betaFeedback

WorksPhysics and Astronomy

Origins of Faceted Anomalous Scaling

Abstract

Fluctuating interfaces are typically characterized by their global, spectral, and local roughness exponents. Faceted anomalous scaling, characterized by αs>1α_s>1, αs≠αα_s\neqα, and αloc=1α_{\mathrm{loc}}=1, owes its name to the tendency of such interfaces to develop long, nearly linear segments called facets. Its statistical origin, and whether facets are required, have remained unclear. We show that this exponent structure is determined by the spatial statistics of the slope field. Slope increments split into typical fluctuations and rare, larger jumps. This anomalous scaling arises either when the typical slope increment displays genuine system-size-dependent anomalous scaling or when the rare jumps dominate the second moment. The two contributions can coexist, and their competition leads to bifractal scaling of the slopes. We establish this framework analytically in the generalized anharmonic Larkin model and the columnar Kardar–Parisi–Zhang equation, and we derive exactly the global, spectral, and local roughness exponents in both cases. The Larkin model realizes both mechanisms and produces anomalous interfaces with or without facets, whereas columnar KPZ belongs to the jump-dominated case and yields αs=3/2α_s=3/2.

Scale invariance underlies the classification of fluctuating interfaces. In the Family–Vicsek picture, a self-affine height field h(x,t)h(x,t) is characterized by a single roughness exponent α\alpha and a growing correlation length ξ(t)∼t1/z\xi(t) \sim t^{1/z} [1]. This description includes paradigmatic models such as the Edwards–Wilkinson (EW) [2] and Kardar–Parisi–Zhang (KPZ) [3] equations with space-time white noise. We consider stationary interfaces, for which ξ(t)≃L\xi(t) \simeq L and the time argument can be omitted. Even then, roughness may require three distinct exponents [4]. The global roughness exponent α\alpha is defined by

W2(L)≡⟨1L∫0Ldx [h(x)−hˉ]2⟩∼L2α,(1)W^2(L) \equiv\left\langle\frac{1}{L}\int_0^L dx\,[h(x)-\bar h]^2 \right\rangle\sim L^{2\alpha}, \tag*{(1)}

where hˉ=L−1∫0Ldx h(x)\bar h = L^{-1}\int_0^L dx\,h(x).

The local exponent αloc\alpha_{\mathrm{loc}} follows from height increments over ℓ≪L\ell\ll L [5] [4],

G2(ℓ,L)≡⟨∣h(x+ℓ)−h(x)∣2⟩∼ℓ2αlocL2α−2αloc.(2)G_2(\ell,L) \equiv\left\langle|h(x+\ell)-h(x)|^2 \right\rangle\sim\ell^{2\alpha_{\mathrm{loc}}}L^{2\alpha-2\alpha_{\mathrm{loc}}}. \tag*{(2)}

Finally, defining h^(k)=L−1/2∫0Ldx e−ikx[h(x)−hˉ]\hat h(k)=L^{-1/2}\int_0^L dx\,e^{-ikx}[h(x)-\bar h], the spectral exponent αs\alpha_s is given by

S(k,L)≡⟨∣h^(k)∣2⟩∼k−(1+2αs)L2(α−αs)(3)S(k,L) \equiv\left\langle|\hat h(k)|^2 \right\rangle\sim k^{-(1+2\alpha_s)}L^{2(\alpha-\alpha_s)} \tag*{(3)}

for 1/L≪k≪11/L \ll k \ll1.

Family–Vicsek scaling corresponds to α=αloc=αs\alpha=\alpha_{\mathrm{loc}}=\alpha_s. Two well-understood departures from Family–Vicsek scaling are known. In the first, αs=αloc<1\alpha_s=\alpha_{\mathrm{loc}}<1 while

α≠αs\alpha\ne\alpha_s [5, 6, 7, 8, 9, 10, 11, 12] [4]. In super-rough interfaces, such as the quenched Edwards–Wilkinson equation at the depinning threshold [13, 14] and linear molecular beam epitaxy [15, 16], αs=α>1\alpha_s=\alpha>1 while αloc=1\alpha_{\mathrm{loc}}=1.

A third departure, known as faceted anomalous scaling, is characterized by αs>1\alpha_s>1, αs≠α\alpha_s\ne\alpha, and αloc=1\alpha_{\mathrm{loc}}=1. First observed in the Sneppen model [17], this exponent structure has since been reported in growth models involving localization [18], temporally correlated noise [19], and nonlinear elasticity [20, 21], as well as experimentally in dissolving metal surfaces and epitaxially grown semiconductor films [22, 23]. Unlike the other two departures, however, its statistical origin remains unclear. The term “faceted” refers to the frequent appearance of long, nearly linear segments, but morphology alone neither explains why αs\alpha_s differs from both α\alpha and αloc\alpha_{\mathrm{loc}} nor establishes whether facets are required for this exponent structure. The key observation is that this exponent structure is controlled by the spatial statistics of the slope field m(x)≡∂xh(x)m(x)\equiv\partial_xh(x): slope increments separate into typical fluctuations and rare, sharper events, and their competition governs all three roughness exponents.

In this Letter, we establish the statistical origin of this exponent structure by treating the stationary local slope, m(x)≡∂xh(x)m(x)\equiv\partial_xh(x), as a fluctuating interface in its own right. We show that the separation of αs\alpha_s from both α\alpha and αloc\alpha_{\mathrm{loc}} can originate from two contributions to slope-increment moments: typical increments with genuine system-size-dependent scaling, and rare large increments whose sampling probability may vanish as L→∞L\to\infty. When the latter dominate the second moment, the associated large slope changes generate a pronounced faceted morphology; when the anomalous typical contribution dominates, the same exponent structure does not require facets. The two contributions may coexist and exchange dominance with moment order, naturally producing bifractal slope statistics. Their relation to height morphology is summarized in Figure 1. One-dimensional columnar KPZ isolates the rare-increment-dominated limit and yields αs=3/2\alpha_s=3/2. The generalized anharmonic Larkin model instead tunes the competition between the two contributions and the degree of visible faceting. For both models, we derive exact expressions for the global, spectral, and local roughness exponents.

Phenomenological slope-field classification of stationary one-dimensional rough interfaces

Figure 1. Phenomenological slope-field classification of stationary one-dimensional rough interfaces with αloc=1\alpha_{\mathrm{loc}}=1 and αs>1\alpha_s>1, according to the contribution controlling the second moment of slope increments. Here Δℓ=m(x+ℓ)−m(x)\Delta_\ell=m(x+\ell)-m(x), and Eq. (7) decomposes ⟨∣Δℓ∣2⟩\langle|\Delta_\ell|^2\rangle into typical and rare contributions, the latter weighted by the probability Pr⁡(ℓ,L)\operatorname{Pr}(\ell,L) of sampling a large rare increment. Standard LL-independent typical fluctuations yield a super-rough, nonfaceted height field (top). “Faceted” anomalous scaling has two origins, illustrated by the anharmonic Larkin model: dominant anomalous typical fluctuations produce weakly faceted (1/2<p<11/2<p<1) or nonfaceted (p>1p>1) morphologies (middle), whereas dominant rare large increments produce strong faceting (0<p<1/20<p<1/2; bottom). Columnar KPZ also realizes the rare-dominated class and is analyzed separately in Figure 2.

Spectral constraints and slope scaling. Using Eq. (3) in the Fourier representation of the height increment gives

G2(ℓ,L)∝∫1/L1dk [1−cos⁡(kℓ)]S(k,L)∼{ℓ2αsL2(α−αs),αs<1,ℓ2L2(α−1),αs>1.(4)G_2(\ell,L)\propto\int_{1/L}^{1}dk\,[1-\cos(k\ell)]S(k,L) \sim \begin{cases} \ell^{2\alpha_s}L^{2(\alpha-\alpha_s)}, & \alpha_s<1,\\ \ell^2L^{2(\alpha-1)}, & \alpha_s>1. \end{cases} \tag*{(4)}

Hence αloc=αs\alpha_{\mathrm{loc}}=\alpha_s for αs<1\alpha_s<1, whereas αloc=1\alpha_{\mathrm{loc}}=1 for αs>1\alpha_s>1, with logarithmic corrections at αs=1\alpha_s=1. Let αm\alpha_m, αs,m\alpha_{s,m}, and αloc,m\alpha_{\mathrm{loc},m} denote the corresponding slope-field exponents. The slope-field analogue of Eq. (2) then reads

G2m(ℓ,L)=⟨∣Δℓ∣2⟩∼ℓ2αs,mL2(αm−αs,m),(5)G_2^m(\ell,L)=\langle|\Delta_\ell|^2\rangle\sim\ell^{2\alpha_{s,m}}L^{2(\alpha_m-\alpha_{s,m})}, \tag*{(5)}

where Δℓ=m(x+ℓ)−m(x)\Delta_\ell=m(x+\ell)-m(x); the slope exponents are identified by matching this scaling. For periodic boundaries, m^k=ikh^k\hat{m}_k=ik\hat{h}_k. For free boundaries, the corresponding cosine modes of hh and sine modes of mm satisfy mn(s)=−knhn(c)m_n^{(s)}=-k_nh_n^{(c)}. In either case, Eq. (4) applied to both fields gives

α=1+αm,\alpha=1+\alpha_m,
αs=1+αs,m,\alpha_s=1+\alpha_{s,m},
αloc=1(0<αloc,m=αs,m<1),(6)\alpha_{\mathrm{loc}}=1\quad(0<\alpha_{\mathrm{loc},m}=\alpha_{s,m}<1), \tag*{(6)}

where the last relation applies to the range 1<αs<21<\alpha_s<2 considered below. Both super-rough and faceted anomalous interfaces obey these relations. For super-rough interfaces, standard scaling of typical slope fluctuations implies αm=αs,m=αloc,m>0\alpha_m=\alpha_{s,m}=\alpha_{\mathrm{loc},m}>0, and hence α=αs>1\alpha=\alpha_s>1 [5, 24, 25]. Faceted anomalous scaling instead requires αm≠αs,m\alpha_m\ne\alpha_{s,m}, and therefore nonstandard slope statistics. To identify their possible origins, define Δℓ=m(x+ℓ)−m(x)\Delta_\ell=m(x+\ell)-m(x) and decompose its second moment, at the scaling level, as

⟨∣Δℓ∣2⟩∼∣Δℓ∣typ2+Pr⁡(ℓ,L)∣Δℓ∣r2.(7)\langle|\Delta_\ell|^2\rangle\sim|\Delta_\ell|_{\mathrm{typ}}^2+\operatorname{Pr}(\ell,L)|\Delta_\ell|_r^2. \tag*{(7)}

The first term represents typical slope increments; the second represents rare large increments of characteristic amplitude ∣Δℓ∣r|\Delta_\ell|_r, sampled with a probability Pr⁡(ℓ,L)\operatorname{Pr}(\ell,L) that may vanish as L→∞L\to\infty. Faceted anomalous scaling can thus arise either from an LL-dependent typical contribution or from dominance of the rare contribution. We isolate the latter mechanism in columnar KPZ before studying their competition in the generalized Larkin model.

Scaling dominated by rare large increments: columnar KPZ. We illustrate this limit with the one-dimensional columnar KPZ equation [18],

∂th(x,t)=∂x2h(x,t)+λ[∂xh(x,t)]2+η(x),(8)\partial_t h(x,t)=\partial_x^2h(x,t)+\lambda[\partial_xh(x,t)]^2+\eta(x), \tag*{(8)}

whose defining feature is that the random force η(x)\eta(x) is independent of time. Under the Cole–Hopf transformation Z=eλhZ=e^{\lambda h}, Eq. (8) becomes the imaginary-time Schrödinger equation

∂tZ(x,t)=−HZ(x,t),H=−∂x2−λη(x),(9)\partial_t Z(x,t)=-\mathcal{H}Z(x,t),\qquad\mathcal{H}=-\partial_x^2-\lambda\eta(x), \tag*{(9)}

where the time-independent random operator H\mathcal{H} is the one-dimensional Anderson Hamiltonian. Let Hψn=Enψn\mathcal{H}\psi_n=E_n\psi_n, with E0E_0 the ground-state energy. The spectral decomposition then gives

Z(x,t)=∑ncnψn(x)e−Ent→t→∞c0ψ0(x)e−E0t,(10)Z(x,t)=\sum_n c_n\psi_n(x)e^{-E_nt}\xrightarrow{t\to\infty}c_0\psi_0(x)e^{-E_0t}, \tag*{(10)}

so that the interface reaches a stationary profile in a frame moving at velocity v∞=−E0/λv_\infty=-E_0/\lambda:

H(x)≡h(x,t)−v∞t≃1λln⁡ψ0(x)+const.(11)H(x)\equiv h(x,t)-v_\infty t\simeq\frac{1}{\lambda}\ln\psi_0(x)+\mathrm{const}. \tag*{(11)}

In one dimension, the ground state is exponentially localized around a random position xcx_c [26, 27, 28, 29]. Its logarithm can therefore be decomposed as

ln⁡ψ0(x)=−γ∣x−xc∣+χ(x),(12)\ln\psi_0(x)=-\gamma|x-x_c|+\chi(x), \tag*{(12)}

where γ\gamma is the inverse localization length and χ\chi describes subleading spatial fluctuations. The envelope produces two long, nearly linear segments joined near xcx_c. Their vertical span is of order γL\gamma L; hence its leading algebraic dependence is W(L)∼LW(L)\sim L, giving α=1\alpha=1.

Consider now the stationary slope for free boundaries, m(x)=∂xH(x)=λ−1∂xln⁡ψ0(x)m(x)=\partial_xH(x)=\lambda^{-1}\partial_x\ln\psi_0(x), which has two nearly constant plateaus separated by a single jump at xcx_c. Denoting its amplitude by J≡m(xc+)−m(xc−)J\equiv m(x_c^+)-m(x_c^-), an interval of length ℓ≪L\ell\ll L crosses the jump with probability Pr⁡∼ℓ/L\Pr\sim\ell/L. The contribution from rare large increments controls the large-scale behavior and gives

⟨∣Δℓ∣2⟩∼⟨J2⟩ℓL.(13)\left\langle\lvert\Delta_\ell\rvert^2\right\rangle\sim\langle J^2\rangle\frac{\ell}{L}. \tag*{(13)}

The ℓ\ell and LL dependences in Eq. (13) give αloc,m=1/2\alpha_{\mathrm{loc},m}=1/2 and αm=0\alpha_m=0. Equation (6) then yields αs,m=1/2\alpha_{s,m}=1/2 and the complete height-exponent triplet

(α,αs,αloc)=(1,32,12).(14)(\alpha,\alpha_s,\alpha_{\mathrm{loc}})=\left(1,\frac{3}{2},\frac{1}{2}\right). \tag*{(14)}

The predicted small-kk decay Sh(k,L)∼k−4S_h(k,L)\sim k^{-4}, generated by the macroscopic slope jump, is confirmed in Figure 2. The k−2k^{-2} regime at larger kk, associated with the Brownian-like fluctuations χ(x)\chi(x), is discussed in the Supplemental Material. The argument extends beyond columnar KPZ whenever a finite number of macroscopic slope jumps dominates the second moment, including in heterogeneous elastic lines [10]. As detailed in the End Matter, the same result extends to any dimension under the geometric assumption of a finite number of macroscopic planar facets. We now turn to the anharmonic Larkin model, where the two terms in Eq. (7) can coexist and compete.

Figure 2. Stationary height structure factor Sh(k,L)S_h(k,L) of the one-dimensional columnar KPZ equation for L=214L=2^{14}, averaged over 360 disorder realizations, with free boundary conditions. The dashed segments indicate the k−4k^{-4} behavior at small kk, generated by the macroscopic slope jump, and the k−2k^{-2} behavior at larger kk, associated with the short-scale fluctuations χ(x)\chi(x) of the localized state. Insets: centered height profile h(x,t)−h‾(t)h(x,t)-\overline{h}(t) (upper) and corresponding slope profile m(x,t)=∂xh(x,t)m(x,t)=\partial_xh(x,t) (lower) for a representative stationary configuration. The two finite slope plateaus separated by a macroscopic jump produce the faceted height profile. The structure factor is computed with the cosine transform appropriate to free boundary conditions, and its oscillatory sub-branches, which arise from profiles with one or a few kinks, do not affect the k−4k^{-4} envelope.

Competing slope fluctuations: anharmonic Larkin model

We consider the one-dimensional anharmonic Larkin model [20],

∂th(x,t)=∂x[(∂xh)2n−1]+η(x),n=2,3,…,(15)\partial_t h(x,t)=\partial_x\left[(\partial_xh)^{2n-1}\right]+\eta(x),\qquad n=2,3,\ldots, \tag*{(15)}

where η(x)\eta(x) is a quenched spatial white noise. For a stationary translating solution h(x,t)=vt+H(x)h(x,t)=vt+H(x), with m(x)=∂xH(x)m(x)=\partial_xH(x), spatial averaging fixes v=ηˉv=\bar{\eta}, where ηˉ=L−1∫0Lη(x) dx\bar{\eta}=L^{-1}\int_0^L\eta(x)\,\mathrm{d}x. One integration in the comoving frame gives

m(x)2n−1=m(0)2n−1+B(x),B(x)≡∫0x[ηˉ−η(y)] dy.(16)m(x)^{2n-1}=m(0)^{2n-1}+B(x),\qquad B(x)\equiv\int_0^x[\bar{\eta}-\eta(y)]\,\mathrm{d}y. \tag*{(16)}

Since B(0)=B(L)=0B(0)=B(L)=0, B(x)B(x) is a Brownian bridge and Eq. (16) implies m(L)=m(0)m(L)=m(0). We will consider Free-end Neumann boundary conditions, m(0)=m(L)=0m(0)=m(L)=0. The physical model therefore corresponds to p=(2n−1)−1p=(2n-1)^{-1} in the generalized stationary process

m(x)=sgn⁡[B(x)]∣B(x)∣p,p>0.(17)m(x)=\operatorname{sgn}[B(x)]\lvert B(x)\rvert^p,\qquad p>0. \tag*{(17)}

We now use the Brownian mapping in Eq. (17) to derive the scaling properties of the associated height field, H(x)−H(0)=∫0xm(y) dyH(x)-H(0)=\int_0^x m(y)\,\mathrm{d}y. Under x→bxx \to bx, the Brownian bridge scales as B→b1/2BB \to b^{1/2}B, so m→bp/2mm \to b^{p/2}m. Spatial integration, H(x)−H(0)=∫0xm(y) dyH(x)-H(0)=\int_0^x m(y)\,\mathrm{d}y, then gives

αm=p2,α=1+p2.(18)\alpha_m=\frac{p}{2},\qquad\alpha=1+\frac{p}{2}. \tag*{(18)}

To evaluate the two terms in Eq. (7), let δℓB(x)=B(x+ℓ)−B(x)\delta_\ell B(x)=B(x+\ell)-B(x). Writing B=B(x)B=B(x), Eq. (17) gives the exact relation

Δℓ=sgn⁡[B+δℓB]∣B+δℓB∣p−sgn⁡(B)∣B∣p.\Delta_\ell=\operatorname{sgn}[B+\delta_\ell B]|B+\delta_\ell B|^p-\operatorname{sgn}(B)|B|^p.

For ℓ≪L\ell\ll L, ∣δℓB∣∼ℓ1/2|\delta_\ell B|\sim\ell^{1/2}, whereas at a typical bulk point ∣B(x)∣∼L1/2|B(x)|\sim L^{1/2}. Thus typical points satisfy ∣δℓB∣≪∣B∣|\delta_\ell B|\ll|B|, and linearizing the exact expression gives Δℓ≃p∣B∣p−1δℓB\Delta_\ell\simeq p|B|^{p-1}\delta_\ell B, hence ∣Δℓ∣typ2∼ℓLp−1|\Delta_\ell|^2_{\mathrm{typ}}\sim\ell L^{p-1}. At points where ∣B(x)∣≲∣δℓB∣∼ℓ1/2|B(x)|\lesssim|\delta_\ell B|\sim\ell^{1/2}, however, the linearization fails and the exact expression gives ∣Δℓ∣r2∼∣δℓB∣2p∼ℓp|\Delta_\ell|^2_{\mathrm{r}}\sim|\delta_\ell B|^{2p}\sim\ell^p. These points therefore have a different scaling from typical ones. For p<1p<1, their increments are sharper by a factor ∣Δℓ∣r/∣Δℓ∣typ∼(L/ℓ)(1−p)/2≫1|\Delta_\ell|_{\mathrm{r}}/|\Delta_\ell|_{\mathrm{typ}}\sim(L/\ell)^{(1-p)/2}\gg1. The one-point distribution of B(x)B(x) has width L1/2L^{1/2} and hence a density of order L−1/2L^{-1/2} near the origin, so Pr(ℓ,L)∼(ℓ/L)1/2P_{\mathrm{r}}(\ell,L)\sim(\ell/L)^{1/2}. These rare points are thus more frequent than the macroscopic jump of columnar KPZ, whose crossing probability is ℓ/L\ell/L. Substituting the typical and rare estimates into Eq. (7) gives

⟨∣Δℓ∣2⟩∼ℓLp−1+ℓp+1/2L−1/2.(19)\langle|\Delta_\ell|^2\rangle\sim\ell L^{p-1}+\ell^{p+1/2}L^{-1/2}. \tag*{(19)}

The first term is the typical contribution and is always present. The second is the zero-crossing contribution; for p<1p<1 it corresponds to rare large increments and becomes dominant for p<1/2p<1/2. At the marginal value p=1/2p=1/2, the two contributions have the same algebraic scaling, with a logarithmic correction that causes αs\alpha_s to converge to 3/23/2 anomalously slowly as 1/ln⁡L1/\ln L (End Matter, “Finite-size convergence at p=12p=\frac{1}{2}”). Each term can be matched to the anomalous-scaling form ⟨∣Δℓ∣2⟩∼ℓ2αs,mL2(αm−αs,m)\langle|\Delta_\ell|^2\rangle\sim\ell^{2\alpha_{s,m}}L^{2(\alpha_m-\alpha_{s,m})}, giving αs,m=p/2+1/4\alpha_{s,m}=p/2+1/4 when the rare contribution dominates and αs,m=1/2\alpha_{s,m}=1/2 when the typical contribution dominates. Equation (6) then yields all three height exponents in a single expression:

(α,αs,αloc)={(1+p2,54+p2,1),0<p<12,(1+p2,32,1),p>12.(20)(\alpha,\alpha_s,\alpha_{\mathrm{loc}})= \begin{cases} \left(1+\frac{p}{2},\frac{5}{4}+\frac{p}{2},1\right), & 0<p<\frac{1}{2},\\ \left(1+\frac{p}{2},\frac{3}{2},1\right), & p>\frac{1}{2}. \end{cases} \tag*{(20)}

At p=1/2p=1/2, the two branches meet at (α,αs,αloc)=(5/4,3/2,1)(\alpha,\alpha_s,\alpha_{\mathrm{loc}})=(5/4,3/2,1), with logarithmic corrections to scaling. For the physical anharmonic Larkin family, p≤1/3p\leq1/3, rare large increments dominate and Eq. (20) gives the exact prediction αs=α+1/4\alpha_s=\alpha+1/4. This explains the numerical estimate αs−α≃0.23\alpha_s-\alpha\simeq0.23 previously reported for this model [20]. For the numerical comparison, we sample free-BC ground states on both sides of p=1/2p=1/2, generating the Brownian bridge B(x)B(x) with B(0)=B(L)=0B(0)=B(L)=0.

and setting m(x)=sgn⁡[B(x)]∣B(x)∣pm(x)=\operatorname{sgn}[B(x)]|B(x)|^p directly. Figure 3 compares exact finite-size scaling of W2(L)W^2(L) and αs\alpha_s extracted from the height structure factor Sh(k,L)S_h(k,L) with the exact predictions in Eq. (20).

Figure 3. Global and spectral roughness exponents of the generalized stationary slope process as a function of pp. Blue circles show αs\alpha_s extracted from the height structure factor, for system sizes L=214L=2^{14} (lightest) to L=226L=2^{26} (darkest); the spread near p=12p=\frac{1}{2} reflects the logarithmically slow finite-size convergence (End Matter). Black diamonds show α\alpha obtained from the finite-size scaling of W2(L)W^2(L) over the same size range. The dashed and solid lines show the predictions α=1+p/2\alpha=1+p/2 and αs=min⁡(54+p2,32)\alpha_s=\min(\frac{5}{4}+\frac{p}{2},\frac{3}{2}), respectively. Free boundary conditions are used; the exponents are independent of this choice (Figure 5).

The visibility of facets correlates with, but is not defined by, which contribution controls the second moment. For p<1/2p<1/2, rare large increments dominate and produce a pronounced faceted morphology. For 1/2<p<11/2<p<1, they remain present and dominate sufficiently high moments, but their morphological signature is weaker; the associated bifractality is discussed in the End Matter. At p=1p=1, typical slope fluctuations recover standard scaling and the interface is super-rough. For p>1p>1, all positive moments are controlled by typical fluctuations: the interface is nonfaceted and nonbifractal, although its three height exponents remain distinct.

Conclusion. We have shown that interfaces with αloc=1\alpha_{\mathrm{loc}}=1 and αs>1\alpha_s>1 admit a natural classification based on the scaling of slope-field increments. The separation of their global, spectral, and local exponents originates from LL-dependent typical fluctuations, rare large increments whose sampling probability vanishes with system size, or the competition between these contributions. Columnar KPZ isolates the limit dominated by rare large increments, giving αs=3/2\alpha_s=3/2, whereas the generalized

⟦?⟧ Larkin model displays their coexistence and exchange of dominance for 0<p<10 < p < 1, and yields exact exponents across pp. Dominance by rare large increments produces facets, whereas dominance by typical fluctuations need not. When both scaling sectors coexist, their exchange of dominance with moment order yields bifractal statistics (End Matter, “General mechanism”), so qq- and LL-resolved slope increments diagnose the underlying mechanism.

Acknowledgments

We thank Christophe Texier for bringing the paper by Dumaz to our attention and for discussions on columnar KPZ.

End Matter

Higher dimensions

For clarity, consider d=2d = 2. Figure 4 shows the time evolution of a columnar-KPZ interface for a realization in which the dynamics becomes localized. Several localized structures compete at early times, producing cone-like profiles, while at late times a single localized structure dominates.

Figure 4. Time evolution of a two-dimensional columnar-KPZ interface with columnar disorder for L=256L = 256. The centered height h(x,t)−h‾(t)h(x,t) - \overline{h}(t) is shown at three times. Several localized structures compete initially, while a single structure dominates at late times.

When a finite number of localized structures contributes, neighboring cones meet along lines across which the slope m=∇h\mathbf{m} = \nabla h jumps. The total length of these lines scales as LL, so a segment of length ℓ\ell crosses one of them with probability

Pr(ℓ,L)∼ℓLL2∼ℓL.(21)P_{\mathrm{r}}(\ell,L) \sim\frac{\ell L}{L^2} \sim\frac{\ell}{L}. \tag*{(21)}

Consequently, for a displacement ℓ\ell of magnitude ℓ\ell,

⟨∣m(x+ℓ)−m(x)∣2⟩r∼⟨∣Δm∣2⟩ℓL.(22)\left\langle\lvert\mathbf{m}(\mathbf{x}+\ell)-\mathbf{m}(\mathbf{x})\rvert^2 \right\rangle_{\mathrm{r}} \sim\langle\lvert\Delta\mathbf{m}\rvert^2\rangle\frac{\ell}{L}. \tag*{(22)}

This is the same ℓ/L\ell/L dependence as in one dimension and therefore gives αs=3/2\alpha_{\mathrm{s}} = 3/2.

More generally, in dimension dd, the same result holds whenever the interface contains a finite number of macroscopic facets separated by boundaries of total size Ld−1L^{d-1}. Their crossing probability remains proportional to ℓ/L\ell/L, giving αs=3/2\alpha_{\mathrm{s}} = 3/2, independently of dd.

General mechanism

The two-sector decomposition in Eq. (7) extends directly to the qqth slope-increment moment:

⟨∣Δℓ∣q⟩∼∣Δℓ∣typq+Pr(ℓ,L)∣Δℓ∣rq.(23)\left\langle\lvert\Delta_\ell\rvert^q\right\rangle\sim\lvert\Delta_\ell\rvert_{\mathrm{typ}}^q + P_{\mathrm{r}}(\ell,L)\lvert\Delta_\ell\rvert_{\mathrm{r}}^q. \tag*{(23)}

Here Pr(ℓ,L)P_{\mathrm{r}}(\ell,L) is the sampling probability of rare large increments introduced in Eq. (7); only the characteristic amplitudes of typical fluctuations and rare large increments are raised to the power qq. When each sector is governed by a single characteristic scale, their exchange of dominance with qq produces two scaling branches: low-order moments probe typical fluctuations, whereas sufficiently high-order moments select rare large increments. The resulting slope statistics are bifractal.

Generalized Larkin model

As an exactly solvable example, consider the generalized process in Eq. (17) with 0<p<10 < p < 1. Set X=B(x)X = B(x), Y=B(x+ℓ)−B(x)Y = B(x+\ell)-B(x), and Δℓ=fp(X+Y)−fp(X)\Delta_\ell= f_p(X+Y)-f_p(X), with fp(b)=sgn⁡(b)∣b∣pf_p(b)=\operatorname{sgn}(b)\lvert b\rvert^p. At a bulk point, ∣X∣∼L1/2\lvert X\rvert\sim L^{1/2} and ∣Y∣∼ℓ1/2\lvert Y\rvert\sim\ell^{1/2}. The typical sector follows from linearization, Δℓ≃p∣X∣p−1Y\Delta_\ell\simeq p\lvert X\rvert^{p-1}Y, and contributes

∣Δℓ∣typq∼ℓq/2Lq(p−1)/2.(24)\lvert\Delta_\ell\rvert_{\mathrm{typ}}^q \sim\ell^{q/2}L^{q(p-1)/2}. \tag*{(24)}

The linearization fails in the region producing rare large increments, ∣X∣≲ℓ1/2\lvert X\rvert\lesssim\ell^{1/2}, whose probability is (ℓ/L)1/2(\ell/L)^{1/2}. There ∣Δℓ∣∼ℓp/2\lvert\Delta_\ell\rvert\sim\ell^{p/2}, giving

Pr(ℓ,L)∣Δℓ∣q∼ℓ(1+pq)/2L−1/2.(25)P_{\mathrm{r}}(\ell,L)\lvert\Delta_\ell\rvert^q \sim\ell^{(1+pq)/2}L^{-1/2}. \tag*{(25)}

The two contributions exchange dominance at qc=1/(1−p)q_{\mathrm{c}} = 1/(1-p):

⟨∣Δℓ∣q⟩1/q∼{ℓ1/2L(p−1)/2,q<qc,ℓp/2+1/(2q)L−1/(2q),q>qc.(26)\left\langle\lvert\Delta_\ell\rvert^q\right\rangle^{1/q} \sim \begin{cases} \ell^{1/2}L^{(p-1)/2}, & q < q_{\mathrm{c}},\\ \ell^{p/2+1/(2q)}L^{-1/(2q)}, & q > q_{\mathrm{c}}. \end{cases} \tag*{(26)}

At q=qcq=q_{\mathrm{c}}, a factor [log⁡(L/ℓ)]1/qc[\log(L/\ell)]^{1/q_{\mathrm{c}}} multiplies the common power law. In particular, for q=2q=2 and p=1/2p=1/2, this gives the logarithmic correction mentioned in the main text. The two branches in Eq. (26) constitute a bifractal spectrum. For the second moment, q=2q=2, rare large increments dominate for p<1/2p<1/2, whereas typical fluctuations dominate for p>1/2p>1/2, reproducing Eq. (19). At p=1p=1, m=Bm=B and all increments are Gaussian. For p>1p>1, zero crossings are smoother rather than sharper than typical fluctuations; the second term in Eq. (23) no longer defines a competing sector of rare large increments, and all positive moments are controlled by the typical contribution. There is therefore no bifractality.

Second moment and numerical protocol

Using the Fourier representation of slope increments, the q=2q=2 results in Eqs. (24) and (25) imply

Sm(k,L)∼{L−1/2k−(p+3/2),0<p<1/2,Lp−1k−2,p>1/2.(27)S_m(k,L) \sim \begin{cases} L^{-1/2}k^{-(p+3/2)}, & 0 < p < 1/2,\\ L^{p-1}k^{-2}, & p > 1/2. \end{cases} \tag*{(27)}

At p=1/2p=1/2, the same algebraic k−2k^{-2} decay is accompanied by the marginal logarithmic correction identified above. Thus αs,m=1/4+p/2\alpha_{\mathrm{s},m}=1/4+p/2 for p<1/2p<1/2 and αs,m=1/2\alpha_{\mathrm{s},m}=1/2 for p>1/2p>1/2. Equation (6) then gives αs=1+αs,m\alpha_{\mathrm{s}}=1+\alpha_{\mathrm{s},m}, while spatial integration gives α=1+p/2\alpha=1+p/2. The exponents are insensitive to the choice of boundary conditions, as confirmed in Figure 5. At p=1/2p=1/2, the finite-size corrections to αs\alpha_{\mathrm{s}} decay as A/ln⁡LA/\ln L; Figure 6 shows that corrections decay faster on both sides of the crossover, for p<1/2p<1/2 and p>1/2p>1/2. The data in Figure 3 were obtained from independent discrete Brownian bridges for L=214,…,226L=2^{14},\ldots,2^{26}, with B(0)=B(L)=0B(0)=B(L)=0 and m(0)=0m(0)=0, leading to free boundary conditions; the periodic-BC data in Figure 5 use m(0)m(0) chosen numerically to ensure ∫0Lm(x) dx=0\int_0^L m(x)\,\mathrm{d}x=0 which enforces

Figure 5. Global (α\alpha, diamonds) and spectral (αs\alpha_s, circles) roughness exponents versus pp for periodic boundary conditions, for system sizes L=214L = 2^{14} to 2222^{22} (light to dark blue). Both exponents agree with the theoretical predictions (lines), confirming that the exponents are insensitive to the choice of boundary conditions.

Figure 6. Finite-size convergence of αs\alpha_s near the crossover p=12p = \frac{1}{2}, for free boundary conditions and 2000 disorder realizations per (p,L)(p,L) pair. Deviation αs(p)−αs(p,L)\alpha_s(p) - \alpha_s(p,L) from the asymptotic prediction versus LL, for pp straddling 12\frac{1}{2} (blue: p<12p < \frac{1}{2}; amber: p=12p = \frac{1}{2}; red: p>12p > \frac{1}{2}). The dashed line shows A/ln⁡LA/\ln L: corrections at p=12p = \frac{1}{2} track it closely, while corrections decay faster for both p<12p < \frac{1}{2} and p>12p > \frac{1}{2}.

h(0)=h(L)h(0) = h(L). We fitted α\alpha from the finite-size scaling of W2(L)W^2(L) and αs\alpha_s from Sh(k,L)S_h(k,L) over the wavevector range 0.015<k<0.150.015 < k < 0.15.

References

Supplemental Material

Columnar KPZ with Gaussian columnar disorder

Localized eigenvectors and Gaussian fluctuations

The low-lying spectrum and localized eigenvectors of the white-noise Anderson Hamiltonian H\mathcal{H}, defined in the main text, were studied in Refs. [28, 29]. The result needed here is that a low-lying eigenvector has a dominant localized peak with an exponential envelope and Gaussian fluctuations around it. Under the Cole–Hopf transformation, the logarithm of the envelope produces the faceted height profile, while the Gaussian fluctuations produce the large-kk regime of its structure factor. We briefly recall how both features follow from the Riccati representation. To describe its Dirichlet spectrum, consider for each aa the solution of

Hfa=−afa,fa(0)=0,fa′(0)=1.(28)\mathcal{H}f_a=-af_a,\qquad f_a(0)=0,\qquad f_a'(0)=1. \tag*{(28)}

The value −a-a is an eigenvalue if and only if fa(L)=0f_a(L)=0. Introduce the Riccati transform Xa(x)=fa′(x)/fa(x)X_a(x)=f_a'(x)/f_a(x). It starts from +∞+\infty and satisfies

dXa=(a−Xa2) dx+g dWx,(29)dX_a=(a-X_a^2)\,dx+g\,dW_x, \tag*{(29)}

where WxW_x is Brownian motion and gg fixes the disorder strength. Whenever faf_a vanishes, XaX_a reaches −∞-\infty and is immediately restarted at +∞+\infty. The eigenvalue En=−anE_n=-a_n is selected by requiring the (n+1)(n+1)-st such explosion to occur at x=Lx=L; equivalently, ψn\psi_n has nn interior zeros. When a=ana=a_n, fanf_{a_n} is proportional to the eigenvector ψn\psi_n.

For a>0a>0, the drift in Eq. (29) has a stable fixed point +a+\sqrt{a} and an unstable one −a-\sqrt{a}. For fixed nn, the low-energy scale satisfies an∼aLa_n\sim a_L, with aL∼(log⁡L)2/3a_L\sim(\log L)^{2/3}, and the Riccati process spends most of its time close to the stable fixed point. A localized peak is created by the rare transition from +aL+\sqrt{a_L} to −aL-\sqrt{a_L}. As shown in Ref. [29], this transition yields X(x)≃−aLtanh⁡[aL(x−U)]X(x)\simeq-\sqrt{a_L}\tanh[\sqrt{a_L}(x-U)]. Its inverse Riccati transform gives

∣ψ(U+x)∣∣ψ(U)∣≃1cosh⁡(aLx).(30)\frac{|\psi(U+x)|}{|\psi(U)|}\simeq\frac{1}{\cosh(\sqrt{a_L}x)}. \tag*{(30)}

Thus each peak has exponential tails.

The same fixed-point picture gives the fluctuations around the exponential envelope. Writing Xa=a+yX_a=\sqrt{a}+y on a long stable segment and linearizing Eq. (29) yields

dy=−2a y dx+g dWx,(31)dy=-2\sqrt{a}\,y\,dx+g\,dW_x, \tag*{(31)}

an Ornstein–Uhlenbeck process. Consequently,

ln⁡∣ψ(x)∣≃−a∣x−U∣+χ(x),χ(x)=∫xy(s) ds.(32)\ln|\psi(x)|\simeq-\sqrt{a}|x-U|+\chi(x),\qquad\chi(x)=\int^x y(s)\,ds. \tag*{(32)}

Since yy is stationary with a finite correlation length, its integral is Gaussian and becomes Brownian on larger scales. Since the stationary columnar-KPZ profile is proportional to ln⁡∣ψ∣\ln|\psi|, these Brownian-like fluctuations produce the additional k−2k^{-2} contribution to its structure factor.

Finite-time competition and facet coarsening

At finite time, several eigenvectors contribute to Eq. (10). Each mode has a dominant peak at its own random center UnU_n; if NN modes remain relevant, their NN competing exponential envelopes produce a faceted profile, with slope changes where neighboring envelopes exchange dominance. For any pair Ek<EnE_k<E_n, the relative temporal weight of mode nn with respect to mode kk decays as (cn/ck)e−(En−Ek)t(c_n/c_k)e^{-(E_n-E_k)t}. Higher-energy modes and their localization centers therefore disappear progressively, and neighboring facets merge.

To estimate this coarsening, consider two relevant modes k<nk<n, centered at UkU_k and UnU_n and separated by R=∣Un−Uk∣R=|U_n-U_k|. For Gaussian white disorder, one has aR∼(log⁡R)2/3a_R\sim(\log R)^{2/3} and the eigenvector tails decay as e−aRRe^{-\sqrt{a_R}R}, while the characteristic gap between the competing modes is ΔER≡En−Ek∼aR−1/2\Delta E_R\equiv E_n-E_k\sim a_R^{-1/2} [29]. At UnU_n, the contribution of the lower-energy mode kk is spatially suppressed but gains temporal weight relative to mode nn. Neglecting nonuniversal prefactors and the ratio of initial overlaps ck/cnc_k/c_n, their ratio scales as

∣ckψk(Un)∣e−Ekt∣cnψn(Un)∣e−Ent∼exp⁡[−aRR+ΔERt].(33)\frac{\left|c_k\psi_k(U_n)\right|e^{-E_k t}}{\left|c_n\psi_n(U_n)\right|e^{-E_n t}} \sim\exp\left[-\sqrt{a_R}R+\Delta E_R t\right]. \tag*{(33)}

The center UnU_n disappears when the two contributions become comparable. This occurs after

tR∼aRRΔER∼aRR∼R(log⁡R)2/3.(34)t_R \sim\frac{\sqrt{a_R}R}{\Delta E_R} \sim a_R R \sim R(\log R)^{2/3}. \tag*{(34)}

Inverting this relation, with log⁡Rt∼log⁡t\log R_t\sim\log t at leading order, gives the typical spacing RtR_t between the centers that remain relevant at time tt, and hence their number:

Rt∼t(log⁡t)2/3,Nrel(t,L)∼max⁡{1,L(log⁡t)2/3t}.(35)R_t\sim\frac{t}{(\log t)^{2/3}},\qquad N_{\mathrm{rel}}(t,L)\sim\max\left\{1,\frac{L(\log t)^{2/3}}{t}\right\}. \tag*{(35)}

The number of macroscopic slope changes follows the same law, until only the principal localized mode remains. To quantify this coarsening, let Nflip(t,L)N_{\mathrm{flip}}(t,L) be the number of positions at which the slope changes sign. Since the final stationary profile contains one such sign change, we define

ρflip(t,L)≡⟨Nflip(t,L)⟩−1L≃Nrel(t,L)−1L∼(log⁡t)2/3t,Nrel(t,L)≫1.(36)\rho_{\mathrm{flip}}(t,L)\equiv\frac{\langle N_{\mathrm{flip}}(t,L)\rangle-1}{L}\simeq\frac{N_{\mathrm{rel}}(t,L)-1}{L}\sim\frac{(\log t)^{2/3}}{t},\qquad N_{\mathrm{rel}}(t,L)\gg1. \tag*{(36)}

In the simulations, the slope at site ii is estimated from the height difference between sites i−4i-4 and i+4i+4. Using points separated by eight lattice spacings prevents small-scale fluctuations from producing spurious sign changes. Figure 7(a) shows the time evolution of the interface and the progressive decrease in the number of peaks. Figure 7(b) shows the measured density, whose scaling is given by Eq. (36).

Figure 7. Finite-time coarsening in the one-dimensional columnar KPZ equation. (a) Interface profiles at times t=200t=200, 10001000, 50005000 for L=210L=2^{10}, vertically translated to display them in the same frame, illustrating the progressive reduction in the number of facets. (b) Decay of the density of slope flips for system sizes L=210,…,214L=2^{10},\ldots,2^{14}. The dashed line indicates the leading t−1t^{-1} term in Eq. (35).

Full slope-increment distribution of the anharmonic Larkin model

In the End Matter, we obtained the slope multiscaling spectrum by comparing typical intervals with rare intervals containing a small value of the Brownian field. Here we derive the full distribution of local slope increments and show how the same scaling follows from its tails. Starting from Eq. (16), we write

m(x)=fp[B(x)],fp(b)=sgn⁡(b)∣b∣p.m(x)=f_p[B(x)],\qquad f_p(b)=\operatorname{sgn}(b)|b|^p.

We take BB to be standard Brownian motion.

Calculation of the distribution

Consider a bulk point x∼Lx \sim L and a separation 1≪ℓ≪x1 \ll\ell\ll x, and define

X=B(x),Y=B(x+ℓ)−B(x),Δℓ=fp(X+Y)−fp(X).(37)X = B(x), \qquad Y = B(x+\ell)-B(x), \qquad\Delta_\ell= f_p(X+Y)-f_p(X). \tag*{(37)}

The variables XX and YY are independent Gaussians with variances xx and ℓ\ell, respectively. Let

gp(u)=fp−1(u)=sgn⁡(u)∣u∣1/p.g_p(u) = f_p^{-1}(u) = \operatorname{sgn}(u)|u|^{1/p}.

For fixed XX, the relation Δ=fp(X+Y)−fp(X)\Delta= f_p(X+Y)-f_p(X) has the unique solution

Y=gp[Δ+fp(X)]−X.Y = g_p[\Delta+f_p(X)]-X.

Changing variables from YY to Δ\Delta therefore gives the exact conditional density

Pℓ(Δ∣X)=∣Δ+fp(X)∣1/p−1p2πℓexp⁡[−{gp[Δ+fp(X)]−X}22ℓ].(38)P_\ell(\Delta|X) = \frac{|\Delta+f_p(X)|^{1/p-1}}{p\sqrt{2\pi\ell}}\exp\left[-\frac{\{g_p[\Delta+f_p(X)]-X\}^{2}}{2\ell}\right]. \tag*{(38)}

Averaging over XX yields the full increment distribution,

Pℓ(Δ;x)=∫−∞∞dX2πxe−X2/(2x)Pℓ(Δ∣X).(39)P_\ell(\Delta;x) = \int_{-\infty}^{\infty}\frac{dX}{\sqrt{2\pi x}}e^{-X^2/(2x)}P_\ell(\Delta|X). \tag*{(39)}

For r=ℓ/x≪1r=\ell/x\ll1, most intervals satisfy ∣Y∣≪∣X∣|Y|\ll|X|, so that

Δℓ≃fp′(X)Y=p∣X∣p−1Y.(40)\Delta_\ell\simeq f'_p(X)Y=p|X|^{p-1}Y. \tag*{(40)}

Conditioned on XX, the increment is then Gaussian with variance p2ℓ∣X∣2p−2p^2\ell|X|^{2p-2}. Introducing

Δtyp=pℓ1/2x(p−1)/2,y=∣Δ∣Δtyp,\Delta_{\mathrm{typ}}=p\ell^{1/2}x^{(p-1)/2}, \qquad y=\frac{|\Delta|}{\Delta_{\mathrm{typ}}},

and averaging over X=x1/2vX=x^{1/2}v gives

Pℓ(Δ;x)≃1ΔtypFp(y),Fp(y)=1π∫0∞dv v1−pexp⁡[−v22−y2v2(1−p)2].(41)P_\ell(\Delta;x)\simeq\frac{1}{\Delta_{\mathrm{typ}}}F_p(y), \qquad F_p(y)=\frac{1}{\pi}\int_0^\infty dv\,v^{1-p}\exp\left[-\frac{v^2}{2}-\frac{y^2v^{2(1-p)}}{2}\right]. \tag*{(41)}

For p<1p<1 and y≫1y\gg1, Eq. (41) is dominated by v≪1v\ll1, corresponding to intervals close to a zero of BB. Dropping e−v2/2e^{-v^2/2} and setting z=y2v2(1−p)/2z=y^2v^{2(1-p)}/2 gives

Fp(y)∼Cpy−(2−p)/(1−p),Cp=2(2−p)/[2(1−p)]2π(1−p)Γ(2−p2(1−p)).(42)F_p(y)\sim C_p y^{-(2-p)/(1-p)}, \qquad C_p=\frac{2^{(2-p)/[2(1-p)]}}{2\pi(1-p)}\Gamma\left(\frac{2-p}{2(1-p)}\right). \tag*{(42)}

The linear approximation requires ∣X∣≫ℓ1/2|X|\gg\ell^{1/2}, so this algebraic tail ends at

ymax⁡∼(xℓ)(1−p)/2,Δmax⁡∼ℓp/2.(43)y_{\max}\sim\left(\frac{x}{\ell}\right)^{(1-p)/2}, \qquad\Delta_{\max}\sim\ell^{p/2}. \tag*{(43)}

Thus the distribution consists of a typical core of width Δtyp\Delta_{\mathrm{typ}} and a broad algebraic sector terminated by the zero-crossing scale Δmax⁡\Delta_{\max}. Taking moments of this distribution recovers Eq. (26): moments below qc=1/(1−p)q_c=1/(1-p) probe the core, whereas moments above qcq_c probe the upper cutoff.

For p>1p>1, the large-yy behavior of Eq. (41) follows from a saddle at

v∗=(p−1)1/(2p)y1/p.v_*=(p-1)^{1/(2p)}y^{1/p}.

Laplace’s method gives

Fp(y)∼Apy−(p−1)/pexp⁡[−cpy2/p],cp=p2(p−1)1/p−1,(44)F_p(y)\sim A_p y^{-(p-1)/p}\exp\left[-c_p y^{2/p}\right], \qquad c_p=\frac{p}{2}(p-1)^{1/p-1}, \tag*{(44)}

where ApA_p depends only on pp. The stretched-exponential tail leaves all positive moments controlled by the typical scale Δtyp\Delta_{\mathrm{typ}}, recovering ζm(q)=1/2\zeta_m(q)=1/2 for every q>0q>0; hence there is no multiscaling for p>1p>1.

References

  1. [5]J. M. López, Scaling approach to calculate critical exponents in anomalous surface roughening, Physical Review Letters 83, 4594 (1999).DOI
  2. [6]M. A. Auger, L. Vázquez, R. Cuerno, M. Castro, M. Jergel, and O. Sánchez, Intrinsic anomalous surface roughening of TiN films deposited by reactive sputtering, Physical Review B 73, 045436 (2006).DOI
  3. [7]A. S. Balankin, O. S. Huerta, R. C. M. de Oca, D. S. Ochoa, J. M. Trinidad, and M. A. Mendoza, Intrinsically anomalous roughness of randomly crumpled thin sheets, Physical Review E 74, 061602 (2006).
  4. [8]J. M. López and J. Schmittbuhl, Anomalous scaling of fracture surfaces, Physical Review E 57, 6405 (1998).DOI
  5. [9]J. Soriano, J. Ramasco, M. Rodríguez, A. Hernández-Machado, and J. Ortín, Anomalous roughening of Hele-Shaw flows with quenched disorder, Physical Review Letters 89, 026102 (2002).DOI
  6. [10]M. M. Bernard, P. Le Doussal, A. Rosso, and C. Texier, Anomalous scaling of heterogeneous elastic lines: A new picture from sample-to-sample fluctuations, Physical Review E 110, 011404 (2024).DOI
  7. [11]M. Bernard, A. A. Fedorenko, P. Le Doussal, and A. Rosso, Stochastic porous-medium equation in one dimension, Physical Review E 112, L043501 (2025).DOI
  8. [12]M. Bernard, Fluctuations and multifractality in stochastic models of interface growth and population dynamics, arXiv preprint arXiv:2609.34468 (2026).arxiv.org/abs/2609.34468
  9. [13]H. Leschhorn and L.-H. Tang, Comment on “elastic string in a random potential”, Phys. Rev. Lett. 70, 2973 (1993).pubmed.ncbi.nlm.nih.gov/10053701
  10. [14]A. Rosso and W. Krauth, Origin of the roughness exponent in elastic strings at the depinning threshold, Physical Review Letters 87, 187002 (2001).
  11. [15]J. G. Amar, P.-M. Lam, and F. Family, Groove instabilities in surface growth with diffusion, Physical Review E 47, 3242 (1993).pubmed.ncbi.nlm.nih.gov/9960376
  12. [16]S. D. Sarma, S. Ghaisas, and J. Kim, Kinetic super-roughening and anomalous dynamic scaling in nonequilibrium growth models, Physical Review E 49, 122 (1994).
  13. [17]K. Sneppen, Self-organized pinning and interface growth in a random medium, Physical Review Letters 69, 3539 (1992).pubmed.ncbi.nlm.nih.gov/10046847
  14. [18]I. G. Szendro, J. M. López, and M. A. Rodríguez, Localization in disordered media, anomalous roughening, and coarsening dynamics of faceted surfaces, Physical Review E 76, 011603 (2007).pubmed.ncbi.nlm.nih.gov/17677460
  15. [19]A. Alés and J. M. López, Faceted patterns and anomalous surface roughening driven by long-range temporally correlated noise, Physical Review E 99, 062139 (2019).DOI
  16. [20]V. H. Purrello, J. L. Iguain, and A. B. Kolton, Roughening of the anharmonic Larkin model, Physical Review E 99, 032105 (2019).DOI
  17. [21]A. Alés and J. M. López, Roughening of the anharmonic elastic interface in correlated random media, Physical Review E 104, 044108 (2021).DOI
  18. [22]P. Córdoba-Torres, T. J. Mesquita, I. N. Bastos, and R. P. Nogueira, Complex dynamics during metal dissolution: From intrinsic to faceted anomalous scaling, Physical Review Letters 102, 055504 (2009).
  19. [23]F. S. Nascimento, S. C. Ferreira, and S. O. Ferreira, Faceted anomalous scaling in the epitaxial growth of semiconductor films, EPL (Europhysics Letters) 94, 68002 (2011).DOI
  20. [24]J. M. López, M. Castro, and R. Gallego, Scaling of local slopes, conservation laws, and anomalous roughening in surface growth, Physical Review Letters 94, 166103 (2005).DOI
  21. [25]A. Alés and J. M. López, Surface super-roughening driven by spatiotemporally correlated noise, Journal of Statistical Mechanics: Theory and Experiment 2020, 033210 (2020).DOI
  22. [26]T. Nattermann and W. Renz, Diffusion in a random catalytic environment, polymers in random media, and stochastically growing interfaces, Physical Review A 40, 4675 (1989).pubmed.ncbi.nlm.nih.gov/9902712
  23. [27]J. Krug and T. Halpin-Healy, Directed polymers in the presence of columnar disorder, Journal de Physique I 3, 2179 (1993).DOI
  24. [28]C. Texier, Individual energy level distributions for one-dimensional diagonal and off-diagonal disorder, Journal of Physics A: Mathematical and General 33, 6095 (2000).
  25. [29]L. Dumaz and C. Labbé, Localization of the continuous Anderson hamiltonian in 1-D, Probability Theory and Related Fields 176, 353 (2020).

Paper details

vlnc:W_2sd3r89v
Version
v2 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 1

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents