Mathematical conjectures: full catalogue

Home · Search the collection · Verification · Lean status

Each entry links to the mathematical argument and its accompanying checks. The claims below retain the scope stated in the notes; formalization status is tracked separately.

Partial results on open problems

  • Erdős essential-component conjecture. Essentiality of the numbers 2m3n2^m3^n is equivalent to uniform one-sided expansion of half-density sets modulo 2k2^k. Expansion is at least of order (logk)3/2/k(\log k)^{3/2}/k, and small boundaries project to shorter moduli under the full projected shift set. Research note · Issue #73.
  • Erdős–Sauer conjecture. For every fixed uniformity, the maximum hypergraph clique-and-singleton decomposition number agrees asymptotically with the corresponding Turán number. The exact bound holds with at most two cliques through each edge. Research note · Issue #72.
  • Bollobás–Erdős–Tuza conjecture. Graphs with all maximal independent sets of linear size can require transversals of order nlogn\sqrt{n\log n}. Bounded modular width gives a constant transversal bound. Research note · Issue #71.
  • Erdős–Oler conjecture. The packing bound holds on unit-separated lattices, and triangular arrays with one site removed have a separation radius independent of their order. In a counterexample, thin Delaunay faces must occur near at least two sides of the containing triangle. Packing and rigidity · Boundary faces · Issue #62.
  • Erdős–Granville–Pomerance–Spiro conjecture. The preimage of every natural density-zero set under the sum-of-proper-divisors map has density zero within a fixed input family of natural density about 53.46%. The arguments use three second-moment estimates for different prime-factor ranges. Scope and proofs · Finite checks.
  • Frankl’s conjecture (partial). Generators supported in four consecutive positions of a common cyclic order guarantee an abundant point. For uniform generators with every point in three generators and incidence girth at least 16, each point occurs in at least 51.8208% of the unions. Local proof · Frequency proof · Replay Both all-point frequency bounds are fully Lean-verified; the cyclic span-four formalization remains incomplete. Checked scope.
  • Lonely Runner (partial). Two arbitrary positive integer speeds can be added to any Goddyn–Wong simultaneous acceleration. The proof gives a prescribed interval for the common witness and an obstruction to the corresponding three-speed assertion. Proof · Replay The complete insertion note is Lean-verified, including the unbounded speeds, acceleration criterion, common-dilation value bound, and literal three-insertion obstruction. Checked scope.
  • Erdős–Gyárfás (partial). Every abelian cover of K4K_4 with one edge subdivided has a cycle of length 4, 8, 16 or 32. A 19-sheeted cyclic example attains the upper bound. The universal voltage-cover theorem, literal 95-vertex sharpness example, and normalization of all seven voltages are Lean-verified. Lean scope. Proof · Replay
  • Hadamard existence conjecture, order 2396 (open). The signed automorphism group of any candidate matrix has order coprime to 479. The fixed-border common multiplier group is cyclic of order dividing four, with at least 151 sign orbits. Lean verifies these statements, both norm-obstruction proofs, the complete conditional fixed-border construction from its original four sign rows, all fourteen size-149 baseline patterns, and the length-598 order-eleven obstruction. Lean also constructs the actual Hadamard matrix from any Legendre pair and proves the invariant quotient transport from the original centered rows. The order and orbit theorems for odd automorphism subgroups on groups of order nine are also checked. For length-1197 Legendre pairs, Lean proves that the common-multiplier order divides 27; the stated sharper bound of nine and the exhaustive quotient/no-lift certificates remain pending. No order-2396 matrix existence is claimed. Checked scope and receipts · Automorphism proof · Sequence proof · Reproduce.

Word-count games

Problem Result Formalization Evidence
Ekhad–Zeilberger’s most unfair bet conjecture Determines the pair of words maximizing the leading difference of win probabilities in Litt’s game for every word length and alphabet size. Includes the sponsor’s full-alphabet convention. Not formalized Proof · Exact checks · Issue 219

Restricted permutations

Problem Result Formalization Evidence
Kauers–Koutschan recurrence conjecture A Schur-character formula for signed adjacency blocks proves the specified order-eight and order-thirteen recurrences for all permutation lengths. Not formalized Proof · Polynomial identities · Issue 218

Pfaffian matching bounds

Problem Result Formalization Evidence
Johnson's conjecture Complete proposed proof of Conjecture 7.1 for every positive nonincreasing log-convex distance kernel, every even order, and every perfect matching. A stronger theorem for products of circular interval factors follows from a complement scaling and adjacent-pair deletion. Johnson's endpoint theorem and kernel uncrossing reduction are credited. Partial: Lean proves the four-node matching-sum Pfaffian case, all-order kernel factorization, diagonal scaling and local uncrossing; the general even-order theorem remains open. Proof · Source and priority audit · Combinatorics review · Analysis review · Root review · Checker · Issue 211

Subtree roots

Problem Result Formalization Evidence
Brown–Mol's conjecture Complete proposed disproof of the universal translated-annulus assertion. An infinite family of diameter-four trees has subtree roots strictly inside the forbidden disk, proved by a local analytic zero argument. A separate exact Rouché certificate verifies a 1001-vertex example. The outer inequality remains separate. Complete: Lean verifies both the explicit 1001-vertex counterexample and the all-sufficiently-large diameter-four counterfamily. Proof · Sources · Priority audit · Analysis review · Root review · Checker · Issue 210

Matrix Hardy norms

Problem Result Formalization Evidence
Kovalev's conjecture Complete proposed disproof of Conjecture 6.1. A uniform positive diagonal matrix family violates the lower Hardy-duality comparison in every order at least 500001. The supporting functional has dual norm exactly one; analytic Gaussian bounds give a strict rational separation. The separate upper inequality is not resolved. Not formalized. Analytic proof with supplementary exact arithmetic. Proof · Sources · Priority audit · Algebra review · Root review · Checker · Issue 209

Radial kernels in BCS theory

Problem Result Formalization Evidence
Roos–Seiringer's conjecture Complete proposed proof that the radial Dirichlet kernel is strictly positive at every positive argument, with the zero endpoint and all positive chemical potentials covered. Analytic tails and an essential exact integer interval certificate prove the full scalar assertion. The unaveraged kernel is separate. Not formalized. Essential computer-assisted proof. Proof · Sources · Priority audit · Algebra review · Root review · Certificate · Issue 204

Intrinsic-volume sequence realizability

Problem Result Formalization Evidence
Dospolova–Germanskov–Zaporozhets's conjecture Complete proposed disproof of the ultra-log-concave sequence realizability conjecture. Every positive factorial sequence c^k/k! satisfies the conjectured inequalities with equality but cannot be an intrinsic-volume sequence. A strict global-range/local-oscillation comparison and Vitale's published limit theorem provide the obstruction. Long's earlier results also permit a short alternative deduction and are explicitly credited. Three independent mathematical reviews passed. Partial: factorial sequence equalities and abstract Gaussian strict-mean obstruction checked; intrinsic-volume links remain. Proof · Source audit · Analysis review · Algebra review · Root review · Source-symbol audit · Issue 199

Macdonald polynomial semirings

Problem Result Formalization Evidence
Chen–Khare–Sahi's conjecture Complete proposed disproof of the majorization-to-Muirhead-semiring implication in Conjecture 5. A rational functional is nonnegative on an exhaustive generating list of 38 products and seven monomials, but strictly negative on an admissible normalized Macdonald difference. Exact eigenoperator and independent Hall-inner-product constructions identify the polynomials. Rectangular translates and a continuous parameter family give further counterexamples to the same assertion. Ordinary evaluation positivity is not refuted. Partial pending full gate: the exact counterexample, rectangular-translate family, and continuous-parameter refinement are formalized. Proof · Parameter family · Source audit · Analysis review · Root review · Exact verifier · Issue 201

Beta-prime self-convolution

Problem Result Formalization Evidence
Ferreira–Simon's conjecture Complete proposed analytic proof of the independent beta-prime self-convolution identity for every positive first shape and every second shape strictly between zero and one half. A quadratic hypergeometric transformation evaluates the mixing factor; a beta integral identifies the Mellin transforms on a nonempty strip. Two independent derivations passed. Not formalized Proof · Source audit · Algebra review · Root review · Exact algebra · Issue 200

Gamma medians

Problem Result Formalization Evidence
Lyon's conjecture Complete proposed computer-assisted proof that the transformed unit-scale gamma median decreases strictly for every positive real shape. Analytic arguments handle both unbounded tails. Exact integer interval arithmetic certifies all 4258 median brackets and 4256 cells covering the remaining compact interval; convexity makes each inequality valid throughout its cell. The certificate is essential, and two complete independent replays passed. Not formalized Proof · Source audit · Algebra review · Root review · Certificate data · Verifier · Issue 196

Lyapunov matrix optimization

Problem Result Formalization Evidence
Han–Sabag–Kostina–Hassibi's conjecture Complete proposed disproof of the Lyapunov positivity conjecture. A rational feasible encoder has power strictly below the infimum over every encoder with one nonzero per column. All sparse assignments are exhausted, and the same failure holds on an open interval of channel gains. The full matrix assertion is refuted; no separate communication-code theorem is claimed. Complete source formalization; merged-package gate pending Proof · Source audit · Algebra review · Analysis review · Checks · Issue 197

Zeros of Gaussian shift sums

Problem Result Formalization Evidence
Gröchenig's problem Complete proposed disproof of the universal upper-density comparison in Problem 4.12. One nonzero Gaussian shift sum with absolutely summable real coefficients has coefficient sign-change upper density one half and zero-set upper density at least one. An infinite block construction preserves sign witnesses at arbitrarily large scales. The separate lower-density assertion remains undecided here. Not formalized Proof · Source audit · Root review · Algebra review · Checks · Issue 191

Zeros of polynomial transforms

Problem Result Formalization Evidence
Vishnyakova's conjecture Complete proposed proof of the full root-radius conjecture, including all positive weights. Every zero lies in a sharper centered disk; the outer radius is attained. A complex-parameter exclusion and Grace–Walsh–Szegő polarization extend the credited earlier equal-parameter result to arbitrary independent parameters. Two internal mathematical reviews passed. Complete: Lean proves the original sampled-sequence theorem and the weighted transform-class theorem for every positive degree and weight. The original class uses every finite ordered Toeplitz minor. Both centered disks and attained exact radii are checked. All 17 modules passed strict sources, the complete axiom audit, and stock-kernel replay. Proof · Source audit · Root review · Analysis review · Checks · Issue 182

Gauss–Legendre quadrature

Result Finding Lean Evidence
López Pouso–Segura's node conjecture Complete proposed disproof of the exact upper inequalities in published Conjectures 3 and 7. An exact common witness refutes both; a positive fourth-order correction proves failure of Conjecture 3 for every sufficiently large even degree. The leading asymptotic constant and separate Bessel-limit results are unaffected. Independent internal mathematical review passed. Partial: Lean checks the exact two-node counterexample; the eventual-even-degree strengthening and combined gate remain pending. Proof · Source audit · Independent review · Issue 184

Bessel zeros and weights

Result Finding Lean Evidence
López Pouso–Segura's conjectures Complete proposed proofs of published Conjectures 2, 5, and 8: positivity or negativity, strict monotonicity, and exact limits of all three Bessel-weight sequences. Inverse-phase convexity gives the signs; a Fourier–Bessel resolvent determines the limits. The linked assertions count as one result. Two independent internal reviews passed. Partial: Lean proves the signs and strict monotonicity of all three sequences from explicit convexity, antiderivative and positive-weight hypotheses; the Bessel identification, convexity and limits remain open. Proof · Source audit · Root review · Algebra review · Checks · Issue 177

Mittag–Leffler functions and stochastic ordering

Result Finding Lean Evidence
Simon's stable-infimum conjecture Complete proposed proof that the mean-normalized fixed-time infima of spectrally negative stable processes increase in convex order throughout the full source parameter interval, including the Brownian endpoint. Increasing logarithmic Lévy measures yield an independent mean-one multiplier. Not formalized Proof · Source audit · Independent review · Issue 175
Ferreira–Simon's stochastic-order conjecture Complete proposed proof of the usual stochastic decrease of the normalized inverse-stable family throughout the full real parameter interval. One Poisson coupling is almost surely strictly decreasing for every parameter pair. The endpoint limits and Mittag–Leffler bounds follow from the same argument. Not formalized Proof · Source audit · Independent review · Issue 170
Ferreira–Simon's diagonal Mittag–Leffler conjecture Complete proposed proof that the diagonal Mittag–Leffler value at minus one strictly increases throughout the unit parameter interval. A classical sine quantile, logarithmic contraction and reciprocal pairing give a bounded pointwise integral comparison. The endpoint and crossing-point consequences are not counted separately. Not formalized Proof · Sources · Priority audit · Independent review · Issue 171

Both complete arguments passed independent internal mathematical review. They are analytic proofs without finite-case inference or computational verification. The source audits record no earlier resolution in the inspected material and state their access and indexing limits. The notes remain unrefereed.

Partition statistics

Result Finding Lean Evidence
McCrea's conjecture Complete proposed proof of the unique linear-density crossover for the two fixed-perimeter Franklin statistics, for every modulus at least three and every permitted density. Positive power-series factorizations give local coefficient asymptotics including the floor; opposite monotonicity of the two growth exponents gives existence and uniqueness. Independently reviewed internally. Not formalized Proof · Priority audit · Review · Checks · Issue 168

Permanent minima

Result Finding Lean Evidence
Pula–Song–Wanless conjecture Complete proposed proof of the cohesive/noncohesive threshold for every block size, including equality, with a unique boundary minimizer at and above the threshold. The new threshold proof repairs the old equalization argument; previously known directions are credited. This is a named subsidiary conjecture, not the full characterization requested by Minc Problems 14 or 15. Not formalized Proof · Source audit · Exact checks · Review

Subpermanent transform inequalities

Result Finding Lean Evidence
Malek's conjecture Complete proposed proof of the negative-transform comparison for every order and every subpermanent degree, with all equality cases. A stronger reciprocal-mean capacity estimate closes every proper degree; the full-permanent endpoint is prior work. This is a named subsidiary in the Minc site's linked survey, not a new numbered Conjecture 3 resolution. Two internal reviews found no gap; priority remains provisional. Not formalized Proof · Source audit · Exact checks · Review

Permanental compounds

Result Finding Lean Evidence
Perron root of permanental compounds A general FPRAS for every nonnegative rational input and every compound degree, with deterministic zero detection. This answers the randomized approximation formulation of Minc Problem 18; exact polynomial-time eigenvalue computation and the characteristic-polynomial problem are not claimed. The proof is unrefereed and priority remains provisional. Not formalized Proof · Source audit · Exact checks · Review

Polynomial identities and positivity

Result Finding Lean Evidence
Novelli–Thibon's conjecture Complete proposed proof of the intended monomial-product Conjecture 5.7 for all partitions and independent parameters. A universal first-derivative identity and the Pieri rule prove the product formula; a plethystic evaluation also gives every hook coefficient. The source index typo is corrected explicitly. Two internal mathematical reviews passed. Not formalized Proof · Source audit · Independent review · Checks · Issue 180
Chen–Kitaev–Zhang's record conjectures Complete proposed proofs of the original Conjectures 1 and 3 for every permitted length. New one-, two- and three-record comparisons combine with the credited earlier Conjecture 2 tail theorem. Independent internal review passed; no separate count is assigned to individual statistics. Complete Lean: actual permutation classes, all record statistics, the credited all-index tail and singleton cases checked. Proof · Source audit · Review · Checks · Issue 157
Zhu's generalized Jacobi–Stirling conjecture Complete disproof of the unweighted coefficientwise Hankel-total-positivity conjecture. The leading three-by-three determinant has coefficient minus four at a squared times x to the fourth power. A symbolic identity also supplies an open region of strictly positive real counterexamples. Independent internal review passed. Complete source formalization; merged-package gate pending Proof · Source audit · Review · Checks · Issue 163
Baricz–Pogány's asymptotic-coefficient positivity conjecture Previously published result package: every asymptotic coefficient of degree at least two exceeds the stated geometric term for every permitted real Bessel order. A formal-series factorization supplies a strict positive lower bound. The source audit records an inaccessible submitted follow-up and does not assert exclusive priority. Complete: full coefficient theorem, explicit lower bound, and uniqueness/positivity for the source recurrence; strict sources, all-owned axiom audit, and stock-kernel replay passed. Combined package gate pending. Proof · Exact diagnostics · Source audit
Ferreira–Simon's Wright log-concavity conjecture The full threshold-to-log-concavity implication is false. Lean proves a continuous boundary family directly from the infinite Wright series and a strict midpoint violation. The written proof also gives interior positive-parameter families. Complete original-conjecture disproof; the additional interior families are not formalized Proof · Lean · Validation · Exact algebra · Literature audit
Karp–Vishnyakova–Zhang's gamma-quotient conjecture The gamma quotient is sign regular at every order for every positive parameter difference. The exact signature and integer rank cutoff are proved; the source's fixed third-order signature is corrected. Not formalized Proof · Exact checks · Literature check
Alexandersson's Fuss–Catalan conjecture A short classical deduction proves real-rootedness for all parameters: changing the Fuss parameter only multiplies the known ordinary-Dyck polynomial by a positive scalar. Prior-resolution audit qualified. Complete: Lean verifies the literal path enumerator, closed form, real-rootedness, exact simple-root count and scalar reduction; transfer of any ordinary-case root relation is proved conditionally on that relation. Proof · Convention checks · Attribution and source limits
Sagan–Swanson conjecture Conjecture 7.8 is proved: all four q-Stirling coefficient distributions are asymptotically normal whenever k/nk/n tends to any value in (0,1)(0,1), using their actual means and variances. Not formalized Proof · Exact model checks · Literature check
Bernardi–Morales's delta-process conjecture Published Conjectures 27–28 are proved for all admissible parameters. The tree count holds for every fixed color map, and the stronger event identity holds for every fixed surjection. Complete: literal cyclic delta map, all-parameter tree counts, full-subset interval-event matrices and the source probability determinant; strict sources, all-owned axiom audit and stock-kernel replay passed. Combined package gate pending. Tree proof · Event proof · Checks · Literature check
Gilmore's crossing–nesting conjecture Conjecture 6.16 is proved for every row: the product-defined matrix's first column enumerates perfect matchings by crossings and nestings. Its additional parameter counts aligned pairs. Complete: literal stabilized matrix product, exhaustive perfect-matching bijection, and crossing/nesting/alignment weights; strict sources, all-owned axiom audit, and stock-kernel replay passed. Combined package gate pending. Proof · Checks · Literature check
Chow's conjecture Ordinary type-D Eulerian polynomials interlace in consecutive ranks, strictly from rank three onward. The weak assertion is a direct deduction from Savage–Visontai's published refinement theorem; the note supplies an additional strictness argument. Complete: Lean proves consecutive interlacing for every rank at least two and strict interlacing from rank three. The source polynomials are literal even-signed permutation enumerators. All 11 modules passed strict sources, the complete axiom audit, and stock-kernel replay. Proof · Checks · Attribution and literature check

Solitaire and combinatorial games

Result Finding Lean Evidence
Generalized Bulgarian solitaire limit shapes Eriksson–Jonsson–Sjoestrand (2020), Conj. 1: fixed-scale counterexample (a_n = 1) and corrected theorem for arbitrary layer sets at every diverging scale. ☑ Lean · Proof
Pham's Bulgarian solitaire denominator degree Pham (2022), §6: primitive period-10 necklace with reduced denominator degree 5. ☑ Proof · Check
Spivey's three-player War Spivey (2010), Q5: three winner-first conventions admit cycles with all three players active. ☑ Proof · Check
CUT Family B: nim-sequence threshold Ellis–Thanatipanonda (2022), Conj. 2: nim-sequence matches the base game iff the least odd cut above 1 exceeds three times the least even cut. ☑ Proof · Check
CUT arithmetic periodicity with a preperiod Dailly et al., Open Problem 4: finite arithmetic-periodicity test; both examples have preperiod 6, period 12, saltus 8. ☑ Proof · Check

Graph theory and discrete geometry

Result Finding Lean Evidence
Mazák–Zrubák conjecture on complete graphs minus an edge Mazák–Zrubák, arXiv Conjecture 2: for every even d>=6, the circular chromatic index of K_{d+1} minus one edge is d+1. By Guichard's theorem a smaller value would give a proper (d+1)-edge-colouring with acyclic digraph of consecutive colours, and the proof excludes such colourings. One independent model referee pass; the scripts check d=4 and d=6. Not formalized Proof · Exact checks
Németh–Seberry–Shu conjecture The number of distinct permanent values of prime-order binary circulants grows superlinearly, contrary to the 1979 conjecture. A refinement gives a quadratic lower bound. ☐ Proof · Quadratic bound · External audit by Yair Lavi · Finite checks
Sonntag–Teichert conjecture The minimal dominating sets of every tournament form a connected hypergraph after deleting isolated vertices. For oriented graphs with missing arcs, the only disconnected supports are two disjoint stars. Complete source formalization; merged-package gate pending. Proof · Checks · Literature check
Horňák–Woźniak conjecture Admissibility suffices for every tripod and, with the induced tripod admissible, for every four-arm tree with a one-edge arm. Three of the five arithmetic tests suffice. Partial Lean: tripods with short arm at most two under source tests, and general large-part and short-arm-subset cases; remaining all-small and four-arm cases remain. Proof · Checks · Literature check
Graffiti.pc Conjecture 318 DeLaViña (2007): every total dominating set contains a dominating pair under the given even-distance bound. Credited to @JesterInABarrel. ☑ Proof · Check · Lean
Li's chromatic symmetric function signs 31-vertex non-spider, non-caterpillar tree: [e₂¹⁴e₃]X_T=29, [e₂⁸e₃⁵]X_T=−6304, all 6,842 partition types, exactly eight partitions of type (3,2^14), and 2^30 connected vertex partitions in total. ☑ Proof · Check · Lean scope
Circulants with circular chromatic index in the upper half Mazák–Zrubák, arXiv:2603.08822, Section 8: no graph with Δ>=8 and circular chromatic index in (Δ+1/2, Δ+1) was known. For every g>=2 the circulant C_{6g+1}(1,...,2g-2, 2g+1, 3g) is 4g-regular and 4g-connected with χ'_c = 4g+2/3, and C_25(1,2,3,7,...,12) is 18-regular and 18-connected with χ'_c = 75/4. This answers the non-trivial reading of their Problem 6 for d ≡ 0 mod 4, d>=8, and for d=18; Problem 7 is not answered. Computer-certified further examples include C_31(1,2,3,7,...,12) with χ'_c = 93/5 and non-regular graphs with Δ ≡ 1, 2 mod 4, 25<=Δ<=54. One independent model referee pass. Partial overall: Theorems 1–2 and the C31/C33 regular examples have complete Lean proofs of regularity, vertex-deletion connectivity, and the actual real circular chromatic index. Strict sources, all-owned axiom audit, and stock-kernel replay passed. The nonregular Section 7 certificates and SAT exclusions remain outside this checkpoint. Proof · Exact checks · Certificates
Erdős Problem 607: line-size sets Sharp counting asymptotic for records of actual finite real-plane configurations, with the exact leading prefactor. Lean verifies the geometric count, formal and analytic generating functions, Fourier extraction, all limit and domination inputs, and the final stated asymptotic. Complete Proof · Formalization and receipts · Finite check
Distance graphs with three missing differences Carraher et al., arXiv Conjecture 38 proved for all admissible parameters. Conjecture 39 fails for every k=50,59,68,…. The corrected formula gives all densities for gaps five and seven. ☑ Strip verifier and counterexample verifier
Oravcová–Soták conjecture for five to eight distances For a primitive distance set of size k in {5,6,7,8}, the chromatic number is k+1 exactly for {1,...,k-1,ks}, {1,2,3,5,8}, {1,4,5,6,7} and {1,2,3,4,5,7,12}. The classification is conditional on the repository's five- to eight-speed spectrum results. Contributed by @aadrianvv. Partial: actual integer distance graphs, all literal certificates and infinite coloring families, and Theorem A relative to explicit unproved spectrum classifications R5–R8. Strict new sources, complete owned-declaration audit and stock-kernel replay passed. Proof · Certificates · Exact checks
Johnson–Kaya–Matzke conjecture on Babai spectra of paths Johnson–Kaya–Matzke, arXiv Conjecture 2.1: for all n>=2 and 1<=k<=n-1, the chromatic numbers of the k-element distance graphs of the path P_n are exactly the integers from ceil(n/(n-k)) to k+1. The missing lower bound comes from the bound n-k on independent sets. Complete: actual finite-graph chromatic number, both bounds, and construction of every spectrum value; strict sources, all-owned axiom audit, and stock-kernel replay passed. Combined package gate pending. Proof · Brute-force checks

Combinatorics on words

Result Finding Lean Evidence
Currie's conjecture on complexity 2n+1 and critical exponent Currie, Electron. J. Combin. 33(3) #P3.31, Conjecture 19 (arXiv Conjecture 1): each of the ten listed morphic images of the fixed point G of [01,2,02] or of its reversal has factor complexity 2n+1 and critical exponent exactly μ = 2+1/(λ²-1), where λ is the real root of x³-2x²+x-1; the supremum is not attained. The proof reduces the words to three languages, describes all their bispecial factors and proves the final inequalities by exact arithmetic in Q(λ). One independent model referee pass. Not formalized Proof · Exact checks

Number theory, packing and partitions

Result Finding Lean Evidence
Batista's Egyptian fractions OEIS A398581: the lexicographically first triple uniquely maximizes the last denominator for admissible k ≢ 1 (mod 5). ☑ Proof · Check
Rectangle partition asymptotics Gajdzica–Zakarczemny, Remark 5.9: uniform asymptotic for log p(h,w) as h → ∞, w ≥ h. ☑ Proof · Check
Two-difference Motzkin density The exact character formula for every pair in every compact abelian group, including noncyclic finite subgroups. Lean verifies the exhaustive relation-basis classification, compact and cyclic formulas, exact evaluator with explicit complexity bounds, and literal attaining witnesses. The cyclic maximum-independent-set problem has a prior 2026 solution. Complete Proof · Formalization · Finite checks
Liu–Zhu packing and Natalchenko–Sagdeev covering Sharp densities for every symmetric four-point set of integers. For odd coprime steps the proof determines the longest interval that can be covered exactly once. ☑ Proof and finite checks
Natalchenko–Sagdeev nonmixed-gap packing Natalchenko–Sagdeev’s main packing conjecture for arbitrary gap counts. The same formula gives the circle parameter in their Section 4 question. ☑ Proof and finite checks
Haralambis’s three-difference conjecture For every triple of positive integer differences, the extremal density equals the circle rotation bound. One pair sum is a period of an optimal set. ☑ Proof and certificate checks

Lonely runners

Result Finding Lean Evidence
Loneliness on a fixed relation hyperplane Explicit conditional limiting law for all admissible normals and shifts, with a uniform O(1/N)O(1/N) fixed-tolerance tail. Complete Proof · Exact checks
Zhang's Question 8.4: deletion values of the lonely runner For n >= 3, 1 <= r <= n-1, and V = [n-1] \ {r}, LR(V) = 1/r when 2r > n-1, and LR(V) = 2/D_r when 2r <= n-1, where D_r is the least integer at least n+r coprime to r. In particular the strict inequality holds for all n >= 4 and 1 <= r <= n-2. Contributed by @aadrianvv. Complete: Lean checks the strict inequality and the full exact-value formula; seven-module kernel replay passed, full repository gate pending. Proof · Exact checks
Five-speed Loneliness Spectrum Conjecture The values below 1/5 are exactly s/(5s+1) for positive integers s. Cordella’s twelve sporadic primitive quintuples exhaust the exceptions to the two known families. Lean verifies the lower bound 1/6 and both primitive tight-five directions; the full spectrum classification remains incomplete. Partial Proof · Tight-five proof and receipt · Exhaustive verifier
Eight-speed Loneliness Spectrum Conjecture The values below 1/8 are exactly s/(8s+1). Three infinite families and 17 sporadic tuples exhaust the primitive cases. ◐ Proof and exhaustive verifier
Seven-speed Amended Loneliness Spectrum Conjecture The values below 1/7 are exactly s/(7s+1), 3/23, 5/37 and 9/65. One infinite family and 27 sporadic tuples exhaust the primitive cases. Lean supplies the proved five-speed bound in the critical-plane and large-norm implications; PrimitiveSixClassification remains an explicit premise. ◐ Proof and exhaustive verifier
Six-speed Loneliness Spectrum Conjecture The values below 1/6 are exactly (P−1)/(6P) for P ≥ 7 congruent to ±1 modulo 6. Cordella’s nine sporadic sextuples exhaust the exceptions to the three known families. Lean verifies realization of every proposed value by six distinct positive primitive speeds; actual rational-lattice, Gram and integer-factor components are proved, and the registered norm wrappers use the proved four-speed bound and retain hplane. Universal classification remains incomplete. Lean also proves the conditional two-parameter radius estimate; the finite plane enumeration remains incomplete. Partial Proof · Realization proof and receipt · Exhaustive verifier
Tight six-speed sets and seven-coordinate critical subtori The consecutive set is the unique primitive six-speed equality case. The seven-coordinate critical plane is unique up to signs and permutations. ◐ Proof and exhaustive verifier
Eight lonely runners, all equality cases Exactly three primitive seven-speed sets attain 1/8. They are the consecutive speeds and the two examples in Goddyn and Wong’s census. ◐ Proof and exhaustive verifier
Nine lonely runners, unique equality case The consecutive speeds 1 through 8 are the only primitive set attaining 1/9. ◐ Proof and exhaustive verifier
Eleven through fourteen lonely runners, all equality cases Only consecutive primitive speeds attain equality for 11–13 runners. For 14 runners the Goddyn–Wong set is the only further case. Partial; conditional norm reduction checked Proof and verifier
Ten lonely runners, unique equality case The consecutive speeds 1 through 9 are the only primitive set attaining 1/10. Partial; conditional norm reduction checked Proof and exhaustive verifier

☑ Lean formalization of the stated result · ◐ Partial formalization with explicit remaining dependencies · ☐ No Lean formalization

See docs/lean-status.md for exact formalized statements and gaps.

Precise scope

The gamma-quotient result proves the full all-order assertion in Karp–Vishnyakova–Zhang's Section 3, item 11 for all positive gamma arguments and every c−d>0c-d>0. The parameter-dependent signs and integer finite-rank cases belong to this single theorem. The subsidiary order-three sign correction is not counted separately.

The Sagan–Swanson result proves the full four-family asymptotic-normality Conjecture 7.8, with no rate restriction on convergence of k/nk/n to a fixed value in (0,1)(0,1). The proof controls the conditioning event and derives the actual conditional means and variances. It does not claim the separate coefficient log-concavity, parity log-concavity, bottom-heaviness, or boundary-regime questions.

The Bernardi–Morales result proves the complete delta-process tree-probability conjecture and its stronger event-determinant conjecture: published numbers 27–28, preprint numbers 21–22. The tree proof allows arbitrary fixed color maps, including nonsurjections; the event proof uses a fixed surjection, as in the source. Both statements are recorded under one named investigation. The two exact verifiers check finite instances independently of the general determinant arguments.

Gilmore's matching identity concerns his product-defined matrix and proves the complete published Conjecture 6.16 (arXiv Conjecture 6.11). The neighboring shifted reverse Bessel Hankel-total-positivity conjecture remains open.

Chow's weak consecutive-interlacing assertion follows immediately from established stronger refinement results. The note credits those results and does not claim a new compatibility theorem. The current literature audit found no earlier explicit resolution of the consecutive assertion; that is a bounded search, not proof of priority.

The L-solitaire counterexample relies on the original paper's Definitions 2--4 allowing positive, not necessarily diverging, horizontal scaling factors. It does not refute the natural corrected assertion for diverging scales. The note supplies a proof of that corrected assertion for all admissible layer sets.

The necklace result refutes only Pham's exact-degree guess. It does not refute the distinct Harris--Nguyen (2025) conjecture that a necklace and its reverse-complement dual admit a common denominator of degree equal to their period. The included dual calculation is consistent with that conjecture.

The Li example is neither a spider nor a caterpillar and refutes the non-spider sign prediction. Earlier spider constructions already refute the unrestricted coefficient assertion; likewise, earlier literature contains trees with two degree-4 vertices. The newly reported result is explicitly limited to the non-spider class, and novelty is provisional pending independent review (see the novelty audit). This example does not refute Dahlberg--She--van Willigenburg's general non-e-positivity conjecture.

These are mathematical claims with explicit proofs and computer checks, not peer-reviewed publications. A literature search cannot establish that a counterexample or stronger theorem is the first discovery. Please cite the primary publications when discussing the original conjectures.

Primary sources

For the other entries, primary publications and conjecture identifiers are cited in their linked proof notes.