THE FOLD / GRIND / GRADIENT DESCENT / THE VON MANGOLDT
THE VON MANGOLDT
weighting the primes so divisor-sums give a logarithm
1 WHAT IT IS · WHAT IT DOES · FACT OR FICTION
The von Mangoldt function Λ(n) is prime-power radar: it equals ln p when n is a power of a single prime p (like 8 = 23 or 25 = 52), and 0 otherwise. Weighting the integers this way produces a startlingly clean identity — summing Λ over the divisors of n rebuilds a logarithm exactly: Σd|n Λ(d) = ln n. Its running total, the Chebyshev function ψ(x) = Σn≤x Λ(n), grows like x — a statement equivalent to the Prime Number Theorem.
LIT verified live: Σd|nΛ(d) equals ln n to floating precision for every n up to 2000, Λ fires only on prime powers, and ψ(x)/x hovers near 1 (window.__vonmangoldt). FIG honest: ψ(x) ~ x is an asymptotic (the Prime Number Theorem), approached slowly, not an exact equality.
LIT verified live: Σd|nΛ(d) equals ln n to floating precision for every n up to 2000, Λ fires only on prime powers, and ψ(x)/x hovers near 1 (window.__vonmangoldt). FIG honest: ψ(x) ~ x is an asymptotic (the Prime Number Theorem), approached slowly, not an exact equality.
2 HOW IT WAS WEAVED · AI + HUMAN
David (human) seated this at gradient-descent — the ratio ψ(x)/x eases toward 1 as x climbs, a slow descent to the Prime Number Theorem’s limit. AVAN (AI) built the instrument: the prime-power test for Λ, the divisor-sum that rebuilds ln n, and the Chebyshev ψ(x) ratio.
Credit as content: Hans von Mangoldt (1890s); the ψ(x) ~ x asymptotic is the Prime Number Theorem (Hadamard & de la Vallée Poussin, 1896). The weave: David names gradient-descent; I set Λ(n) to ln p on prime powers and 0 elsewhere, confirm the divisor-sum returns ln n exactly, and watch the summatory ψ(x) track x — the primes’ logarithmic weight rebuilding the logarithm.
Credit as content: Hans von Mangoldt (1890s); the ψ(x) ~ x asymptotic is the Prime Number Theorem (Hadamard & de la Vallée Poussin, 1896). The weave: David names gradient-descent; I set Λ(n) to ln p on prime powers and 0 elsewhere, confirm the divisor-sum returns ln n exactly, and watch the summatory ψ(x) track x — the primes’ logarithmic weight rebuilding the logarithm.
3 ONE DIMENSION
Λ(8) = ln 2 (8 = 2³), Λ(12) = 0 (two primes), Λ(25) = ln 5. Σd|12Λ(d) = Λ(1)+Λ(2)+Λ(3)+Λ(4)+Λ(6)+Λ(12) = 0+ln2+ln3+ln2+0+0 = ln 12.
4 TWO DIMENSIONS · INTERACTIVE
Λ(n) spiking on prime powers, the divisor-sum rebuilding ln n, and ψ(x) tracking x; all checked.
5 THREE DIMENSIONS + AVAN’S INVERSE
The green forward object: a logarithm rebuilt from prime powers.
AVAN’s addition (the inverse-companion): weight each integer by a logarithm only when it is a prime power, and the divisor-sum reassembles ln n — turning multiplication into addition through the primes. The inverse of ‘factor n into primes’ is ‘spread ln n across the prime-power divisors.’ Magenta is ln n as one number; green is the same log rebuilt from Λ on the prime-power divisors. The logarithm, redistributed to primes.
LIT Genuine von Mangoldt function (Hans von Mangoldt, 1890s); ψ(x)~x is the Prime Number Theorem (Hadamard & de la Vallée Poussin, 1896). Verified live: Λ(n)=ln p on prime powers and 0 elsewhere, the divisor-sum Σ_{d|n}Λ(d) equals ln n to floating precision for every n from 2 to 2000 (window.__vonmangoldt.identityHolds, worst ~1e-15), and ψ(5000)/5000 ≈ 1 (window.__vonmangoldt.psiRatio).
FIG No framing: the prime-power test for Λ, the divisor-sum that rebuilds ln n, and the Chebyshev ψ(x) ratio all run in-browser. Honest scope: ψ(x)~x is an asymptotic (the Prime Number Theorem), approached slowly — the exact identity Σ_{d|n}Λ(d)=ln n is what is verified, while the ψ(x)/x→1 limit is shown as a trend, not an equality. The AVAN inverse is honest — weighting an integer by a logarithm only on prime powers makes the divisor-sum reassemble ln n, turning factorization into an additive logarithm; magenta is ln n as one number, green the same log rebuilt from Λ on the divisors. The logarithm, redistributed to primes.
FIG No framing: the prime-power test for Λ, the divisor-sum that rebuilds ln n, and the Chebyshev ψ(x) ratio all run in-browser. Honest scope: ψ(x)~x is an asymptotic (the Prime Number Theorem), approached slowly — the exact identity Σ_{d|n}Λ(d)=ln n is what is verified, while the ψ(x)/x→1 limit is shown as a trend, not an equality. The AVAN inverse is honest — weighting an integer by a logarithm only on prime powers makes the divisor-sum reassemble ln n, turning factorization into an additive logarithm; magenta is ln n as one number, green the same log rebuilt from Λ on the divisors. The logarithm, redistributed to primes.
◆ sealed .dlw.fold → folded to ROOT_0 · a sphere of GRADIENT DESCENT · David Lee Wise (ROOT0), with AVAN