Constructive realization of Wilson's 1971 differential renormalization group for the Ising model below four dimensions: analytic flow functions, fixed-point convergence, and the critical scaling limit

In his 1971 formulation of the renormalization group (RG) for critical phenomena, Kenneth G. Wilson wrote differential flow equations for the Kadanoff block-spin couplings and showed that the Widom-Kadanoff scaling laws follow if the flow functions $u$ and $v$ are analytic at the critical point, while stating that there was "at present no hope" of proving that analyticity - without it, the equations "become a tautology". Wilson's 1971 formulation bundles five unresolved issues: (1) finiteness and analyticity of the flow functions at criticality; (2) an explicit microscopic derivation of the equations; (3) the asymptotic behavior of many-coupling RG flows, including possible limit cycles; (4) the marginal (zero-eigenvalue) case with logarithmic scaling violations; (5) the below-$T_c$, ordered-phase computation. Later literature resolved issue (2) at the level of physics - Wilson's paper II, the $\varepsilon$ expansion, the functional RG, and the conformal bootstrap now determine three-dimensional (3D) Ising critical exponents with high precision - and turned issue (4) into the standard theory of marginal operators, while limit cycles are now known to occur in specific renormalization-group systems (three-body physics, DNA, toy Hamiltonians). What survives as an open problem is the constructive, rigorous realization of the differential RG for the actual (non-hierarchical) lattice Ising model in $2<d<4$, with $d=3$ the case of the source paper: prove that the flow functions are finite and analytic at a non-Gaussian fixed point, that critical flows converge to that fixed point (excluding limit cycles, runaways, and irregular behavior), and that the construction extends to the ordered phase. Rigorous renormalization-group theory currently covers only hierarchical models and $d\geq 4$; the 3D construction is a recognized central open problem of mathematical physics.

Background

Near a continuous phase transition, equilibrium systems develop nonanalytic thermodynamic behavior. For the ferromagnetic nearest-neighbor Ising model - spins $s_n=\pm1$ on a $d$-dimensional cubic lattice with dimensionless coupling $K=J/(k_B T)$ and field variable $h$ - the infinite-volume free-energy density $F(K,h)$ and the correlation length $\xi(K,h)$ (the distance over which spin fluctuations remain correlated) are singular at the critical point $(K_c,0)$, although every term of the partition function is analytic in $K$ and $h$. Kadanoff's block-spin phenomenology (reviewed in Kadanoff et al. 1967) proposed that on intermediate length scales $1\ll L\ll\xi$ the coarse-grained system again looks like an Ising model with $L$-dependent effective couplings $K_L$ and $h_L$; this self-similarity underlies the Widom-Kadanoff scaling laws, in which the singular parts of $F$ and $\xi$ are powers of the reduced temperature multiplied by functions of a single scaling combination of field and temperature.

Wilson's 1971 paper I (the source of this problem) turned the block picture into differential equations, treating the coarse-graining scale $L$ as a continuous variable:

$$\frac{dK_L}{dL}=\frac{1}{L}\,u(K_L,h_L^2),\qquad \frac{dh_L}{dL}=\frac{1}{L}\,h_L\,v(K_L,h_L^2),$$

whose right-hand sides define the RG flow functions $u$ and $v$. If $u$ and $v$ are differentiable at the critical fixed point $(K_c,0)$ - a fixed point being a stationary point of the flow, i.e. $u(K_c,0)=0$ - linearization gives $K_L-K_c\sim\epsilon L^y$ and $h_L\sim h L^x$, and integrating the flow up to $L\sim\xi$ reproduces the scaling laws, with the critical exponents (power-law indices such as $\gamma$ for the susceptibility and $\nu$ for the correlation length) determined by the eigenvalues of the linearized flow. The paper's guiding image is a ball rolling on a smooth hill: an analytic generator can still produce nonanalytic behavior in the infinite-time limit, the singularity of $F$ arising from the infinite flow time $L\to\infty$ rather than from singular coefficients.

The catch that generates this problem is stated in the paper itself. Differentiating the scaling relations $F(K,h)=L^{-d}F(K_L,h_L)$ and $\xi(K,h)=L\,\xi(K_L,h_L)$ yields an explicit formula for $u$ in terms of $F$ and $\xi$ and their derivatives - but $F$ and $\xi$ are themselves singular at criticality, so this construction makes the RG equations a tautology: they cannot explain the critical singularities if their own coefficients are singular. Wilson wrote that "there is at present no hope" of showing $u$ and $v$ analytic for these two-coupling equations, and proposed generalizing the RG to realistic transformations involving an infinite number of scale-dependent couplings. In such a many-coupling space, the eigenvalues of the linearized flow classify perturbations: relevant directions (positive eigenvalue), such as temperature and field, must be tuned to reach criticality; irrelevant directions (negative eigenvalue) die out under coarse graining - the RG explanation of universality, the observed independence of critical behavior from microscopic details. A zero eigenvalue (the marginal case, $z=0$ in the paper's notation) was flagged as exceptional, with possible logarithmic violations of the scaling laws and, sometimes, the need to tune a third thermodynamic quantity. The paper also noted that in a many-coupling space the flow could in principle approach a limit cycle (a closed recurrent orbit), run off to infinity, or oscillate irregularly as $L\to\infty$, in which cases ordinary power-law scaling would fail; and that below $T_c$ - in the ordered phase, on the $h=0$ boundary between the two magnetized phases - the RG picture additionally requires reliable noncritical computations of the free energy and correlation length on the phase boundary at coarse-grained scales.

At the level of physics, the subsequent program delivered much of this agenda. Paper II (published back-to-back with paper I) supplied the promised explicit realization: a momentum-shell ("phase-space cell") recursion for a Landau-Ginzburg-type model - a continuous-spin generalization in which the interaction is a functional of a smooth spin field with gradient and potential terms - with infinitely many couplings, giving qualitative three-dimensional exponents $\nu\simeq0.61$ for the correlation length, $\gamma\simeq1.22$ for the susceptibility, and $\eta=0$ for the anomalous dimension (the correction to the power-law decay of the critical two-point spin correlation), while explicitly declining to treat $T<T_c$. The $\varepsilon$ expansion (Wilson and Fisher 1972; systematized in Wilson and Kogut 1974) computes exponents as series in $\varepsilon=4-d$ about a non-Gaussian (that is, interacting rather than free-field) fixed point in $4-\varepsilon$ dimensions, where the flow functions are analytic by construction. Resummed at $\varepsilon=1$, and combined with the functional renormalization group (an exact flow equation for a scale-dependent coarse-grained free energy; Wetterich 1993), Monte Carlo renormalization, and the conformal bootstrap - a numerical method that determines the correlation functions of a conformal field theory, i.e. a scale-invariant continuum theory, from consistency conditions (Kos et al. 2016) - these methods yield today's high-precision 3D Ising exponents. The marginal case $z=0$ became the textbook theory of marginal operators: logarithmic corrections at marginality were derived by Wegner and Riedel (1973), and the marginal-flow structure of the two-dimensional XY model - a line of fixed points with essential singularities - was analyzed by Kosterlitz (1974). Renormalization-group limit cycles proved real, though in systems other than equilibrium critical points: the Efimov effect, an infinite geometric tower of three-body bound states, realizes an ultraviolet limit cycle (Bedaque, Hammer and van Kolck 1999; Braaten and Hammer 2003), and soluble Hamiltonian models (Głazek and Wilson 2002, 2004) and polymer systems (Pal, Sadhukhan and Bhattacharjee 2013) exhibit them, while monotonicity theorems for relativistic RG flows (the two-dimensional c-theorem and four-dimensional a-theorem; Komargodski and Schwimmer 2011) forbid closed flows in unitary Lorentz-invariant quantum field theories in two and four dimensions.

What remains unresolved is the constructive, mathematical realization of the differential RG for the actual (non-hierarchical) lattice Ising model in the dimension range $2<d<4$. Rigorous renormalization-group theory exists for hierarchical models - models in which the lattice is replaced by an exactly self-similar ultrametric structure, making the RG map explicit - including proofs of a non-Gaussian fixed point (Collet and Eckmann 1978; Koch and Wittwer 1986); and for $d\geq 4$, where the critical scaling limit is Gaussian: the logarithmic corrections at $d=4$ for the weakly self-avoiding walk and related $\phi^4$-type models are now theorems (Bauerschmidt, Brydges and Slade 2015, 2019), and the marginal triviality of the 4D scaling limits has been proven (Aizenman and Duminil-Copin 2021). For $2<d<4$, no construction of a non-Gaussian RG fixed point for a short-range lattice model, no proof of analyticity of its flow functions, and no proof of convergence of critical flows (as opposed to limit cycles, runaways, or irregular behavior) is known. Constructing the fixed point together with a reconstruction theorem connecting it to the lattice model's correlation functions corresponds, by the standard heuristic dictionary between RG fixed points and continuum limits, to constructing the critical scaling limit of the Ising model: the continuum, interacting, scale-invariant theory believed to be conformally invariant, known as the 3D Ising conformal field theory, whose operator content and exponents the bootstrap computes with high precision without constructing the theory. This is a recognized central open problem of mathematical physics; Aizenman and Duminil-Copin (2021) explicitly describe the rigorous description of critical 3D Ising behavior as the goal at which their tools are aimed.

Problem Statement

For the ferromagnetic nearest-neighbor Ising model on the $d$-dimensional hypercubic lattice in the dimension range $2<d<4$ ($d=3$ being the case addressed in the source paper), construct the infinitesimal renormalization group as a well-defined mathematical dynamical system on the space of scale-dependent effective Hamiltonians - which necessarily involves infinitely many coupling constants - and prove the statements that Wilson's 1971 formulation had to assume: (i) finiteness and real-analyticity of the RG flow functions at criticality, so that all critical singularities arise from the infinite coarse-graining limit $L\to\infty$ rather than from singular coefficients - the assumption whose absence, in Wilson's words, would make the renormalization-group equations a tautology; (ii) existence of a non-Gaussian fixed point of the flow and convergence to it of every flow starting on the critical manifold (the set of initial Hamiltonians tuned to criticality), so that the asymptotic attractor is a fixed point rather than a limit cycle, a runaway, or irregular (ergodic) behavior - with the eigenvalues of the linearized flow at the fixed point then yielding the critical exponents, the Widom-Kadanoff scaling laws, and universality (independence of the microscopic short-range interaction details); (iii) extension of the construction to nonzero field $h\neq 0$ and to the ordered phase below $T_c$, where the coarse-grained free energy and correlation length on the $h=0$ phase boundary are shown to be computable and to reproduce the below-$T_c$ critical singularities (coexistence curve and ordered-phase equation of state); and (iv) a reconstruction theorem connecting the constructed fixed point and its correlation functions to the ferromagnetic nearest-neighbor lattice Ising model, thereby identifying the construction with the lattice model's critical scaling limit. The four requirements (i)-(iv) are jointly posed and interlock, as aspects of a single constructive realization of the differential renormalization group.

References

1. Kenneth G. Wilson, "Renormalization Group and Critical Phenomena. I. Renormalization Group and the Kadanoff Scaling Picture", Physical Review B 4(9), 3174-3183 (1971). DOI: 10.1103/PhysRevB.4.3174; URL: https://doi.org/10.1103/PhysRevB.4.3174
2. Kenneth G. Wilson, "Renormalization Group and Critical Phenomena. II. Phase-Space Cell Analysis of Critical Behavior", Physical Review B 4(9), 3184-3205 (1971). DOI: 10.1103/PhysRevB.4.3184; URL: https://doi.org/10.1103/PhysRevB.4.3184
3. Kenneth G. Wilson and Michael E. Fisher, "Critical Exponents in 3.99 Dimensions", Physical Review Letters 28(4), 240-243 (1972). DOI: 10.1103/PhysRevLett.28.240; URL: https://doi.org/10.1103/PhysRevLett.28.240
4. Kenneth G. Wilson and John Kogut, "The renormalization group and the epsilon expansion", Physics Reports 12(2), 75-199 (1974). DOI: 10.1016/0370-1573(74)90023-4; URL: https://doi.org/10.1016/0370-1573(74)90023-4
5. Kenneth G. Wilson, "The renormalization group and critical phenomena", Reviews of Modern Physics 55(3), 583-600 (1983). DOI: 10.1103/RevModPhys.55.583; URL: https://doi.org/10.1103/RevModPhys.55.583
6. Michael E. Fisher, "Renormalization group theory, the epsilon expansion and Ken Wilson as I knew him", International Journal of Modern Physics B 29(8), 1530006 (2015). DOI: 10.1142/S0217979215300066; URL: https://doi.org/10.1142/S0217979215300066
7. Leo P. Kadanoff, Wolfgang Götze, David Hamblen, Robert Hecht, E. A. S. Lewis, V. V. Palciauskas, Martin Rayl, J. Swift, David Aspnes, and Joseph Kane, "Static Phenomena Near Critical Points: Theory and Experiment", Reviews of Modern Physics 39(2), 395-431 (1967). DOI: 10.1103/RevModPhys.39.395; URL: https://doi.org/10.1103/RevModPhys.39.395
8. P. F. Bedaque, H.-W. Hammer, and U. van Kolck, "Renormalization of the Three-Body System with Short-Range Interactions", Physical Review Letters 82(3), 463-467 (1999). DOI: 10.1103/PhysRevLett.82.463; URL: https://doi.org/10.1103/PhysRevLett.82.463
9. Eric Braaten and H.-W. Hammer, "An Infrared Renormalization Group Limit Cycle in QCD", Physical Review Letters 91(10), 102002 (2003). DOI: 10.1103/PhysRevLett.91.102002; URL: https://doi.org/10.1103/PhysRevLett.91.102002
10. Stanisław D. Głazek and Kenneth G. Wilson, "Limit Cycles in Quantum Theories", Physical Review Letters 89(23), 230401 (2002). DOI: 10.1103/PhysRevLett.89.230401; URL: https://doi.org/10.1103/PhysRevLett.89.230401
11. Stanisław D. Głazek and Kenneth G. Wilson, "Universality, marginal operators, and limit cycles", Physical Review B 69(9), 094304 (2004). DOI: 10.1103/PhysRevB.69.094304; URL: https://doi.org/10.1103/PhysRevB.69.094304
12. Jean-François Fortin, Benjamín Grinstein, and Andreas Stergiou, "Limit cycles in four dimensions", Journal of High Energy Physics 2012(12), 112 (2012). arXiv:1206.2921. DOI: 10.1007/JHEP12(2012)112; URL: https://doi.org/10.1007/JHEP12(2012)112
13. Michael Aizenman and Hugo Duminil-Copin, "Marginal triviality of the scaling limits of critical 4D Ising and $\phi_4^4$ models", Annals of Mathematics 194(1), 163-235 (2021). arXiv:1912.07973. DOI: 10.4007/annals.2021.194.1.3; URL: https://doi.org/10.4007/annals.2021.194.1.3
14. Michael Aizenman, Hugo Duminil-Copin, and Vladas Sidoravicius, "Random Currents and Continuity of Ising Model's Spontaneous Magnetization", Communications in Mathematical Physics 334(2), 719-742 (2015). DOI: 10.1007/s00220-014-2093-y; URL: https://doi.org/10.1007/s00220-014-2093-y
15. Roland Bauerschmidt, David C. Brydges, and Gordon Slade, "Introduction to a Renormalisation Group Method", Lecture Notes in Mathematics 2242, Springer, Singapore (2019). DOI: 10.1007/978-981-32-9593-3; URL: https://doi.org/10.1007/978-981-32-9593-3
16. Roland Bauerschmidt, David C. Brydges, and Gordon Slade, "Logarithmic Correction for the Susceptibility of the 4-Dimensional Weakly Self-Avoiding Walk: A Renormalisation Group Analysis", Communications in Mathematical Physics 337(2), 817-877 (2015). DOI: 10.1007/s00220-015-2352-6; URL: https://doi.org/10.1007/s00220-015-2352-6
17. Franz J. Wegner and Eberhard K. Riedel, "Logarithmic Corrections to the Molecular-Field Behavior of Critical and Tricritical Systems", Physical Review B 7(1), 248-256 (1973). DOI: 10.1103/PhysRevB.7.248; URL: https://doi.org/10.1103/PhysRevB.7.248
18. R. Kenna, D. A. Johnston, and W. Janke, "Scaling Relations for Logarithmic Corrections", Physical Review Letters 96(11), 115701 (2006). DOI: 10.1103/PhysRevLett.96.115701; URL: https://doi.org/10.1103/PhysRevLett.96.115701
19. Christof Wetterich, "Exact evolution equation for the effective potential", Physics Letters B 301(1), 90-94 (1993). DOI: 10.1016/0370-2693(93)90726-X; URL: https://doi.org/10.1016/0370-2693(93)90726-X
20. J. Berges, N. Tetradis, and C. Wetterich, "Critical Equation of State from the Average Action", Physical Review Letters 77(5), 873-876 (1996). DOI: 10.1103/PhysRevLett.77.873; URL: https://doi.org/10.1103/PhysRevLett.77.873
21. Jürgen Berges, Nikolaos Tetradis, and Christof Wetterich, "Non-perturbative renormalization flow in quantum field theory and statistical physics", Physics Reports 363(4-6), 223-386 (2002). DOI: 10.1016/S0370-1573(01)00098-9; URL: https://doi.org/10.1016/S0370-1573(01)00098-9
22. Zohar Komargodski and Adam Schwimmer, "On renormalization group flows in four dimensions", Journal of High Energy Physics 2011(12), 099 (2011). DOI: 10.1007/JHEP12(2011)099; URL: https://doi.org/10.1007/JHEP12(2011)099
23. Tanmoy Pal, Poulomi Sadhukhan, and Somendra M. Bhattacharjee, "Renormalization Group Limit Cycle for Three-Stranded DNA", Physical Review Letters 110(2), 028105 (2013). DOI: 10.1103/PhysRevLett.110.028105; URL: https://doi.org/10.1103/PhysRevLett.110.028105
24. Pierre Collet and Jean-Pierre Eckmann, "A Renormalization Group Analysis of the Hierarchical Model in Statistical Mechanics", Lecture Notes in Physics 74, Springer, Berlin, Heidelberg (1978). DOI: 10.1007/BFb0017107; URL: https://doi.org/10.1007/BFb0017107
25. Hans Koch and Peter Wittwer, "A non-Gaussian renormalization group fixed point for hierarchical scalar lattice field theories", Communications in Mathematical Physics 106(3), 495-532 (1986). DOI: 10.1007/BF01207260; URL: https://doi.org/10.1007/BF01207260
26. Filip Kos, David Poland, David Simmons-Duffin, and Alessandro Vichi, "Precision islands in the Ising and O(N) models", Journal of High Energy Physics 2016(8), 036 (2016). arXiv:1603.04436. DOI: 10.1007/JHEP08(2016)036; URL: https://doi.org/10.1007/JHEP08(2016)036
27. J. M. Kosterlitz, "The critical properties of the two-dimensional xy model", Journal of Physics C: Solid State Physics 7(6), 1046-1060 (1974). DOI: 10.1088/0022-3719/7/6/005; URL: https://doi.org/10.1088/0022-3719/7/6/005