For ferromagnetic Ising models on the square lattice with translation-invariant finite-range couplings, the interaction graph is non-planar and every exact solution method fails, yet universality predicts the same critical behavior as the solvable nearest-neighbor model. The paper that established the first rigorous universality property for this class — the asymptotic emergence of Pfaffian (fermionic) structure in boundary spin correlations at criticality — explicitly listed four remaining problems: (i) prove that the spontaneous magnetization vanishes at the critical point (continuity of the phase transition); (ii) prove polynomial mixing of the critical sourceless random current; (iii) prove Russo–Seymour–Welsh-type uniform crossing estimates for the percolation of positive-current edges; and (iv) strengthen the asymptotic Pfaffian relation for bulk order–disorder correlations to an error term relative to the order–disorder correlation itself. All four problems remain open at the source's generality; adjacent cases (planar biperiodic graphs, dimension three and higher, weak perturbations of the nearest-neighbor model, and supercritical regimes) are settled or partially settled and delineate the open core precisely.

Background

The Ising model on the vertex set of the square lattice Z2\mathbb{Z}^2 assigns spins σx∈{−1,+1}\sigma_x\in\{-1,+1\} to each site with Gibbs weight proportional to exp⁡(β∑x,yJx,yσxσy)\exp(\beta\sum_{x,y}J_{x,y}\sigma_x\sigma_y), where β≥0\beta\ge 0 is the inverse temperature and the couplings Jx,yJ_{x,y} are nonnegative (ferromagnetic). For nearest-neighbor couplings the interaction graph is the planar square lattice and the model is exactly solvable: Onsager computed the free energy and Yang the spontaneous magnetization. For finite-range couplings — Jx,y=J(x−y)≥0J_{x,y}=J(x-y)\ge 0 vanishing whenever ∥x−y∥>R\|x-y\|>R but not restricted to nearest neighbors — the interaction graph contains diagonals and longer edges and is therefore no longer planar, and all integrability methods break down. Universality, a central conjecture of statistical mechanics, predicts that the critical behavior is nevertheless identical across all ferromagnetic, translation-invariant, finite-range models in two dimensions.

The phase transition of such a model is described by the spontaneous magnetization m∗(β)m^*(\beta), the infinite-volume expectation of the spin at the origin in the state obtained with ++ boundary conditions. Sharpness results guarantee a well-defined critical inverse temperature βc\beta_c separating exponential decay of truncated correlations (for β<βc\beta<\beta_c) from magnetized states (for β>βc\beta>\beta_c). The transition is called continuous if m∗(β)→0m^*(\beta)\to 0 as β↓βc\beta\downarrow\beta_c, and discontinuous (first-order) if m∗(βc)>0m^*(\beta_c)>0. Continuity is known for the nearest-neighbor model on Z2\mathbb{Z}^2 (Yang), for Ising models on planar biperiodic graphs (Duminil-Copin–Lis, via a map between double random currents and dimers), and for nearest-neighbor models on Zd\mathbb{Z}^d in all dimensions d≥3d\ge 3 (Aizenman–Duminil-Copin–Sidoravicius, with d≥4d\ge 4 earlier by Aizenman–Fernández). The two-dimensional finite-range case is open; it is a genuinely delicate question, since first-order transitions do occur for one-dimensional Ising models with long-range 1/∣x−y∣21/|x-y|^2 couplings (Aizenman–Chayes–Chayes–Newman), so finite range is an essential hypothesis.

The random current representation expands the Boltzmann weight in powers of βJx,yσxσy\beta J_{x,y}\sigma_x\sigma_y: a current configuration is an integer-valued function n\mathbf{n} on edges, its sources ∂n\partial\mathbf{n} are the sites at which the sum of incident current values is odd, and the correlation ⟨σA⟩\langle\sigma_A\rangle equals the ratio of the total current weight with sources AA to that with no sources. The sourceless measure PZ2,β∅\mathbf{P}^{\emptyset}_{\mathbb{Z}^2,\beta} exists in infinite volume, and the combinatorial switching lemma for pairs of currents converts statements about correlations into statements about the stochastic geometry of the (duplicated) currents. The percolation formed by the edges carrying positive current — the current percolation — governs which sites are connected in a current configuration, and intersection properties of long current paths encode the structure of multi-point correlations. Two quantitative properties of this geometry at criticality are central here: Russo–Seymour–Welsh (RSW) crossing estimates, which in Bernoulli percolation state that the probability of crossing a long thin rectangle in the hard direction is bounded away from zero uniformly in the scale, and mixing, meaning decay of dependence between events depending on edge sets separated by large annuli. For the sourceless current both are open in the finite-range setting, the obstruction being that this measure has no known positive association (FKG inequality) and lives on a non-planar graph, so planar technology does not apply.

Finally, the Pfaffian of a triangular array (Ai,j)i<j(A_{i,j})_{i<j} is the signed sum over pairings of the products of paired entries, Pf(A)=∑πsgn(π)∏kAπ(2k−1),π(2k)\mathrm{Pf}(A)=\sum_{\pi}\mathrm{sgn}(\pi)\prod_k A_{\pi(2k-1),\pi(2k)}; it is the fermionic analogue of the Wick rule by which Gaussian (bosonic) correlations factorize into two-point functions. In any planar Ising model, the 2n2n-point correlation of spins at cyclically ordered boundary points equals exactly the Pfaffian of the two-point correlations (Groeneveld–Boel–Kasteleyn, restated as Theorem 1.1 of the source paper). Disorder operators τ\tau (often called Kadanoff–Ceva operators) are dual variables obtained by flipping couplings along a path from the dual site to a boundary; order–disorder correlations ⟨τ1⋯τ2n⟩\langle\tau_1\cdots\tau_{2n}\rangle pair spin (order) variables with such disorder variables and probe the fermionic structure in the bulk. The source paper proved that for finite-range models at βc\beta_c the Pfaffian structure of boundary spin correlations emerges asymptotically at large separations (Theorem 1.2, the emergent-planarity theorem) and derived the crossing input needed for it only along an infinite sequence of scales (Theorem 1.3); its Corollary 6.5 extends the Pfaffian relation to bulk order–disorder correlations, but with an error term relative to the spin correlation ⟨σx1⋯σx2n⟩\langle\sigma_{x_1}\cdots\sigma_{x_{2n}}\rangle rather than to the order–disorder correlation itself. The Discussion section of that paper lists the four problems considered here.

Problem Statement

Consider the ferromagnetic Ising model on Z2\mathbb{Z}^2 with couplings Jx,y=J(x−y)≥0J_{x,y}=J(x-y)\ge 0 that are translation-invariant, of finite range (vanishing beyond some fixed distance RR), and such that the induced interaction graph is connected; for statements on the upper half-plane, the reflection symmetry of the couplings assumed in Theorem 1.2 of the source paper is additionally part of the standing hypotheses. Let βc\beta_c be the critical inverse temperature, m∗(β)m^*(\beta) the spontaneous magnetization, and PZ2,βc∅\mathbf{P}^{\emptyset}_{\mathbb{Z}^2,\beta_c} the infinite-volume sourceless random current measure at criticality. The problem is to extend the planar critical structure to this non-planar class by proving the following four statements, listed as Questions 7.1–7.4 in the source paper.

(i) Continuity of the order parameter: m∗(βc)=0m^*(\beta_c)=0, i.e. the phase transition of every such two-dimensional finite-range ferromagnetic Ising model is continuous.

(ii) Polynomial mixing of random currents: there exist constants c,C>0c,C>0 such that for every pair of events AA (depending on edges inside a box of size nn) and BB (depending on edges outside a concentric box of size N≥nN\ge n),

∣PZ2,βc∅[A∩B]−PZ2,βc∅[A] PZ2,βc∅[B]∣ ≤ C (n/N)c PZ2,βc∅[A] PZ2,βc∅[B].\bigl|\mathbf{P}^{\emptyset}_{\mathbb{Z}^2,\beta_c}[A\cap B]-\mathbf{P}^{\emptyset}_{\mathbb{Z}^2,\beta_c}[A]\,\mathbf{P}^{\emptyset}_{\mathbb{Z}^2,\beta_c}[B]\bigr|\ \le\ C\,(n/N)^{c}\,\mathbf{P}^{\emptyset}_{\mathbb{Z}^2,\beta_c}[A]\,\mathbf{P}^{\emptyset}_{\mathbb{Z}^2,\beta_c}[B].

(iii) Russo–Seymour–Welsh crossing estimates for the current percolation: there exists c>0c>0 such that for every nn, the probability, under PZ2,βc∅\mathbf{P}^{\emptyset}_{\mathbb{Z}^2,\beta_c}, that an n×2nn\times 2n rectangle is crossed in the hard direction (the crossing spanning the side of length 2n2n) by a path of edges carrying positive current is at least cc (the source paper proves this only along an infinite sequence of scales).

(iv) Full asymptotic Pfaffian relation for order–disorder correlations: strengthen Corollary 6.5 of the source paper so that the error term in the asymptotic Pfaffian relation for bulk order–disorder correlations is o(1) ∣⟨τ1⋯τ2n⟩Z2,βc∣o(1)\,\bigl|\langle\tau_1\cdots\tau_{2n}\rangle_{\mathbb{Z}^2,\beta_c}\bigr|, in place of the currently available o(1) ⟨σx1⋯σx2n⟩Z2,βco(1)\,\langle\sigma_{x_1}\cdots\sigma_{x_{2n}}\rangle_{\mathbb{Z}^2,\beta_c}, with the o(1)o(1) asymptotics taken, as in the source, along sequences of configurations with min⁡i<j∣xi−xj∣→∞\min_{i<j}|x_i-x_j|\to\infty, thereby establishing the fermionic Wick-type structure for bulk order–disorder fields in finite-range models.

The four statements share one scientific objective — the universality of the planar critical Ising structure beyond the exactly solvable and planar cases — and are linked: (ii) and (iii) are the missing stochastic-geometric properties of the critical sourceless current from which the physical conclusions (i) and (iv) are expected to be derivable, with (iv) described by the source as the most natural question its analysis leaves open. An answer may resolve any subset of the four parts, provided each resolved part is established at the stated generality over the whole class of finite-range couplings.

References

  1. M. Aizenman, H. Duminil-Copin, V. Tassion, S. Warzel, Emergent planarity in two-dimensional Ising models with finite-range interactions, Inventiones mathematicae 216, 661–743 (2019). DOI: 10.1007/s00222-018-00851-4, arXiv:1801.04960, https://arxiv.org/abs/1801.04960
  2. C. N. Yang, The Spontaneous Magnetization of a Two-Dimensional Ising Model, Physical Review 85, 808–816 (1952). DOI: 10.1103/PhysRev.85.808, https://doi.org/10.1103/PhysRev.85.808
  3. M. Aizenman, H. Duminil-Copin, V. Sidoravicius, Random Currents and Continuity of Ising Model's Spontaneous Magnetization, Communications in Mathematical Physics 334, 719–742 (2015). DOI: 10.1007/s00220-014-2093-y, https://doi.org/10.1007/s00220-014-2093-y
  4. H. Duminil-Copin, M. Lis, On the double random current nesting field, Probability Theory and Related Fields 175, 937–955 (2019). DOI: 10.1007/s00440-019-00899-0, arXiv:1712.02305, https://arxiv.org/abs/1712.02305
  5. M. Aizenman, R. Fernández, On the critical behavior of the magnetization in high-dimensional Ising models, Journal of Statistical Physics 44, 393–454 (1986). DOI: 10.1007/BF01011304, https://doi.org/10.1007/BF01011304
  6. M. Aizenman, J. T. Chayes, L. Chayes, C. M. Newman, Discontinuity of the magnetization in one-dimensional 1/|x−y|^2 Ising and Potts models, Journal of Statistical Physics 50, 1–40 (1988). DOI: 10.1007/BF01022985, https://doi.org/10.1007/BF01022985
  7. H. Duminil-Copin, M. Lis, W. Qian, Conformal invariance of double random currents II: tightness and properties in the discrete, arXiv:2107.12880 (2021), https://arxiv.org/abs/2107.12880
  8. H. Duminil-Copin, M. Lis, W. Qian, Conformal invariance of double random currents I: identification of the limit, Proceedings of the London Mathematical Society 130 (2025). DOI: 10.1112/plms.70022, https://doi.org/10.1112/plms.70022
  9. U. T. Hansen, F. Ravn Klausen, The Supercritical Loop O(1) and Random Current models: Uniqueness and Mixing, arXiv:2603.28520 (2026), https://arxiv.org/abs/2603.28520
  10. U. T. Hansen, F. Ravn Klausen, The End of the Double Random Current, arXiv:2607.07616 (2026), https://arxiv.org/abs/2607.07616
  11. H.-B. Chen, J. Xia, Conformal invariance of random currents: a stability result, arXiv:2306.10625 (2023), https://arxiv.org/abs/2306.10625
  12. G. Cava, A. Giuliani, R. L. Greenblatt, The scaling limit of boundary spin correlations in non-integrable Ising models, Journal of Mathematical Physics 66, 023301 (2025). DOI: 10.1063/5.0235381, arXiv:2408.12923, https://arxiv.org/abs/2408.12923
  13. H. Duminil-Copin, 100 Years of the (Critical) Ising Model on the Hypercubic Lattice, arXiv:2208.00864 (2022), https://arxiv.org/abs/2208.00864
  14. A. Giuliani, Scaling limits and universality of Ising and dimer models, arXiv:2110.09372 (2021), https://arxiv.org/abs/2110.09372