Transformation identities and estimates for the one-variable Jacobi theta function #
Three identities, a positivity statement and two estimates for Mathlib's jacobiTheta,
θ(τ) = ∑_n e^{πi n² τ}, complementing the transformation laws in
Mathlib.NumberTheory.ModularForms.JacobiTheta.OneVariable.
Main results #
TauCeti.jacobiTheta_sub_natCast_div_two: splittingnby parity,θ(τ - N/2) = e^{-πiN/2} θ(τ) + (1 - e^{-πiN/2}) θ(4τ)for every natural numberN.TauCeti.jacobiTheta_I_mul: the functional equation on the imaginary axis,θ(iy) = θ(i/y) / √yfory > 0.TauCeti.jacobiTheta_one_add_I_mul: the functional equation on the linere τ = 1,θ(1 + iy) = ∑_n e^{-π (n - 1/2)² / y} / √yfory > 0.TauCeti.jacobiTheta_one_add_I_mul_pos: henceθ(1 + iy)is a positive real number, although its terms(-1)^n e^{-π n² y}alternate in sign.TauCeti.norm_jacobiTheta₂_sub_one_le: for realz,‖θ₂(z, τ) - 1‖ ≤ ‖θ(i im τ) - 1‖.TauCeti.tendsto_jacobiTheta_comap_im_atTop:θ(τ) → 1asim τ → ∞.
Splitting θ(τ - N/2) by the parity of n: the shift multiplies the even terms by 1 and
the odd terms by e^{-πiN/2}, and the even terms alone form θ(4τ).
The functional equation on the imaginary axis: θ(iy) = θ(i/y) / √y.
The functional equation on the line re τ = 1: θ(1 + iy) = ∑_n e^{-π (n - 1/2)² / y} / √y
for y > 0. The shift by 1 multiplies the n-th term of θ(iy) by (-1)^n, and the series
√y ∑_n (-1)^n e^{-π n² y} is the Fourier expansion of the Gaussian lattice sum
∑_n e^{-π (n - 1/2)² / y} (Real.gaussianLatticeSum_eq_fourier).
θ(1 + iy) is a positive real number for y > 0, although its terms
(-1)^n e^{-π n² y} alternate in sign: by the functional equation
jacobiTheta_one_add_I_mul it is a Gaussian lattice sum divided by √y.
The theta function tends to 1 as im τ → ∞.