In degree 104, two rank-324 Hecke operators over Q(−3), together with classical tests, leave four candidate primes. An integer evaluation matrix of rank 2,925 constructed using Q(−7) has determinant coprime to their product. Together with the certificates for earlier degrees, these computations imply K4k(Z)=0 for 1≤k≤26, assuming the strict Selmer comparison, coverage of the admissible representations, and a simultaneous residual trace condition. The supplement contains the matrices or their generators, exact residues, and replay code.
Author comments: Two pages. The linked computational supplement contains code and certificates for the stated range, with a full reconstruction of the degree-104 matrices and a transcript of the fresh-copy replay. https://github.com/leopintar05/hecke-k-theory-104/releases/tag/v1.0
Primarily AI-generated textHuman understanding: some parts
In the proposed proof of the generalized Riemann hypothesis for Dirichlet L-functions by Alpöge and Claude, theta lifts give an upper bound for a smoothed prime sum on certain long intervals. We simplify three steps linking the lifts to the zeros of L-functions. Our main result compares the character sums at a common scale and allows greater variation in their weights while preserving the exponent in the prime-sum bound. We also remove a logarithmic loss in the comparison of lattice slopes and give a short deduction of the zero-free conclusion from a classical oscillation theorem, including for families with a possible real zero. The construction of the lifts and the estimates producing the intervals enter as inputs.
Author comments: 10 pages. The paper simplifies three steps conditional on the arithmetic inputs to the Alpöge–Claude argument. The source includes two exact checks of the finite character calculations. The manuscript and code are available at https://github.com/twinheterodox/three-simplifications-dirichlet-grh
We study an unweighted MAX-HORN-2SAT class, which is also a class of MAX 2-SAT instances, in which every Boolean variable has exactly three literal occurrences. For a simple cubic graph H=(V,E) the formula TH satisfies OPT(TH)=4∣V∣+∣E∣+α(H), and every assignment of value A is repaired in polynomial time to an independent set of size at least A−(4∣V∣+∣E∣). This gives a polynomial-time 83/84 approximation from a cubic independent-set approximation and a near-bipartite cut algorithm, an L-reduction from cubic MIS, and APX-completeness. For simple d-regular graphs the optimum is (d+1)∣V∣+∣E∣+α(H) and the factors are: (3d+2)/(3d+3) from the all-true assignment, which is at least 83/84 for d≥27; 83/84 from the Berman--Fujito guarantee for d≥22 and, with a semidefinite cut rounding, for d=21; and (21d+20)/(21d+21) from any 6/7-approximate independent set. For arbitrary simple graphs the optimum is n+3m+α(H).
MAX-CUT on a d-regular graph G reduces to MAX 2-SAT with the clauses (xu∨xv) and (¬xu∨¬xv) per edge, of value m+cut(S). The conflict graph H(G) is (d+1)-regular and α(H(G))=m+maxcut(G). A Goemans--Williamson cut gives on TH(G) the factor 1−6d+121−αε≥83/84 for every d≥1, and a γ-dependent analysis gives 1−6d+11.50.1053, where αε=αGW−10−5 and γ=maxcut(G)/m. For arbitrary simple graphs without isolated vertices the factor exceeds 0.9932. A cut guarantee cd on graphs of maximum degree d gives 1−6d+121−cd, which exceeds 0.9977 in expectation for d=3. A ratio on TH(G) and a cut guarantee are equivalent, so the cut recovered is the Goemans--Williamson or bounded-degree cut used, and the MAX-CUT guarantee is not improved.
The main consequence is that on the class {TH(G)} of three-occurrence Horn formulas obtained from MAX-CUT instances the proved factor 83/84 exceeds the Unique Games thresholds ≈0.94616 and ≈0.94017 that hold for all instances of weighted MAX-HORN-2SAT and of MAX 2-SAT.
For a restricted hard-split family with unit vertex rewards and edge-clause weights at least one we obtain 27/28, and 257/260 with edge weights four. The Unique Games thresholds of Brakensiek--Huang--Zwick (≈0.94616), Lewin--Livnat--Zwick (≈0.94017) for MAX 2-SAT, and Khot--Kindler--Mossel--O'Donnell (≈0.87856) for MAX-CUT quantify over all instances; the guarantees above concern classes with three occurrences per variable, exceed the first two thresholds, and are consistent with the conjecture. Max-Ones 2-SAT for negative (Horn) formulas equals Maximum Independent Set and Min-Ones 2-SAT for positive formulas equals Minimum Vertex Cover, with the same optima, so the published hardness of these two problems applies to them, and OPT(TH)=4∣V∣+∣E∣+MaxOnes(NH) for the cubic negative formula NH. The Lean~4 development formalizes the formula TH for every degree and every degree sequence, its score decomposition, repair and optimum identity, the arithmetic transfers, the MAX-CUT clause count and lifting step, and the hard-score optima 4m+α and m+α; the Goemans--Williamson, Berman--Fujito and Unique Games statements are explicit hypotheses taken from the literature.
Author comments: This submission is supported by a Lean 4 formalization, available at https://github.com/frankvegadelgado/paloma. The formalization covers the exact optimum identities, the repair and lifting steps, and the arithmetic of the approximation ratios. The Goemans–Williamson, Berman–Fujito and Trevisan guarantees are taken from the literature and are stated as explicit hypotheses; the Unique Games thresholds are cited in the paper and are not formalized. The formalization will be submitted to Palomar, and its acceptance is not guaranteed.
Primarily AI-generated textHuman understanding: all partsMax-SATUnique Games Conjectureapproximate-computingindependent-set-problemlean4np-hardsatisfiability
Contributed by Pengcheng Fang, Tao Hu, Yanyang Li, Quanyu Tang, Haiqi Zhang
Br\"uck conjectured that if a nonconstant entire function f of finite hyper-order not equal to a positive integer shares a finite value a counting multiplicities with its derivative, then f′−a=c(f−a) for some nonzero constant c. We prove this conjecture.
Author comments: First version. The main theorem has been formalized in Lean 4. The formalization and verification instructions are available at https://github.com/QuanyuTang/bruck-conjecture-formalization.
A mix of human-written and AI-generated textHuman understanding: some partsBrück conjectureentire functionsvalue sharing
In August 1960 the Simulmatics Corporation ran a computer simulation for the Democratic
National Committee. It ranked 32 non-Southern states by Kennedy’s likely strength if religion
dominated the campaign. Pool and Abelson (1961) reported that this ranking correlated .82
with the November result, using no political data later than October 1958. This paper reconstructs
the mathematics behind that result from the published record: the 1961 Public Opinion
Quarterly article and the list of 480 voter types later printed in the appendix of Burdick’s The
480 (1964).
We make four contributions. First, we write the model as a complete system: a 480-type
typology, a pooled issue-cluster estimator, a synthetic-state aggregator and nine cross-pressure
equation sets. Second, we show that the model’s one non-trivial step, estimating how many
1958 Democrats were anti-Catholic, is an odds-ratio constrained 2 × 2 table with a unique
feasible solution for every odds ratio, and we derive its comparative statics. Third, we prove
that the authors’ “lucky guess” benchmark (add 34 per cent of the Catholic share to the 1956
Democratic vote) is the in-sample ratio of two least-squares coefficients, so its correlation is a
two-parameter multiple correlation fitted with hindsight. Fourth, we recompute the published
benchmarks from public returns. The authors’ printed ranking has a Spearman correlation of
0.755 with Kennedy’s 1960 share across the 32 states (p < 10−6). The model’s own list of
failures, the Mountain states, reappears in our data. Six animations, collected in an online story
map, accompany the paper..
Primarily human-written textHuman understanding: all partsSimulmatics Corporation; computer simulation of voting; history of computing
We prove that for n≥1062+4, every closed smooth Riemannian n-manifold admits a smooth isometric embedding into Rsn+5, where sn=n(n+1)/2.
A mix of human-written and AI-generated textHuman understanding: some partsHopf spheresclosed Riemannian manifolddifferential right inverseglobal isometric embeddingh-principlephase regularity
Contributed by Matija Kazalicki, Siniša Slijepčević
We prove that the coefficients of
h=Δ/E4=∑n≥1b(n)qn/2 satisfy the
Atkin--Swinnerton-Dyer congruence
b(mpr)−ap(g)b(mpr−1)+p3b(mpr−2)≡0(modp3r)
for every prime p≥5, m≥1 prime to p, and r≥1,
with b(x)=0 for nonintegral x. Here ap(g) is the pth Hecke
eigenvalue of the unique normalised newform g of weight four and
level 24 with trivial character. Since h branches at the elliptic
point ρ=e2πi/3, it is not a modular form. In a local
uniformising chart of the double cover s2=1/j above ρ, the associated
differential extends holomorphically across the ramification point.
It defines a parabolic cohomology class with
coefficients in the symmetric square of the elliptic local system.
More generally, we study modular differentials twisted by rational
powers of a Hauptmodul. When the differential is nonzero and square-integrable
on the Kummer cover and the twisted parabolic cohomology consists
of two lines of types (k−1,0) and (0,k−1), we prove three-term
congruences modulo p(k−1)r at good primes of residue degree one.
A comparison of crystalline and \'etale Frobenius identifies the
polynomial of the recurrence with the Frobenius polynomial of a
two-dimensional Galois representation. Our local criteria produce
further examples, including E4j−a for every rational a with
1/3<a<2/3.
Let L be an even hyperbolic lattice. Its exceptional lattice E(L) consists of the vectors with finite orbit under the stabilizer of a chamber of the (−2)-reflection group. We prove that the largest rank for which E(L) can be nonzero is 26, answering a question of Grabbel, Martin, Mezzedimi, Raitz von Frentz and Schmidt. Conway's lattice II1,25 attains the bound.
In rank r≥27, we construct in each chamber a finite set of deficient cusps f (meaning dimSf<r−1, where Sf is the rational span of f and the roots perpendicular to f) with ⋂fSf=0. We use these cusps to construct a Zariski-dense subgroup of the chamber group generated by at most r−1 Eichler transformations. If L contains a (−2)-root, this subgroup is thin and no chamber admits a Weyl vector. For the groups obtained by this cusp construction in rank at least three, we study the principal congruence subgroups defined by the action on L. At all sufficiently large levels, the expanding eigenvalue of every systolic element has degree two or four. In rank three the systole divided by the logarithm of the level has lower limit 2 and upper limit 4. Every complex projective K3 surface with Picard lattice U⊕⟨−4m⟩⊕2 (m≥3) or ⟨2⟩⊕⟨−4⟩⊕D8(−2) admits a symplectic automorphism of maximal Salem degree.
A mix of human-written and AI-generated textHuman understanding: all partsExceptional latticeK3 surfaceLeech latticeSalem numbercongruence systolehyperbolic reflection group
Let ν be the offspring variable of a Bienaymé tree without leaves,
with finite mean m>1. We prove that the speed v(λ) of the
λ-biased random walk is continuously differentiable on [0,m),
with v′(λ)<0 for every 0<λ<m. In particular v is
strictly decreasing, which answers a question posed by Lyons, Pemantle
and Peres in 1996. Heng Ma recently gave an
independent proof of strict monotonicity. Our proof is entirely
analytic when E[1/ν]≤31/49. In general it also uses two
interval-arithmetic computations, for biases between 1 and 9/5.
The main theorem, including both computations, has been formalized
in Lean.
Author comments: 25 pages. Mathematics and text produced with AI systems under the direction of the contributor; see the section on AI use. AA Lean 4 formalization of the main theorem, including both interval-arithmetic certificates, has been completed and is available at https://github.com/igor-kortchemski/monotonicity-speed-biased-random-walk
Primarily AI-generated textHuman understanding: some partsBienaymé–Galton–Watson treesGalton–Watson treesLyons–Pemantle–Peres conjecturebiased random walksmonotonicityspeed
How macroscopic irreversibility emerges from time-reversible Hamiltonian dynamics remains one of the deepest foundational problems in theoretical physics. The difficulty is structural, not technical. Microscopic equations of motion conserve phase-space volume and obey strict time-reversal invariance. Under the Poincaré recurrence theorem, nearly every trajectory returns arbitrarily close to its initial state; moreover, unitary (quantum) or Liouvillian (classical) evolution preserves relative entropy with respect to any invariant measure. Consequently, ensemble averaging alone cannot single out a thermodynamic arrow of time. Conventional derivations have long navigated this barrier through a split modeling strategy: either dissipation is imposed by hand at the microscopic scale, or uncontrolled approximations are introduced mid-derivation, implicitly injecting time asymmetry into the dynamics they seek to explain. As a result, reversible mechanics built on the principle of least action and dissipative relaxation governed by free-energy gradient flows have been treated as incompatible variational frameworks. Separately, standard system–bath constructions fail globally when extended from Euclidean space to compact manifolds under linear coupling. In this work, we resolve this split paradigm by constructing a unified, end-to-end variational and geometric framework that derives macroscopic irreversibility directly from time-reversible dynamics, without any a priori entropy-production inequality. Rather than treating dissipation as a phenomenological add-on, we show that directed relaxation is encoded structurally from the outset by three intertwined ingredients: conditional initial preparation, retarded bath response, and controlled coarse-graining. Our construction unites the two variational principles long kept apart: least action furnishes the reversible Hamiltonian backbone, while equilibrium statistical mechanics supplies the free-energy reference geometry. We model the environment as a Euclidean bath of harmonic oscillators linearly coupled to the core system and conditionally equilibrated to the core's instantaneous initial configuration. This preparation transforms bath modes into independent, centered Gaussians, cleanly cancelling the spurious initial-slip force while retaining thermal fluctuations and memory kernels. This is the precise point where initial-state conditioning, not the bare Hamiltonian, imprints the arrow of time. On this basis, systematic model reduction follows a rigorous, non-commuting asymptotic hierarchy. Exact bath elimination yields a generalized Langevin equation with nonlocal memory. A Lorentzian spectral density combined with the continuum limit converts this memory into an exponential kernel. Subsequent short-memory and small-mass limits produce instantaneous friction and overdamped diffusion, respectively. Critically, the ordering of these limits is an intrinsic feature of the theorem, not an arbitrary analytical convention, enforced by distinct scale separations. When we extend this framework to compact manifolds such as the torus, we uncover a fundamental geometric obstruction: global linear coupling cannot simultaneously preserve translation invariance and constant mobility. We resolve this obstruction by introducing periodic paired couplings, which restore uniform mobility by averaging over translated copies of the coupling profile. At the macroscopic level, the emergent dynamics selects probability currents under the Onsager–Rayleigh variational principle, which is equivalent to a weighted Wasserstein gradient flow—that is, the probability current follows steepest descent of free energy in a mobility-weighted Wasserstein metric. Lifting the analysis back to full underdamped phase space exposes the geometric mechanism of hypocoercivity: reversible Hamiltonian transport shears perturbations into ever finer filaments, so that thermal diffusion acts efficiently on small scales; together they drive exponential relaxation toward equilibrium. The Markov data-processing inequality then ties the entire multiscale hierarchy together. Equilibrium, we conclude, is not merely a passive asymptotic endpoint. It functions as a persistent reference geometry—the zero of relative entropy with respect to that reference—that organizes kinetic transport, dissipation metrics, and path asymmetry across scales.
A mix of human-written and AI-generated textHuman understanding: some parts
We study the free analogues of the real algebraic curves and surfaces of small degree, defined by one polynomial equation in N self-adjoint noncommuting variables. In degree 2 and 3 this brings together known results, coming from several sources. A free quadric corresponds to a Hermitian form H=Q+iP, the free conics are the quadratic relations classified by Ostrovsky\u\i\ and Samo\u\i lenko, which include the free circle, the quantum discs and planes, and the Heisenberg relation, and the free cubics in separated variables are fibered into duals of free products of cyclic groups. Our main new result concerns the elliptic quadrics, for which 3 notions of compactness, which coincide in the classical case, are shown to be different, with sharp thresholds. We also show that any compact free variety, given by finitely many equations, is homeomorphic to a quartic, and that the compact quartic curves, which contain the dynamics of the logistic map, cannot be classified by a list.
Primarily AI-generated textHuman understanding: some partsFree conicFree quadric
We prove the sharp Holmes--Thompson systolic inequality for every Finsler metric on the real projective plane, with no reversibility assumption. A smooth strongly convex metric attains equality exactly when its geodesic flow is Zoll. The proof cuts along an oriented systole and relates a quadratic boundary-distance integral to that of the symmetrized boundary distance. A reversible metric built from chords and a collar realizes the latter integral in the limit while preserving the systolic lower bound. The result implies the sharp Viterbo inequality for convex bodies in C2 invariant under multiplication by i, with rigidity for smooth equality cases.
We realize the Cartan algebras of restricted Kirillov--Reshetikhin modules as Cox rings of affine Schubert varieties, with Cartan degree given by the primitive divisor class. This yields upper cluster structures with non-inverted frozen variables for untwisted and diagram-twisted current algebras. In untwisted type, translation of the loop parameter gives a second cluster structure whose frozen variables are indexed by the affine Dynkin nodes. We construct charts adapted to the finite Chevalley operators and both shifted affine simple-root operators.
For every untwisted finite simple type, every fundamental node, and every Cartan degree, the translated theta basis carries a finite regular affine crystal. The construction uses a uniform reddening theorem and perfectness of the current operators on theta functions. A degree--level identity obtained from tropical Donaldson--Thomas duality determines the minimum level in divisible degrees and gives a finite minimization formula in all degrees. The crystals are simple, their tensor products are connected, and the divisible-degree crystals have a common limit. The affine theta crystals have the finite characters of the corresponding restricted KR modules; they are identified with quantum KR crystals in every classical type and at the exceptional minuscule and adjoint nodes covered by the available recognition theorems.
Author comments: 44 pages, comments are welcome
A mix of human-written and AI-generated textHuman understanding: all partsCluster AlgebraCox ringKR moduleTheta basiscrystalpolyhedral model
We prove Conjecture 8.12 of Fomin and Pylyavskyy on direction independence of their cluster structures. We also prove their surface Conjecture 5: the natural map from the tensor-diagram skein algebra to the ring of SL_3-invariants is an isomorphism.
Primarily AI-generated textHuman understanding: some parts
A collector repeatedly draws independent packages of at most s coupons from a universe of n coupons. When s∣n, we prove that the uniform law on the blocks of an s-set partition uniquely minimizes the expected completion time, resolving the Fang–Chang optimal-package conjecture. The partition law is stochastically optimal in Poissonized time, also for every common collection multiplicity. The proof combines independent package clocks, Finner's inequality, and strict concavity; an incidence-class tail argument characterizes equality. The same bound proves the lower-bound conjecture of Grunbaum and Yaakobi without their regularity assumptions. A separate elementary discrete-time bound identifies the unique optimal complementary-partition laws whenever s<n and n−s∣n. For fixed-size packages, a four-coupon perturbation preserves singleton marginals and strictly increases the completion probability at every feasible draw count. This gives an alternative proof and a stochastic strengthening of Schilling's conjecture, previously proved in expectation by Berend and Sher. More generally, perturbations formed from d disjoint pairs preserve all inclusion probabilities below order d and are strictly stochastically monotone in their absolute amplitude, with the direction determined by the parity of d. Their inclusion–exclusion coefficients count covering families. Combined with the Johnson-scheme decomposition, this yields the complete alternating signature of the expectation Hessian at the uniform law. The general fixed-draw partition comparison is left as a conjecture; its first feasible draw count follows from the Poissonized bound.
Author comments: 17 pages. Resolves the Fang–Chang expected-time optimum; the general fixed-draw partition comparison remains open. Internal audit and exact finite checks, without proof-assistant formalization or independent refereeing. Source and reproducibility records: https://github.com/long-mathematics/optimal-coupon-packages/tree/f11c52cc42b82575fff5ed1c699717976c8546d0
A mix of human-written and AI-generated textHuman understanding: some parts
Let L(xayb)=a!b!. We prove factorial-moment nonvanishing for nonzero polynomials in two variables whose highest homogeneous part is not a scalar multiple of (x+y)D. The proof uses an exact radial Mittag–Leffler transform, polynomial monodromy, and Lindemann–Weierstrass. The transform produces vanishing leaf germs containing a pole density and a Volterra correction. The corrections continue with their full integration histories retained and have no negative Puiseux coefficients at a nonzero terminal critical value. The resulting weighted coefficient annihilates the top rational monodromy component, while the unweighted inverse-derivative vector belongs to that component by a zero-trace identity on every proper composition block. Conjugate pairing excludes all nonzero critical values, and endpoint leaf germs dispose of the remaining pure powers. Combining the nonflat theorem with the explicitly stated single-integral nonvanishing theorem of EIT I gives the two-variable Factorial Conjecture. As a formal consequence, the zero-integral polynomials for the product-exponential measure e−x−ydxdy form a Mathieu–Zhao subspace; more strongly, a polynomial whose factorial moments vanish for all sufficiently large powers is zero. Thus the two-dimensional product-exponential case of van den Essen's Integral Conjecture follows. The nonflat argument is independent of EIT I.
Author comments: 20 pages. The general flat case uses EIT I, Corollary 1.4 (hexagon:2611.00736v1); the nonflat argument is independent of EIT I. Source and reproducibility records: https://github.com/long-mathematics/factorial-conjecture-two-variables/tree/237c1a7c1e07076e8117f20bade591a4a9fe05a9
A mix of human-written and AI-generated textHuman understanding: some parts
We give an outline of the proof by L. Alpöge, in a formal development in Lean 4, of the Riemann hypothesis and of the generalised Riemann hypothesis for Dirichlet L-functions. The main theorem is that L(s,χ)=0 for Res>21 and every Dirichlet character χ. A positivity argument reduces it to one-sided bounds, on windows of scales [Y,Y3/2], for positive combinations of Riesz-weighted Chebyshev functions twisted by the powers of a primitive character. These bounds come from hermitian lattices attached to theta lifts over auxiliary abelian CM fields: the inert prime sums of these fields are bounded, prime by prime, by local gains of the lattices, and slope inequalities bound the lattice side. The losses in the inequalities are controlled by bounds, polynomial in the conductor, for Tate--Shafarevich groups of Tate twists, which are proved by Eisenstein congruences; the Galois representations that these need are constructed by theta lifts. We state the argument in standard language as far as we understand it, give many proofs in outline, and say what we do not yet understand. The paper was prepared with the help of Aleph, an AI math research assistant.
Author comments: This is a best-effort attempt to give a faithful, 100-pages outline of the GRH, AI-assisted proof by Alpöge, at the moment when the only available source is the Lean 4 formal development. This is work in progress, focusing on clarity and understanding, with as complete references as possible to used prior knowledge.
Primarily AI-generated textHuman understanding: some parts
The main purpose of this preliminary report is to begin looking for some of the sources, provenance, and mathematical history behind the proposed proof of the generalized Riemann hypothesis announced by Levent Alp\"{o}ge using a Claude model. The mathematics involved reflects many lifetimes of work, so I would like to recognize as many as possible of the people whose work may have contributed to it: the mathematicians working today, our teachers, and all of our predecessors. This is only a beginning, and I apologize for all the people, papers, and other contributions I may have missed.
Author comments: 65 pages, 4 figures
A mix of human-written and AI-generated textHuman understanding: some partsArakelov theoryDirichlet L-functionsGeneralized Riemann HypothesisHistory of mathematicsautomorphic formstheta lifting
Assuming the Riemann hypothesis, we show that for every real x ≥ 9.6 × 10³⁷ the open interval of length 384√x (log x)² above x contains at least (14/3)√x distinct products of two primes of size comparable to √x. Their contribution to the sum of the least prime factor divided by the integer is at least 14/9. For every fixed C > 0, we also prove that the composite sum in an interval of length C√x (log x)² is bounded above unconditionally and bounded away from zero under the Riemann hypothesis. We present a computer-assisted proof, verified in Lean 4, that the mean-square estimate of Chirre and Hagen implies the explicit count and weight bounds. The results have not been independently reviewed.
Author comments: The proposed Lean statements and proof plan are available in the Tau Ceti roadmap.
https://github.com/TauCetiProject/TauCetiRoadmap/commit/37a574a19ac4e127a147ab5a3ee4d456568a3f80
Primarily AI-generated textHuman understanding: no parts
Put λ1=log(1+1/N) and λ2=log(1+2/N). We prove that λ1λ2 is irrational for every integer N≥84 and every even N≥52, and more strongly that 1,λ1,λ2,λ1λ2 are linearly independent over Q on the same range. We also prove that 1,λ1,λ2,λ22 are Q-linearly independent for N≥58, that 1,λ1,λ2,λ1λ2,λ22 are for N≥88, and that 1,λ1,λ2,λ12,λ1λ2,λ22 are for N≥151: for N≥151 the point (λ1,λ2) lies on no curve of degree at most two defined over Q. These are instances, not conditional on the conjecture and with rational coefficients, of what the strong four exponentials conjecture predicts for two logarithms. To our knowledge no published result states any of them for N below about 4.5⋅1011. The proofs use explicit two-level Rodrigues forms of type II Hermite-Padé type, linear recurrences in the index proved by exact creative-telescoping certificates, Pituk's theorem on the growth of solutions of Poincaré-type recurrences (read only in restatements, so all results are stated modulo it), exact or ball-arithmetic certificates for the roots of the characteristic polynomials, and a 2-adic content lemma at odd points that lowers the threshold for even N. The irrationality of λ1λ2 and the four-term independence for N≥103, and the six-term (conic) independence for N≥151, are also checked in Lean 4 with Mathlib, by arguments that do not use Pituk's theorem.
Primarily AI-generated textHuman understanding: no partsHermite–Padé approximationcreative telescopingirrationalitylinear independencelinear recurrencesproducts of logarithmsstrong four exponentials conjecture
Over any field, we construct an essentially small hereditary extriangulated category with enough injectives that admits no triangulated embedding, that is, no fully faithful extriangulated functor into a triangulated category. In particular, not every extriangulated category is equivalent as an extriangulated category to a full extension-closed subcategory of a triangulated category. Nevertheless, every countable family of objects lies in a full extension-closed subcategory equivalent as an extriangulated category to a full extension-closed subcategory of an algebraic triangulated category. To construct this example, we build an extriangulated category from any inverse system of vector spaces with surjective transition maps over a nonempty ordinal. We prove that it admits a triangulated embedding if and only if every projection from the inverse limit is surjective. We obtain the counterexample from such an inverse system over the countable ordinals with nonzero initial term and vanishing inverse limit.
Two preparation experiments can each fit a heterogeneous coverage model and still fail a shared-environment explanation. We characterize joint compatibility for finite-label probability representations with half-step source probes and pointwise calibration error. Absolute weaker-source void probabilities require the weighted calibration marginals to share mass at least κ=(β1+β2−1)+. Combining this observed budget with exposure moments and a shared calibration band gives an exact criterion for the closure of the representation. A deterministic-anchor example requires error radius 0.10376 for separate fits and 0.33048 for joint compatibility. On an explicit three-moment family, matching bounds identify the first-order heterogeneous coefficient κ; four environments preserve the anchors and give second-order stability of the minimum radius over unknown centers. A calibration-preserving moment repair and response-opening geometry yield sharp local inverse bounds with unknown, unequal group masses and first anchors. For prescribed mass and standard-deviation envelopes pn,hn with hn2≤bpn, the matching radius-estimation risk is min{1,[n−1/2+min(hn,n−1/4)]/pn}. Uniform consistency has a corresponding two-scale necessary and sufficient criterion. Unknown envelopes introduce distinct adaptation costs. An interval honest over the full envelope union has matching expected diameter min{1,n−1/4/pn} even on the deterministic subclass. Under uniform protection of the deterministic regular risk, point estimators pay a sharp logarithmic cost elsewhere: one scale-free thresholded projection attains min{1,[n−1/2+min(hn,(log(en)/n)1/4)]/pn}, and a constrained-risk lower bound matches it. The larger mass-unrestricted moment class admits no uniformly consistent radius estimator. Simultaneous Bernoulli confidence events also give finite-sample compatibility certificates.
Author comments: Published English–Chinese preprint: https://doi.org/10.5281/zenodo.24412395. Joint compatibility, calibration-radius stability and inference for heterogeneous finite-label coverage with shared calibration. Companion preprints: doi:10.5281/zenodo.23912425 and doi:10.5281/zenodo.23911961. Reproducibility scripts and synthetic fixtures are included under anc/.
A mix of human-written and AI-generated textHuman understanding: some partsfinite-sample testingheterogeneous mixturesoverlapping measurespartial identificationshared calibrationtruncated moment problem
For the induced complex on primitive integer pairs in a triangular lattice, we identify first homology with interior components of a half-scale midpoint bond graph. Its complementary planar dual has the visible-site graph as its line graph, giving a bijection between finite visible islands and bounded nonsingleton dual faces.
In particular the hole density h is a local Euler-product term plus the island density b0:
h=3J2−2J3−J1+b0, with Jj=∏p(1−j/p2).
Each odd prime produces a matching, two primes produce a small forest, and diameter exclusion bounds cycle birth. We refine the bounds by retaining the actual earlier prime sieve at each step.
A computer-assisted certificate, including a nonnegative coverage recurrence for islands, yields 0.10917978<h<0.10923225.
A separate length-six finite-path CRT method gives h>0.10914952, still below the island certificate, and an exact decomposition distinguishes old-component chords from cycles born among new cross-component bonds.
Assuming the random-coprime nonpercolation theorem of Le Fourn--Liu--Martineau, the relative cycle increment of a new prime converges to a conditional connection probability. The planar correspondences, density existence, and numerical bounds do not require this probabilistic input.
A mix of human-written and AI-generated textHuman understanding: some partsBetti densityEuler productsmidpoint bond graphsplanar topologyprime sievetriangular latticevisible lattice points
A reader's guide to L. Alpöge's Lean 4 formalization, announced on 20 November 2026, of the Riemann hypothesis for ζ and for every Dirichlet L-function. It states in ordinary notation the five theorems of the archive's Main.lean, then follows the chain of declarations behind them. The chain runs through the reduction to an inequality that excludes zeros to the right of 21, the analytic argument through ramp sums and Landau's theorem on windows, the origin of that inequality, and a bound for Tate–Shafarevich groups of Tate twists. The guide compares the classical inputs with their archive forms and sets Alpöge's twenty dated steps against the chain. The theorems and their proofs are Alpöge's, and the guide is a reading of his archive. A. Sutherland has compiled the archive and reports that Lean's kernel and two independent kernels accept it.
Primarily AI-generated textHuman understanding: no partsDirichlet L-functionsFormalized mathematicsLean 4MathlibRiemann Hypothesis
We work in Ford and Maynard's framework for sieve problems with Type I and Type II information. To Type I information at level 1/2 and Type II information for factors up to xν, we add Hooley's condition: the parity bias of the sequence in arithmetic progressions is small on average over moduli up to xα. We construct prime-free sequences that satisfy all three for every Type II range ν<1/30 and every twisted level α<4/15, and strictly below each of four further corners (η,α′) of a staircase in the (ν,α)-plane, from (1/11,1/10) to (1/20,1/5), that is, for every ν<η and every α<α′. The constructions add to Ford and Maynard's continuum model one function of the total size of the small prime factors for each number of them; these functions solve the twisted conditions exactly. Each case rests on a computer-assisted certificate in exact or directed-rounding integer arithmetic. Independent audits read the certificates' code for every case except (1/11,1/10) and (1/14,1/8), whose certificates had an independent review of their formulas and inputs but not of their code. The cases (1/11,1/10), (1/14,1/8) and (1/30,4/15) each rest on a single implementation. Conversely, the term d=1 of Hooley's condition, global parity balance, already forces a positive proportion of primes once ν≥0.116. For α>1/3 we exhibit twisted rows that the added functions cannot enter, and we conjecture, on floating-point evidence, that constructions of exactly this form cannot pass α=1/3 (Conjecture 8.2). We also record what the constructions imply for sieve approaches to Goldbach's problem.
Primarily AI-generated textHuman understanding: no parts
A d-dimensional simplex is a k-reptile if it can be tiled by k mutually congruent simplices similar to it. Matoušek and Safernová conjectured that for d≥3 a k-reptile d-simplex exists only when k=md for an integer m, and proved this for d=3. Kynčl and Patáková proved that in dimension four k must be a square, noted that every k-reptile simplex known in dimension d≥3 has k=md, and asked whether a four-dimensional k-reptile simplex exists with k a square but not a fourth power. We show that the simplex with vertices (0,0,0,0), (1,0,0,0), (21,21,0,0), (32,31,31,0) and (43,41,41,41), the fundamental alcove of the affine Weyl group of type F4, is a 4-reptile: it is tiled by four congruent copies of itself scaled by 1/2, two of them mirror images. This answers the question of Kynčl and Patáková with k=4 and disproves the conjecture of Matoušek and Safernová in dimension four. The subdivision itself is due to Felikson and Tumarkin, who found in 2004 that the affine reflection group of type F4 has a reflection subgroup of index 4 whose fundamental simplex is similar to the alcove. What we add is the observation that this settles the reptile question, a short direct proof of the tiling, and exact computer checks of it. This work has not yet been reviewed by experts. The proofs, the computer checks and the text were produced by AI models, working under the author's direction.
Author comments: Zenodo archive: https://doi.org/10.5281/zenodo.24128596. Paper text license: CC BY 4.0. The published code and data are governed by the licenses in the tagged release. Release archive folders: checks, data.
A graph is 4-flat if the complex obtained by filling every simple cycle with a separate disk embeds piecewise linearly in four-dimensional space. The Heawood family consists of 78 graphs obtained from K7 and K3,3,1,1 by triangle–star and star–triangle transformations. Van der Holst proved that these graphs are not 4-flat; Georgakopoulos and Winter completed the proof that all 78 are excluded minors and conjectured that there are no other excluded minors for 4-flatness. We show that the known eleven-vertex, twenty-two-edge graph H is not 4-flat and has no Heawood-family minor, resolving the completeness question negatively. Mattman and Pierce found by computer search that every proper minor of H is 2-apex; this implies that H itself is a further excluded minor. Mattman and Pierce found H, subsequently studied by Kim and Mattman. An earlier paper by the author proved that its one-edge extension G1 has Colin de Verdière invariant 6 and no Heawood-family minor. This made 4-flatness completeness incompatible with van der Holst's proposed equivalence between 4-flatness and invariant at most 5. The new step is an explicit mod-two deleted-product cycle with invariant one. Van der Holst and Pendavingh's obstruction theorem therefore excludes an even map, and hence an embedding, of a cycle subcomplex of H into four-dimensional space. Two separately written exact implementations evaluate the certificate, using moment-curve intersections and the recursive cellular invariant, respectively. This work has not yet been reviewed by experts. The proofs, the computer checks and the text were produced by AI models, working under the author's direction.
Author comments: Zenodo archive: https://doi.org/10.5281/zenodo.24146253. Paper text license: CC BY 4.0. The published code and data are governed by the licenses in the tagged release. Release archive folders: checks, data.
Primarily AI-generated textHuman understanding: some parts4-flat graphsColin de Verdière invariantGeorgakopoulos–Winter conjectureHeawood familydeleted productexcluded minors
We prove that every even integer other than 2, 12, 14 and 18 is the combinatorial period, the number of bounces before repeating, of a periodic billiard trajectory in the regular pentagon that does not have five-fold rotational symmetry. Davis and Lelièvre conjectured this in 2018 and checked it up to period 4000, and Kontorovich and Zhang proved it for almost every even number. As they explain, a theorem of Davis, Fuchs and Tabachnikov turns the question into one about vectors. Each vector (a+bφ,c+dφ) obtained from (0,1) by products of four explicit 2×2 matrices, where φ is the golden ratio and a,b,c,d are integers, has the half-period ℓ=a+b+c+d, and when a congruence modulo 5 holds it gives such a trajectory of period n=2ℓ. Varying the power of one of the two triangular matrices in such a product gives half-periods in an arithmetic progression. We give 379 of these progressions, which together contain every half-period ℓ≥2 except nine values. Six of the nine come from explicit trajectories checked in exact arithmetic; the other three are halves of excluded periods. The covering argument is checked in Lean, assuming the cited theorem. Whether the four excluded periods occur is not treated. This work has not yet been reviewed by experts. The proofs, the Lean formalization and the text were produced by AI models, working under the author's direction.
Author comments: Zenodo archive: https://doi.org/10.5281/zenodo.24119275. Paper text license: CC BY 4.0. The published code and data are governed by the licenses in the tagged release. Release archive folders: lean, checks, data.
Primarily AI-generated textHuman understanding: some partsDavis–Fuchs–Tabachnikov theoremDavis–Lelièvre conjectureGolden Ratiobilliards in the regular pentagoncombinatorial periodperiodic billiard trajectories
Let cn,3 be the largest size of a subset of {0,1,2}n containing no combinatorial line, a triple obtained by varying a nonempty set of coordinates through the same letter while fixing the rest. We give a computer-assisted proof that c7,3=1302, confirming Polymath1's tentative conjecture and closing its bounds 1302≤c7,3≤1348, recorded by Tao in 2009 and O'Bryant in 2015. The witness realizes Polymath1's known lower bound. Three auxiliary SAT bounds, slicing and global alphabet symmetry give sixteen main slice obligations. Refutations of all nineteen formulas were accepted by FRAT-rs, and in a separate earlier production campaign by drat-trim's lrat-check. Formal verification applies to certificate checking: cake_lpr, verified in HOL4 and compiled with CakeML, accepted refutations of all nineteen formulas across two regeneration campaigns. The SAT encodings, cap induction, symmetry reduction, case cover and counting interpretation are justified by the written proof and separate executable checks; they are outside that formal verification.
Author comments: Zenodo archive: https://doi.org/10.5281/zenodo.23752609. Paper text license: CC BY 4.0. The published code and data are governed by the licenses in the tagged release. Release archive folders: checks, data.
Primarily AI-generated textHuman understanding: some partsA156989LRATSATcombinatorial linescomputer-assisted proofdensity Hales–Jewett
For integers r≥2 and d≥1, Solus's base-rd-simplex is the convex hull of the standard basis vectors of Rd and the point −(r−1)(1,r,…,rd−1). Solus introduced these simplices through numeral systems, proved that their h∗-polynomials are real-rooted, and observed that computations suggest they are also Ehrhart positive. Fu Liu recorded this as Solus's conjecture that every base-rd-simplex is Ehrhart positive, and Liu and Solus restated it for all r≥2 and d≥1. We show that the conjecture fails: the base-145-simplex has Ehrhart polynomial \[ 1-\tfrac{268}{15}t+\tfrac{3955}{12}t^2+\tfrac{5833}{2}t^3+\tfrac{10343}{12}t^4+\tfrac{67228}{15}t^5, \] whose linear coefficient is negative. The proof is an explicit bijection between the lattice points of each dilate and pairs consisting of a residue modulo 145 and a vector of nonnegative integers, which reduces the count to the distribution of one integer-valued height function over the 537,824 residues. Lean checks the whole statement from the definition of the convex hull, including the finite height distribution, by kernel evaluation. This work has not yet been reviewed by experts. The proofs, the Lean formalization and the text were produced by AI models, working under the author's direction.
Author comments: Zenodo archive: https://doi.org/10.5281/zenodo.23875145. Paper text license: CC BY 4.0. The published code and data are governed by the licenses in the tagged release. Release archive folders: lean, checks.
We construct smooth odd periodic data for the 2D inviscid Boussinesq equations whose temperature gradient becomes unbounded at the origin in finite time. On an accelerating shear with inclination δ, a wave amplifies its temperature gradient by a factor of at least exp(c/δ). At the first return of its vorticity to zero, the gradient points downwards with inclination asymptotic to δ/2. We use the successive zeros of the leading wave vorticity to select return times on successively shorter intervals. An upper bound on the pressure Hessian permits correction of each periodic wave with both particle endpoints prescribed. A transverse Green function bounds the contribution of late forcing to the initial velocity by the inverse of the preceding amplification, yielding smooth initial data for the limiting solution.
Author comments: This version corresponds to https://doi.org/10.6084/m9.figshare.35081110 and https://doi.org/10.5281/zenodo.24409261
A mix of human-written and AI-generated textHuman understanding: all parts
The Boltzmann weights of the symmetric eight-vertex model with fixed invariants Δ and Γ form a curve of genus one, the intersection of two quadrics in $\PP^3$. When Δ and Γ are rational, this curve is an elliptic curve over $\Q$ with at least eight rational points and full rational 2-torsion, and the crossing parameter of Baxter's elliptic parametrization corresponds to a rational point on it. By Mazur's theorem, if this point is a torsion point (an elliptic root of unity), its order is 3, 4, 6, or 8. We give the loci of each exact order explicitly; over $\C$ they lie in finite-order conditions found by Boukraa and Maillard. Consequently the finite-order loci of orders 5, 7, 10, 12, and 20 contain no rational nondegenerate (Δ,Γ), and the order-8 case, which occurs with rational weights, has no rational counterpart in the six-vertex limit. Away from these loci the crossing point has infinite order, so the curve has positive rank, and its multiples produce infinite families of commuting inhomogeneous transfer matrices with rational weights.
Primarily AI-generated textHuman understanding: all partsBaxter's curveElliptic CurvesMazur's theoremXYZ modeleight-vertex modelrational pointsroots of unitytorsion pointstransfer matrices
We construct crystal operators on the regular theta functions of skew-symmetrizable cluster algebras with polynomial frozen variables. A frozen vertex that becomes a sink under mutation defines a locally nilpotent derivation of the upper cluster algebra. These derivations satisfy the Serre relations. Under mutable reddening, the regular theta basis of the middle algebra is perfect for both the sink and the source derivations. In each seed the g-vectors of its elements are the integral points of a rational polyhedral cone.
A tropical criterion detects the sink and source operators that form Chevalley pairs. Such pairs give integrable Kac--Moody actions and, under reddening, normal crystals on the regular theta basis. In the skew-symmetric case with mutable reddening and finite-dimensional Jacobian algebra, we compute the order of the theta function attached to a supported presentation weight along each frozen divisor. We illustrate this theory on various examples.
Author comments: 49 pages, comments are welcome
Primarily human-written textHuman understanding: all partsCluster AlgebraTheta basiscystal structure
We write out in full the proof, sketched by L. Alpöge in a footnote of his note of 20 November 2026, of Deligne's bound for the eigenvalues of Frobenius. Let X be a separated scheme of finite type over a finite field with q elements, and let F be a constructible ℓ-adic sheaf on X whose Frobenius eigenvalues at each closed point x have absolute value at most ∣κ(x)∣w/2. Then the eigenvalues of Frobenius on Hci(X,F) have absolute value at most q(w+i)/2, and they have absolute value exactly q(w+i)/2 if X is smooth and proper and F is lisse and pure of weight w. The proof is short. Granted the formalism of étale cohomology (the trace formula, the Künneth formula, Poincaré duality and Artin's vanishing theorem) and a weak Lefschetz theorem for generic hyperplane sections of affine varieties, it takes less than six pages, and its only estimate is the Cauchy–Schwarz inequality for a character sum over a finite field. Lefschetz pencils, monodromy and Rankin's method are not used. The proof was produced by GPT-6 Astra (ChatGPT Pro mode) from that sketch. It was checked by Claude Opus 5.5 at effort high and, independently, by Claude Fable 5.2 at effort max, which also wrote this text. No human has checked it.
Author comments: This note writes out, with full proofs, the argument sketched in footnote 3 of L. Alpöge's note of 20 November 2026 (https://alpo.ge/remark.pdf); its ideas are his. The weak Lefschetz theorem it uses is Deligne's, from the appendix of N. Katz, Affine cohomological transforms, perversity, and monodromy, J. Amer. Math. Soc. 6 (1993).
Checks: every step was checked by Claude Opus 5.5 at effort high and re-proved independently by Claude Fable 5.2 at effort max; no human has checked the proof. The prompt, the output of the run and the check records are at https://github.com/KokunoYumeto/grothendieck-montpellier-archive/tree/main/research/grh_20261120/deligne_sign_only
Primarily AI-generated textHuman understanding: no partsAI-generated mathematicsFrobenius eigenvaluesRiemann hypothesis over finite fieldsWeil conjecturescharacter sumsexposition of a proof sketchtensor powersweak Lefschetz theoremweightsétale cohomology
Under Dirichlet GRH, the Li--Liu spectral hypothesis R fails. For every fixed D>0 and 0<ε<1, each sufficiently large interval [X,2X] contains at least Xexp(−(logX)ε) even integers N for which N∥TN∥>(logN)D. We construct these integers unconditionally and use GRH only at the modulus N to obtain the norm bound. Each has at least (logN)D representations as a sum of two distinct primes. The proof isolates a matrix entry in which the continuous model cancels and the two arithmetic terms have different Buchstab densities.
Primarily AI-generated textHuman understanding: no partsBuchstab functionDirichlet GRHGoldbach representationsprimes in short intervals
Contributed by Will Howe, Grant Huffman, Jensen Kohlmeyer, Liam Kruer
For every ε>0, every sufficiently large integer N is a sum of two positive integers all of whose prime factors are at most Nε. We prove this smooth-summand conjecture of Erdős using weighted dilation graphs, a uniform character-energy estimate, and a positive marked principal term.
Author comments: Manuscript accompanying Justin Sun Prize submission JSP-000277. Pinned manuscript, original Lean statement, development, and verification evidence: https://github.com/Ghuffy1/erdos-334/tree/0f66e1d94275854de3a6bee7d109c4d2b5cecb35
Primarily AI-generated textHuman understanding: no partsDirichlet L-functionsadditive representationsdilation graphssmooth numbers
Contributed by Will Howe, Grant Huffman, Jensen Kohlmeyer, Liam Kruer
We construct a fixed finite bipartite graph G such that every set of at least two vertices spans at most twice its cardinality minus three edges, while the ratio R(G,H)/e(H) is unbounded as H ranges over graphs without isolated vertices. For all sufficiently large m, R(G,K2570m)>21144m−1. This gives a negative answer to Erdős problem 566. The construction uses a layered two-parent graph and a random subset of a Hamming cube. A conditional-entropy inequality excludes the fixed graph. A bound on the negative Krawtchouk spectrum and an independent-set encoding exclude large independent sets.
Author comments: Manuscript accompanying Justin Sun Prize submission JSP-000456. Pinned source and Lean verification materials: https://github.com/Ghuffy1/erdos-566/tree/ac0b12b5cb264a092ef1cdf6304760dcdc4b6328
Primarily AI-generated textHuman understanding: no partsHamming cubeRamsey numbersentropy
We construct a permutation of the natural numbers containing no increasing or decreasing four-term arithmetic progression as a subsequence. This gives a negative answer to Erdős problem 196. The construction extends finite prefixes while preserving closure and consistency of binary preferences. To extend a prefix, we order a new finite block by an adapted binary order and impose saturation at the nodes where the old prefix branches. Finiteness follows from the weight w(n)=2ρ(n)n, where ρ is a finite binary residue rank. The extension can include any prescribed finite set. Iteration gives a bijection of order type ω on the natural numbers. The finite extension theorem, the resulting bijection, and the exact negative answer are formalized in Lean 4.
Author comments: TeX edition of the manuscript accompanying Justin Sun Prize submission JSP-000185. Pinned source and Lean verification materials: https://github.com/lkruer/conjectures/tree/bf058b903350eb8f873a22ab4a54b2e0faa9727d/problems/erdos-196
Primarily AI-generated textHuman understanding: no partsarithmetic progressionsbinary ordersinfinite permutations
Contributed by Will Howe, Grant Huffman, Jensen Kohlmeyer, Liam Kruer
Let μ be an uncountable cardinal of countable cofinality and suppose that 2μ=μ+. We construct a self-avoiding set mapping on μ+ whose values have cardinality at most μ, whose distinct values have intersection of cardinality less than μ, and which has no free set of cardinality μ+. The construction combines a countable neighborhood base with an enumeration of the μ-sized subsets and a schedule having fewer than μ requirements at each bounded depth. Under the generalized continuum hypothesis, this gives a negative answer to the free-set question of Erdős and Hajnal at ℵω+1.
Author comments: Manuscript accompanying Justin Sun Prize submission JSP-000978. Pinned source and Lean verification materials: https://github.com/Ghuffy1/erdos-1173/tree/7c7a362c0f426ddb3e55e9e43a5a51f859a4e0ca
Primarily AI-generated textHuman understanding: no partsfree setsset mappingssingular cardinals
Contributed by Will Howe, Grant Huffman, Jensen Kohlmeyer, Liam Kruer
We study the graph-target partition relation ω12→(ω1⋅ω,H)2 from Erdős problem 597. In ZFC we construct a connected bipartite graph of cardinality ℵ1 and diameter at most three, containing no Kω,ω, for which this relation fails. We also construct a fixed finite K4-free graph H0 for which the relation fails under the continuum hypothesis. Under Martin's axiom at ℵ1, it holds for every finite graph. Consequently the universal finite-target assertion, and the assertion for H0 itself, are independent of ZFC, relative to the consistency of ZFC.
Author comments: Manuscript accompanying Justin Sun Prize submission JSP-000484. Pinned source and Lean verification materials: https://github.com/Ghuffy1/erdos-597/tree/279825a39739918a2553880a32ec9ab1dee1f780
Primarily AI-generated textHuman understanding: no partsOrdinal partition relationsindependenceinfinite graphs
Contributed by Will Howe, Grant Huffman, Jensen Kohlmeyer, Liam Kruer
We give sufficient conditions for the signed differences occurring infinitely often in a set of natural numbers to have bounded gaps. The hypotheses use pairs of finite arithmetic progressions with coprime step quotients, positive normalized overlap, and bounded mutual distance. Bounded perturbations of the progressions are allowed. A finite circular-gap criterion gives the explicit bound 2gH+4C, including equality at the overlap threshold. Exact centered pairs yield a finite-intersection theorem. Separated constructions realize these conditions at zero upper Banach density and give examples with no recurrent three-point pattern, no infinite arithmetic progression of recurrent differences, and infinitely many pairwise almost disjoint translates. Further examples describe the rational limiting threshold and the order of the quantitative gap bound.
Author comments: Manuscript accompanying Justin Sun Prize submission JSP-000276. Pinned source and Lean verification materials: https://github.com/Ghuffy1/erdos-332/tree/cf1e6ab6401ce288dcb4194fdfe314c2edb2fc84
Primarily AI-generated textHuman understanding: no partsarithmetic progressionsrecurrent differencessyndetic sets
Contributed by Jensen Kohlmeyer, Liam Kruer, Will Howe, Grant Huffman
Let f(N) be the largest size of a subset of {1,…,N} admitting a signed zero sum of its reciprocals with no proper nonempty signed zero subsum. We prove f(N)=N−o(N) using the subset-sum theorem of Liu and Sawhney. The construction joins finitely many large-prime groups by one CRT-chosen unit fraction. The resulting positive sum can be arbitrarily large, although the density cost of excluding those primes is arbitrarily small. A dense coprime-denominator expansion of that sum provides the opposite side and preserves minimality. Earlier public work claims this asymptotic and the protected-prime construction. This version gives an alternative approach through Liu and Sawhney, with an accompanying Lean formalization.
Author comments: Manuscript accompanying Justin Sun Prize submission JSP-000265. Pinned source and Lean verification materials: https://github.com/H0W4RD-dev/erdos319-density-one/tree/ed555ed4120225800a6c642e6ba81704d5ed6b11
Primarily AI-generated textHuman understanding: no partsegyptian fractionsprimitive relationsreciprocal sums
The spectrum of a free-fermionic operator is built from a few single-particle quantities. For operators with rational entries this confines, as is known, the Galois group of the characteristic polynomial of each sector to a hyperoctahedral group, up to elementary abelian 2-groups. We turn this into a test for free-fermionic spectra that uses only integer arithmetic: Frobenius cycle types modulo primes can prove that a spectrum is not free-fermionic, and an exact recovery of the single-particle data, certified by power sums, confirms it when it is. As a worked example, the test recovers Onsager's dispersion relation from the characteristic polynomial of the Ising transfer matrix. As applications, we prove that the periodic Fendley chain with generic couplings admits no free-fermion solution of the Chapman--Elman--Mann form for a range of sizes, bearing on whether a simplicial clique is necessary for free-fermion solvability, and we detect hidden tensor-product and matchgate structure in any basis. The Galois group grades spectral structure: hyperoctahedral for free fermions, symmetric on subsets with particle-number conservation, apparently imprimitive for the partially decoupled Bethe equations of the exclusion process, and alternating within sectors for the interacting models tested.
Primarily AI-generated textHuman understanding: all partsfree fermionsintegrable models
Contributed by Will Howe, Grant Huffman, Jensen Kohlmeyer, Liam Kruer
We give a detailed exposition of the OpenAI–Galán argument that resolves Erdős problem 706, including character rigidity, its compact-tower and radial arguments, and the transfer from arbitrary colorings to weak measurable colorings for algebraic distances. Galán’s application of Bukh’s theorem gives finite planar graphs with r forbidden distances and chromatic number greater than 4.0485r. We also prove a uniform product estimate for normalized compact distance clusters and give quantitative fixed-ratio bounds from weighted unit-distance graphs. The linked Lean verification covers the original negative answer and the elementary bounds 2r≤L(r)≤7r.
Author comments: Revised manuscript accompanying Justin Sun Prize submission JSP-000576. Original source package, attribution and verification scope: https://github.com/Ghuffy1/erdos-706-lean/tree/2a22f736eea81c04ec696c1b1d7f770b4417f5f7
Primarily AI-generated textHuman understanding: no partscharacter rigiditydistance graphsgeometric graph coloringindependence density
Contributed by Grant Huffman, Jensen Kohlmeyer, Will Howe, Liam Kruer
Let ℓ(n) be the largest size of a subset guaranteed in every n-element set of integers in which no element is a sum of at least two distinct other elements. We give an upper bound of order n/(logn)2 by propagating mixed averages through a layered construction and lifting repeated-summand relations to distinct summands. Together with Samuel Korsky’s lower bound, this gives matching upper and lower bounds. A linked Lean development proves the upper bound and matching order, with an independent statement bridge.
Author comments: TeX edition of the public matching-order manuscript for JSP-000649. Pinned publication and contributor credits: https://github.com/Ghuffy1/erdos790-jsp/tree/7c3d195fa10d7b52f4fa4a2165b336396d4b21c0 . Pinned proof and statement bridge: https://github.com/Ghuffy1/erdos790-jsp/tree/0a69771b18df655fe47af2d5fb78a7d125fcd5f0
Primarily AI-generated textHuman understanding: no partsadditive combinatoricsarithmetic progressionsdistinct summandssum-free sets
We show that among all convex planar polygons of prescribed perimeter with at most four sides, the square uniquely maximizes the first positive Steklov eigenvalue. If k∈(0,π/2) denotes the unique root of tanktanhk=1, then Lσ1(Ω)≤8ktanhk, with equality if and only if Ω is a square. By the quadrilateral isoperimetric inequality, the square also uniquely maximizes the area-normalized eigenvalue ∣Ω∣σ1(Ω) in the same class.
Our proof splits the four-dimensional moduli space of convex quadrilaterals into a neighborhood of the square and its complement. Near the square, a weighted combination of the two first Steklov eigenfunctions of the square produces a negative definite Rayleigh defect; its sign on an explicit parameter box is established via an exact degree-22 Taylor expansion in ball arithmetic together with an analytic Cauchy remainder estimate. Far from the square, a boundary-variance inequality along a diagonal reduces the remaining shapes to a compact region, which we cover by 940 closed subboxes using affine and harmonic quadratic trial functions checked in outward-rounded integer interval arithmetic.
Author comments: The authors’ PDF and the source files and computational certificates for this version are available at https://doi.org/10.5281/zenodo.24409123.
A mix of human-written and AI-generated textHuman understanding: all parts
Let dt be the natural density of integers n for which t is a sum of distinct positive divisors of n. We prove liminft→∞(−logdt)/loglogt=δ and (logt)δdt→0, where δ=1−(1+loglog2)/log2. Together these exclude every pure logarithmic power asymptotic with a positive leading coefficient. This exposition gives the exact finite prime model and the density and change-of-measure reductions. It explains the divisor-interval saving and the arithmetic construction and links to the complete formal proof.
Author comments: TeX edition of the expanded Conjectures.io exposition linked from Justin Sun Prize submission JSP-000712. Original expanded PDF: https://conjectures.io/papers/erdos859.pdf?v=20260921-expanded . Accepted Lean source and verification: https://conjectures.io/results/e060355f-a728-4b06-af11-6302b5780d19
Primarily AI-generated textHuman understanding: no partsdivisor intervalsdivisor sumsexponential tiltingnatural density
Let f(n) be the largest size of a subset of {1,…,n} in which no element divides two distinct other elements. We prove the exact formula f(n)=∑m≤n,(m,6)=1(R(n/m)−1B(n/m)), where R is the two-prime grid rank and B is an explicit union of intervals. A matching construction follows from a cutoff recursion whose finite certificates are checked in Lean. The limit c=limn→∞f(n)/n=0.67296565119948934747… is irrational. The proof uses rotation returns, a two-cutoff 2-adic nonvanishing argument, and a three-place Subspace Theorem. The linked Lean development verifies the exact formula, convergence, and irrationality.
Author comments: TeX edition of the manuscript accompanying Justin Sun Prize submission JSP-000882. Pinned manuscript, Lean proof and exact full-statement audit: https://github.com/TheJustinSunPrize/awards/pull/4905
Primarily AI-generated textHuman understanding: no partsSubspace Theoremdivisibilityextremal densityfork-free sets
Contributed by Will Howe, Grant Huffman, Jensen Kohlmeyer, Liam Kruer
We characterize pairs of positive-density subsets of the positive integers for which the ordinary density of their sumset equals the sum of their densities. The modular boundary is controlled by literal residue supports. Above that boundary, the classification consists of parallel Bohr intervals with finite backgrounds, two periodic templates with a literal fringe test, or one sparse exceptional section with an exact terminal condition. This overview states the characterization and explains the proof structure. The linked development gives the detailed mathematical arguments and verifies the complete result in Lean using the original positive-integer counts.
Author comments: TeX edition of the four-page overview accompanying Justin Sun Prize submission JSP-000278. Pinned overview, detailed mathematical notes, and Lean verification: https://github.com/Ghuffy1/erdos-335/tree/29e64ce67b4999f76657bb4b51e306076f59e689
Primarily AI-generated textHuman understanding: no partsasymptotic densityinverse theoremsperiodicitysumsets
For every positive integer r, we construct a finite four-chromatic graph such that deleting any vertex lowers the chromatic number, and deleting any set of at most r edges preserves it. The proof combines punctured cyclic three-colorings, an integral height character, an Andrásfai winding bound, and a Chinese remainder construction with an exact classification of its small-deficit characters. Together with the higher-color result of Skottová and Steiner, this gives finite existence for every chromatic number at least four. The accompanying Lean formalization checks the finite-existence theorem.
Author comments: TeX edition of the manuscript accompanying Justin Sun Prize submission JSP-000785. Pinned source, proof audit and credit for the separate prescribed-order extension: https://github.com/TheJustinSunPrize/awards/pull/5110