Compilation··3:30·Probability and statistical mechanics
Benjamini–Schramm Conjecture, Mézard–Parisi Formula, Voronoi Percolation and 7 More
Watch on YouTubeAbout the problem →
About this video
10 results from OpenAI's public mathematics catalogue, one per chapter. Each chapter names the problem, shows the real objects, and states what OpenAI claims; none of it is peer reviewed yet.
Chapters:
0:00 No Bigeodesics in First-Passage Percolation
0:21 Benjamini–Schramm Conjecture
0:41 XY Model at the BKT Point
1:03 Sherrington–Kirkpatrick Spin Glass
1:24 Ising Model Universality
1:45 Random Walk in Random Environment
2:07 Mézard–Parisi Formula
2:28 Random-Cluster Interfaces
2:49 Voronoi Percolation
3:10 Tree Reconstruction Threshold
No Bigeodesics in First-Passage Percolation: Give every edge of the square grid a random travel time and let a fire spread from the centre along the fastest routes: this is first-passage percolation.
Explained: gradientdamage.com/p/212
Paper: github.com/openai/math/blob/main/preprints/No-bigeodesics-in-…
Benjamini–Schramm Conjecture: Open each link of an infinite graph independently with probability p.
Explained: gradientdamage.com/p/214
Paper: github.com/openai/math/blob/main/preprints/Nonuniqueness-of-p…
XY Model at the BKT Point: The planar XY model puts a spinning arrow on every lattice site.
Explained: gradientdamage.com/p/216
Paper: github.com/openai/math/blob/main/preprints/The-critical-logar…
Sherrington–Kirkpatrick Spin Glass: How much does the free energy of a spin glass wobble from one random sample to the next? In the Sherrington–Kirkpatrick model every pair of n spins is joined by a random link, and at low temperature physicists predicted…
Explained: gradientdamage.com/p/217
Paper: github.com/openai/math/blob/main/preprints/The-low-temperatur…
Ising Model Universality: Every site of a square grid points up or down.
Explained: gradientdamage.com/p/218
Paper: github.com/openai/math/blob/main/preprints/Quenched-SLE3-limi…
Random Walk in Random Environment: Drop a walker into a random world where every site has its own random odds of stepping in each direction.
Explained: gradientdamage.com/p/220
Paper: github.com/openai/math/blob/main/preprints/A-directional-zero…
Mézard–Parisi Formula: A diluted spin glass is a sparse random network of tiny magnets, each link asking its pair to be aligned or opposed.
Explained: gradientdamage.com/p/221
Paper: github.com/openai/math/blob/main/preprints/The-Mezard-Parisi-…
Random-Cluster Interfaces: Colour each edge of a square grid open or closed, weighting every connected cluster by q, at the critical point p = √q / (1 + √q).
Explained: gradientdamage.com/p/223
Paper: github.com/openai/math/blob/main/preprints/Square-lattice-FK-…
Voronoi Percolation: Scatter random dots, split the square into Voronoi cells, and colour each cell gold with probability p.
Explained: gradientdamage.com/p/224
Paper: github.com/openai/math/blob/main/preprints/From-critical-cros…
Tree Reconstruction Threshold: Colour the root of a tree with one of q colours, then let every child copy its parent's colour with strength λ or otherwise pick a colour at random.
Explained: gradientdamage.com/p/229
Paper: github.com/openai/math/blob/main/preprints/The-exact-reconstr…
The animations illustrate the results; they are not proofs.
Gradient Damage: mathematics you can see, equations you can hear.
youtube.com/@gradientdamage
Chapters
One problem per chapter. Each name links to the result's own page, each time to that moment on YouTube.
- No Bigeodesics in First-Passage Percolation Probability and statistical mechanics
- Benjamini–Schramm Conjecture Probability and statistical mechanics
- XY Model at the BKT Point Probability and statistical mechanics
- Sherrington–Kirkpatrick Spin Glass Probability and statistical mechanics
- Ising Model Universality Probability and statistical mechanics
- Random Walk in Random Environment Probability and statistical mechanics
- Mézard–Parisi Formula Probability and statistical mechanics
- Random-Cluster Interfaces Probability and statistical mechanics
- Voronoi Percolation Probability and statistical mechanics
- Tree Reconstruction Threshold Probability and statistical mechanics
Chapter 1: about the problem
Probability and statistical mechanics·OpenAI Math·Result 212
Planar first-passage geometry and the absence of bigeodesics
Proves that planar first-passage percolation has no doubly infinite geodesic for iid nonnegative nonatomic edge weights when the minimum of four weights has finite second moment. For exponential weights, the limit shape is strictly convex with C1 boundary. Differentiability also holds for every Gamma law with positive shape and rate. (Lean)
Chapter 2: about the problem
Probability and statistical mechanics·OpenAI Math·Result 214·from 0:21
The Benjamini–Schramm nonuniqueness conjecture
Proves pc < pu for Bernoulli bond percolation on every infinite connected locally finite nonamenable quasi-transitive graph, resolving the Benjamini–Schramm nonuniqueness conjecture. Thus there is a nonempty range of probabilities with infinitely many infinite clusters. A stronger operator bound also establishes the critical triangle condition. (Lean)
Chapter 3: about the problem
Probability and statistical mechanics·OpenAI Math·Result 216·from 0:41
Critical and near-critical XY scaling and BKT universality
For the square-lattice nearest-neighbor cosine XY model, proves critical axis correlations Cβc(r) ∼ Ar-1/4(log r)1/8 and the Berezinskii–Kosterlitz–Thouless essential singularity √(βc-β) log ξ(β) → B, with A,B > 0 after the free-box thermodynamic limit. For finite square-symmetric interactions containing nearest neighbors, discrete Gaussian heights converge to Gaussian fields throughout the rough phase, including its threshold, along geometric torus sizes. Critical center-magnetization and spin-field conclusions retain their stated height, renormalization, and field-input assumptions.
Chapter 4: about the problem
Probability and statistical mechanics·OpenAI Math·Result 217·from 1:03
The low-temperature Sherrington–Kirkpatrick fluctuation law
For every fixed inverse temperature β > 1, determines the fluctuation scale and limiting law of the zero-field Gaussian Sherrington–Kirkpatrick log partition function. Its variance is asymptotic to cβ n1/3, with cβ > 0, confirming the predicted n1/6 standard-deviation scale. Exact centering and standardization give full-sequence convergence to a uniquely characterized nondegenerate law.
Chapter 5: about the problem
Probability and statistical mechanics·OpenAI Math·Result 218·from 1:24
Conformal universality for weakly interacting and random-bond Ising models
Weak finite-range square-symmetric even multispin perturbations of the square-lattice Ising model preserve critical bulk spin and energy limits; weak square-symmetric contour interactions also yield chordal SLE3 interface limits. With sufficiently weak iid bond disorder of any fixed bounded nondegenerate mean-zero law, critical spin interfaces converge to the same law in probability over environments. (Lean)
Chapter 6: about the problem
Probability and statistical mechanics·OpenAI Math·Result 220·from 1:45
Directional zero–one laws beyond iid environments and iid ballisticity
On ℤd, d ≥ 3, directional escape has probability zero or one for iid strictly elliptic nearest-neighbor environments, and for stationary ergodic finite-range-dependent environments under uniform ellipticity. In iid uniformly elliptic environments with d ≥ 2, almost-sure directional transience implies a deterministic limiting velocity with positive projection in that direction, resolving the ballisticity conjecture. (Lean)
Chapter 7: about the problem
Probability and statistical mechanics·OpenAI Math·Result 221·from 2:07
The Mézard–Parisi formula for diluted spin glasses
Proves the Mézard–Parisi hierarchical cavity formula for Poisson-diluted even-arity Ising models satisfying the Panchenko–Talagrand factorization and positivity assumptions, with only first-moment integrability. The limiting free energy equals the infimum over finite-depth hierarchical trial laws. This includes the Viana–Bray model, symmetric diluted even-spin models, and weighted soft even-K satisfiability. (Lean)
Chapter 8: about the problem
Probability and statistical mechanics·OpenAI Math·Result 223·from 2:28
Random-cluster interfaces: critical, disordered, thermal, and natural-time scaling
Proves chordal SLEκ limits for critical square-lattice random-cluster interfaces for 0 < q ≤ 4, with κ=4π/arccos(-√q/2): bounded Jordan domains are allowed for q ≥ 1, and smooth Jordan domains for q < 1, under the stated marked-boundary approximations. For 1 ≤ q ≤ 4, complete nested plane loops converge to CLEκ.
Chapter 9: about the problem
Probability and statistical mechanics·OpenAI Math·Result 224·from 2:49
Critical and quenched near-critical universality for Poisson–Voronoi percolation
Proves Cardy's formula for annealed critical Poisson–Voronoi crossing probabilities in every bounded Jordan quadrilateral. With each model normalized by its own expected unit-square pivotal count, the conditional joint near-critical crossing-threshold laws for rational polygonal quads converge in environment probability to the triangular-lattice reference law. This establishes quenched near-critical universality for crossing thresholds.
Chapter 10: about the problem
Probability and statistical mechanics·OpenAI Math·Result 229·from 3:10
Exact three- and four-state reconstruction thresholds and four-state tree capacity
Proves the exact reconstruction threshold dλ2 > 1, with nonreconstruction at equality, for three-state symmetric and four-state ferromagnetic broadcasting on regular trees (d ≥ 2) and observed Poisson trees (mean d > 1 and d > 0, respectively), with Poisson advantage averaged without conditioning on survival. The three-state theorem allows both signs of λ and gives the exact weak-recovery threshold for the symmetric three-community stochastic block model. (Lean)
Topics
first-passage percolationBenjamini-Schramm conjectureXY modelSherrington-Kirkpatrick modelIsing modelrandom walk in random environmentMézard–Parisi formularandom-cluster modelVoronoi percolationreconstruction problemProbability and statistical mechanicsOpenAI mathAI proofAI mathematicsmathsGradient Damage