence to attach to either.

A control that caught its own failure

The first control ensemble was defective, and the control is what exposed it. Streams were seeded sm64(n ⊕ seed); for small seeds, XOR with n only permutes n inside a short block, so every seed produced a stream with the same bit multiset — all 32 "controls" reported L(108) = +16362, to within 2. Ensemble dispersion collapsed, and against that fake error bar λ's popcount slope read as a 4.4σ anomaly — a discovery that existed entirely inside the miscalibration. Reseeded as sm64(sm64(seed·φ) + n), the streams decorrelate (L(108) = −5552, +5176, −2372, +7388, −3638, +1142, …) and the identical λ measurement is −0.44σ. Left in the record on purpose: an ensemble of controls that is secretly one control is exactly the way this instrument would have lied.

10.4 What "certified" does not mean here, and the reproducibility defect

Machine: 4 cores, 15 GB RAM, no other load. The k = 31,32,33 run reached N31 at 1718 s, N32 at 3774 s, N33 at 7387 s (sieve 1701 s, bitmap probes 5677 s); steady state ≈ 390 s per 1010 values of n. Reproduce (no NumPy required):

artifactcontents
lambda_coverage.cexact segmented λ sieve + coverage/first-occurrence census; --prng control stream, --firstocc full spectrum, --lcheck L(10j)
verify_coverage.pyindependent trial-division verifier: Nk from scratch, single-window, endgame batches
firstocc_stats.c / firstocc_analysis.pyConway-normalised R(w); λ vs control ensemble: mean, popcount slope, overlap classes
coverage_model.pycoupon-collector comparison for Nk, self-overlap correction bound, Gumbel z