The Clausen function and Catalan's constant #
The Clausen function Cl₂ θ = ∑_{n ≥ 1} sin(nθ) / n² is the imaginary part of the dilogarithm on the
unit circle. Its real part is the Bernoulli parabola ∑ cos(nθ)/n² = π²/6 - πθ/2 + θ²/4 on
[0, 2π], derived here from Mathlib's Fourier expansion of the Bernoulli polynomials. Abel summation
and the Fejér–Jackson inequality give Cl₂ θ > 0 on (0, π).
Main definitions #
Dilog.Cl₂: the Clausen function.Dilog.catalanConst: Catalan's constantG = ∑ (-1)^k / (2k + 1)².Dilog.zetaStateRe: the cosine series∑ cos(nθ) / n².
Main results #
Dilog.Cl₂_neg,Dilog.Cl₂_two_pi_sub: symmetries.Dilog.Cl₂_pi_div_two:Cl₂ (π/2) = G.Dilog.Cl₂_two_mul: duplicationCl₂(2θ) = 2 Cl₂(θ) - 2 Cl₂(π - θ);Dilog.Cl₂_two_pi_div_three.Dilog.hasSum_cos_div_sq: the Bernoulli parabola.Dilog.Cl₂_pos:0 < Cl₂ θforθ ∈ (0, π).Dilog.zetaState_never_orthogonal: forθ ∈ (0, 2π)the two series∑ cos(nθ)/n²and∑ sin(nθ)/n²do not vanish together.Dilog.zetaState_meanEnergy_not_summable,Dilog.zetaState_secondMoment_not_summable: the populations6/(π² n²)on the levelsnhave infinite mean energy and infinite second moment.
References #
- L. Lewin, Polylogarithms and Associated Functions, North-Holland, 1981, Chapter 4.
Cl₂(π/2) = G — Catalan's constant is the Clausen function at the quarter
circle. (Even-indexed sines vanish, odd-indexed give the alternating series.)
The Bernoulli parabola: for θ ∈ [0, 2π],
∑_{n≥1} cos(nθ)/n² = π²/6 − πθ/2 + θ²/4 — the real part of Li₂ on the unit circle
(the weight-2 Fourier expansion of the Bernoulli polynomial B₂). Note the n = 0
term of the ℕ-indexed sum is 0 by the 1/0 = 0 convention.
Positivity of Cl₂ on (0, π) #
Abel rearrangement of the partial sums of the Cl₂ series in terms of the
Fejér–Jackson sums, then fejerSum_pos makes every piece positive. The arrangement is
asymptotics-free: every partial sum is at least sin θ / 2.
The clock that never ticks #
The weight-2 zeta state — populations ∝ 1/n² on equally spaced energy levels — has
autocorrelation S(θ) = (6/π²)·Li₂(e^{−iθ}), with real part zetaStateRe (the Bernoulli
parabola) and imaginary part −Cl₂. Despite infinite mean energy and infinite energy
variance (so the Margolus–Levitin and Mandelstam–Tamm bounds say nothing), the state
never reaches an orthogonal state: S has no zero on (0, 2π).
At the antipode θ = π the real part is −π²/12 ≠ 0 (Bernoulli parabola).
The clock that never ticks: the weight-2 zeta-state autocorrelation never
vanishes on (0, 2π) — its real part (zetaStateRe) and imaginary part (−Cl₂) are
never simultaneously zero. The quantum state with populations ∝ 1/n² on equally
spaced levels has infinite mean energy, yet never reaches an orthogonal state.