Fabian Arevalo

Computational research note · Gilbreath's conjecture

The fine structure of the Chase–Hunter–Tao constants

Exact values c4–c6, a two-parameter submask law, and a two-channel lower bound.
July 28, 2026 · updated July 29 with the microscope program: thirteen proved lenses on the primes, an impossibility theorem for statistical proof strategies, and a new open problem.
Produced in Claude Code research sessions · code, data & certificates in the companion repository.

Start here — no maths required

Write down the prime numbers. Underneath each pair, write the gap between them — always as a positive number. Then do the same thing to that row. And again. And again.

Something strange happens. Every single row you make starts with the number 1. Not sometimes. Every time, as far as anyone has ever checked — which is now hundreds of millions of rows deep.

Nobody knows why, and nobody has been able to prove it always will. That is Gilbreath's conjecture, noticed by Norman Gilbreath in 1958 while doodling on a napkin. Press play below and watch it happen.

Row 0 — the primes themselves.
The Gilbreath triangle of the primes. Each row is the absolute differences of the row above it, so every row is one entry shorter. The leading entry is boxed in black: from row 1 down it is always 1, and that unbroken column is the entire conjecture. Odd / even switches to the halved gaps and keeps only their parity; since |a − b| mod 2 is exactly XOR, what you get is Pascal's triangle mod 2 — the Sierpiński gasket. That fractal is the engine behind sections 3 and 5.

What this note adds

The primes are hard, so in July 2026 Chase, Hunter and Tao published a practice version: replace the primes with random numbers and do the same subtracting. Now you can ask a clean question — after i rows, how big is a typical number? Call that average ci. If it shrinks fast enough, rows of ones are inevitable. They could compute the first four of these numbers exactly and guessed the rest shrink like 1/i.

This note does four things.

  1. Three more exact answers. c4, c5 and c6 as exact fractions, each with a machine-checked certificate. The last one is a 153-digit fraction, and computing it required splitting space into 2,097,152 pieces.
  2. The pattern is not smooth — it is binary. ci depends on how i looks written in ones and zeros. Depth 511 and depth 512 sit next door to each other and behave almost five times differently. We give a two-knob formula that predicts this, and it survived a test two octaves beyond the data it was fitted on.
  3. So the 1/i guess cannot be right here. Under that guess the running totals would grow like a logarithm. They grow like a fifth root instead — a completely different shape.
  4. A better proof. The published lower bound follows one path down the triangle. The triangle is symmetric, so there are two. Using both doubles the bound.

Update, July 29 — the microscope

The Tao paper reduced the whole conjecture to one job: show that a few very specific rigid patterns never occur among the prime gaps. Proving that needs a tool that can see individual arrangements of primes inside microscopically thin windows of the number line — a tool nobody has. We started building it, and three more things came out.

  1. Most of the dangerous patterns are now provably impossible. Old-fashioned divisibility does it: a long evenly-spaced run of primes would need a spacing divisible by 2, 3, 5, 7, … at once, which makes it astronomically wide. Thirteen lemmas of this kind kill every dangerous pattern except the ones built from powers of two — and when we checked the surviving bounds against the first half-billion primes, the primes press against them exactly, like a crowd against a fence.
  2. A whole class of proof strategies is provably dead. We built a fake sequence with perfect prime-like statistics that still breaks Gilbreath's pattern infinitely often — one gigantic gap, hidden so rarely it is statistically invisible. So no proof can rest on gap statistics alone, however precise. The figure in section 7 lets you plant the sabotage yourself and watch the pattern break at exactly the predicted row.
  3. One sharp new question marks the frontier. The sabotage needs a gap as big as its position, and real primes provably never do that. So: do good statistics plus modest gaps force Gilbreath's pattern? Either answer would be a publishable result.

Everything below is the technical version. The interactive figures in sections 3, 5 and 7 are the fastest way to see points 2, 4 and 6.

AbstractSummary of results

For the continuous model of Gilbreath's conjecture introduced by Chase, Hunter and Tao (arXiv 2607.08712, July 2026) — iterated absolute differences of i.i.d. Exp(1) data — we compute the expected depth-i entry ci exactly for i = 4, 5, 6, extending CHT's exact values, each with a computer-verified exactness certificate in ℚ. We prove the model's sign arrangement attains the maximal 2i(i+1)/2 chambers. A Monte Carlo to depth 4095, anchored on the exact values, shows that ci is governed by the binary expansion of i through a two-parameter submask law that explains the Lucas-theorem oscillation CHT observed, fits with R² ≈ 0.98 where a pure power law fails entirely, and renders their tentative conjecture ci ≍ 1/i untenable in the accessible range: partial sums grow like n1/5, not log n. On the rigorous side we prove a two-channel sharpening of CHT's key inequality: cn ≥ 2 exp(−Σk<n ck) for n ≥ 2, whence Σi≤n ci ≥ log(2n − 2 + e²).

July 29 update. Against CHT's deterministic criterion (their Theorem 1.6) we prove thirteen covering-congruence lenses on the actual primes: depth-1 zero-blocks are progressions of consecutive primes and die by the primorial bound (length ≤ 1.25 log x); top-row {0,d}-blocks are impossible for every d of odd part > 1, with the bounds attained exactly in the data; a dichotomy confines depth-2 structures to residue-class prisons or second differences that are exact powers of two. The survivors provably lie in the kernel of every covering argument and below the resolution of every sieve. We then prove: (R1) a Cramér-type bound plus a residual pattern axiom — strictly narrower than CHT's hypotheses — implies Gilbreath for all but finitely many rows; (R2) no axiom system of fixed-order gap statistics implies eventual Gilbreath: a sequence exists with perfect fixed-order Cramér statistics whose leads fail infinitely often. Whether fixed-order statistics plus o(n) entries suffice is posed as the sharp open problem (R3).

Section 1Setting

CHT define the Gilbreath array by a(0,j) = aj (i.i.d. Exp(1)) and a(i+1,j) = |a(i,j) − a(i,j+1)|; by stationarity ci := 𝐄 a(i,j) depends only on i. They compute c0 = c1 = 1, c2 = 7/9, c3 = 227/288, prove Σi≤n ci ≥ log(n+e) and the key inequality cn ≥ exp(−Σi<n ci), state that they cannot prove boundedness of (ci), and tentatively conjecture ci ≍ 1/i from a 106-sample simulation to depth 20 (their Remark 1.5).

Section 2Exact values, with certificates

Fixing the sign of every intermediate difference splits ℝi+1≥0 into polyhedral cones on which the depth-i entry is linear; exponential integrals over simplicial cones have closed rational forms. Rays come from exact integer double description; triangulations use floating Delaunay for combinatorics only, cross-checked by a second independent projection and repaired, on disagreement, by an exact pulling triangulation via integer rank tests. Certificate: the exponential measures of all chambers must sum to exactly 1 in ℚ — this identity holds exactly in every reported run, and the pipeline reproduces CHT's c2 and c3.

Proposition 1 — all sign histories occur

Every one of the 2i(i+1)/2 sign patterns of the intermediate differences is realized on a full-dimensional cone of initial data.

Proof. Build the array bottom-up: set the bottom entry to 1; given row t+1 > 0 and the desired signs, put At1 = Tt and Atj+1 = Atj − st+1jAt+1j; taking Tt large makes row t positive and all constraints strict. ∎

The continuous model therefore has no combinatorial obstruction: cooling is irreducibly statistical. Counts 2, 8, 64, 1024, 32768, 2097152 confirmed computationally.

iciStatus
27 / 9 = 0.7778CHT
3227 / 288 = 0.78819CHT
4778959731701 / 1447295850000
= 0.53821735…
NewCertified
514008668886481596262550223816901 / 25320304994525128311856832700000
= 0.55325830…
NewCertified
60.44838867213…
exact 153-digit numerator in c6_certified.txt
NewCertified

The c6 computation covered 2,097,152 chambers; 685,625 needed the exact fallback. A first run without it silently lost 1.7×10−7 of measure and produced a value wrong by 1.25×10−7 — caught by the algebraic certificate, invisible to the Monte Carlo (both values sat 0.8σ from it). Denominators are smooth (largest prime factors 17, 47, 331); numerators contain enormous primes, so no closed form is apparent. None of these numbers appear in the OEIS. Note c6 < c4 < c5 < c3: the dyadic non-monotonicity deepens.

Section 3The submask law

A tiered Monte Carlo (40M samples to depth 63, decreasing to 150k at 4095; all exact anchors reproduced within 1.2σ) shows that log(i·ci) regressed on the individual binary digits of i yields R² = 0.95 with a cascade of per-bit factors:

Bit j (scale 2j)01234≥ 5
Measured factor1.6961.4611.2141.0711.024≈ 1
1 + q2^j, q = 0.6851.6851.4691.2201.0481.0021.000

The one-parameter match in the second row is the statement that a binary submask m of i contributes weight qm, via the identity Σm⊆i qm = Πj∈i(1 + q2^j). Fitting directly:

(⋆)
ci ≈ C · i−α · Σm ⊆ i qm
q ≈ 0.68 · α ≈ 0.80 · C ≈ 1.14  —  R² = 0.980 on i ∈ [64, 1023]
(10-parameter free-bit model: 0.953; pure power law: R² < 0)
q0.685 α0.800
Hover or tap the curve to read off a depth.
Law (⋆) evaluated live — this is the model, not the Monte Carlo data. Every point is computed in your browser from the formula above; the dashed envelopes join the all-ones depths 2k−1 (filled squares, top) to the powers of two 2k (hollow, bottom), and the widening gap between them is the phenomenon. The teeth are self-similar, which is why the wider range looks like the narrow one. Pull q to zero and everything collapses onto the straight line a pure power law would give — that line is the one fitting the data with R² < 0. Note the honest direction of the residual: at q = 0.685 the model puts c511/c512 at about 3.2, while the measurement is 4.7. The real oscillation is larger than (⋆), and grows with scale.

Section 4Consequences for CHT's Remark 1.5

Partial sums fit Sn = 10.84·n1/5 − 11.43 to three decimals at every dyadic checkpoint n = 63 … 1023 (where log(n+e) ≤ 6.94, the CHT lower bound). Fitted on n ≤ 1023, the same formula then predicted S2047 = 38.37 and S4095 = 45.78; the deep run measured 38.33 and 45.60 — 0.1–0.4% accuracy two octaves out of sample.

At n = 4095 the empirical fit stands at 45.8; the best proved lower bound is 9.01.
The gap between what is proved and what is measured. All three curves are closed forms, drawn directly. The proved bounds are logarithms and stay under 10 across the whole computed range; the observed partial sums follow a fifth-root power law. Under ci ≍ 1/i the measured curve would itself be logarithmic — flat on this scale.

Under ci ≍ 1/i one would have Sn ~ log n; the observed growth is a clean power, consistent with the typical exponent 1 − α ≈ 0.2. The conjecture of Remark 1.5 is untenable in the computationally accessible range — though CHT's theorems only force Sn ≳ log n, so a turnover beyond this range remains logically possible.

Section 5Mechanism: Lucas channels, activated by the tail

A large initial value at offset m propagates through |diff| dynamics as an exact Pascal-mod-2 pattern and reaches the apex iff C(i, m) is odd — iff m ⊆ i bitwise (Lucas). The submask sum in (⋆) is a sum over these open channels, with qm the measured attenuation of the channel at lateral offset m.

Depth i31
Which starting positions can still be felt at the apex. The top strip is every lateral offset m = 0 … i; a filled square means the channel is open. Click one and the XOR propagation runs down to the apex — open channels arrive, closed ones cancel themselves out on the way. The 31 → 32 button is the whole paper in one gesture: at depth 31 = 11111₂ every one of the 32 channels is open; add a single row and 32 = 100000₂ leaves exactly two. That discontinuity, repeated at every scale, is the sawtooth in section 3 — and why c511 and c512 are worlds apart.

Two confirmations of the mechanism:

  1. Starve the channels and the law dies. With Uniform[0,1] initial data (bounded tail) the oscillation nearly vanishes — c127 ≈ c128 within 1.5%, against ×3.1 apart for exponential data — and decay is a clean power law, α ≈ 0.92. Since the Cramér model of prime gaps has an exponential-type tail, the exponential model — where the law is strong — is the Gilbreath-relevant one.
  2. CHT's key inequality is the m = 0 channel — and the two outer channels can be used at once. Their Prop. 2.1 lower bound runs on a1 − Σk a(k,2): exactly the leftmost channel.
Theorem 2 — two-channel lower bound

For every n ≥ 2,  cn ≥ 𝐄 e−(SL−+B) + 𝐄 e−(SR−+A) ≥ 2 exp(−Σk<n ck), where A = a(n−1,1), B = a(n−1,2), SL− = Σk≤n−2 a(k,2), SR− = Σk≤n−2 a(k,n−k). Consequently Σi≤n ci ≥ log(2n − 2 + e²), an asymptotic additive log 2 over CHT's Theorem 1.4.

Proof. The apex is exactly |A − B| = (A−B)+ + (B−A)+. CHT's telescoping at depth n−1 gives A ≥ a1 − SL−; its mirror (the triangle is reversal-invariant) gives B ≥ an+1 − SR−. Now W = SL− + B involves only a2,…,an+1, so it is independent of a1, and 𝐄[(a1 − W)+ | W] = e−W; Jensen with 𝐄SL− + 𝐄B = Σk<n ck gives the first term, and the mirrored argument the second. ∎

Elementary — CHT's own techniques plus reversal symmetry — but a strict sharpening of their Prop. 2.1, and the first bound using two Lucas channels simultaneously. Every step was additionally verified numerically at n = 6 on 4×106 samples. The natural conjecture suggested by (⋆): all open channels contribute, cn ≳ (Σm⊆n qm)·exp(−Σk<n ck).

Section 6The microscope: proved lenses on the primes

CHT's deterministic criterion (their Theorem 1.6, whose proof we read in full) reduces Gilbreath at row N to excluding two families of rigid structures from the gap array: 0-valued blocks of length ≈ log10N at any depth, and {0,d}-valued blocks vastly longer than they are deep. The following are unconditional theorems of this program (proofs in MICROSCOPE.md; every bound verified against the primes below 10⁹):

Hover or tap a bar to read d, its factorisation, the proved cap, and the measured maximum.
The fence and the crowd. For each d, the bar is the longest run of consecutive prime gaps drawn from {2, 2d+2} found anywhere below 10⁹ (measurement, from lens_check.py); the black tick is the cap q₀ − 2 proved by the covering lemma, where q₀ is the smallest odd prime factor of d (theorem, computed live). Wherever a cap exists the measurement touches it or sits under it — at d = 5, 7, 10, 14 it touches exactly. The bars with no tick are the powers of two: the covering kernel, where no such cap can exist and the remaining difficulty lives.

Why the survivors survive — two structural principles. Covering arguments run on the one residue forbidden to primes, 0 mod q; modulo powers of two, primes forbid only evenness, so no covering argument can touch a pattern whose modulus support is 2-adic. And sieve bounds exceed trivial counts in polylog-length windows, so density methods are structurally blind at exactly the scale where the dangerous patterns live. Every surviving enemy sits in the intersection of those two blind spots — which is a precise statement of what a new tool must do.

Section 7Two theorems about proof strategies, and the frontier

Theorem R1 (conditional reduction; proved). Cramér's gap bound plus a residual pattern axiom P — strictly narrower than CHT's own hypotheses, because the lenses of section 6 discharge depth-1 zero-blocks and all odd-part-d top-row blocks unconditionally — implies Gilbreath's conjecture for all but finitely many rows. Statement and proof in REDUCTION.md.

Theorem R2 (insufficiency of fixed-order statistics; proved). There is a gap sequence whose every fixed-order pattern count matches the i.i.d. Cramér model to within Ok(log log x) — so every Hardy–Littlewood-analogue asymptotic holds, at any error quality — and whose Gilbreath leads exceed 1 infinitely often. The construction plants a single value V = m² at doubly-exponentially sparse positions m: one entry changed per plant leaves all fixed-order statistics intact, but the plant rides the always-open diagonal Pascal channel (C(t,t) = 1), and the mirrored CHT telescope bounds its erosion by the background sum, so it reaches the left edge with room to spare. Consequently no axiom system consisting of fixed-order gap statistics can imply eventual Gilbreath — the first rigorous form of CHT's remark that their hypotheses "look difficult to establish rigorously, even if one assumes strong conjectures on the primes such as the Hardy–Littlewood prime tuples conjecture."

Sabotage with one entry. Computed live in your browser from the Cramér random model (a model, not the primes): 1,600 geometric gaps, the lead of every row plotted on a log scale. The clean background settles into the conjectured {0,1} within a dozen rows. Planting a single value 3m at position m = 800 — statistically invisible, one entry among 1,600 — derails the lead at exactly row m = 800, the depth at which the plant's light cone first reaches the left edge — precisely where the theorem's diagonal Pascal channel says it must arrive. This interaction is the proof mechanism of Theorem R2.

Open Problem R3 (the frontier). The plant is forced to be huge — a gap comparable to its position, grossly non-Cramér — and this is provable: under any o(n) entry bound the plant erodes away before arriving, while extended "highway" structures leave fixed-order twin-density fingerprints of exactly the kind the statistics forbid. So: does fixed-order statistics + an = o(n) imply eventual Gilbreath? The o(n) bound is known unconditionally for primes (Baker–Harman–Pintz), so an affirmative answer would make Gilbreath a consequence of the standard conjectural statistics of prime gaps — conditional, since those contain the twin prime conjecture, but the first route of any kind from standard conjectures to Gilbreath. A negative answer would strengthen R2. Either resolution is a paper.

Section 8Open questions

  1. Resolve R3: fixed-order gap statistics + o(n) entries ⇒ eventual Gilbreath?
  2. Prove the twin-clustering bound of Cor. M9 — the weakest single statement that would finish the depth-0 lens for the surviving 2-adic d.
  3. The aperiodic 2-adic depth-2 sector: any mechanism beyond covering and density.
  4. Prove any strict digit-dependence of ci (e.g. c2^k+1 > c2^k for large k) — nothing of this kind is currently known.
  5. Extend Theorem 2 to middle channels or beat its constant — e.g. does cn ≥ (2+κ)e−Σc hold for odd n (channel m = 1 open)?
  6. Does (⋆) persist as i → ∞, or is it a transient? What are the true asymptotics of q and α?
  7. Exact c7 (228 chambers) — needs symmetry reduction or a smarter integration scheme.

Section 9Related session results on Gilbreath's conjecture proper

The same session verified Gilbreath's conjecture rigorously for the first 455,052,510 rows (primes < 1010, via the propagation-lemma certificate), audited the two dangerous structures of CHT's deterministic Theorem 1.6 on the real primes — worst zero-block 31 vs threshold ~3×1012 and worst {0,d}-block ratio 11 vs ~1026, both growing at coin-flip rates — and measured the prime array's defect dynamics (×0.936/row decay under an exactly critical mean-field, with period-4 Sierpiński modulation). Details in the companion session write-up.