ConceptComplete

Zeta and L-Functions - Examples and Constructions

Explicit computation and special values of LL-functions reveal arithmetic structure and test conjectures.

ExampleDirichlet's Theorem on Primes in Arithmetic Progressions

For coprime a,na, n, there are infinitely many primes p≑a(modn)p \equiv a \pmod{n}.

Proof via LL-functions: Consider L(s,Ο‡)L(s, \chi) for characters modulo nn. The non-vanishing L(1,Ο‡)β‰ 0L(1, \chi) \neq 0 for non-principal Ο‡\chi implies: βˆ‘p≑a(modn)1ps\sum_{p \equiv a \pmod{n}} \frac{1}{p^s}

diverges as s→1+s \to 1^+, proving infinitude of primes.

With more care, obtain density: primes ≑a(modn)\equiv a \pmod{n} have natural density 1/Ο†(n)1/\varphi(n).

ExampleQuadratic $L$-Functions

For K=Q(d)K = \mathbb{Q}(\sqrt{d}) with quadratic character Ο‡d\chi_d: L(s,Ο‡d)=∏p11βˆ’Ο‡d(p)pβˆ’sL(s, \chi_d) = \prod_p \frac{1}{1 - \chi_d(p)p^{-s}}

Special values:

  • L(1,Ο‡βˆ’3)=Ο€33L(1, \chi_{-3}) = \frac{\pi}{3\sqrt{3}}
  • L(1,Ο‡βˆ’4)=Ο€4L(1, \chi_{-4}) = \frac{\pi}{4}
  • L(1,Ο‡5)=log⁑(Ο•)5L(1, \chi_5) = \frac{\log(\phi)}{\sqrt{5}} where Ο•=(1+5)/2\phi = (1+\sqrt{5})/2

These relate to class numbers via: hKL(1,Ο‡d)=2Ο€wK∣d∣h_K L(1, \chi_d) = \frac{2\pi}{w_K\sqrt{|d|}} for imaginary quadratic, and similar formulas for real quadratic (involving regulators).

ExampleElliptic Curve $L$-Functions

For an elliptic curve E/QE/\mathbb{Q}, the LL-function is: L(E,s)=∏p11βˆ’appβˆ’s+p1βˆ’2sL(E, s) = \prod_p \frac{1}{1 - a_p p^{-s} + p^{1-2s}}

where ap=p+1βˆ’#E(Fp)a_p = p + 1 - \#E(\mathbb{F}_p). The Birch and Swinnerton-Dyer conjecture predicts: ords=1L(E,s)=rank(E(Q))\text{ord}_{s=1}L(E, s) = \text{rank}(E(\mathbb{Q}))

and gives an exact formula for lim⁑sβ†’1(sβˆ’1)βˆ’rL(E,s)\lim_{s \to 1}(s-1)^{-r}L(E, s) in terms of regulator, Shafarevich-Tate group, and Tamagawa numbers.

For E:y2=x3βˆ’xE: y^2 = x^3 - x (CM by Z[i]\mathbb{Z}[i]): L(E,1)β‰ 0L(E, 1) \neq 0, so rank(E(Q))=0\text{rank}(E(\mathbb{Q})) = 0 (BSD predicts, proven for CM curves).

ExampleZeros and Primes

The Riemann zeta function satisfies: βˆ’ΞΆβ€²(s)ΞΆ(s)=βˆ‘nΞ›(n)ns-\frac{\zeta'(s)}{\zeta(s)} = \sum_n \frac{\Lambda(n)}{n^s}

where Ξ›\Lambda is the von Mangoldt function. The zeros of ΞΆ(s)\zeta(s) control the distribution of primes via: Ο€(x)=Li(x)βˆ’βˆ‘ΟLi(xρ)+lowerΒ orderΒ terms\pi(x) = \text{Li}(x) - \sum_\rho \text{Li}(x^\rho) + \text{lower order terms}

where the sum is over nontrivial zeros ρ\rho of ΞΆ(s)\zeta(s). RH implies optimal error bounds for Ο€(x)\pi(x).

ExampleComputing $L$-Function Values

Modern algorithms compute L(1,Ο‡)L(1, \chi) using:

  1. Direct summation: Accelerate convergence with Euler-Maclaurin
  2. Functional equation: Relate L(1,Ο‡)L(1, \chi) to rapidly convergent series
  3. Fast Fourier methods: For character sums and Gauss sums

For Ο‡\chi modulo mm: Complexity O(m)O(\sqrt{m}) or better with precomputation.

Pari/GP example:

\\ L(1, chi) for chi of conductor 5
lfun(lfuncreate([5, 1, [0], 1, 0]), 1)
\\ Output: 0.728882...
Remark

The Grand Riemann Hypothesis (GRH) asserts that zeros of all global LL-functions (Dedekind zeta, Dirichlet LL, Artin LL, etc.) on the critical strip lie on the critical line.

This would imply:

  • Optimal prime number theorems
  • Effective versions of class number bounds (Bach's bound)
  • Correctness of many probabilistic algorithms in number theory

GRH is used conditionally in computational number theory for efficiency.