The theta series of a lattice #
For a ℤ-submodule L of a real normed space E and τ in the upper half-plane ℍ, the
theta series of L is
Θ_L(τ) = ∑_{v ∈ L} exp (π i ‖v‖² τ),
and for γ : E the theta series of the translate γ + L is
θ_{γ+L}(τ) = ∑_{v ∈ L} exp (π i ‖γ + v‖² τ).
In the applications E is a real inner product space, so that ‖v‖² is the positive definite
quadratic form of the inner product; the definitions and the results here use only the norm.
The exponent is π i ‖v‖² τ rather than 2 π i ‖v‖² τ: for an even lattice the q-expansion
in q = exp (2 π i τ) is then indexed by the half-norms ‖v‖² / 2. In rank one, for L = ℤ ⊆ ℝ,
Θ_L is the Jacobi theta function jacobiTheta. When γ lies in the dual lattice, θ_{γ+L} is
the coset theta series of the coset γ + L.
This file defines both series and proves that they converge and are holomorphic, for a discrete
subgroup L (of any rank, not necessarily full) of a finite-dimensional real normed space. The
term of index v has norm exp (-π ‖γ + v‖² im τ), so on every half-plane {τ | y ≤ im τ},
0 < y, the terms are dominated by the summable Gaussian exp (-π y ‖γ + v‖²) of
ZLattice.summable_exp_neg_mul_norm_add_sq. Hence the series converge uniformly on these
half-planes and locally uniformly on ℍ, and are holomorphic, both as functions of z : ℂ on
{z | 0 < im z} and in the manifold sense on ℍ, which is what a bundled modular form requires.
The series θ_{γ+L} depends only on the class of γ modulo L, and θ_{-γ+L} = θ_{γ+L}. For a
subgroup M ≤ L, splitting L into the classes of L modulo M writes θ_{γ+L} as the sum of
the series θ_{γ+y+M} over those classes, and in particular Θ_L as the sum of the θ_{y+M}.
Main definitions #
ZLattice.thetaSeries L: the theta seriesΘ_L : ℍ → ℂ.ZLattice.thetaCoset L γ: the theta seriesθ_{γ+L} : ℍ → ℂof the translateγ + L.
Main statements #
Complex.norm_exp_pi_mul_I_mul_ofReal_mul:‖exp (π i t z)‖ = exp (-π t im z)for realt, so the term of indexvhas normexp (-π ‖γ + v‖² im τ).ZLattice.thetaCoset_eq_of_sub_mem,ZLattice.thetaCoset_neg:θ_{γ+L}depends only onγmoduloL, andθ_{-γ+L} = θ_{γ+L}.ZLattice.summable_norm_thetaCoset,ZLattice.summable_thetaCoset: the series converge absolutely at everyτ : ℍ.ZLattice.hasSum_thetaCoset_add_out,ZLattice.hasSum_thetaCoset_out: forM ≤ L,θ_{γ+L}is the sum of theθ_{γ+y+M}, andΘ_Lthe sum of theθ_{y+M}, over the classesyofLmoduloM.ZLattice.hasSumUniformlyOn_thetaCoset: uniform convergence on{τ | y ≤ im τ}for0 < y.ZLattice.hasSumLocallyUniformly_thetaCoset: locally uniform convergence onℍ.ZLattice.differentiableOn_thetaCoset,ZLattice.mdifferentiable_thetaCoset: holomorphy.
Each statement about thetaCoset has a thetaSeries counterpart, obtained from
ZLattice.thetaCoset_zero : thetaCoset L 0 = thetaSeries L.
References #
- J.-P. Serre, A Course in Arithmetic, Ch. VII §6.
- W. Ebeling, Lattices and Codes, Ch. 2.
The theta series Θ_L(τ) = ∑_{v ∈ L} exp (π i ‖v‖² τ) of L : Submodule ℤ E. It converges
on ℍ when L is discrete and E is finite-dimensional (ZLattice.summable_thetaSeries).
Instances For
The theta series θ_{γ+L}(τ) = ∑_{v ∈ L} exp (π i ‖γ + v‖² τ) of the translate γ + L of
L : Submodule ℤ E. For γ in the dual lattice this is the coset theta series of the coset
γ + L.
Instances For
The value of the theta series at τ: Θ_L(τ) = ∑_{v ∈ L} exp (π i ‖v‖² τ).
The value of the theta series of the translate γ + L at τ:
θ_{γ+L}(τ) = ∑_{v ∈ L} exp (π i ‖γ + v‖² τ).
If the Gaussian exp (-π y ‖γ + v‖²) is summable over v ∈ L for no height y > 0, then the
series defining θ_{γ+L}(τ) is not summable for any τ, so θ_{γ+L} is identically 0 (the value
of a tsum that is not summable).
The theta series of the translate 0 + L is the theta series of L.
Translating γ by a vector of L does not change the theta series of the translate γ + L:
the series is reindexed by v ↦ w + v.
The theta series of the translate γ + L depends only on the class of γ modulo L.
For γ ∈ L the translate γ + L is L itself, with the same theta series.
The theta series of the translates -γ + L and γ + L agree: the series is reindexed by
v ↦ -v.
The sum of the terms of θ_{γ+L} over the fiber of the quotient map L → L / M is the theta
series of the translate of M by γ plus a representative of that class.
The sum of theta terms over the fiber of the quotient map L → L / M is the theta
series of the translate of M by a representative of that class.
The theta series of a translate of a discrete subgroup converges absolutely at every point
of the upper half-plane: the norms exp (-π ‖γ + v‖² im τ) of its terms are summable.
The theta series of a translate of a discrete subgroup converges at every point of the upper half-plane.
The theta series of a discrete subgroup converges at every point of the upper half-plane.
Splitting the theta series of a translate into cosets. For M ≤ L and γ : E, the theta
series of γ + L is the sum of the theta series of the translates γ + y + M over the classes y
of L modulo M, each class represented by Quotient.out (by thetaCoset_eq_of_sub_mem the
representative does not matter). The sum is finite when M has finite index in L.
Splitting a theta series into cosets. For M ≤ L, the theta series of L is the sum of
the theta series of the translates y + M over the classes y of L modulo M, each class
represented by Quotient.out. The sum is finite when M has finite index in L.
The theta series of a translate converges uniformly on each half-plane {τ | y ≤ im τ} with
0 < y.
The theta series converges uniformly on each half-plane {τ | y ≤ im τ} with 0 < y.
The theta series of a translate converges locally uniformly on the upper half-plane.
The theta series converges locally uniformly on the upper half-plane.
The theta series of a translate is holomorphic on the upper half-plane, as a function of
z : ℂ.
The theta series is holomorphic on the upper half-plane, as a function of z : ℂ.
The theta series of a translate is holomorphic on the upper half-plane ℍ.
The theta series is holomorphic on the upper half-plane ℍ.