FormalConjectures/ErdosProblems/388.lean
Explanation legend Lean4 Domain Explanation Language (L4DEL)
L4DEL explanations are a separate layer. They do not change the Lean source. Intuition is explanatory prose; numbered items are a reading path; sources link the wording back to evidence.
- L4DEL
- Lean4 Domain Explanation Language
Credit: The upstream authors and contributors retain credit for the project and mathematics. lean4-explainer only renders their pinned source and adds the separately disclosed explanation layer.
Project sources: Pinned project documentation, Formal Conjectures paper, arXiv v1
3 declarations, 3 contain sorry, 0 declared as axioms
Erdős Problem 388
References:
- erdosproblems.com/388
- [Er76d] Erdős, P., Problems and results on number theoretic properties of consecutive integers and related questions. Proceedings of the Fifth Manitoba Conference on Numerical Mathematics (Univ. Manitoba, Winnipeg, Man., 1975) (1976), 25-44.
- [ErGr80] Erdős, P. and Graham, R., Old and new problems and results in combinatorial number theory. Monographies de L'Enseignement Mathematique (1980).
- [Er92e] Erdős, Pál, Some Unsolved problems in Geometry, Number Theory and Combinatorics. Eureka (1992), 44-48.
Imports (1)
- theorem
Erdos388.erdos_388Are there only finitely many solutions of $$\prod_{1\leq i\leq k_1}(m_1+i)=\prod_{1\leq j\leq k_2}(m_2+j)$$ with $k_1,k_2>3$ and $m_1+k_1\leq m_2$? sorry line 49 sourceSource documentation
Are there only finitely many solutions of $$\prod_{1\leq i\leq k_1}(m_1+i)=\prod_{1\leq j\leq k_2}(m_2+j)$$ with $k_1,k_2>3$ and $m_1+k_1\leq m_2$?
Here $m_1, m_2 \geq 0$, so every factor is positive and the condition $m_1 + k_1 \leq m_2$ makes the two blocks disjoint. Negative $m_1$ would give trivial solutions such as $(-5)(-4)(-3)(-2) = 2 \cdot 3 \cdot 4 \cdot 5$. The source also asks to classify all solutions; see
Erdos388.erdos_388.variants.classification.depends on sorry what this means
Statement uses:
Iff,True,Set.Finite,Prod,Nat,Set.ofPred,Erdos388.erdos_388.match_1,And,LT.lt,OfNat.ofNat,LE.le,HAdd.hAdd,Eq,Finset.prod,Finset.IccProof uses:
sorryAx,Lean.Name,Bool.false,Lean.Name.num,Lean.Name.str,Lean.Name.anonymous16 instances, coercions and auxiliary constants left out.
Tactic trace: 1 steps
Open to load the checked before-and-after goal states.
theorem erdos_388 : answer(sorry) ↔ {(m₁, k₁, m₂, k₂) : ℕ × ℕ × ℕ × ℕ | 3 < k₁ ∧ 3 < k₂ ∧ m₁ + k₁ ≤ m₂ ∧ ∏ i ∈ Finset.Icc 1 k₁, (m₁ + i) = ∏ j ∈ Finset.Icc 1 k₂, (m₂ + j)}.Finite := by … - theorem
Erdos388.erdos_388.variants.classificationCan one classify all solutions of $$\prod_{1\leq i\leq k_1}(m_1+i)=\prod_{1\leq j\leq k_2}(m_2+j)$$ with $k_1,k_2>3$ and $m_1+k_1\leq m_2$? sorry line 64 sourceSource documentation
Can one classify all solutions of $$\prod_{1\leq i\leq k_1}(m_1+i)=\prod_{1\leq j\leq k_2}(m_2+j)$$ with $k_1,k_2>3$ and $m_1+k_1\leq m_2$?
A solution must supply the set of all solutions $(m_1, k_1, m_2, k_2)$. Whether a description of this set is a satisfactory classification is up to human judgement.
depends on sorry what this means
Statement uses:
Eq,Set,Prod,Nat,Set.ofPred,Erdos388.erdos_388.match_1,And,LT.lt,OfNat.ofNat,LE.le,HAdd.hAdd,Finset.prod,Finset.Icc,sorryAx,Lean.Name,Bool.false,Lean.Name.num,Lean.Name.str,Lean.Name.anonymous16 instances, coercions and auxiliary constants left out.
Tactic trace: 1 steps
Open to load the checked before-and-after goal states.
theorem erdos_388.variants.classification : {(m₁, k₁, m₂, k₂) : ℕ × ℕ × ℕ × ℕ | 3 < k₁ ∧ 3 < k₂ ∧ m₁ + k₁ ≤ m₂ ∧ ∏ i ∈ Finset.Icc 1 k₁, (m₁ + i) = ∏ j ∈ Finset.Icc 1 k₂, (m₂ + j)} = answer(sorry) := by … - theorem
Erdos388.erdos_388.variants.generalMore generally, if $k_1>2$ then for fixed $a$ and $b$ the equation $$a\prod_{1\leq i\leq k_1}(m_1+i)=b\prod_{1\leq j\leq k_2}(m_2+j)$$ should have only finitely many solutions sorry line 85 sourceSource documentation
More generally, if $k_1>2$ then for fixed $a$ and $b$ the equation $$a\prod_{1\leq i\leq k_1}(m_1+i)=b\prod_{1\leq j\leq k_2}(m_2+j)$$ should have only finitely many solutions.
The source states no condition on $a$, $b$, $k_2$ or the position of the blocks, so this is our reading: $a$ and $b$ are positive integers, we keep the condition $m_1+k_1\leq m_2$ of the main problem (the $k_1$-block lies below the $k_2$-block), and we assume $k_2 > 2$ as well. Without these conditions there are infinite families of solutions:
- for $k_2 = 1$ and $b = 1$, take $m_2 + 1 = a\prod_{1\leq i\leq k_1}(m_1+i)$;
- for $k_2 = 2$, $a = 1$ and $b = 4$, the identity $(n+1)(n+2)(n+3)(n+4) = Y(Y+2)$ with $Y = (n+1)(n+4)$ even gives a solution with $m_2 + 1 = Y/2$ for every $n \geq 2$;
- overlapping blocks give $2\prod_{1\leq i\leq k}(k-1+i)=\prod_{1\leq j\leq k}(k+j)$ for every $k \geq 2$.
depends on sorry what this means
Statement uses:
Nat,LT.lt,OfNat.ofNat,Set.Finite,Prod,Set.ofPred,Erdos388.erdos_388.match_1,And,LE.le,HAdd.hAdd,Eq,HMul.hMul,Finset.prod,Finset.IccProof uses:
sorryAx,Lean.Name,Bool.false,Lean.Name.num,Lean.Name.str,Lean.Name.anonymous20 instances, coercions and auxiliary constants left out.
Tactic trace: 1 steps
Open to load the checked before-and-after goal states.
theorem erdos_388.variants.general (a b : ℕ) (ha : 0 < a) (hb : 0 < b) : {(m₁, k₁, m₂, k₂) : ℕ × ℕ × ℕ × ℕ | 2 < k₁ ∧ 2 < k₂ ∧ m₁ + k₁ ≤ m₂ ∧ a * ∏ i ∈ Finset.Icc 1 k₁, (m₁ + i) = b * ∏ j ∈ Finset.Icc 1 k₂, (m₂ + j)}.Finite := by …