Constructions of thick B_h-sets
DefinitionGeneralized Bose set#✓fully proved
Let \(F\) be a finite field with \(q\) elements, \(K\) a field extension of \(F\), \(h\) and \(b\) natural numbers and \(\theta \in K\). The generalized Bose set \(\mathrm{Bose}_h(q, b)\) is the set of residues \(a \in \mathbb{Z}/(q^h - 1)\mathbb{Z}\) such that \(\theta^a = \theta^b + v\) for some \(v \in F\), where \(\theta^a\) is computed with the least nonnegative representative of \(a\) (for \(h = 0\) the modulus is \(0\), \(a\) ranges over \(\mathbb{Z}\) and \(\theta^a\) means \(\theta^{|a|}\)).
In the paper \(K\) has degree \(h\) over \(F\) and \(\theta\) generates \(K^\times\); the definition does not need these assumptions.
public def boseSet (F : Type*) {K : Type*} [Field F] [Fintype F] [Field K] [Algebra F K]
(h : ℕ) (θ : K) (b : ℕ) : Set (ZMod (Fintype.card F ^ h - 1)) :=
{a | ∃ v : F, θ ^ a.val = θ ^ b + algebraMap F K v}
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, Definition 1.
TheoremMembership in a Bose set#✓fully proved
Let \(F\) be a finite field with \(q\) elements, let \(K\) be a field extension of \(F\), let \(\theta \in K\), and let \(h\) and \(b\) be natural numbers. A residue \(a \in \mathbb{Z}/(q^h - 1)\mathbb{Z}\) lies in the Bose set \(\mathrm{Bose}_h(q, b)\) built from \(F\), \(K\) and \(\theta\) if and only if \(\theta^a = \theta^b + v\) for some \(v \in F\), where \(\theta^a\) means \(\theta\) raised to the least nonnegative representative of \(a\) (or to \(|a|\) when \(h = 0\), since then \(\mathbb{Z}/(q^h - 1)\mathbb{Z} = \mathbb{Z}\)).
public theorem mem_boseSet {F K : Type*} [Field F] [Fintype F] [Field K] [Algebra F K]
(h : ℕ) (θ : K) (b : ℕ) (a : ZMod (Fintype.card F ^ h - 1)) :
a ∈ boseSet F h θ b ↔ ∃ v : F, θ ^ a.val = θ ^ b + algebraMap F K v
public def boseSet (F : Type*) {K : Type*} [Field F] [Fintype F] [Field K] [Algebra F K]
(h : ℕ) (θ : K) (b : ℕ) : Set (ZMod (Fintype.card F ^ h - 1)) :=
{a | ∃ v : F, θ ^ a.val = θ ^ b + algebraMap F K v}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:42
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, Definition 1.
Uses 1
DefinitionGeneralized Singer set#✓fully proved
Let \(F\) be a finite field with \(q\) elements, let \(K\) be a field extension of \(F\), let \(\theta \in K\), let \(h\) and \(b\) be natural numbers, and put \(M = (q^{h+1} - 1)/(q - 1)\). The generalized Singer set \(\mathrm{Singer}_h(q, b)\) is the set of residues \(a \in \mathbb{Z}/M\mathbb{Z}\) such that \(\theta^a = u\theta^b + v\) for some \(u, v \in F\), where \(\theta^a\) means \(\theta\) raised to the least nonnegative representative of \(a\).
The construction is intended for \([K : F] = h + 1\) and \(\theta\) a generator of \(K^\times\); the definition itself does not assume either.
public def singerSet (F : Type*) {K : Type*} [Field F] [Fintype F] [Field K] [Algebra F K]
(h : ℕ) (θ : K) (b : ℕ) :
Set (ZMod ((Fintype.card F ^ (h + 1) - 1) / (Fintype.card F - 1))) :=
{a | ∃ u v : F, θ ^ a.val = algebraMap F K u * θ ^ b + algebraMap F K v}
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, Definition 2.
TheoremMembership in a Singer set#✓fully proved
Let \(F\) be a finite field with \(q\) elements, let \(K\) be a field extension of \(F\), let \(\theta \in K\), let \(h\) and \(b\) be natural numbers, and put \(M = (q^{h+1} - 1)/(q - 1)\). A residue \(a \in \mathbb{Z}/M\mathbb{Z}\) lies in the Singer set \(\mathrm{Singer}_h(q, b)\) built from \(F\), \(K\) and \(\theta\) if and only if \(\theta^a = u\theta^b + v\) for some \(u, v \in F\), where \(\theta^a\) means \(\theta\) raised to the least nonnegative representative of \(a\).
public theorem mem_singerSet {F K : Type*} [Field F] [Fintype F] [Field K] [Algebra F K]
(h : ℕ) (θ : K) (b : ℕ)
(a : ZMod ((Fintype.card F ^ (h + 1) - 1) / (Fintype.card F - 1))) :
a ∈ singerSet F h θ b ↔
∃ u v : F, θ ^ a.val = algebraMap F K u * θ ^ b + algebraMap F K v
public def singerSet (F : Type*) {K : Type*} [Field F] [Fintype F] [Field K] [Algebra F K]
(h : ℕ) (θ : K) (b : ℕ) :
Set (ZMod ((Fintype.card F ^ (h + 1) - 1) / (Fintype.card F - 1))) :=
{a | ∃ u v : F, θ ^ a.val = algebraMap F K u * θ ^ b + algebraMap F K v}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:62
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, Definition 2.
Uses 1
DefinitionAffine equivalence of sets of residues#✓fully proved
For a natural number \(M\), sets \(A_1, A_2 \subseteq \mathbb{Z}/M\mathbb{Z}\) are affinely equivalent, written \(A_1 \sim A_2\), if there are a unit \(d\) of \(\mathbb{Z}/M\mathbb{Z}\) and an element \(s \in \mathbb{Z}/M\mathbb{Z}\) such that \(A_2 = \{d a + s : a \in A_1\}\).
public def ZModAffinelyEquivalent {M : ℕ} (A₁ A₂ : Set (ZMod M)) : Prop :=
∃ d s : ZMod M, IsUnit d ∧ A₂ = (fun a => d * a + s) '' A₁
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, the definition before Theorem 4.
TheoremAffine equivalence is reflexive#✓fully proved
For every natural number \(M\), every subset \(A\) of \(\mathbb{Z}/M\mathbb{Z}\) is affinely equivalent to itself (take \(d = 1\) and \(s = 0\)).
public theorem zModAffinelyEquivalent_refl {M : ℕ} (A : Set (ZMod M)) :
ZModAffinelyEquivalent A A
public def ZModAffinelyEquivalent {M : ℕ} (A₁ A₂ : Set (ZMod M)) : Prop :=
∃ d s : ZMod M, IsUnit d ∧ A₂ = (fun a => d * a + s) '' A₁
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:83
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, the definition before Theorem 4.
Uses 1
DefinitionMaximum size of a \(B_h\)-set in \(\{1, \dots, n\}\)#✓fully proved
For natural numbers \(h\) and \(n\), \(R_h(n)\) is the largest number of elements of a \(B_h\)-set of natural numbers contained in \(\{1, \dots, n\}\). If no \(B_h\)-set is contained in \(\{1, \dots, n\}\) (for example when \(n = 0\)), then \(R_h(n) = 0\).
public noncomputable def bhMaxCard (h n : ℕ) : ℕ :=
sSup {k | ∃ A : Finset ℕ, A ⊆ Finset.Icc 1 n ∧
AdditiveCombinatorics.IsBhSet (A : Set ℕ) h ∧ A.card = k}
public def IsBhSet {G : Type*} [AddCommSemigroup G] (A : Set G) (h : ℕ) : Prop :=
A.Nonempty ∧ 0 < h ∧
∀ (l₁ l₂ : List G),
(∀ x ∈ l₁, x ∈ A) →
(∀ x ∈ l₂, x ∈ A) →
l₁.length = h →
l₂.length = h →
∀ (e₁ : l₁ ≠ []) (e₂ : l₂ ≠ []),
sumNE l₁ e₁ = sumNE l₂ e₂ → l₁.Perm l₂
MathlibExt/Combinatorics/Additive/BhSet.lean:62
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, the paragraph before Theorem 5.
Uses 1 · Used by 10
DefinitionShortest interval containing a \(B_h\)-set of given size#✓fully proved
For natural numbers \(h\) and \(k\), \(R_h^{-1}(k)\) is the smallest natural number \(n\) such that some \(B_h\)-set of natural numbers with exactly \(k\) elements is contained in \(\{1, \dots, n\}\); it is \(0\) when no such \(n\) exists, for example when \(k = 0\), since a \(B_h\)-set is nonempty.
public noncomputable def bhMinLength (h k : ℕ) : ℕ :=
sInf {n | ∃ A : Finset ℕ, A ⊆ Finset.Icc 1 n ∧
AdditiveCombinatorics.IsBhSet (A : Set ℕ) h ∧ A.card = k}
public def IsBhSet {G : Type*} [AddCommSemigroup G] (A : Set G) (h : ℕ) : Prop :=
A.Nonempty ∧ 0 < h ∧
∀ (l₁ l₂ : List G),
(∀ x ∈ l₁, x ∈ A) →
(∀ x ∈ l₂, x ∈ A) →
l₁.length = h →
l₂.length = h →
∀ (e₁ : l₁ ≠ []) (e₂ : l₂ ≠ []),
sumNE l₁ e₁ = sumNE l₂ e₂ → l₁.Perm l₂
MathlibExt/Combinatorics/Additive/BhSet.lean:62
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, the paragraph before Theorem 5.
Uses 1 · Used by 9
Theorem\(R_h(n)\) is at most \(n\)#✓fully proved
For all natural numbers \(h\) and \(n\), let \(R_h(n)\) be the largest size of a \(B_h\)-set contained in \(\{1, \dots, n\}\), or \(0\) if there is no such set (for example when \(h = 0\) or \(n = 0\)). Then \(R_h(n) \le n\).
public theorem bhMaxCard_le (h n : ℕ) : bhMaxCard h n ≤ n
public noncomputable def bhMaxCard (h n : ℕ) : ℕ :=
sSup {k | ∃ A : Finset ℕ, A ⊆ Finset.Icc 1 n ∧
AdditiveCombinatorics.IsBhSet (A : Set ℕ) h ∧ A.card = k}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:98
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, the definition ofR_h(n)before Theorem 5.
Uses 2 · Used by 1
Theorem\(R_h(1) = 1\)#✓fully proved
For every natural number \(h \ge 1\), the largest size \(R_h(1)\) of a \(B_h\)-set contained in \(\{1\}\) equals \(1\).
public theorem bhMaxCard_one {h : ℕ} (hh : 0 < h) : bhMaxCard h 1 = 1
public noncomputable def bhMaxCard (h n : ℕ) : ℕ :=
sSup {k | ∃ A : Finset ℕ, A ⊆ Finset.Icc 1 n ∧
AdditiveCombinatorics.IsBhSet (A : Set ℕ) h ∧ A.card = k}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:98
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, the definition ofR_h(n)before Theorem 5.
Uses 4
TheoremThe shortest interval holding a one-element \(B_h\)-set#✓fully proved
For every natural number \(h \ge 1\), the smallest natural number \(n\) such that \(\{1, \dots, n\}\) contains a \(B_h\)-set with exactly one element is \(1\); that is, \(R_h^{-1}(1) = 1\).
public theorem bhMinLength_one {h : ℕ} (hh : 0 < h) : bhMinLength h 1 = 1
public noncomputable def bhMinLength (h k : ℕ) : ℕ :=
sInf {n | ∃ A : Finset ℕ, A ⊆ Finset.Icc 1 n ∧
AdditiveCombinatorics.IsBhSet (A : Set ℕ) h ∧ A.card = k}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:107
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, the paragraph before Theorem 7 ("Clearly,R_h⁻¹(1) = 1").
Uses 3
TheoremBose's construction of \(B_h\)-sets#●ready to prove
Let \(h \ge 2\), let \(F\) be a finite field with \(q\) elements, let \(K\) be a field extension of \(F\) of degree \(h\), let \(\theta\) be a generator of the multiplicative group \(K^\times\), and let \(b\) be a natural number such that \(\theta^b\) has degree \(h\) over \(F\). Then the generalized Bose set \(\mathrm{Bose}_h(q, b)\), a subset of \(\mathbb{Z}/(q^h - 1)\mathbb{Z}\), is a \(B_h\)-set with exactly \(q\) elements.
public theorem_wanted isBhSet_boseSet {F K : Type*} [Field F] [Fintype F] [Field K]
[Fintype K] [Algebra F K] {h : ℕ} (hh : 2 ≤ h) (hK : Module.finrank F K = h) {θ : K}
(hθ : orderOf θ = Fintype.card K - 1) {b : ℕ}
(hb : (minpoly F (θ ^ b)).natDegree = h) :
AdditiveCombinatorics.IsBhSet (boseSet F h θ b) h ∧
(boseSet F h θ b).ncard = Fintype.card F
public def IsBhSet {G : Type*} [AddCommSemigroup G] (A : Set G) (h : ℕ) : Prop :=
A.Nonempty ∧ 0 < h ∧
∀ (l₁ l₂ : List G),
(∀ x ∈ l₁, x ∈ A) →
(∀ x ∈ l₂, x ∈ A) →
l₁.length = h →
l₂.length = h →
∀ (e₁ : l₁ ≠ []) (e₂ : l₂ ≠ []),
sumNE l₁ e₁ = sumNE l₂ e₂ → l₁.Perm l₂
MathlibExt/Combinatorics/Additive/BhSet.lean:62
public def boseSet (F : Type*) {K : Type*} [Field F] [Fintype F] [Field K] [Algebra F K]
(h : ℕ) (θ : K) (b : ℕ) : Set (ZMod (Fintype.card F ^ h - 1)) :=
{a | ∃ v : F, θ ^ a.val = θ ^ b + algebraMap F K v}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:42
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, Theorem 3 (first sentence) (jis-2024-24.1.2-thm-3).
Uses 2
TheoremSinger's construction of \(B_h\)-sets#●ready to prove
Let \(h \ge 2\), let \(F\) be a finite field with \(q\) elements, let \(K\) be a field extension of \(F\) of degree \(h + 1\), let \(\theta\) be a generator of the multiplicative group \(K^\times\), and let \(b\) be a natural number such that \(\theta^b\) has degree \(h + 1\) over \(F\). Then the generalized Singer set \(\mathrm{Singer}_h(q, b)\), a subset of \(\mathbb{Z}/M\mathbb{Z}\) with \(M = (q^{h+1} - 1)/(q - 1)\), is a \(B_h\)-set with exactly \(q + 1\) elements.
public theorem_wanted isBhSet_singerSet {F K : Type*} [Field F] [Fintype F] [Field K]
[Fintype K] [Algebra F K] {h : ℕ} (hh : 2 ≤ h) (hK : Module.finrank F K = h + 1) {θ : K}
(hθ : orderOf θ = Fintype.card K - 1) {b : ℕ}
(hb : (minpoly F (θ ^ b)).natDegree = h + 1) :
AdditiveCombinatorics.IsBhSet (singerSet F h θ b) h ∧
(singerSet F h θ b).ncard = Fintype.card F + 1
public def IsBhSet {G : Type*} [AddCommSemigroup G] (A : Set G) (h : ℕ) : Prop :=
A.Nonempty ∧ 0 < h ∧
∀ (l₁ l₂ : List G),
(∀ x ∈ l₁, x ∈ A) →
(∀ x ∈ l₂, x ∈ A) →
l₁.length = h →
l₂.length = h →
∀ (e₁ : l₁ ≠ []) (e₂ : l₂ ≠ []),
sumNE l₁ e₁ = sumNE l₂ e₂ → l₁.Perm l₂
MathlibExt/Combinatorics/Additive/BhSet.lean:62
public def singerSet (F : Type*) {K : Type*} [Field F] [Fintype F] [Field K] [Algebra F K]
(h : ℕ) (θ : K) (b : ℕ) :
Set (ZMod ((Fintype.card F ^ (h + 1) - 1) / (Fintype.card F - 1))) :=
{a | ∃ u v : F, θ ^ a.val = algebraMap F K u * θ ^ b + algebraMap F K v}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:62
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, Theorem 3 (second sentence) (jis-2024-24.1.2-thm-3).
Uses 2
TheoremAffine equivalences between Bose sets#●ready to prove
Let \(F\) be a finite field with \(q\) elements and characteristic \(p\), let \(h \ge 2\), let \(K\) be a field extension of \(F\) of degree \(h\), and let \(\theta \in K\) have order \(|K| - 1\), that is, generate \(K^\times\). Let \(b\) and \(e\) be natural numbers such that the minimal polynomials of \(\theta^b\) and \(\theta^e\) over \(F\) both have degree \(h\). Suppose that (i) \(b \equiv e \pmod{(q^h - 1)/(q - 1)}\), or (ii) \(\theta^b - \theta^e \in F\), or (iii) \(b \equiv p^i e \pmod{q^h - 1}\) for some natural number \(i\). Then the Bose sets \(\mathrm{Bose}_h(q, b)\) and \(\mathrm{Bose}_h(q, e)\) built from \(F\), \(K\) and \(\theta\) are affinely equivalent.
public theorem_wanted zModAffinelyEquivalent_boseSet {F K : Type*} [Field F] [Fintype F]
[Field K] [Fintype K] [Algebra F K] {h : ℕ} (hh : 2 ≤ h) (hK : Module.finrank F K = h)
{θ : K} (hθ : orderOf θ = Fintype.card K - 1) {b e : ℕ}
(hb : (minpoly F (θ ^ b)).natDegree = h) (he : (minpoly F (θ ^ e)).natDegree = h)
(hbe : b ≡ e [MOD (Fintype.card F ^ h - 1) / (Fintype.card F - 1)] ∨
θ ^ b - θ ^ e ∈ Set.range (algebraMap F K) ∨
∃ i : ℕ, b ≡ ringChar F ^ i * e [MOD Fintype.card F ^ h - 1]) :
ZModAffinelyEquivalent (boseSet F h θ b) (boseSet F h θ e)
public def ZModAffinelyEquivalent {M : ℕ} (A₁ A₂ : Set (ZMod M)) : Prop :=
∃ d s : ZMod M, IsUnit d ∧ A₂ = (fun a => d * a + s) '' A₁
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:83
public def boseSet (F : Type*) {K : Type*} [Field F] [Fintype F] [Field K] [Algebra F K]
(h : ℕ) (θ : K) (b : ℕ) : Set (ZMod (Fintype.card F ^ h - 1)) :=
{a | ∃ v : F, θ ^ a.val = θ ^ b + algebraMap F K v}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:42
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, Theorem 4 (items (i)–(iii)) (jis-2024-24.1.2-thm-4).
Uses 2
TheoremAffine equivalences between Singer sets#●ready to prove
Let \(F\) be a finite field with \(q\) elements and characteristic \(p\), let \(h \ge 2\), let \(K\) be a field extension of \(F\) of degree \(h + 1\), and let \(\theta \in K\) have order \(|K| - 1\), that is, generate \(K^\times\). Put \(M = (q^{h+1} - 1)/(q - 1)\). Let \(b\) and \(e\) be natural numbers such that the minimal polynomials of \(\theta^b\) and \(\theta^e\) over \(F\) both have degree \(h + 1\). Suppose that (iv) \(b \equiv e \pmod{M}\), or (v) \(b \equiv p^i e \pmod{q^{h+1} - 1}\) for some natural number \(i\), or (vi) \(\theta^b - \theta^e \in F\), or (vii) \(r\theta^b + t\theta^{e+b} + w\theta^e \in F\) for some \(r, t, w \in F\) with \(r\) and \(t\) not both \(0\). Then the Singer sets \(\mathrm{Singer}_h(q, b)\) and \(\mathrm{Singer}_h(q, e)\) built from \(F\), \(K\) and \(\theta\) are affinely equivalent.
public theorem_wanted zModAffinelyEquivalent_singerSet {F K : Type*} [Field F] [Fintype F]
[Field K] [Fintype K] [Algebra F K] {h : ℕ} (hh : 2 ≤ h)
(hK : Module.finrank F K = h + 1) {θ : K} (hθ : orderOf θ = Fintype.card K - 1) {b e : ℕ}
(hb : (minpoly F (θ ^ b)).natDegree = h + 1)
(he : (minpoly F (θ ^ e)).natDegree = h + 1)
(hbe : b ≡ e [MOD (Fintype.card F ^ (h + 1) - 1) / (Fintype.card F - 1)] ∨
(∃ i : ℕ, b ≡ ringChar F ^ i * e [MOD Fintype.card F ^ (h + 1) - 1]) ∨
θ ^ b - θ ^ e ∈ Set.range (algebraMap F K) ∨
∃ r t w : F, (r ≠ 0 ∨ t ≠ 0) ∧
algebraMap F K r * θ ^ b + algebraMap F K t * θ ^ (e + b) +
algebraMap F K w * θ ^ e ∈ Set.range (algebraMap F K)) :
ZModAffinelyEquivalent (singerSet F h θ b) (singerSet F h θ e)
public def ZModAffinelyEquivalent {M : ℕ} (A₁ A₂ : Set (ZMod M)) : Prop :=
∃ d s : ZMod M, IsUnit d ∧ A₂ = (fun a => d * a + s) '' A₁
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:83
public def singerSet (F : Type*) {K : Type*} [Field F] [Fintype F] [Field K] [Algebra F K]
(h : ℕ) (θ : K) (b : ℕ) :
Set (ZMod ((Fintype.card F ^ (h + 1) - 1) / (Fintype.card F - 1))) :=
{a | ∃ u v : F, θ ^ a.val = algebraMap F K u * θ ^ b + algebraMap F K v}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:62
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, Theorem 4 (items (iv)–(vii)) (jis-2024-24.1.2-thm-4).
Uses 2
TheoremUnconditional bounds for \(R_h(n)\) and \(R_h^{-1}(k)\)#●ready to prove
Let \(h \ge 2\). Write \(R_h(n)\) for the largest size of a \(B_h\)-set contained in \(\{1, \dots, n\}\), with \(R_h(0) = 0\), and \(R_h^{-1}(k)\) for the least \(n\) such that \(\{1, \dots, n\}\) contains a \(B_h\)-set with \(k\) elements, with \(R_h^{-1}(0) = 0\) (a \(B_h\)-set is nonempty). Then every natural number \(n\) satisfies \(R_h(n) \ge n^{1/h} - 2^{44} n^{154/(155h)}\), and every natural number \(k\) satisfies \(R_h^{-1}(k) \le k^h + 3^{155h} k^{h - 1/155}\).
public theorem_wanted bhMaxCard_bhMinLength_bounds_unconditional {h : ℕ} (hh : 2 ≤ h) :
(∀ n : ℕ, (n : ℝ) ^ (1 / (h : ℝ)) - 2 ^ 44 * (n : ℝ) ^ ((154 : ℝ) / (155 * h)) ≤
bhMaxCard h n) ∧
∀ k : ℕ, (bhMinLength h k : ℝ) ≤
(k : ℝ) ^ h + 3 ^ (155 * h) * (k : ℝ) ^ ((h : ℝ) - 1 / 155)
public noncomputable def bhMaxCard (h n : ℕ) : ℕ :=
sSup {k | ∃ A : Finset ℕ, A ⊆ Finset.Icc 1 n ∧
AdditiveCombinatorics.IsBhSet (A : Set ℕ) h ∧ A.card = k}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:98
public noncomputable def bhMinLength (h k : ℕ) : ℕ :=
sInf {n | ∃ A : Finset ℕ, A ⊆ Finset.Icc 1 n ∧
AdditiveCombinatorics.IsBhSet (A : Set ℕ) h ∧ A.card = k}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:107
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, Theorem 5(a) (jis-2024-24.1.2-thm-5).
TheoremBounds for \(R_h(n)\) and \(R_h^{-1}(k)\) beyond \(e^{e^{34}}\)#●ready to prove
Let \(h \ge 2\). Write \(R_h(n)\) for the largest size of a \(B_h\)-set contained in \(\{1, \dots, n\}\) and \(R_h^{-1}(k)\) for the least \(n\) such that \(\{1, \dots, n\}\) contains a \(B_h\)-set with \(k\) elements. Then every natural number \(n \ge e^{e^{34}}\) satisfies \(R_h(n) \ge n^{1/h} - 7 n^{2/(3h)}\), and every natural number \(k \ge e^{e^{34}}\) satisfies \(R_h^{-1}(k) \le k^h + (3k)^{h - 1/3}\).
public theorem_wanted bhMaxCard_bhMinLength_bounds_large {h : ℕ} (hh : 2 ≤ h) :
(∀ n : ℕ, Real.exp (Real.exp 34) ≤ n →
(n : ℝ) ^ (1 / (h : ℝ)) - 7 * (n : ℝ) ^ ((2 : ℝ) / (3 * h)) ≤ bhMaxCard h n) ∧
∀ k : ℕ, Real.exp (Real.exp 34) ≤ k →
(bhMinLength h k : ℝ) ≤ (k : ℝ) ^ h + (3 * k : ℝ) ^ ((h : ℝ) - 1 / 3)
public noncomputable def bhMaxCard (h n : ℕ) : ℕ :=
sSup {k | ∃ A : Finset ℕ, A ⊆ Finset.Icc 1 n ∧
AdditiveCombinatorics.IsBhSet (A : Set ℕ) h ∧ A.card = k}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:98
public noncomputable def bhMinLength (h k : ℕ) : ℕ :=
sInf {n | ∃ A : Finset ℕ, A ⊆ Finset.Icc 1 n ∧
AdditiveCombinatorics.IsBhSet (A : Set ℕ) h ∧ A.card = k}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:107
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, Theorem 5(b) (jis-2024-24.1.2-thm-5).
TheoremEventual bounds for \(R_h(n)\) and \(R_h^{-1}(k)\)#●ready to prove
Let \(h \ge 2\). Write \(R_h(n)\) for the largest size of a \(B_h\)-set contained in \(\{1, \dots, n\}\) and \(R_h^{-1}(k)\) for the least \(n\) such that \(\{1, \dots, n\}\) contains a \(B_h\)-set with \(k\) elements. Then \(R_h(n) \ge n^{1/h} - n^{21/(40h)}\) for all sufficiently large natural numbers \(n\), and \(R_h^{-1}(k) \le k^h + 2^h k^{h - 19/40}\) for all sufficiently large natural numbers \(k\).
public theorem_wanted bhMaxCard_bhMinLength_bounds_eventually {h : ℕ} (hh : 2 ≤ h) :
(∀ᶠ n : ℕ in atTop,
(n : ℝ) ^ (1 / (h : ℝ)) - (n : ℝ) ^ ((21 : ℝ) / (40 * h)) ≤ bhMaxCard h n) ∧
∀ᶠ k : ℕ in atTop,
(bhMinLength h k : ℝ) ≤ (k : ℝ) ^ h + 2 ^ h * (k : ℝ) ^ ((h : ℝ) - 19 / 40)
public noncomputable def bhMaxCard (h n : ℕ) : ℕ :=
sSup {k | ∃ A : Finset ℕ, A ⊆ Finset.Icc 1 n ∧
AdditiveCombinatorics.IsBhSet (A : Set ℕ) h ∧ A.card = k}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:98
public noncomputable def bhMinLength (h k : ℕ) : ℕ :=
sInf {n | ∃ A : Finset ℕ, A ⊆ Finset.Icc 1 n ∧
AdditiveCombinatorics.IsBhSet (A : Set ℕ) h ∧ A.card = k}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:107
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, Theorem 5(c) (jis-2024-24.1.2-thm-5).
TheoremRiemann Hypothesis bounds for \(R_h(n)\) and \(R_h^{-1}(k)\)#●ready to prove
Assume the Riemann Hypothesis and let \(h \ge 2\). Write \(R_h(n)\) for the largest size of a \(B_h\)-set contained in \(\{1, \dots, n\}\), with \(R_h(0) = 0\) (a \(B_h\)-set is nonempty), and \(R_h^{-1}(k)\) for the least \(n\) such that \(\{1, \dots, n\}\) contains a \(B_h\)-set with \(k\) elements, and let \(\log\) be the natural logarithm, with the convention \(\log 0 = 0\). Then every natural number \(k \ge 1\) satisfies \(R_h^{-1}(k) < k^h + \log(20k)\, k^{h - 1/2} + 2 k^{h - 1} (\log(20k))^{2h}\), and every natural number \(n\) satisfies \(R_h(n) \ge n^{1/h} - (7 + (\log n)/h)\, n^{1/(2h)}\).
public theorem_wanted bhMaxCard_bhMinLength_bounds_of_RH (hRH : RiemannHypothesis) {h : ℕ}
(hh : 2 ≤ h) :
(∀ k : ℕ, 1 ≤ k → (bhMinLength h k : ℝ) <
(k : ℝ) ^ h + Real.log (20 * k) * (k : ℝ) ^ ((h : ℝ) - 1 / 2) +
2 * (k : ℝ) ^ (h - 1) * Real.log (20 * k) ^ (2 * h)) ∧
∀ n : ℕ, (n : ℝ) ^ (1 / (h : ℝ)) - (7 + Real.log n / h) * (n : ℝ) ^ (1 / (2 * (h : ℝ))) ≤
bhMaxCard h n
public noncomputable def bhMinLength (h k : ℕ) : ℕ :=
sInf {n | ∃ A : Finset ℕ, A ⊆ Finset.Icc 1 n ∧
AdditiveCombinatorics.IsBhSet (A : Set ℕ) h ∧ A.card = k}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:107
public noncomputable def bhMaxCard (h n : ℕ) : ℕ :=
sSup {k | ∃ A : Finset ℕ, A ⊆ Finset.Icc 1 n ∧
AdditiveCombinatorics.IsBhSet (A : Set ℕ) h ∧ A.card = k}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:98
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, Theorem 5(d) (jis-2024-24.1.2-thm-5).
TheoremCully-Hugill's theorem for 155th powers#●ready to prove
For every natural number \(n \ge 1\) there is a prime \(p\) with \(n^{155} < p < (n + 1)^{155}\).
public theorem_wanted exists_prime_between_pow_155 (n : ℕ) (hn : 1 ≤ n) :
∃ p : ℕ, p.Prime ∧ n ^ 155 < p ∧ p < (n + 1) ^ 155
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, Theorem 6 (Cully-Hugill) (jis-2024-24.1.2-thm-6).
TheoremUnconditional bounds for \(R_h^{-1}(k)\), \(k \ge 4\), and \(R_h(n)\), \(n \ge h + 3\)#●ready to prove
Let \(h \ge 2\). Write \(R_h(n)\) for the largest size of a \(B_h\)-set contained in \(\{1, \dots, n\}\) and \(R_h^{-1}(k)\) for the least \(n\) such that \(\{1, \dots, n\}\) contains a \(B_h\)-set with \(k\) elements. Then every natural number \(k \ge 4\) satisfies \(R_h^{-1}(k) < k^h + 3^{155h} k^{h - 1/155}\), and every natural number \(n \ge h + 3\) satisfies \(R_h(n) \ge n^{1/h} - 2^{44} n^{154/(155h)}\).
public theorem_wanted bhMinLength_lt_and_bhMaxCard_ge_explicit {h : ℕ} (hh : 2 ≤ h) :
(∀ k : ℕ, 4 ≤ k → (bhMinLength h k : ℝ) <
(k : ℝ) ^ h + 3 ^ (155 * h) * (k : ℝ) ^ ((h : ℝ) - 1 / 155)) ∧
∀ n : ℕ, h + 3 ≤ n →
(n : ℝ) ^ (1 / (h : ℝ)) - 2 ^ 44 * (n : ℝ) ^ ((154 : ℝ) / (155 * h)) ≤ bhMaxCard h n
public noncomputable def bhMinLength (h k : ℕ) : ℕ :=
sInf {n | ∃ A : Finset ℕ, A ⊆ Finset.Icc 1 n ∧
AdditiveCombinatorics.IsBhSet (A : Set ℕ) h ∧ A.card = k}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:107
public noncomputable def bhMaxCard (h n : ℕ) : ℕ :=
sSup {k | ∃ A : Finset ℕ, A ⊆ Finset.Icc 1 n ∧
AdditiveCombinatorics.IsBhSet (A : Set ℕ) h ∧ A.card = k}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:98
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, Theorem 7 (jis-2024-24.1.2-thm-7).
TheoremCully-Hugill's theorem for cubes#●ready to prove
For every natural number \(n > \exp(\exp(32.537))\) there is a prime \(p\) with \(n^3 < p < (n + 1)^3\).
public theorem_wanted exists_prime_between_cubes_of_large (n : ℕ)
(hn : Real.exp (Real.exp 32.537) < n) :
∃ p : ℕ, p.Prime ∧ n ^ 3 < p ∧ p < (n + 1) ^ 3
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, Theorem 8 (Cully-Hugill) (jis-2024-24.1.2-thm-8).
TheoremStrict bounds for thick \(B_h\)-sets beyond \(e^{e^{34}}\)#●ready to prove
Let \(h \ge 2\). For natural numbers \(n\) and \(k\), let \(R_h(n)\) be the largest size of a \(B_h\)-set of natural numbers contained in \(\{1, \dots, n\}\), and let \(R_h^{-1}(k)\) be the smallest \(n\) such that \(\{1, \dots, n\}\) contains a \(B_h\)-set with exactly \(k\) elements. Then every natural number \(k > e^{e^{34}}\) satisfies \(R_h^{-1}(k) < k^h + (3k)^{h - 1/3}\), and every natural number \(n > e^{e^{34}}\) satisfies \(R_h(n) > n^{1/h} - 7 n^{2/(3h)}\).
The printed theorem leaves \(n\) unquantified; the hypothesis on \(n\) is taken from Theorem 5(b).
public theorem_wanted bhMinLength_lt_and_bhMaxCard_gt_large {h : ℕ} (hh : 2 ≤ h) :
(∀ k : ℕ, Real.exp (Real.exp 34) < k →
(bhMinLength h k : ℝ) < (k : ℝ) ^ h + (3 * k : ℝ) ^ ((h : ℝ) - 1 / 3)) ∧
∀ n : ℕ, Real.exp (Real.exp 34) < n →
(n : ℝ) ^ (1 / (h : ℝ)) - 7 * (n : ℝ) ^ ((2 : ℝ) / (3 * h)) < bhMaxCard h n
public noncomputable def bhMinLength (h k : ℕ) : ℕ :=
sInf {n | ∃ A : Finset ℕ, A ⊆ Finset.Icc 1 n ∧
AdditiveCombinatorics.IsBhSet (A : Set ℕ) h ∧ A.card = k}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:107
public noncomputable def bhMaxCard (h n : ℕ) : ℕ :=
sSup {k | ∃ A : Finset ℕ, A ⊆ Finset.Icc 1 n ∧
AdditiveCombinatorics.IsBhSet (A : Set ℕ) h ∧ A.card = k}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:98
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, Theorem 9 (jis-2024-24.1.2-thm-9).
TheoremBaker–Harman–Pintz theorem#●ready to prove
There is a real number \(x_0\) such that for every real \(x \ge x_0\) there is a prime \(p\) with \(x - x^{21/40} \le p \le x\) and a prime \(p'\) with \(x \le p' \le x + x^{21/40}\).
This is the form quoted in the paper, which prints the first interval as \([x - x^{21/40}, n]\) with \(n\) undefined; the right endpoint is read as \(x\).
public theorem_wanted eventually_exists_prime_short_intervals :
∀ᶠ x : ℝ in atTop,
(∃ p : ℕ, p.Prime ∧ x - x ^ ((21 : ℝ) / 40) ≤ p ∧ (p : ℝ) ≤ x) ∧
∃ p : ℕ, p.Prime ∧ x ≤ p ∧ (p : ℝ) ≤ x + x ^ ((21 : ℝ) / 40)
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, Theorem 10 (Baker, Harman and Pintz) (jis-2024-24.1.2-thm-10).
TheoremEventual strict upper bound for \(R_h^{-1}(k)\) and lower bound for \(R_h(n)\)#●ready to prove
Let \(h \ge 2\). Write \(R_h(n)\) for the largest size of a \(B_h\)-set contained in \(\{1, \dots, n\}\) and \(R_h^{-1}(k)\) for the least \(n\) such that \(\{1, \dots, n\}\) contains a \(B_h\)-set with \(k\) elements. Then \(R_h^{-1}(k) < k^h + 2^h k^{h - 19/40}\) for all sufficiently large natural numbers \(k\), and \(R_h(n) \ge n^{1/h} - n^{21/(40h)}\) for all sufficiently large natural numbers \(n\).
public theorem_wanted bhMinLength_lt_and_bhMaxCard_ge_eventually {h : ℕ} (hh : 2 ≤ h) :
(∀ᶠ k : ℕ in atTop,
(bhMinLength h k : ℝ) < (k : ℝ) ^ h + 2 ^ h * (k : ℝ) ^ ((h : ℝ) - 19 / 40)) ∧
∀ᶠ n : ℕ in atTop,
(n : ℝ) ^ (1 / (h : ℝ)) - (n : ℝ) ^ ((21 : ℝ) / (40 * h)) ≤ bhMaxCard h n
public noncomputable def bhMinLength (h k : ℕ) : ℕ :=
sInf {n | ∃ A : Finset ℕ, A ⊆ Finset.Icc 1 n ∧
AdditiveCombinatorics.IsBhSet (A : Set ℕ) h ∧ A.card = k}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:107
public noncomputable def bhMaxCard (h n : ℕ) : ℕ :=
sSup {k | ∃ A : Finset ℕ, A ⊆ Finset.Icc 1 n ∧
AdditiveCombinatorics.IsBhSet (A : Set ℕ) h ∧ A.card = k}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:98
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, Theorem 11 (jis-2024-24.1.2-thm-11).
TheoremDudek–Grenié–Molteni theorem#●ready to prove
Assume the Riemann Hypothesis. Then for every natural number \(n \ge 2\) there is a prime \(p\) with \(n^2 < p < \bigl(n + (1 + 1/\log n)^2 \log n\bigr)^2\), where \(\log\) is the natural logarithm.
public theorem_wanted exists_prime_between_sq_of_RH (hRH : RiemannHypothesis) (n : ℕ)
(hn : 2 ≤ n) :
∃ p : ℕ, p.Prime ∧ ((n : ℝ) ^ 2 < p) ∧
(p : ℝ) < ((n : ℝ) + (1 + 1 / Real.log n) ^ 2 * Real.log n) ^ 2
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, Theorem 12 (jis-2024-24.1.2-thm-12).
TheoremRiemann Hypothesis bounds for \(R_h^{-1}(k)\), \(k \ge 4\), and \(R_h(n)\), \(n \ge h + 3\)#●ready to prove
Assume the Riemann Hypothesis and let \(h \ge 2\). Write \(R_h(n)\) for the largest size of a \(B_h\)-set contained in \(\{1, \dots, n\}\) and \(R_h^{-1}(k)\) for the least \(n\) such that \(\{1, \dots, n\}\) contains a \(B_h\)-set with \(k\) elements, and let \(\log\) be the natural logarithm. Then every natural number \(k \ge 4\) satisfies \(R_h^{-1}(k) < k^h + \log(20k)\, k^{h - 1/2} + 2 k^{h - 1} (\log(20k))^{2h}\), and every natural number \(n \ge h + 3\) satisfies \(R_h(n) \ge n^{1/h} - (7 + (\log n)/h)\, n^{1/(2h)}\).
public theorem_wanted bhMinLength_lt_and_bhMaxCard_ge_of_RH (hRH : RiemannHypothesis) {h : ℕ}
(hh : 2 ≤ h) :
(∀ k : ℕ, 4 ≤ k → (bhMinLength h k : ℝ) <
(k : ℝ) ^ h + Real.log (20 * k) * (k : ℝ) ^ ((h : ℝ) - 1 / 2) +
2 * (k : ℝ) ^ (h - 1) * Real.log (20 * k) ^ (2 * h)) ∧
∀ n : ℕ, h + 3 ≤ n →
(n : ℝ) ^ (1 / (h : ℝ)) - (7 + Real.log n / h) * (n : ℝ) ^ (1 / (2 * (h : ℝ))) ≤
bhMaxCard h n
public noncomputable def bhMinLength (h k : ℕ) : ℕ :=
sInf {n | ∃ A : Finset ℕ, A ⊆ Finset.Icc 1 n ∧
AdditiveCombinatorics.IsBhSet (A : Set ℕ) h ∧ A.card = k}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:107
public noncomputable def bhMaxCard (h n : ℕ) : ℕ :=
sSup {k | ∃ A : Finset ℕ, A ⊆ Finset.Icc 1 n ∧
AdditiveCombinatorics.IsBhSet (A : Set ℕ) h ∧ A.card = k}
WantedExt/Combinatorics/Additive/ThickBhSetsWanted.lean:98
Sources
- Kevin O'Bryant, Constructing Thick
B_h-Sets, Journal of Integer Sequences 27 (2024), Article 24.1.2, Theorem 13 (jis-2024-24.1.2-thm-13).
Loading dependency explorer…
The dependency explorer needs JavaScript and its graph data.