Opener Explorer Deep Analysis Geometric Canvas ζ(2) GCD Lift Paper K Constant Spectral D_χ Explorer
C ≈ 0.530711806246
Lift Theory
ref · Chart 07 01 Lift Friction Map — local factor φ(M+1)/[(M+1)·C] deviation

Each integer M contributes a local lift factor L(M) = φ(M+1)/(M+1). The theoretical constant C ≈ 0.5307 is the limit R(N) converges to. Friction measures the signed relative deviation: f(M) = L(M)/C − 1. Smooth numbers (products of small primes) create deep negative friction; primes spike sharply positive. The histogram shows the full deviation distribution — a fat-tailed landscape that reveals where the convergence R(N) → C is most turbulent, and where potential counter-examples to the density claims would most naturally hide.

N 3,000
Positive friction (primes)
Negative friction (smooth)
Running R(N)/C − 1 → 0
Zero line (f = 0)
N
C
Max |friction|
Mean |friction|
M at max friction
Skewness
The friction distribution is heavily right-skewed: rare but extreme positive spikes (primes) and a wide negative tail (smooth numbers with many small factors). These extremes are the "rough patches" in convergence — the arithmetic analogue of turbulence in a laminar flow approaching C.
ref · Opener Ring 02 Ghost Residues — Stable vs Transient Primes in the Primorial Sieve

At primorial level M = p₁·…·pₖ, most coprime residues r in [1, M) are genuine primes — but some are composites of primes > pₖ that only appear prime-like. These are the ghost composites: they pass the gcd(r, M) = 1 test but will be eliminated when M advances to M·p_{k+1}. The transient residues are those that fail the very next sieve step. Twin-prime pair candidates where at least one member is a ghost are the "composite interference paths" — pairs that look like twin primes at this level but are not.

Primorial level Twin pairs
Modulus M
30
= 2 · 3 · 5
Coprime count φ(M)
gcd(r, M) = 1
Stable primes
coprime ∧ prime — genuine survivors
Ghost composites
coprime ∧ composite — not yet sieved
Next-level transients
r ≡ 0 mod p_{k+1} — expelled next round
Genuine twin pairs
both members prime
Ghost twin pairs
≥1 member composite — interference
Stable prime (large dot)
Ghost composite (medium dot)
Next-level transient (small dot)
Unit (r = 1)
Eliminated
At M = 6 every coprime residue in [2, 6) is already prime. Ghost composites appear from M = 30 onward (e.g., 49 = 7² is coprime to 30 but composite). Their prevalence grows with M — the sieve must advance through infinitely many primorial levels to eliminate all ghost composites and isolate the genuine twin primes.
Sign Change Dynamics
ref · Chart 06 03 Sign-Change Waiting Time — log-scale gap distribution & Skewes context

Chart 06 shows the raw gap between consecutive sign changes of E(N). Here the gaps are plotted on a logarithmic scale — log₁₀(gap_k) vs sign-change index k — exposing any long-term growth trend in waiting time. A rising OLS regression line indicates sign changes grow rarer at larger N, connecting the small-N behaviour of E(N) to Littlewood's theorem about π(x) − li(x): the changes must occur infinitely often, but with astronomically increasing spacing as N → ∞.

Skewes Number Context
Skewes (1933) proved under RH that π(x) first exceeds li(x) before x = 10101034. Computational evidence places the first crossing near e727.95… ≈ 1.40 × 10316. Our E(N) = R(N) − C differs from π(x) − li(x); its sign changes also exhibit growing gaps, though no analytic rate has been established for E(N). The log-scale scatter makes this rarity visible even over the small-N range accessible to a browser — a window into a structure that only becomes fully dramatic at astronomical scale.
N 5,000
log₁₀(gap_k)
OLS trend
Gap distribution
Sign changes
Median gap
Max gap
OLS slope (log)
SC / π(N)
Skewes gap ratio
A positive OLS slope means successive sign changes take longer on average — the functional analogue of Skewes-type behaviour. Even if the slope is small in this range, extrapolated to N ≈ 10³¹⁶ it predicts waiting times that dwarf observable scales. The histogram shows the gap distribution is log-normal, not geometric — another deviation from a simple Poisson process.
Parity Obstruction
ref · Opener / Sieve 04 Parity Interference — Liouville function & prime-factor parity balance

The parity problem in sieve theory: standard multiplicative sieves cannot distinguish integers with an even number of prime factors (Liouville λ = +1) from those with an odd number (λ = −1). The top chart shows L(N) = Σ λ(n) — the cumulative Liouville sum — whose smallness relative to N reflects near-perfect parity balance. The bottom chart tracks, in real time, the fraction of coprimes to M = 30 with odd Ω. If this fraction deviated detectably from ½, a sieve could exploit the imbalance to count twin primes exactly. It does not — and the chart shows why.

N 5,000
L(N) = Σ λ(n)
Fill area below zero
Odd-Ω fraction (coprimes to 30)
Primorial thresholds
L(N)
L(N) / N
Min L(N)
Max L(N)
Odd-Ω fraction
The Pólya conjecture (L(N) ≤ 0 for all N > 1) held computationally until Haselgrove disproved it in 1958; the first explicit counterexample lies near 1.845 × 10³⁶¹. In the range shown, L(N) remains negative and the odd-Ω fraction among coprimes to 30 hugs ½ with deviations invisible to any linear sieve. The Jump Theorem's doubling law works within the structured environment of M = 30·2ⁿ, where additional arithmetic constraints may break this balance in ways that standard sieve weights cannot.
Jump Theorem Prime Checker
ref · Jump Theorem · Layer 05 05 Jump Theorem Prime Checker — D(x) detects primes through exact jump sizes

The log-derivative D(x) = d/dx log C(x) is smooth everywhere except at the primes. At x = p−1, D(x) jumps by exactly 1/(p(p−1)). Composites pass in silence. Given any jump size s, the prime is recovered exactly: p = (1 + √(1 + 4/s)) / 2. The discriminant 1 + 4p(p−1) = (2p−1)² is always a perfect square for integer primes — the algebraic fingerprint of the theorem. The golden ratio φ is the supremum of the jump spectrum: s = 1 is the only value no integer prime reaches. Enter any integer to verify.

Jump Spectrum — φ boundary to primes
Presets
n = n=1 gives φ
Display
Enter n ≥ 1 and press Check — or click a preset above.
n
Classification
Jump size s
Discriminant
Recovery p
The recovery formula p = (1 + √(1+4/s)) / 2 acts as an arithmetic filter. For every integer prime p, the discriminant 1 + 4p(p−1) = (2p−1)² — always a perfect square, always returning p exactly. At s = 1 (the φ boundary), the discriminant is 5 — not a perfect square — and the formula returns φ = (1+√5)/2. This is the unique point where the integer-producing property breaks down. The set {1/2, 1/6, 1/20, 1/42, …} has φ-position s = 1 as its least upper bound.