Selberg's eigenvalue conjecture
DefinitionCongruence subgroup of the modular group#✓fully proved
A subgroup \(\Gamma\) of \(\mathrm{SL}(2, \mathbb{Z})\) is a congruence subgroup if there is a positive integer \(N\) such that every \(\gamma \in \mathrm{SL}(2, \mathbb{Z})\) whose entries are congruent modulo \(N\) to those of the identity matrix lies in \(\Gamma\); that is, \(\Gamma\) contains the principal congruence subgroup of level \(N\).
def IsCongruenceSubgroup (Γ : Subgroup SL(2, ℤ)) : Prop :=
∃ N : ℕ, 0 < N ∧ ∀ γ : SL(2, ℤ),
(∀ i j, γ.1 i j ≡ (1 : Matrix (Fin 2) (Fin 2) ℤ) i j [ZMOD (N : ℤ)]) →
γ ∈ Γ
Used by 15
TheoremThe modular group is a congruence subgroup#✓fully proved
The full group \(\mathrm{SL}(2, \mathbb{Z})\), as a subgroup of itself, is a congruence subgroup.
It contains the principal congruence subgroup of level \(1\), which is the whole group.
theorem isCongruenceSubgroup_top : IsCongruenceSubgroup ⊤
def IsCongruenceSubgroup (Γ : Subgroup SL(2, ℤ)) : Prop :=
∃ N : ℕ, 0 < N ∧ ∀ γ : SL(2, ℤ),
(∀ i j, γ.1 i j ≡ (1 : Matrix (Fin 2) (Fin 2) ℤ) i j [ZMOD (N : ℤ)]) →
γ ∈ Γ
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:64
Uses 1 · Used by 4
DefinitionZero extension from the upper half plane#✓fully proved
For a function \(f\) from the upper half plane to \(\mathbb{C}\), its zero extension is the function from \(\mathbb{C}\) to \(\mathbb{C}\) that takes the value \(f(z)\) at every \(z\) with \(\operatorname{Im} z > 0\) and the value \(0\) at every other point.
It is used to read coordinate derivatives of \(f\) in the interior.
def ext (f : UpperHalfPlane → ℂ) : ℂ → ℂ :=
fun z => if hz : 0 < z.im then f ⟨z, hz⟩ else 0
DefinitionPartial derivative in the real direction#✓fully proved
For a function \(f \colon \mathbb{H} \to \mathbb{C}\) and a point \(z \in \mathbb{H}\), \(\partial_x f(z)\) is the real Fréchet derivative at \(z\) of the zero extension of \(f\) to \(\mathbb{C}\), applied to the vector \(1\). When the zero extension is real-differentiable at \(z\) this is the partial derivative of \(f\) with respect to \(x = \operatorname{Re} z\); otherwise the value is \(0\).
Real, not complex, differentiability is essential: with complex derivatives the two second derivatives in the hyperbolic Laplacian would cancel by the Cauchy–Riemann equations, and the Laplacian would vanish identically.
def partialRe (f : UpperHalfPlane → ℂ) (z : UpperHalfPlane) : ℂ :=
fderiv ℝ (ext f) z (1 : ℂ)
def ext (f : UpperHalfPlane → ℂ) : ℂ → ℂ :=
fun z => if hz : 0 < z.im then f ⟨z, hz⟩ else 0
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:75
Uses 1 · Used by 3
DefinitionPartial derivative in the imaginary direction#✓fully proved
For a function \(f \colon \mathbb{H} \to \mathbb{C}\) and a point \(z \in \mathbb{H}\), \(\partial_y f(z)\) is the real Fréchet derivative at \(z\) of the zero extension of \(f\) to \(\mathbb{C}\), applied to the vector \(i\). When the zero extension is real-differentiable at \(z\) this is the partial derivative of \(f\) with respect to \(y = \operatorname{Im} z\); otherwise the value is \(0\).
def partialIm (f : UpperHalfPlane → ℂ) (z : UpperHalfPlane) : ℂ :=
fderiv ℝ (ext f) z Complex.I
def ext (f : UpperHalfPlane → ℂ) : ℂ → ℂ :=
fun z => if hz : 0 < z.im then f ⟨z, hz⟩ else 0
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:75
Uses 1 · Used by 3
DefinitionHyperbolic Laplacian#✓fully proved
For a function \(f\) from the upper half plane to \(\mathbb{C}\) and a point \(z = x + iy\) of the upper half plane, the hyperbolic Laplacian of \(f\) at \(z\) is
where, for a function \(g\) from the upper half plane to \(\mathbb{C}\), \(\partial_x g(z)\) and \(\partial_y g(z)\) are the real Fréchet derivatives at \(z\) of the zero extension of \(g\) to \(\mathbb{C}\), applied to the vectors \(1\) and \(i\) respectively (taken to be \(0\) where that extension is not real differentiable).
On smooth functions this agrees with the coordinate formula \(-y^2 (\partial^2 f / \partial x^2 + \partial^2 f / \partial y^2)\) at every interior point.
def hyperbolicLaplacian (f : UpperHalfPlane → ℂ) (z : UpperHalfPlane) : ℂ :=
-((z.im : ℂ) ^ 2) * (partialRe (partialRe f) z + partialIm (partialIm f) z)
def partialRe (f : UpperHalfPlane → ℂ) (z : UpperHalfPlane) : ℂ :=
fderiv ℝ (ext f) z (1 : ℂ)
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:83
def partialIm (f : UpperHalfPlane → ℂ) (z : UpperHalfPlane) : ℂ :=
fderiv ℝ (ext f) z Complex.I
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:88
TheoremThe hyperbolic Laplacian kills constants#✓fully proved
For every complex number \(c\) and every point \(z\) of the upper half plane, the hyperbolic Laplacian of the constant function with value \(c\) at \(z\) is \(0\).
theorem hyperbolicLaplacian_const (c : ℂ) (z : UpperHalfPlane) :
hyperbolicLaplacian (fun _ => c) z = 0
def hyperbolicLaplacian (f : UpperHalfPlane → ℂ) (z : UpperHalfPlane) : ℂ :=
-((z.im : ℂ) ^ 2) * (partialRe (partialRe f) z + partialIm (partialIm f) z)
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:94
Uses 4
TheoremHyperbolic Laplacian of the squared imaginary part#✓fully proved
Let \(f\) be the function on the upper half plane given by \(f(u) = (\operatorname{Im} u)^2\), viewed as a complex number. Then for every point \(z\) of the upper half plane the hyperbolic Laplacian of \(f\) at \(z\) equals \(-2 (\operatorname{Im} z)^2\).
This is a non-holomorphic check: the value is not zero, whereas with complex derivatives the computation would collapse by Cauchy--Riemann.
theorem hyperbolicLaplacian_im_sq (z : UpperHalfPlane) :
hyperbolicLaplacian (fun u : UpperHalfPlane => (((u : ℂ).im ^ 2 : ℝ) : ℂ)) z =
-2 * (((z : ℂ).im ^ 2 : ℝ) : ℂ)
def hyperbolicLaplacian (f : UpperHalfPlane → ℂ) (z : UpperHalfPlane) : ℂ :=
-((z.im : ℂ) ^ 2) * (partialRe (partialRe f) z + partialIm (partialIm f) z)
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:94
Uses 4
DefinitionTwice continuously differentiable functions on the upper half-plane#✓fully proved
A function \(f \colon \mathbb{H} \to \mathbb{C}\) is smooth (of order two) when it is the restriction of a function \(F \colon \mathbb{C} \to \mathbb{C}\) that is twice continuously differentiable, in the real sense, on the open set \(\{z \in \mathbb{C} : \operatorname{Im} z > 0\}\).
Differentiability is real, not complex: eigenfunctions of the Laplacian need not be holomorphic.
def IsSmooth (f : UpperHalfPlane → ℂ) : Prop :=
∃ F : ℂ → ℂ, ContDiffOn ℝ 2 F {z | 0 < z.im} ∧ ∀ z : UpperHalfPlane, f z = F z
Used by 1
DefinitionStandard fundamental domain of the modular group#✓fully proved
The standard fundamental domain for \(\mathrm{SL}(2, \mathbb{Z})\) is the set of points \(z\) of the upper half plane with \(|\operatorname{Re} z| \le 1/2\) and \(|z| \ge 1\).
def fundamentalDomain : Set UpperHalfPlane :=
{z | |z.re| ≤ 1 / 2 ∧ 1 ≤ ‖(z : ℂ)‖}
Used by 1
DefinitionFinite coset covers of the modular group#✓fully proved
A finite set \(s \subseteq \mathrm{SL}(2, \mathbb{Z})\) covers \(\mathrm{SL}(2, \mathbb{Z})\) by cosets of a subgroup \(\Gamma\) when every \(g \in \mathrm{SL}(2, \mathbb{Z})\) can be written as \(g = \gamma a\) with \(\gamma \in \Gamma\) and \(a \in s\), that is, when the cosets \(\Gamma a\) with \(a \in s\) cover \(\mathrm{SL}(2, \mathbb{Z})\). The elements of \(s\) need not lie in distinct cosets.
def IsCosetCover (Γ : Subgroup SL(2, ℤ)) (s : Finset SL(2, ℤ)) : Prop :=
∀ g : SL(2, ℤ), ∃ a ∈ s, ∃ γ ∈ Γ, g = γ * a
Used by 1
DefinitionSquare-integrable functions on a modular quotient#✓fully proved
For a subgroup \(\Gamma\) of \(\mathrm{SL}(2, \mathbb{Z})\), a function \(f \colon \mathbb{H} \to \mathbb{C}\) is square-integrable on \(\Gamma \backslash \mathbb{H}\) when there is a finite set \(s \subseteq \mathrm{SL}(2, \mathbb{Z})\) whose cosets \(\Gamma a\), \(a \in s\), cover \(\mathrm{SL}(2, \mathbb{Z})\), such that
where \(\mathcal{F} = \{z \in \mathbb{H} : |\operatorname{Re} z| \le 1/2,\ |z| \ge 1\}\) is the standard fundamental domain of \(\mathrm{SL}(2, \mathbb{Z})\), \(a \cdot \mathcal{F}\) is its image under the Möbius action of \(a\), and \(\mu\) is the hyperbolic area measure \(dx \, dy / y^2\) on \(\mathbb{H}\).
def IsL2Quotient (Γ : Subgroup SL(2, ℤ)) (f : UpperHalfPlane → ℂ) : Prop :=
∃ s : Finset SL(2, ℤ), IsCosetCover Γ s ∧
(∫⁻ z in ⋃ a ∈ (s : Set SL(2, ℤ)), (fun z => a • z) '' fundamentalDomain,
‖f z‖ₑ ^ 2) < ∞
def IsCosetCover (Γ : Subgroup SL(2, ℤ)) (s : Finset SL(2, ℤ)) : Prop :=
∀ g : SL(2, ℤ), ∃ a ∈ s, ∃ γ ∈ Γ, g = γ * a
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:216
def fundamentalDomain : Set UpperHalfPlane :=
{z | |z.re| ≤ 1 / 2 ∧ 1 ≤ ‖(z : ℂ)‖}
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:211
Uses 2 · Used by 1
DefinitionEigenvalues of the hyperbolic Laplacian on a modular quotient#✓fully proved
Let \(\Gamma\) be a subgroup of \(\mathrm{SL}(2, \mathbb{Z})\). A real number \(\lambda\) is an eigenvalue of the hyperbolic Laplacian \(\Delta\) for \(\Gamma\) when there is a function \(f \colon \mathbb{H} \to \mathbb{C}\) on the upper half plane \(\mathbb{H}\) such that \(f\) is not identically zero, \(f\) is the restriction of a function on \(\mathbb{C}\) that is twice continuously differentiable in the real sense on the upper half plane, \(f(\gamma z) = f(z)\) for every \(\gamma \in \Gamma\) and every \(z \in \mathbb{H}\) (Möbius action), \(f\) is square-integrable on \(\Gamma \backslash \mathbb{H}\), and \(\Delta f(z) = \lambda f(z)\) for every \(z \in \mathbb{H}\).
def IsEigenvalue (Γ : Subgroup SL(2, ℤ)) (lam : ℝ) : Prop :=
∃ f : UpperHalfPlane → ℂ, f ≠ 0 ∧ IsSmooth f ∧
(∀ γ : SL(2, ℤ), γ ∈ Γ → ∀ z : UpperHalfPlane, f (γ • z) = f z) ∧
IsL2Quotient Γ f ∧
∀ z : UpperHalfPlane, hyperbolicLaplacian f z = (lam : ℂ) * f z
def IsSmooth (f : UpperHalfPlane → ℂ) : Prop :=
∃ F : ℂ → ℂ, ContDiffOn ℝ 2 F {z | 0 < z.im} ∧ ∀ z : UpperHalfPlane, f z = F z
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:206
def IsL2Quotient (Γ : Subgroup SL(2, ℤ)) (f : UpperHalfPlane → ℂ) : Prop :=
∃ s : Finset SL(2, ℤ), IsCosetCover Γ s ∧
(∫⁻ z in ⋃ a ∈ (s : Set SL(2, ℤ)), (fun z => a • z) '' fundamentalDomain,
‖f z‖ₑ ^ 2) < ∞
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:224
def hyperbolicLaplacian (f : UpperHalfPlane → ℂ) (z : UpperHalfPlane) : ℂ :=
-((z.im : ℂ) ^ 2) * (partialRe (partialRe f) z + partialIm (partialIm f) z)
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:94
Uses 3 · Used by 15
DefinitionGap contributions of a subgroup of the modular group#✓fully proved
For a subgroup \(\Gamma\) of \(\mathrm{SL}(2, \mathbb{Z})\), the set of gap contributions of \(\Gamma\) is the set of real numbers consisting of \(0\) together with \(\sqrt{1/4 - \lambda}\) for every real eigenvalue \(\lambda\) of the hyperbolic Laplacian for \(\Gamma\) with \(0 < \lambda < 1/4\) (an exceptional eigenvalue).
def exceptionalGaps (Γ : Subgroup SL(2, ℤ)) : Set ℝ :=
{0} ∪ {y | ∃ lam : ℝ, IsEigenvalue Γ lam ∧ 0 < lam ∧ lam < 1 / 4 ∧
y = Real.sqrt (1 / 4 - lam)}
def IsEigenvalue (Γ : Subgroup SL(2, ℤ)) (lam : ℝ) : Prop :=
∃ f : UpperHalfPlane → ℂ, f ≠ 0 ∧ IsSmooth f ∧
(∀ γ : SL(2, ℤ), γ ∈ Γ → ∀ z : UpperHalfPlane, f (γ • z) = f z) ∧
IsL2Quotient Γ f ∧
∀ z : UpperHalfPlane, hyperbolicLaplacian f z = (lam : ℂ) * f z
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:232
Uses 1 · Used by 8
DefinitionGap parameter of a subgroup of the modular group#✓fully proved
For a subgroup \(\Gamma\) of \(\mathrm{SL}(2, \mathbb{Z})\), the gap parameter \(\theta(\Gamma)\) is the supremum of the set of gap contributions of \(\Gamma\): the number \(0\) together with \(\sqrt{1/4 - \lambda}\) for every real eigenvalue \(\lambda\) of the hyperbolic Laplacian for \(\Gamma\) with \(0 < \lambda < 1/4\).
noncomputable def exceptionalGap (Γ : Subgroup SL(2, ℤ)) : ℝ :=
sSup (exceptionalGaps Γ)
def exceptionalGaps (Γ : Subgroup SL(2, ℤ)) : Set ℝ :=
{0} ∪ {y | ∃ lam : ℝ, IsEigenvalue Γ lam ∧ 0 < lam ∧ lam < 1 / 4 ∧
y = Real.sqrt (1 / 4 - lam)}
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:240
Uses 1 · Used by 8
DefinitionSelberg's congruence spectral-gap constant#✓fully proved
Selberg's constant \(C_{61}\) is the supremum of the gap parameters \(\theta(\Gamma)\) over all congruence subgroups \(\Gamma\) of \(\mathrm{SL}(2, \mathbb{Z})\), where \(\theta(\Gamma)\) is the supremum of \(0\) and of the numbers \(\sqrt{1/4 - \lambda}\) for the eigenvalues \(\lambda\) of \(\Gamma\) with \(0 < \lambda < 1/4\).
noncomputable def selbergConstant : ℝ :=
sSup {y | ∃ Γ : Subgroup SL(2, ℤ), IsCongruenceSubgroup Γ ∧
y = exceptionalGap Γ}
def IsCongruenceSubgroup (Γ : Subgroup SL(2, ℤ)) : Prop :=
∃ N : ℕ, 0 < N ∧ ∀ γ : SL(2, ℤ),
(∀ i j, γ.1 i j ≡ (1 : Matrix (Fin 2) (Fin 2) ℤ) i j [ZMOD (N : ℤ)]) →
γ ∈ Γ
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:64
noncomputable def exceptionalGap (Γ : Subgroup SL(2, ℤ)) : ℝ :=
sSup (exceptionalGaps Γ)
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:245
Uses 2 · Used by 8
DefinitionSelberg's eigenvalue conjecture#✓fully proved
No congruence subgroup \(\Gamma\) of \(\mathrm{SL}(2, \mathbb{Z})\) has an eigenvalue \(\lambda\) with \(0 < \lambda < 1/4\): for every congruence subgroup \(\Gamma\) and every eigenvalue \(\lambda\) of \(\Gamma\), either \(\lambda \le 0\) or \(\lambda \ge 1/4\).
Here eigenvalues are realized by smooth square-integrable eigenfunctions, which captures the discrete spectrum of the hyperbolic Laplacian.
def SelbergEigenvalueConjecture : Prop :=
∀ Γ : Subgroup SL(2, ℤ), IsCongruenceSubgroup Γ → ∀ lam : ℝ,
IsEigenvalue Γ lam → ¬ (0 < lam ∧ lam < 1 / 4)
def IsCongruenceSubgroup (Γ : Subgroup SL(2, ℤ)) : Prop :=
∃ N : ℕ, 0 < N ∧ ∀ γ : SL(2, ℤ),
(∀ i j, γ.1 i j ≡ (1 : Matrix (Fin 2) (Fin 2) ℤ) i j [ZMOD (N : ℤ)]) →
γ ∈ Γ
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:64
def IsEigenvalue (Γ : Subgroup SL(2, ℤ)) (lam : ℝ) : Prop :=
∃ f : UpperHalfPlane → ℂ, f ≠ 0 ∧ IsSmooth f ∧
(∀ γ : SL(2, ℤ), γ ∈ Γ → ∀ z : UpperHalfPlane, f (γ • z) = f z) ∧
IsL2Quotient Γ f ∧
∀ z : UpperHalfPlane, hyperbolicLaplacian f z = (lam : ℂ) * f z
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:232
Uses 2 · Used by 1
TheoremExceptional eigenvalues bound the gap parameter#✓fully proved
Let \(\Gamma\) be a subgroup of \(\mathrm{SL}(2, \mathbb{Z})\) and let \(\lambda\) be a real eigenvalue of the hyperbolic Laplacian on \(L^2(\Gamma \backslash \mathbb{H})\) with \(0 < \lambda < 1/4\). Then \(\sqrt{1/4 - \lambda}\) is at most the gap parameter \(\theta(\Gamma)\) of \(\Gamma\).
theorem sqrt_le_exceptionalGap {Γ : Subgroup SL(2, ℤ)} {lam : ℝ}
(hlam : IsEigenvalue Γ lam) (hpos : 0 < lam) (hlt : lam < 1 / 4) :
Real.sqrt (1 / 4 - lam) ≤ exceptionalGap Γ
def IsEigenvalue (Γ : Subgroup SL(2, ℤ)) (lam : ℝ) : Prop :=
∃ f : UpperHalfPlane → ℂ, f ≠ 0 ∧ IsSmooth f ∧
(∀ γ : SL(2, ℤ), γ ∈ Γ → ∀ z : UpperHalfPlane, f (γ • z) = f z) ∧
IsL2Quotient Γ f ∧
∀ z : UpperHalfPlane, hyperbolicLaplacian f z = (lam : ℂ) * f z
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:232
noncomputable def exceptionalGap (Γ : Subgroup SL(2, ℤ)) : ℝ :=
sSup (exceptionalGaps Γ)
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:245
Uses 3 · Used by 1
TheoremCongruence gaps are bounded by Selberg's constant#✓fully proved
For every congruence subgroup \(\Gamma\) of \(\mathrm{SL}(2, \mathbb{Z})\), the gap parameter \(\theta(\Gamma)\) is at most the Selberg constant \(C_{61}\).
theorem exceptionalGap_le_selbergConstant {Γ : Subgroup SL(2, ℤ)}
(hΓ : IsCongruenceSubgroup Γ) : exceptionalGap Γ ≤ selbergConstant
def IsCongruenceSubgroup (Γ : Subgroup SL(2, ℤ)) : Prop :=
∃ N : ℕ, 0 < N ∧ ∀ γ : SL(2, ℤ),
(∀ i j, γ.1 i j ≡ (1 : Matrix (Fin 2) (Fin 2) ℤ) i j [ZMOD (N : ℤ)]) →
γ ∈ Γ
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:64
noncomputable def exceptionalGap (Γ : Subgroup SL(2, ℤ)) : ℝ :=
sSup (exceptionalGaps Γ)
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:245
noncomputable def selbergConstant : ℝ :=
sSup {y | ∃ Γ : Subgroup SL(2, ℤ), IsCongruenceSubgroup Γ ∧
y = exceptionalGap Γ}
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:250
TheoremSelberg's constant is at most one half#✓fully proved
Selberg's constant \(C_{61}\) satisfies \(C_{61} \le 1/2\).
This elementary bound follows directly from the definition.
theorem selbergConstant_le_half : selbergConstant ≤ 1 / 2
noncomputable def selbergConstant : ℝ :=
sSup {y | ∃ Γ : Subgroup SL(2, ℤ), IsCongruenceSubgroup Γ ∧
y = exceptionalGap Γ}
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:250
Uses 6
TheoremExceptional eigenvalues bound Selberg's constant#✓fully proved
Let \(\Gamma\) be a congruence subgroup of \(\mathrm{SL}(2, \mathbb{Z})\) and let \(\lambda\) be a real eigenvalue of the hyperbolic Laplacian on \(L^2(\Gamma \backslash \mathbb{H})\) with \(0 < \lambda < 1/4\). Then \(\sqrt{1/4 - \lambda}\) is at most Selberg's constant \(C_{61}\).
theorem sqrt_le_selbergConstant {Γ : Subgroup SL(2, ℤ)}
(hΓ : IsCongruenceSubgroup Γ) {lam : ℝ} (hlam : IsEigenvalue Γ lam)
(hpos : 0 < lam) (hlt : lam < 1 / 4) :
Real.sqrt (1 / 4 - lam) ≤ selbergConstant
def IsCongruenceSubgroup (Γ : Subgroup SL(2, ℤ)) : Prop :=
∃ N : ℕ, 0 < N ∧ ∀ γ : SL(2, ℤ),
(∀ i j, γ.1 i j ≡ (1 : Matrix (Fin 2) (Fin 2) ℤ) i j [ZMOD (N : ℤ)]) →
γ ∈ Γ
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:64
def IsEigenvalue (Γ : Subgroup SL(2, ℤ)) (lam : ℝ) : Prop :=
∃ f : UpperHalfPlane → ℂ, f ≠ 0 ∧ IsSmooth f ∧
(∀ γ : SL(2, ℤ), γ ∈ Γ → ∀ z : UpperHalfPlane, f (γ • z) = f z) ∧
IsL2Quotient Γ f ∧
∀ z : UpperHalfPlane, hyperbolicLaplacian f z = (lam : ℂ) * f z
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:232
noncomputable def selbergConstant : ℝ :=
sSup {y | ∃ Γ : Subgroup SL(2, ℤ), IsCongruenceSubgroup Γ ∧
y = exceptionalGap Γ}
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:250
Uses 6 · Used by 2
TheoremSelberg's constant vanishes exactly under Selberg's eigenvalue conjecture#✓fully proved
Selberg's constant \(C_{61}\) equals \(0\) if and only if Selberg's eigenvalue conjecture holds, that is, if and only if no congruence subgroup of \(\mathrm{SL}(2, \mathbb{Z})\) has an eigenvalue \(\lambda\) with \(0 < \lambda < 1/4\).
theorem selbergConstant_eq_zero_iff :
selbergConstant = 0 ↔ SelbergEigenvalueConjecture
noncomputable def selbergConstant : ℝ :=
sSup {y | ∃ Γ : Subgroup SL(2, ℤ), IsCongruenceSubgroup Γ ∧
y = exceptionalGap Γ}
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:250
def SelbergEigenvalueConjecture : Prop :=
∀ Γ : Subgroup SL(2, ℤ), IsCongruenceSubgroup Γ → ∀ lam : ℝ,
IsEigenvalue Γ lam → ¬ (0 < lam ∧ lam < 1 / 4)
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:256
Uses 9
TheoremSelberg's 3/16 theorem#●ready to prove
Let \(\Gamma\) be a congruence subgroup of \(\mathrm{SL}(2, \mathbb{Z})\) and let \(\lambda\) be a real eigenvalue of the hyperbolic Laplacian \(\Delta\) for \(\Gamma\), that is, \(\Delta f = \lambda f\) for some nonzero, twice continuously differentiable, \(\Gamma\)-invariant function \(f\) on the upper half plane that is square-integrable on \(\Gamma \backslash \mathbb{H}\). Then \(\lambda\) does not lie in the open interval \((0, 3/16)\): either \(\lambda \le 0\) or \(\lambda \ge 3/16\).
theorem_wanted selberg_threeSixteenths_wanted
(Γ : Subgroup SL(2, ℤ)) (hΓ : IsCongruenceSubgroup Γ) (lam : ℝ)
(hlam : IsEigenvalue Γ lam) :
¬ (0 < lam ∧ lam < 3 / 16)
def IsCongruenceSubgroup (Γ : Subgroup SL(2, ℤ)) : Prop :=
∃ N : ℕ, 0 < N ∧ ∀ γ : SL(2, ℤ),
(∀ i j, γ.1 i j ≡ (1 : Matrix (Fin 2) (Fin 2) ℤ) i j [ZMOD (N : ℤ)]) →
γ ∈ Γ
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:64
def IsEigenvalue (Γ : Subgroup SL(2, ℤ)) (lam : ℝ) : Prop :=
∃ f : UpperHalfPlane → ℂ, f ≠ 0 ∧ IsSmooth f ∧
(∀ γ : SL(2, ℤ), γ ∈ Γ → ∀ z : UpperHalfPlane, f (γ • z) = f z) ∧
IsL2Quotient Γ f ∧
∀ z : UpperHalfPlane, hyperbolicLaplacian f z = (lam : ℂ) * f z
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:232
Sources
- Selberg, as recorded in Luo--Rudnick--Sarnak (1995): every eigenvalue of a congruence subgroup in the model is either
0-side excluded or at least3/16; stated here as the absence of eigenvalues in(0, 3/16).
TheoremSelberg's bound on the spectral-gap constant#●ready to prove
Selberg's congruence spectral-gap constant \(C_{61}\) is at most \(1/4\).
theorem_wanted selbergUpperBound_wanted :
selbergConstant ≤ 1 / 4
noncomputable def selbergConstant : ℝ :=
sSup {y | ∃ Γ : Subgroup SL(2, ℤ), IsCongruenceSubgroup Γ ∧
y = exceptionalGap Γ}
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:250
Sources
- Selberg's theorem
λ₁(Γ) ≥ 3/16, catalogue upper rowC₆₁ ≤ 1/4.
Uses 1
TheoremKim–Sarnak eigenvalue bound#●ready to prove
Let \(\Gamma\) be a congruence subgroup of \(\mathrm{SL}(2, \mathbb{Z})\) and let \(\lambda\) be a real eigenvalue of the hyperbolic Laplacian \(\Delta\) for \(\Gamma\), that is, \(\Delta f = \lambda f\) for some nonzero, twice continuously differentiable, \(\Gamma\)-invariant function \(f\) on the upper half plane that is square-integrable on \(\Gamma \backslash \mathbb{H}\). Then \(\lambda\) does not lie in the open interval \((0, 975/4096)\): either \(\lambda \le 0\) or \(\lambda \ge 975/4096\).
theorem_wanted kimSarnak975_wanted
(Γ : Subgroup SL(2, ℤ)) (hΓ : IsCongruenceSubgroup Γ) (lam : ℝ)
(hlam : IsEigenvalue Γ lam) :
¬ (0 < lam ∧ lam < 975 / 4096)
def IsCongruenceSubgroup (Γ : Subgroup SL(2, ℤ)) : Prop :=
∃ N : ℕ, 0 < N ∧ ∀ γ : SL(2, ℤ),
(∀ i j, γ.1 i j ≡ (1 : Matrix (Fin 2) (Fin 2) ℤ) i j [ZMOD (N : ℤ)]) →
γ ∈ Γ
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:64
def IsEigenvalue (Γ : Subgroup SL(2, ℤ)) (lam : ℝ) : Prop :=
∃ f : UpperHalfPlane → ℂ, f ≠ 0 ∧ IsSmooth f ∧
(∀ γ : SL(2, ℤ), γ ∈ Γ → ∀ z : UpperHalfPlane, f (γ • z) = f z) ∧
IsL2Quotient Γ f ∧
∀ z : UpperHalfPlane, hyperbolicLaplacian f z = (lam : ℂ) * f z
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:232
Sources
- Kim--Sarnak (2003), as recorded in Kim--Ramakrishnan (2003): no eigenvalue of a congruence subgroup lies in
(0, 975/4096).
TheoremKim–Sarnak bound on the spectral-gap constant#●ready to prove
Selberg's congruence spectral-gap constant \(C_{61}\) is at most \(7/64\).
theorem_wanted kimSarnakUpperBound_wanted :
selbergConstant ≤ 7 / 64
noncomputable def selbergConstant : ℝ :=
sSup {y | ∃ Γ : Subgroup SL(2, ℤ), IsCongruenceSubgroup Γ ∧
y = exceptionalGap Γ}
MathlibExt/NumberTheory/ModularForms/SelbergSpectralGap.lean:250
Sources
- Kim--Sarnak bound
λ₁(Γ) ≥ 975/4096 = 1/4 - (7/64)^2, catalogue best upper rowC₆₁ ≤ 7/64.
Uses 1
Loading dependency explorer…
The dependency explorer needs JavaScript and its graph data.