arXiv subjects

Sergio Chibbaro

Publications and source records attributed to Sergio Chibbaro.

At least 19 recordsLinked to original sources

arXiv2026

Neural Renormalization Group Flow for Percolation

Machine learning offers a possible route to data-driven real-space renormalization when the relevant observables are nonlocal and difficult to prescribe explicitly. We explore this idea for two-dimensional site percolation developping a supervised, scale-shared neural architecture. The model recursively applies the same learned coarse-graining rule across scales, producing a latent field from which the crossing probability is predicted, while a corresponding fine-graining decoder reconstructs the largest-cluster mask. Trained only on small lattices, the model extrapolates to substantially larger systems, recovers the spanning cluster with high fidelity, and produces observables obeying the expected finite-size scaling near the critical point. We observe that to get such performance it is key that the learned latent representation exhibits critical fluctuations and scale-dependent flows consistent with the renormalization-group structure of percolation.

cond-mat.dis-nn↗

arXiv2026

Intermittency in Shell Models of Turbulent Cascades: from Single-Branch to Multi-Branch

Intermittency is one of the central features of turbulent transfer: the multi-scale energy cascade is mediated by rare and intense fluctuations. We investigate this phenomenon in a multi-branch shell model, which combines quasi-local triadic nonlinear interactions with a branching structure that mimics the growth of degrees of freedom toward small scales. Comparison with the standard Sabra model shows that branching enhances intermittency, as measured by anomalous scaling exponents of energy-flux structure functions. We further use multiplier statistics and large deviation estimates to characterize the multiplicative nature of the cascade. Our results suggest that reduced descriptions of turbulent intermittency should retain both nonlinear dynamics and geometrical organization. Implications on Navier-Stokes turbulence are discussed.

physics.flu-dyn↗

arXiv2026

Sub-Kolmogorov Intermittency and Multifractal Dissipation in Multiphase Turbulence

Multiphase turbulence displays stronger intermittency than its single-phase counterpart, yet the origin and geometrical organization of its most intense small-scale fluctuations remain poorly understood. Using direct numerical simulations of the incompressible Navier--Stokes equations with surface tension, we show that the local dissipative cutoff broadens strongly in the presence of interfaces, with dissipative events extending deep into the sub-Kolmogorov range. These events are spatially concentrated around topology-changing interfacial regions, namely breakup and coalescence. A multifractal analysis of the dissipation field further reveals that, while the spectrum above the Kolmogorov length, $η_K$, remains close to the single-phase case except for the most singular tail, the near- and sub-Kolmogorov range develops a markedly broader singularity spectrum supported on sparse intense structures. Our results show that breakup and coalescence do not simply perturb turbulence locally, but imprint a distinct multifractal organization on dissipation in multiphase turbulence.

physics.flu-dyn↗

arXiv2026

Resonant interactions in the $α$-FPUT lattice with site-dependent coefficients

The wave turbulence framework has proven to be an effective tool for analyzing certain features of nonlinear energy transfer in one-dimensional nonlinear chains. In this work, we extend this approach to the $α$-FPUT problem when the spring stiffness $χ$ and the nonlinear coefficient $α$ are site-dependent. Although three-wave interactions are non-resonant for constant coefficients, their spatial modulation gives rise to a non-trivial resonant manifold. In this framework, we derive a new kinetic equation that suggests the possibility of substantially faster thermalization with respect to the constant coefficient case. The new kinetic equation includes also an extra term that can be associated to the Bragg-scattering mechanism, which promotes the isotropization of the wave-action spectral density function.

cond-mat.stat-mech↗

arXiv2026

Two-point enstrophy dynamics in homogeneous isotropic turbulence

In the present work we investigate the multiscale dynamics of enstrophy in homogeneous isotropic turbulence by exploiting the two-point formalism provided by the Kármán-Howarth-Monin-Hill approach. The study is conducted on direct numerical simulations with a Taylor-based Reynolds number in the range of $140 \lesssim Re_λ \lesssim 400$. The two-point enstrophy budget at scales $r > 10 η$ appears to be entirely determined by production via vortex stretching, which balances enstrophy destruction, and to be dominated by the diffusive transport at smaller scales, thus preventing the emergence of a range dominated by the inertial transport of enstrophy. The decomposition in longitudinal and transverse contributions also highlights a dual nature of the inertial enstrophy flux. In particular, enstrophy appears to be transferred across scales through a non-trivial combination of direct and reverse interscale transfer. It is shown that the dual nature of this transfer is strictly related to the vortex stretching mechanism, which, in addition to producing enstrophy through vorticity amplification, also transfers longitudinal vorticity towards larger scales (by stretching the vortical elements) and transverse vorticity towards smaller scales (by contracting these vortical elements in the radial direction). The sum of these two contributions results in an overall transfer of enstrophy from large towards small scales. We propose the use of the pressure transport term as a proxy to obtain some information on the dynamics of relevant events of inertial energy and enstrophy transport. The new findings highlight the relevance of inertial compression events in longitudinal energy transport. At the same time, a good correlation between transverse energy transport events and the radial contraction of vortical elements due to vortex stretching mechanisms is also found.

physics.flu-dyn↗

arXiv2026

Turbulent pair dispersion with Stochastic Generative Diffusion Models

Recent advances in data-driven modeling have shown that diffusion models can successfully generate synthetic Lagrangian trajectories in turbulent flows. Building on this progress, we extend the method to the joint generation of pairs of Lagrangian velocity trajectories, enabling a fully data-driven representation of turbulent pair dispersion, a long-standing fundamental problem with broad relevance in fluid dynamics. We demonstrate that diffusion models accurately reproduce the evolution of particle-pair separation, including deviations from Richardson's classical scaling law, while simultaneously preserving all key single-particle statistical properties reported in previous studies. These findings underscore the potential of diffusion-based generative models to emulate high-dimensional, multi-scale turbulent dynamics, further establishing them as a powerful tool for scientific modeling and for future geophysical and astrophysical applications.

physics.flu-dyn↗

arXiv2026

Multi-branch Shell Models of Two-Dimensional Turbulence exhibit Dual Energy-Enstrophy Cascades

Classical shell models of turbulence do not display dual cascade - inverse of energy and direct of enstrophy - because they fail to reproduce the right thermal spectra. We propose here a multi-branch shell model, including a geometry hierarchically organized across scales, in order to overcome this limitation. For this model, we demonstrate numerically both the agreement of the thermal spectra with those of two-dimensional fluid equations and the emergence of a statistically stationary dual cascade. This construction also allows us to study local transfers and to investigate both self-similarity and non-Gaussianity.

physics.flu-dyn↗

arXiv2026

Ensemble of Fixed Points in Multi-branch Shell Models of Turbulent Cascades

Stationary solutions of a shell model of turbulence defined on a dyadic tree topology are studied. Each node's amplitude is expressed as the product of amplitude multipliers associated with its ancestors, providing a recursive representation of the cascade process. A geometrical rule governs the tree growth, and we prove the existence of a continuum of fixed points, including the Kolmogorov solution, that sustain a strictly forward energy cascade. Sampling along randomly chosen branches defines a homogeneous Markov chain, enabling a stochastic characterization of extended self-similarity and intermittency through the spectral properties of the associated Feynman-Kac operators. Numerical simulations confirm the theoretical predictions, showing that multi-branch shell models offer a minimal yet physically rich framework for exploring the complexity of nonlinear energy transfer across scales.

physics.flu-dyn↗

arXiv2025

Building causation links in stochastic nonlinear systems from data

Causal relationships play a fundamental role in understanding the world around us. The ability to identify and understand cause-effect relationships is critical to making informed decisions, predicting outcomes, and developing effective strategies. However, deciphering causal relationships from observational data is a difficult task, as correlations alone may not provide definitive evidence of causality. In recent years, the field of machine learning (ML) has emerged as a powerful tool, offering new opportunities for uncovering hidden causal mechanisms and better understanding complex systems. In this work, we address the issue of detecting the intrinsic causal links of a large class of complex systems in the framework of the response theory in physics. We develop some theoretical ideas put forward by [1], and technically we use state-of-the-art ML techniques to build up models from data. We consider both linear stochastic and non-linear systems. Finally, we compute the asymptotic efficiency of the linear response based causal predictor in a case of large scale Markov process network of linear interactions.

cond-mat.stat-mech↗

arXiv2025

Fluctuations around Turbulence Models

Numerical simulations of turbulent flows at realistic Reynolds numbers generally rely on filtering out small scales from the Navier Stokes equations and modeling their impact through the Reynolds stress tensor $τ_{ij}$. Traditional models approximate $τ_{ij}$ solely as a function of the filtered velocity gradient, leading to deterministic subgrid scale closures. However, small scale fluctuations can locally exhibit instantaneous values whose deviation from the mean can have a significant influence on flow dynamics. In this work, we investigate these effects by employing direct numerical simulations combined with Gaussian filtering to quantify subgrid scale effects and evaluating the local energy flux in both space and time. The mean performance of the canonical Clark model is assessed by conditioning the energy flux distributions on the invariants of the filtered velocity gradient tensor, $Q$ and $R$. The Clark model captures to a good degree the mean energy flux. However, the fluctuations around these mean values for given ($Q,R$) are of the order of the mean displaying fat tailed distributions. To become more precise, we examine the joint distributions of true energy flux and the predictions from both the Clark and the Smagorinsky models. This approach mirrors the strategy adopted in early stochastic subgrid scale models. Clear non Gaussian characteristics emerge from the obtained distributions, particularly through the appearance of heavy tails. The mean, the variance, the skewness and flatness of these distributions are quantified. Our results emphasize that fluctuations are an integral component of the small scale feedback onto large scale dynamics and should be incorporated into subgrid scale modeling through an appropriate stochastic framework.

physics.flu-dyn↗

arXiv2024

Wave Turbulence and thermalization in one-dimensional chains

One-dimensional chains are used as a fundamental model of condensed matter, and have constituted the starting point for key developments in nonlinear physics and complex systems. The pioneering work in this field was proposed by Fermi, Pasta, Ulam and Tsingou in the 50s in Los Alamos. An intense and fruitful mathematical and physical research followed during these last 70 years. Recently, a fresh look at the mechanisms of thermalization in such systems has been provided through the lens of the Wave Turbulence approach. In this review, we give a critical summary of the results obtained in this framework. We also present a series of open problems and challenges that future work needs to address.

nlin.CD↗

arXiv2024

How small droplets form in turbulent multiphase flows

The formation of small droplets and bubbles in turbulent flows is a crucial process in geophysics and engineering, whose underlying physical mechanism remains a puzzle. In this letter, we address this problem by means of high-resolution numerical simulations, comparing a realistic multiphase configuration with a numerical experiment in which we attenuate the presence of strong velocity gradients either across the whole mixture or in the disperse phase only. Our results show unambiguously that the formation of small droplets is governed by the internal dynamics which occur during the break-up of large drops and that the high vorticity and the extreme dissipation associated to these events are the consequence and not the cause of the breakup.

physics.flu-dyn↗

arXiv2023

Fluctuation Relations at Large Scales in Three-Dimensional Hydrodynamic Turbulence

It has long been conjectured that, in three dimensional turbulence, velocity modes at scales larger than the forcing scale follow equilibrium dynamics. Recent numerical and experimental evidence show that such modes share the same mean energy and therefore support this claim, but equilibrium dynamics does not reduce to equipartition of energy. In this work, a large set of direct numerical simulations is carried out to investigate if fluctuation-dissipation relations and the fluctuation theorem also apply at these scales. These two results link out-of-equilibrium properties of a forced system with its behavior at equilibrium. Both relations are verified quantitatively by the results of our simulations, further supporting that large scale modes display equilibrium dynamics. They provide new tools to characterize both the mean value and the fluctuations of the injected energy by a large scale force acting on turbulence driven by small scale random noise.

physics.flu-dyn↗

arXiv2023

Curriculum learning for data-driven modeling of dynamical systems

The reliable prediction of the temporal behavior of complex systems is key in numerous scientific fields. This strong interest is however hindered by modeling issues: often, the governing equations describing the physics of the system under consideration are not accessible or, if known, their solution might require a computational time incompatible with the prediction time constraints. Not surprisingly, approximating complex systems in a generic functional format and informing it ex-nihilo from available observations has become common practice in the age of machine learning, as illustrated by the numerous successful examples based on deep neural networks. However, generalizability of the models, margins of guarantee and the impact of data are often overlooked or examined mainly by relying on prior knowledge of the physics. We tackle these issues from a different viewpoint, by adopting a curriculum learning strategy. In curriculum learning, the dataset is structured such that the training process starts from simple samples towards more complex ones in order to favor convergence and generalization. The concept has been developed and successfully applied in robotics and control of systems. Here, we apply this concept for the learning of complex dynamical systems in a systematic way. First, leveraging insights from the ergodic theory, we assess the amount of data sufficient for a-priori guaranteeing a faithful model of the physical system and thoroughly investigate the impact of the training set and its structure on the quality of long-term predictions. Based on that, we consider entropy as a metric of complexity of the dataset; we show how an informed design of the training set based on the analysis of the entropy significantly improves the resulting models in terms of generalizability, and provide insights on the amount and the choice of data required for an effective data-driven modeling.

cs.LG↗

arXiv2023

Heat-flux Fluctuations reveals regime transitions in Rayleigh-Bénard convection

The study of the transitions among different regimes in thermal convection has been an issue of paramount importance in fluid mechanics. While the bifurcations at low Rayleigh number, when the flow is laminar or moderately chaotic, have been fully understood for a long time, transitions at higher Rayleigh number are much more difficult to be clearly identified. Here, through a numerical study of the two-dimensional Rayleigh-Bénard convection covering four decades in Rayleigh number for two different Prandtl numbers, we find a clear-cut transition by considering the fluctuations of the heat flux through a horizontal plane, rather than its mean value. More specifically, we have found that this sharp transition is displayed by a jump of the ratio of the root-mean-square fluctuations of the heat flux to its mean value and occurs at $Ra/Pr \approx 10^9$. Above the transition, this ratio is found to be constant in all regions of the flow, while taking different values in the bulk and at the boundaries. Below the transition instead, different behaviors are observed at the boundaries and in the bulk: at the boundaries, this ratio decreases with respect to the Rayleigh number whereas it is found to be constant in the bulk for all values of the Rayleigh number. Through this numerical evidence and an analytical reasoning we confirm what was already observed in experiments, that is the decrease of the ratio of root-mean-square fluctuations of the heat flux to its mean value, observed at the boundaries below the transition, can understood in terms of the law of large numbers.

physics.flu-dyn↗

arXiv2023

Intermittency in turbulent emulsions

We investigate the statistics of turbulence in emulsions of two-immiscible fluids of same density. We compute for the first time velocity increments between points conditioned to be located in the same phase or in different phases and examine their probability density functions (PDF) and the associated structure functions (SF). This enables us to demonstrate that the the presence of the interface reduces the skewness of the PDF at scales below the Kolmogorov-Hinze scale and therefore the magnitude of the energy flux towards the dissipative scales, which is quantified by the third-order SF. The analysis of the higher order SFs shows that multiphase turbulence is more intermittent than single-phase turbulence. In particular, the local scaling exponents of the SFs display a saturation about the Kolmogorov-Hinze scale and below, which indicates the presence of large velocity gradients across the interface. Interestingly, the statistics approach of classic homogeneous isotropic turbulence when significantly increasing the viscosity of the dispersed phase.

physics.flu-dyn↗

arXiv2022

Local fluxes in MHD Turbulence

Using highly resolved direct numerical simulations we examine the statistical properties of the local energy flux rate $Π_\ell(x)$ towards small scales for three isotropic turbulent magnetohydrodynamic flows, which differ in strength and structure of the magnetic field. We analyse the cascade process both in the kinetic and magnetic energy, disentangling the different flux contributions to the overall energy dynamics. The results show that the probability distribution of the local energy flux develops long tails related to extreme events, similar to the hydrodynamic case. The different terms of the energy flux display different properties and show sensitivity on the type of the flow examined. We further examine the joint pdf between the local energy flux and the gradients of the involved fields. The results point out a correlation with the magnetic field gradients, showing however a dispersion much stronger than what is observed in hydrodynamic flows. Finally, it is also shown that the local energy flux shows some dependence on the local amplitude of the magnetic field. The present results have implications for subgrid scale models that we discuss.

physics.flu-dyn↗

arXiv2022

The interaction of droplet dynamics and turbulence cascade

The dynamics of droplet fragmentation in turbulence is described in the Kolmogorov-Hinze framework. Yet, a quantitative theory is lacking at higher concentrations when strong interactions between the phases and coalescence become relevant, which is common in most flows. Here, we address this issue through a fully-coupled numerical study of the droplet dynamics in a turbulent flow at high Reynolds number. By means of time-space spectral statistics, not currently accessible to experiments, we demonstrate that the characteristic scale of the process, the Hinze scale, can be precisely identified as the scale at which the net energy exchange due to capillarity is zero. Droplets larger than this scale preferentially break up absorbing energy from the flow; smaller droplets, instead, undergo rapid oscillations and tend to coalesce releasing energy to the flow. Further, we link the droplet-size-distribution with the probability distribution of the turbulent dissipation. This shows that key in the fragmentation process is the local flux of energy which dominates the process at large scales, vindicating its locality.

physics.flu-dyn↗