Hierarchical $O(N)$ Heisenberg-to-spherical convergence in the local potential approximation and the $1/N$ viscosity expansion of the renormalization-group PDE

The hierarchical $O(N)$ Heisenberg model with long-range hierarchical-Laplacian interaction is known to converge, as $N\to\infty$, to the hierarchical spherical model for integer block size $L^d\ge 2$: free energy for $\beta\ne\beta_c$, and moment generating function with Gaussian block-spin fluctuations for $\beta<\beta_c$ (Conti–Marchetti, arXiv:0804.2704). Both proof ingredients — reflection positivity and the Lee–Yang property — are tied to the integer block structure and fail when $L^d$ is not an integer, i.e. exactly in the local potential approximation $L\downarrow 1$, where the block-spin renormalization group becomes a nonlinear evolution PDE in which the coefficient $2/N$ of the second-derivative term $(2/N)\,x\,u_{xx}$ acts as a vanishing viscosity. The open problem, announced by the source, has two coupled halves: (A) establish weak convergence of the hierarchical $O(N)$ Heisenberg equilibrium measures to the spherical measures in the $L\downarrow 1$ regime — including the critical temperature and the low-temperature regime $\beta>\beta_c$, where zero-mode condensation makes the fluctuation laws non-Gaussian — by a method replacing reflection positivity and Lee–Yang; and (B) complement the explicitly solved $N=\infty$ PDE theory with a rigorous finite-N theory: well-posedness of the renormalization-group PDE for large finite $N$ and control of $u^{(N)}$ as a singular $1/N$ perturbation of $u^{(\infty)}$. As of 2026 the problem is open: the same group solved the $N=\infty$ half for $\beta\le\beta_c$ (J. Stat. Phys. 132, 811, 2008) and obtained Gevrey ($s>2$) and initial-condition summability estimates (PhD thesis 2011; Discrete Dyn. Nat. Soc. 2018).

Background

An $O(N)$ Heisenberg ferromagnet places at each lattice site a unit spin $\sigma_i\in\mathbb{R}^N$ with $|\sigma_i|=1$ (for $N=1,2,3$ this is the Ising, XY, and Heisenberg case), with Gibbs equilibrium favoring alignment. The spherical model of Berlin and Kac replaces the many fixed-length constraints by one global constraint $\sum_i s_i^2=n$ on real-valued spins, which makes the model exactly solvable. Stanley (1968) observed that the spherical model is the $N\to\infty$ limit of the $O(N)$ Heisenberg model, and Kunz and Zumbach (1992) gave a rigorous convergence proof of the free energy and moment generating function for nearest-neighbor interactions, repairing a gap in earlier arguments of Kac and Thompson.

Dyson-type hierarchical models replace the lattice interaction by a hierarchical Laplacian: sites are grouped into nested blocks of $L^d$ sites ($L$ and $d$ such that $L^d$ is an integer $\ge 2$, e.g. the dyadic case $L^d=2$), and the interaction couples block means across scales with geometrically decaying weights. This construction makes the block-spin renormalization-group (RG) transformation — integrating out the finest-scale spins and rescaling — an explicit finite-dimensional recursion. Felder's local potential approximation (LPA) takes $L\downarrow 1$ with the scale index $k\to\infty$ and $t=k\ln L$ fixed, turning the discrete RG recursion into a genuine nonlinear evolution PDE in continuous RG time $t$. Note the built-in tension: $L\downarrow 1$ forces $L^d$ to be noninteger, so in that regime the hierarchical spin system is defined only through such limits of its RG evolution.

Two properties drive all existing large-$N$ analyses of hierarchical $O(N)$ models with integer blocks. Reflection positivity is a correlation inequality stating that the interaction quadratic form stays positive under reflection of the lattice through block boundaries; it yields uniform convergence of the $O(N)$ RG trajectories to the exactly solvable $O(\infty)$ (spherical) trajectory. The Lee–Yang property — the single-site characteristic function belongs to the Laguerre class of entire functions, with zeros on a fixed line — controls the approach to equilibrium; Kozitsky used it for hierarchical vector models at the critical inverse temperature and below it, that is, at temperatures at and above $T_c$ ($d>4$, integer blocks), and Watanabe (building on Hara–Hattori–Watanabe) used both ingredients to control the critical trajectory of the $d=4$ hierarchical $O(N)$ model with large $N$ and block $L^d=2$.

Conti and Marchetti (arXiv:0804.2704) extended this classical theory to the hierarchical Laplacian, whose interaction is not translation invariant: they generalized the spherical-model solution and the Kunz–Zumbach convergence proof to this setting — free-energy convergence for $\beta\ne\beta_c$, moment-generating-function convergence with Gaussian fluctuations for $\beta<\beta_c$ (inverse temperature $\beta$, critical value $\beta_c$) — and examined whether the results survive $L\downarrow 1$. In the companion work, the same authors with Guidi solved the $N=\infty$ LPA evolution equation explicitly at $d=4$ and established two Gaussian block-spin limit laws at the critical temperature and above it ($\beta\le\beta_c$): the abnormal scaling $\gamma=d+2$ at $\beta=\beta_c(4)=4$, with limit $\tfrac12|z|^2$, and the normal scaling $\gamma=d$ for $\beta<\beta_c$, with limit $-|z|^2/2\mu(\beta)$. Zero-mode condensation — macroscopic occupation of the lowest eigenmode of the interaction, which makes fluctuations non-Gaussian — sets in strictly below the critical temperature ($\beta>\beta_c$), a regime those limit theorems do not cover. The source states that for noninteger $L^d$ 'an entirely new method of analysis has to be developed from scratch' to prove the $O(N)\to$ spherical weak convergence, and announces a 'third investigation' of the finite-$N$ RG PDE, in which the coefficient $2/N$ multiplies the second-derivative term $x\,u_{xx}$ and 'plays the role of viscosity' as in the hydrodynamic equations of incompressible fluids. The key analytic objects of that half are the formal $1/N$ series of the trajectory $u^{(N)}(t,x)$ and its summability: a formal power series is $s$-Gevrey if its coefficients grow at most like $(k!)^s$, and Borel $s$-summability in a direction $\theta$ means that its Borel transform of order $s$ converges and analytically continues with exponential bounds in a sector of opening exceeding $s\pi$ around $\theta$, so that the series is the asymptotic expansion of a unique actual function — its sum.

Problem Statement

Complete the rigorous large-$N$ theory of the hierarchical $O(N)$ Heisenberg model in the local potential approximation, in the two coupled directions announced but left open by Conti–Marchetti (arXiv:0804.2704):

(A) Convergence without reflection positivity or Lee–Yang. In the local-potential-approximation regime — block size $L\downarrow 1$, so that $L^d$ is not an integer and neither reflection positivity nor the Lee–Yang property is available — prove that the hierarchical $O(N)$ Heisenberg equilibrium measures converge weakly, as $N\to\infty$, to the corresponding hierarchical spherical equilibrium measures, together with convergence of the free energy and of the block-spin moment generating function at its natural scaling. The high-temperature regime $\beta<\beta_c$ should be recovered without the integer-block hypotheses, and the analysis should extend to the critical point $\beta=\beta_c$ and to the low-temperature regime $\beta>\beta_c$ (temperatures below $T_c$), including the fluctuation limit laws: at $\beta=\beta_c$ these are to be matched to the explicitly known Gaussian $N=\infty$ law at the abnormal scaling $\gamma=d+2$, whereas at $\beta>\beta_c$ zero-mode condensation makes Gaussian fluctuations fail and no explicit $N=\infty$ limit law is currently established in the local potential approximation, so the condensed-phase limit law must itself be derived as part of the convergence proof.

(B) Finite-$N$ control of the renormalization-group PDE. For the nonlinear evolution equation satisfied by the block-spin RG trajectory $u^{(N)}(t,x)$ in the local potential approximation — a nonlinear equation, first order in the RG time $t$ but containing the second-derivative term $(2/N)\,x\,u_{xx}$, whose coefficient $2/N$ 'plays the role of viscosity' and vanishes as $N\to\infty$, and whose nonlocal final term $u_x(t,0)$ enforces the normalization condition $u^{(N)}(t,0)=0$ for all $t\ge 0$ — establish existence, uniqueness, and stability of solutions for finite large $N$, and prove that $u^{(N)}$ is a rigorous singular $1/N$ perturbation of the explicitly known $N=\infty$ solution $u^{(\infty)}$ about the critical trajectory; in particular, determine the Gevrey order of the formal $1/N$ series (the known critical estimates are formulated for the inverse trajectory function $v^{(N)}(t,p)=(u^{(N)}_x)^{-1}(t,p)$ at $d=4$) and prove its Borel summability in an explicit direction (improving the known $s>2$ Gevrey estimates, which are too weak to meet the sectorial definition of summability), identifying its sum with the actual solution.

Parts (A) and (B) are the two halves of one objective: a rigorous renormalization-group theory of the hierarchical $O(N)$ Heisenberg model in the local potential approximation. (A) supplies the probabilistic convergence of the $O(N)$ equilibria to the spherical ($N=\infty$) limit that reflection positivity provided in the integer-block case; (B) supplies the finite-$N$ dynamical control of which the already-solved $N=\infty$ PDE is the viscosity limit, and it draws on the convergence furnished by (A). An answer resolving only one part while leaving the other untouched constitutes substantial partial progress but does not complete the announced program; a narrowing that silently restricts to $\beta<\beta_c$, to integer $L^d$, or to $N=\infty$ does not answer the stated problem.

References

1. William R. P. Conti and Domingos H. U. Marchetti, 'Hierarchical Spherical Model as a Viscosity Limit of Corresponding O(N)Heisenberg Model', arXiv:0804.2704 (2008), doi:10.48550/arXiv.0804.2704, https://arxiv.org/abs/0804.2704
2. Domingos H. U. Marchetti, William R. P. Conti and Leonardo F. Guidi, 'Hierarchical Spherical Model from a Geometric Point of View', Journal of Statistical Physics 132, 811-838 (2008), doi:10.1007/s10955-008-9568-1, https://doi.org/10.1007/s10955-008-9568-1 (preprint arXiv:0802.3317)
3. William Remo Pedroso Conti, PhD thesis 'Grupo de renormalização na aproximação de potencial local para o modelo O(N) de Heisenberg hierárquico: trajetória crítica e somabilidade da expansão 1/N', Universidade de São Paulo (2011), doi:10.11606/T.43.2011.tde-25022012-164421, https://doi.org/10.11606/T.43.2011.tde-25022012-164421
4. William Remo Pedroso Conti, MSc thesis 'Teorema Central do Limite para o modelo O(N) de Heisenberg hierárquico na criticalidade e o papel do limite N -> infinito na dinâmica dos zeros de Lee-Yang', Universidade de São Paulo (2008), doi:10.11606/T.43.2008.tde-26082008-093457, https://doi.org/10.11606/T.43.2008.tde-26082008-093457
5. Domingos H. U. Marchetti and William R. P. Conti, 'Singular Perturbation of Nonlinear Systems with Regular Singularity', Discrete Dynamics in Nature and Society 2018, 5163492 (2018), doi:10.1155/2018/5163492, https://doi.org/10.1155/2018/5163492 (preprint arXiv:1207.4212, 2012, with Conti as first author)
6. Gastão A. Braga and William R. P. Conti, 'A multiscale asymptotic analysis of time evolution equations on the complex plane', Journal of Mathematical Physics 57 (2016), doi:10.1063/1.4959233, https://doi.org/10.1063/1.4959233
7. Yuri V. Kozitsky, 'Hierarchical ferromagnetic vector spin model possessing the Lee-Yang property. Thermodynamic limit at the critical point and above', Journal of Statistical Physics 87, 799-820 (1997), doi:10.1007/BF02181245, https://doi.org/10.1007/BF02181245
8. Hiroshi Watanabe, 'Triviality of Hierarchical O(N) Spin Model in Four Dimensions with Large N', Journal of Statistical Physics 115, 1669-1713 (2004), doi:10.1023/B:JOSS.0000028073.49354.9d, https://doi.org/10.1023/B:JOSS.0000028073.49354.9d
9. Takashi Hara, Tetsuya Hattori and Hiroshi Watanabe, 'Triviality of Hierarchical Ising Model in Four Dimensions', Communications in Mathematical Physics 220, 13-40 (2001), doi:10.1007/s002200100440, https://doi.org/10.1007/s002200100440
10. H. Kunz and G. Zumbach, 'Phase transition in a nematic n-vector model: the large-n limit', Journal of Physics A: Mathematical and General 25, 6155-6162 (1992), doi:10.1088/0305-4470/25/22/036, https://doi.org/10.1088/0305-4470/25/22/036
11. Giovanni Felder, 'Renormalization group in the local potential approximation', Communications in Mathematical Physics 111, 101-121 (1987), doi:10.1007/BF01239018, https://doi.org/10.1007/BF01239018
12. T. H. Berlin and M. Kac, 'The Spherical Model of a Ferromagnet', Physical Review 86, 821-835 (1952), doi:10.1103/PhysRev.86.821, https://doi.org/10.1103/PhysRev.86.821
13. H. E. Stanley, 'Spherical Model as the Limit of Infinite Spin Dimensionality', Physical Review 176, 718-722 (1968), doi:10.1103/PhysRev.176.718, https://doi.org/10.1103/PhysRev.176.718
14. Evgenij Kritchevski, 'Spectral localization in the hierarchical Anderson model', Proceedings of the American Mathematical Society 135, 1431-1440 (2007), doi:10.1090/S0002-9939-06-08614-x, https://doi.org/10.1090/S0002-9939-06-08614-x
15. Fernando L. Metz, Jacopo Rocchi and Pierfrancesco Urbani, 'Statistical mechanics of the spherical hierarchical model with random fields', Journal of Statistical Mechanics: Theory and Experiment (2014) P09018, doi:10.1088/1742-5468/2014/09/p09018, https://doi.org/10.1088/1742-5468/2014/09/p09018
16. Domingos H. U. Marchetti, Manfred Requardt and Walter F. Wreszinski, 'A mathematical theory of the critical point of ferromagnetic Ising systems', Physics Reports 1079, 1-32 (2024), doi:10.1016/j.physrep.2024.05.006, https://doi.org/10.1016/j.physrep.2024.05.006
17. D. Marchetti and W. R. P. Conti, 'Global Formal Series Solution of a Singularly Perturbed First Order Partial Differential Equation: A Problem in Statistical Mechanics (Version 8 - Preliminary)' (2018), unpublished manuscript, https://www.semanticscholar.org/paper/ce241b1777a4d6870757acc25613977159ac560c