The upper critical dimension of the Kardar-Parisi-Zhang universality class

The existence and value of the upper critical dimension $d_u$ of the Kardar-Parisi-Zhang (KPZ) universality class — the substrate dimension above which interfacial fluctuations become irrelevant, $χ=0$, and growing interfaces are asymptotically flat — has been contested for four decades. Pagnani and Parisi's multisurface-coding simulations of the restricted solid-on-solid (RSOS) model in $d=4$ (Phys. Rev. E 87, 010102(R) (2013)) measured $χ=0.2532(5)$ on lattices up to $256^4$ sites and concluded that $d=4$ is not $d_u$, while flagging three reliability caveats: the assumption that the RSOS model belongs to the KPZ class, limited steady-state sampling at the largest size, and an unquantified finite-size-correction ansatz. Subsequent work on ballistic deposition, etching models, directed polymers, and direct KPZ integration confirmed the $d=4$ conclusion and pushed nonzero exponents to $d=15$, with the conjecture $β_d=7/(8d+13)$ implying $d_u=∞$; yet growth on Bethe lattices and fully connected graphs finds flat, Edwards-Wilkinson-like interfaces at $d=∞$, implying a finite $d_u$. The open problem is to determine whether the KPZ class has a finite upper critical dimension and, if so, its value.

Background

The Kardar-Parisi-Zhang (KPZ) equation is the minimal continuum model of nonequilibrium interface growth. It describes the time evolution of the height $h(r,t)$ of an interface above a $d$-dimensional substrate,

$∂_t h = ν ∇² h + λ (∇h)² + η,$

where $ν$ is a surface-tension coefficient, $λ$ the strength of the nonlinear lateral-growth term that breaks the $h → −h$ symmetry, and $η$ Gaussian white noise (Kardar, Parisi, and Zhang, 1986). Setting $λ=0$ gives the linear Edwards-Wilkinson (EW) equation, whose fluctuations are purely diffusive. The KPZ universality class — the set of discrete growth models (the Eden model of irreversible random aggregation, ballistic deposition of particles that stick permanently where they first land, restricted solid-on-solid growth, and etching models mimicking corrosive erosion) and physical systems (front propagation, slow combustion, turbulent liquid crystals) sharing its scaling — also encompasses Burgers turbulence (the equation of a randomly stirred fluid, equivalent to KPZ under a change of variables) and directed polymers in random media (DPRM), the problem of the lowest-energy path of a chain advancing through a random energy landscape, which maps exactly onto KPZ growth with polymer wandering exponent $ζ = 1/(2−χ)$.

Kinetic roughening is characterized by Family-Vicsek scaling of the interface width $w²(L,t)=⟨h²⟩−⟨h⟩²$: $w ∼ t^{β}$ at early times and $w ∼ L^{χ}$ in the steady state, defining the growth exponent $β$, the roughness exponent $χ$ (often written $α$), and the dynamic exponent $z$, with $β = χ/z$. An exact tilt (Galilean) symmetry of the KPZ equation enforces $χ + z = 2$ in every dimension. In $d=1$ the exponents are exact ($χ=1/2$, $z=3/2$); for $d>1$ no exact solution is known. For $d>2$ the KPZ equation additionally admits a weak-coupling smooth (EW-like) phase, and the upper-critical-dimension debate concerns the strong-coupling rough phase reached by standard lattice growth models.

The upper critical dimension $d_u$ is the substrate dimension above which interfacial fluctuations become irrelevant, i.e. $χ=0$ and growth is asymptotically EW-like and flat — in complete analogy with the EW class itself, where $d_u=2$ exactly. Whether the KPZ class has a finite $d_u$, and its value, has been contested since the late 1980s: mode-coupling and several field-theoretic analyses suggested $d_u=4$; the nonperturbative renormalization group (NPRG, a field-theoretic method integrating fluctuations over all scales without expanding in the coupling) of Canet, Chaté, Delamotte, and Wschebor (2010) yields a small but nonzero $χ$ at $d=4$; real-space renormalization arguments give $d_u=∞$; and the width-distribution study of Marinari, Pagnani, Parisi, and Rácz (2002) in $d=2..5$ found smoothly varying scaling functions suggesting $d_u=∞$. Numerically the issue is hard: high-dimensional lattices are expensive, and for lattice KPZ models the asymptotic regime is reached only when the fluctuation scale exceeds the lattice spacing ($w²>1$), which in $d=4$ requires linear sizes $L≳32$; before 2013 the largest steady-state simulation was $L=128$, giving $χ=0.245(5)$ for a directed 4-mer diffusion model (Ódor, Liedke, and Heinig, 2010).

Pagnani and Parisi (2013) performed a multisurface-coding study of the restricted solid-on-solid (RSOS) model. The restricted solid-on-solid (RSOS) model lets a site's height grow by one unit only if it stays within one unit of all nearest-neighbor heights; it is a canonical discrete KPZ-class realization. Using bit-parallel multi-surface and multi-lattice coding, the authors simulated $d=4$ lattices up to $L=128$ in the steady state (and $L=256$ in the dynamic regime, with only three long samples, restricted to $t/L^{z}<8$), fitting the first three height cumulants $w_n$ simultaneously with the finite-size ansatz $w_n = A_n L^{nχ}(1 + B_n L^{−ω})$, where $ω$ is a single leading correction exponent. They obtained $χ=0.2532(5)$, clearly nonzero, and concluded that $d=4$ is not the upper critical dimension — while stating three caveats: the conclusion assumes the RSOS model belongs to the KPZ universality class in $d=4$; the $L=256$ statistics are limited; and systematic errors of the single-$ω$ finite-size form are not proven negligible.

Later work confirmed this conclusion and extended the question: KPZ-consistent exponents and universal height distributions were found up to $d=6$ for RSOS-type models (Alves, Oliveira, and Ferreira, 2014), up to $d=10+1$ for directed polymers (Kim, 2019, 2021), and up to $d=15$ together with the closed-form conjecture $β_d = 7/(8d+13)$ (Oliveira, 2022), which implies $d_u=∞$; direct numerical integration of the KPZ equation with an RSOS-type restriction also finds the strong-coupling phase alive at $d=4$ (Torres and Buceta, 2018). On the other hand, growth on the Bethe lattice (an infinite regular tree, the natural $d→∞$ limit of a lattice) and on the fully connected graph (all-to-all, mean-field topology) yields smooth, EW-like interfaces at $d=∞$ (Oliveira, 2021; Marcos, Meléndez, Cuerno, and Ruiz-Lorenzo, 2026), suggesting a finite $d_u$. These two strands have not been reconciled, which is precisely the open problem.

Problem Statement

Determine the upper critical dimension of the Kardar-Parisi-Zhang universality class: decide whether there exists a finite substrate dimension $d_u$ such that for all $d>d_u$ the KPZ nonlinearity is irrelevant, the roughness exponent vanishes ($χ=0$), and growing interfaces become asymptotically flat (Edwards-Wilkinson-like); if such a finite $d_u$ exists, determine its value. The four-dimensional case is settled — $d=4$ is not $d_u$ — so the open target is the general question, which currently stands as a conflict between: (i) positive measured exponents through $d=15$ and the conjectured closed form $β_d = 7/(8d+13)$ for the growth exponent $β_d$ in $d$ substrate dimensions, implying $d_u=∞$; and (ii) flat, EW-like interfaces found in $d=∞$ limits on Bethe lattices and fully connected graphs, suggesting a finite $d_u$.

References

1. M. Kardar, G. Parisi, and Y.-C. Zhang, 'Dynamic Scaling of Growing Interfaces', Phys. Rev. Lett. 56, 889 (1986). DOI: 10.1103/PhysRevLett.56.889, https://doi.org/10.1103/PhysRevLett.56.889
2. E. Marinari, A. Pagnani, G. Parisi, and Z. Rácz, 'Width distributions and the upper critical dimension of Kardar-Parisi-Zhang interfaces', Phys. Rev. E 65, 026136 (2002). DOI: 10.1103/PhysRevE.65.026136, arXiv:cond-mat/0105158, https://arxiv.org/abs/cond-mat/0105158
3. L. Canet, H. Chaté, B. Delamotte, and N. Wschebor, 'Nonperturbative Renormalization Group for the Kardar-Parisi-Zhang Equation', Phys. Rev. Lett. 104, 150601 (2010). DOI: 10.1103/PhysRevLett.104.150601, https://doi.org/10.1103/PhysRevLett.104.150601
4. G. Ódor, B. Liedke, and K.-H. Heinig, 'Directed d-mer diffusion describing the Kardar-Parisi-Zhang-type surface growth', Phys. Rev. E 81, 031112 (2010). DOI: 10.1103/PhysRevE.81.031112, arXiv:0907.3297, https://arxiv.org/abs/0907.3297
5. M. Schwartz and E. Perlsman, 'Upper critical dimension of the Kardar-Parisi-Zhang equation', Phys. Rev. E 85, 050103(R) (2012). DOI: 10.1103/PhysRevE.85.050103, arXiv:1108.4604, https://arxiv.org/abs/1108.4604
6. A. Pagnani and G. Parisi, 'Multisurface coding simulations of the restricted solid-on-solid model in four dimensions', Phys. Rev. E 87, 010102(R) (2013). DOI: 10.1103/PhysRevE.87.010102, arXiv:1210.2512, https://arxiv.org/abs/1210.2512
7. J. M. Kim and S.-W. Kim, 'Restricted solid-on-solid model with a proper restriction parameter N in 4+1 dimensions', Phys. Rev. E 88, 034102 (2013). DOI: 10.1103/PhysRevE.88.034102, https://doi.org/10.1103/PhysRevE.88.034102
8. S. G. Alves, T. J. Oliveira, and S. C. Ferreira, 'Universality of fluctuations in the Kardar-Parisi-Zhang class in high dimensions and its upper critical dimension', Phys. Rev. E 90, 020103(R) (2014). DOI: 10.1103/PhysRevE.90.020103, arXiv:1405.0974, https://arxiv.org/abs/1405.0974
9. E. A. Rodrigues, F. A. Oliveira, and B. A. Mello, 'On the existence of an upper critical dimension for systems within the KPZ universality class', Acta Phys. Pol. B 46, 1231 (2015). DOI: 10.5506/APhysPolB.46.1231, arXiv:1502.06121, https://arxiv.org/abs/1502.06121
10. T. Halpin-Healy and K. A. Takeuchi, 'A KPZ Cocktail-Shaken, not Stirred: Toasting 30 years of kinetically roughened surfaces', J. Stat. Phys. 160, 794 (2015). DOI: 10.1007/s10955-015-1282-1, arXiv:1505.01910, https://arxiv.org/abs/1505.01910
11. S. G. Alves and S. C. Ferreira, 'Scaling, cumulant ratios, and height distribution of ballistic deposition in 3+1 and 4+1 dimensions', Phys. Rev. E 93, 052131 (2016). DOI: 10.1103/PhysRevE.93.052131, arXiv:1603.08457, https://arxiv.org/abs/1603.08457
12. J. M. Kim, 'Zero-temperature directed polymer in random potential in 4+1 dimensions', Phys. Rev. E 94, 062149 (2016). DOI: 10.1103/PhysRevE.94.062149, https://doi.org/10.1103/PhysRevE.94.062149
13. M. F. Torres and R. C. Buceta, 'Numerical integration of KPZ equation with restrictions', J. Stat. Mech. 2018, 033208 (2018). DOI: 10.1088/1742-5468/aab1b3, arXiv:1707.03011, https://arxiv.org/abs/1707.03011
14. J. M. Kim, 'Directed polymer in random potentials on 5+1 and 6+1 dimensions', J. Stat. Mech. 2019, 123206 (2019). DOI: 10.1088/1742-5468/ab4e8c, https://doi.org/10.1088/1742-5468/ab4e8c
15. T. J. Oliveira, 'Surface growth on tree-like lattices and the upper critical dimension of the KPZ class', Europhys. Lett. 133, 28001 (2021). DOI: 10.1209/0295-5075/133/28001, arXiv:2012.06266, https://arxiv.org/abs/2012.06266
16. J. M. Kim, 'Directed polymer in a random potential in higher dimensions of up to d = 10+1', J. Stat. Mech. 2021, 083202 (2021). DOI: 10.1088/1742-5468/ac0f6f, https://doi.org/10.1088/1742-5468/ac0f6f
17. Y. Nakayama and Y. Nishida, 'Efimov effect at the Kardar-Parisi-Zhang roughening transition', Phys. Rev. E 103, 012117 (2021). DOI: 10.1103/PhysRevE.103.012117, arXiv:2010.15161, https://arxiv.org/abs/2010.15161
18. T. J. Oliveira, 'Kardar-Parisi-Zhang universality class in (d+1)-dimensions', Phys. Rev. E 106, L062103 (2022). DOI: 10.1103/PhysRevE.106.L062103, arXiv:2212.03847, https://arxiv.org/abs/2212.03847
19. M. S. Gomes-Filho, P. de Castro, D. B. Liarte, and F. A. Oliveira, 'Restoring the Fluctuation-Dissipation Theorem in Kardar-Parisi-Zhang Universality Class through a New Emergent Fractal Dimension', Entropy 26, 260 (2024). DOI: 10.3390/e26030260, arXiv:2402.15576, https://arxiv.org/abs/2402.15576
20. J. M. Marcos, J. J. Meléndez, R. Cuerno, and J. J. Ruiz-Lorenzo, 'On the upper critical dimension of the KPZ universality class: KPZ and related equations on a fully connected graph', arXiv:2603.02000 (2026). https://arxiv.org/abs/2603.02000