Proof Wanted

A registry of mathematical theorems and open conjectures that are suitable for formalization in Lean 4 + Mathlib. Each candidate has a precise Lean statement that type-checks against the current version of Mathlib (proofs are sorry). The goal is to identify results where the mathematical infrastructure exists in Mathlib and the proof is within reach.

24 problems
0 verified
8 open
16 theorems

What makes a good candidate

Contributions are via pull request. See Contributing for details.

C-01Draft

Chvatal's conjecture

Let F\mathcal{F} be a downward-closed (ideal) family of finite sets. Then the largest intersecting subfamily of F\mathcal{F} is a star: there exists an element xx such that the maximum size of an intersecting subfamily equals {AF:xA}|\{A \in \mathcal{F} : x \in A\}|.

CombinatoricsUnknown (open problem) lines
C-02Draft

Davenport constant of ℤ_n ⊕ ℤ_n

The Davenport constant D(G)D(G) of a finite abelian group GG is the smallest integer dd such that every multiset of dd elements from GG contains a non-empty zero-sum subsequence.

Conjecture: D(ZnZn)=2n1D(\mathbb{Z}_n \oplus \mathbb{Z}_n) = 2n - 1.

The lower bound D(ZnZn)2n1D(\mathbb{Z}_n \oplus \mathbb{Z}_n) \geq 2n - 1 is easy (take n1n-1 copies of (1,0)(1,0) and n1n-1 copies of (0,1)(0,1)). The upper bound is the content of the conjecture.

Additive Combinatorics / Algebraic Number TheoryUnknown (open in general) lines
C-03Draft

Equational theories of magmas (sample implications)

The equational_theories project catalogues thousands of equational laws for magmas (types with a single binary operation) and maps their logical relationships. Many implications remain undecided: given two equations, it is unknown whether every magma satisfying the first must also satisfy the second.

Sample equations:

  • Equation 46: x(yx)=yx \cdot (y \cdot x) = y
  • Equation 387: xy=(yx)xx \cdot y = (y \cdot x) \cdot x
  • Equation 4512: x(yz)=(xy)zx \cdot (y \cdot z) = (x \cdot y) \cdot z (associativity)
Universal Algebra5-50 per implication lines
C-04Draft

Erdos matching conjecture

The maximum size of a kk-uniform family on [n][n] with no s+1s+1 pairwise disjoint sets is

max{(ks+k1k),(nk)(nsk)}\max\left\{\binom{ks+k-1}{k}, \binom{n}{k} - \binom{n-s}{k}\right\}

CombinatoricsUnknown (open problem) lines
C-05Draft

Erdos-Straus conjecture

For every integer n2n \geq 2, the equation

4n=1x+1y+1z\frac{4}{n} = \frac{1}{x} + \frac{1}{y} + \frac{1}{z}

has a solution in positive integers x,y,zx, y, z.

Number TheoryUnknown (open problem) lines
C-06Draft

Frankl's conjecture for finite upper semimodular lattices

Frankl's union-closed conjecture (lattice-theoretic form): In every finite lattice, there exists a join-irreducible element jj such that jj is above at most half the elements:

2{xL:jx}L2 \cdot |\{x \in L : j \leq x\}| \leq |L|

We state the conjecture restricted to upper semimodular lattices: lattices satisfying aba    baba \wedge b \lessdot a \implies b \lessdot a \vee b.

Combinatorics / Lattice TheoryUnknown (open problem) lines
C-07Draft

Sensitivity conjecture (tight bound)

For every Boolean function f:{0,1}n{0,1}f : \{0,1\}^n \to \{0,1\},

bs(f)s(f)2\mathrm{bs}(f) \leq \mathrm{s}(f)^2

where s(f)\mathrm{s}(f) is the sensitivity and bs(f)\mathrm{bs}(f) is the block sensitivity.

Combinatorics / Complexity TheoryUnknown (open problem) lines
C-08Draft

Sunflower conjecture

There exists a constant CC such that any family of more than CkC^k sets, each of size kk, contains a sunflower of size 3 (i.e., three sets whose pairwise intersections are all equal).

CombinatoricsUnknown (open problem) lines