Super-Brownian limits and the -point function for high-dimensional percolation
Arthur Blanc-Renaudie, Tom Hutchcroft
Resolves multiple 40-year-old central conjectures (Aizenman-Newman 1984, Hara-Slade 1998, van der Hofstad-Slade 2003) with a complete metric-measure scaling limit, a milestone for high-dimensional percolation.
Abstract
We prove that there exist positive constants and such that the high-dimensional critical percolation -point function is given by as , where is a set of isomorphism class representatives of trees with labelled leaves and unlabelled internal vertices all of which have degree and is the lattice Green's function. This verifies a conjecture of Aizenman and Newman (1984) subject to the usual perturbative conditions needed for convergence of the lace expansion. It follows from this theorem that the law of the cluster of the origin, considered as the counting measure on its range, converges under rescaling to the canonical measure of the integrated super-Brownian excursion. By computing the asymptotics of various more complicated variations on the -point function, we also prove the stronger result that the cluster converges as an embedded metric-measure space to the continuum random tree equipped with its Brownian embedding into . This convergence holds simultaneously with respect to the chemical distance, pivotal distance, and resistance distance on the cluster, which we prove are asymptotic to constant multiples of each other. This resolves conjectures of Hara and Slade (1998) and van der Hofstad and Slade (2003). As a corollary of our results we prove that there exists a positive constant such that , answering a question of Heydenreich and van der Hofstad (2017).
AI Impact Assessments
(1 models)This is a landmark contribution in probability theory / mathematical statistical mechanics, resolving multiple long-standing conjectures about the scaling limit of high-dimensional critical percolation.
1. Core Contribution. The paper proves the sharp first-order asymptotics of the critical percolation *k*-point function in high dimensions, verifying a conjecture of Aizenman and Newman dating to 1984. From this it deduces that the rescaled critical cluster converges to integrated super-Brownian excursion (as a measure), and—more strongly—that the cluster converges *as an embedded metric-measure space* to the continuum random tree with its Brownian embedding, simultaneously under the chemical, pivotal, and resistance metrics (which are shown to be asymptotically constant multiples of one another). This resolves conjectures of Hara–Slade (1998) and van der Hofstad–Slade (2003) and yields sharp one-arm asymptotics P(0↔∂[−r,r]^d) ∼ Cr^{−2}, answering a question of Heydenreich–van der Hofstad. These are among the most important open problems in the field; while critical exponents have been understood since Hara–Slade 1990, the full scaling limit for unoriented finite-range percolation had remained open for decades.
2. Methodological Rigor. The work is a rigorous mathematical proof built on the lace expansion, but its key methodological innovation is to isolate the lace expansion's role into two clean inputs: the two-point asymptotics (A) and a "two-blob" mixing estimate (B). Everything else—the scheme-function factorization theorem (Theorem 3.1), the overcount-factor localization, the metric comparisons, and the strengthening to the Gromov–Hausdorff–Prokhorov-function topology—follows non-perturbatively from (A) and (B). This yields a conditional theorem (Theorem 1.12) that decouples the analytic machinery from the geometric/combinatorial arguments, a genuinely valuable structural insight. The proofs are careful, self-contained modulo cited inputs, and include topological positivity arguments (Appendix A) that many authors would gloss over. The path-rerouting lemma (Menger-type) and the careful treatment of degenerate branching trees show attention to potential gaps.
3. Potential Impact. Within high-dimensional statistical mechanics this is a reference result that will be built upon extensively: the scheme-function formalism and the (A)+(B) criterion provide reusable tools that should transfer to lattice trees, the Ising model, oriented percolation, and related lace-expansion models. The framework also feeds directly into adjacent programs mentioned in the text (Carpenter–Werner on the Brownian loop soup; random walk on clusters converging to Brownian motion on the CRT via Croydon's theory). It is pure mathematics with no direct real-world/industrial application, so translational potential is essentially nil.
4. Timeliness & Relevance. The paper sits at the center of a currently very active area, as evidenced by the several independent/parallel works cited (Chatterjee–Chinmay–Hanson–Sosoe; Cabezas–Fribergh–Heydenreich–Járai; Blanc-Renaudie–Nachmias; the author's own long-range percolation series). The simultaneous emergence of multiple approaches signals both the ripeness of the problem and the importance of this particular, most complete, resolution.
5. Strengths & Limitations. Strengths: completeness (measure + metric + embedding, three metrics at once), the elegant reduction to two hypotheses, strong topology of convergence enabling corollaries like one-arm asymptotics and diameter distributions, and clear exposition including a genuinely useful "crash course" on the lace expansion. Limitations: the results hold only under the usual perturbative constraints (very large d, or d>6 with spread-out lattice), so nearest-neighbor Z^d in dimensions 7–10 remains out of reach; the critical dimension d=6 and low dimensions are untouched (as expected). The conclusions confirm long-standing conjectures rather than overturning beliefs, so there is little element of surprise. The overlap with parallel works slightly dilutes originality of certain sub-results, though this paper is the most comprehensive.
Other observations. The paper is a pure theory paper; there is no empirical component to reproduce, so empirical reproducibility does not apply. The barrier to entry is expertise (deep familiarity with lace expansion, superprocesses, CRT theory) rather than compute or data. Foundationally, the (A)+(B) criterion and scheme-function factorization are the kind of primitives that a subfield reuses.
Generated Jul 27, 2026
Comparison History (23)
Paper 1 delivers major, technically deep advances in high-dimensional percolation: it proves long-standing conjectures (Aizenman–Newman; Hara–Slade; van der Hofstad–Slade), establishes super-Brownian/CRT scaling limits in multiple intrinsic metrics, and derives sharp one-arm asymptotics. This is a substantial, broadly influential contribution to probability, statistical physics, and geometric scaling limits with high methodological rigor. Paper 2 resolves an important conjecture with a short proof and has relevance to probability/statistics, but its scope and downstream theoretical reach appear narrower than the comprehensive structural results in Paper 1.
Paper 2 resolves a widely known conjecture in probability and theoretical computer science using an exceptionally novel, AI-driven methodology. Its cross-disciplinary approach bridging statistics and convex geometry, combined with the historic milestone of an AI-generated proof, ensures unprecedented breadth of impact across AI, algorithms, and mathematics, surpassing the highly impressive but domain-specific theoretical impact of Paper 1.
Paper 2 resolves multiple long-standing conjectures (Aizenman-Newman 1984, Hara-Slade 1998, van der Hofstad-Slade 2003) in high-dimensional percolation, establishing convergence to super-Brownian motion and the continuum random tree with striking depth. It answers foundational questions in probability theory with rigorous, comprehensive results spanning chemical, pivotal, and resistance distances. Paper 1 offers solid contributions to sampling algorithms and Langevin diffusions, which are timely for ML applications, but relies on conjectures and perturbative conditions, and represents incremental advances relative to the definitive, landmark nature of Paper 2's resolutions of decades-old open problems.
Paper 2 likely has higher impact: it advances the central KLS conjecture in high-dimensional convex geometry by proving a strong variance inequality for all quadratic forms and improving the best-known bound on the KLS constant to O(log^{1/4} n). This is timely and broadly relevant, with downstream implications for sampling, mixing times, concentration, and algorithms in convex optimization and theoretical computer science, beyond pure mathematics. Paper 1 is very deep and resolves multiple long-standing conjectures in high-dimensional percolation, but its applications and cross-field reach are narrower and more specialized.
Paper 2 resolves multiple long-standing conjectures (Aizenman-Newman 1984, Hara-Slade 1998, van der Hofstad-Slade 2003) in high-dimensional percolation, establishing convergence to super-Brownian excursion and the continuum random tree. This represents a definitive, rigorous breakthrough answering foundational open problems in probability theory with lasting theoretical impact. Paper 1 offers a useful neural operator method with good approximation results, but is more incremental within the crowded neural SDE/operator space and lacks the field-defining significance of resolving decades-old conjectures. Paper 2's methodological depth and breadth of resolved questions give it higher enduring impact.
Paper 1 likely has higher impact: it resolves multiple long-standing conjectures in high-dimensional critical percolation (Aizenman–Newman; Hara–Slade; van der Hofstad–Slade), establishes scaling limits to super-Brownian excursion/CRT with several intrinsic metrics, and derives sharp one-arm asymptotics—results central to probability/statistical physics with broad downstream use. Methodologically, it leverages the lace expansion to obtain detailed multi-point function asymptotics and geometric convergence. Paper 2 is innovative (traffic/free-probability random matrix model for lamplighter groups) but appears narrower, with fewer definitive landmark resolutions and more limited immediate cross-field applications.
Paper 2 resolves multiple major, long-standing conjectures in statistical mechanics and probability dating back to 1984 (Aizenman-Newman) and 1998 (Hara-Slade). Proving the universality of high-dimensional critical percolation and its convergence to super-Brownian limits is a monumental mathematical breakthrough. While Paper 1 provides excellent theoretical advancements for interacting diffusions, Paper 2's resolution of multi-decade open problems guarantees broader historical significance and higher foundational impact across mathematical physics.
Paper 2 resolves multiple long-standing conjectures (Aizenman-Newman 1984, Hara-Slade 1998, van der Hofstad-Slade 2003) and answers open questions in high-dimensional percolation, a central topic in probability and statistical physics. It establishes deep connections to super-Brownian motion, ISE, and continuum random trees with technically formidable lace expansion methods. Paper 1, while rigorous, addresses a more incremental convergence result for stochastic heat equations with narrower scope and impact. Paper 2's breadth, resolution of decades-old problems, and cross-field relevance clearly indicate substantially higher scientific impact.
While Paper 1 resolves long-standing mathematical conjectures, Paper 2 demonstrates higher potential scientific impact due to its broader interdisciplinary reach. By establishing dimension-free propagation of chaos for mean-field stochastic PDEs, Paper 2 directly connects rigorous mathematical frameworks to highly active, real-world applications in machine learning and fluid dynamics. Its relevance to current AI theoretical foundations guarantees wider readership, greater citation potential across multiple fields, and higher practical applicability compared to the purely theoretical focus of Paper 1.
While both papers are exceptional in mathematical physics, Paper 2 has higher potential impact because it resolves multiple foundational, long-standing conjectures in percolation theory. By rigorously proving the 1984 Aizenman-Newman conjecture, along with major open problems from Hara and Slade (1998) and van der Hofstad and Slade (2003), it closes decades-old gaps regarding high-dimensional critical percolation and super-Brownian limits. Paper 1 offers highly innovative results in KPZ universality, but Paper 2's definitive settlement of 40-year-old questions gives it monumental historical and theoretical significance in statistical mechanics.
Resolves multiple 40-year-old central conjectures (Aizenman-Newman 1984, Hara-Slade 1998, van der Hofstad-Slade 2003) with a complete metric-measure scaling limit, a milestone for high-dimensional percolation.