https://arxiv.org/api/tRG0C1xf53MZcnWm8spcte9SqT8 2026-07-20T20:14:52Z 34617 135 15 http://arxiv.org/abs/2607.07925v1 2026-07-08T21:16:10Z

Turbulence in the solar interior and atmosphere plays a crucial role in energy transport, yet modeling its subgrid-scale effects remains a major challenge. This study leverages machine learning (ML) models to predict components of the Reynolds stress tensor using high-resolution StellarBox simulations of the quiet Sun. Previously, we have compared a Multi-Layer Perceptron (MLP) and a 3D Convolutional Neural Network (CNN) against physics-based baselines to achieve a lower Mean Squared Error (MSE) and better generalization across various heights and depths in the solar atmosphere. To enhance learning, in this work, we investigate cluster-weighted training using K-Means and Hierarchical Agglomerative Clustering (HAC). By weighing the loss function based on cluster-specific prediction errors, we direct the model's attention to high-error regions. It significantly improves CNN performance, achieving 34% lower MSE and a significantly higher R2 score indicating that integrating deterministic clustering with ML is a promising technique for modeling subgrid turbulence, in particular, and regression in diverse environments, in general.

2026-07-08T21:16:10Z 9 pages, 8 figures, 2 tables, SABID workshop at 2025 IEEE International Conference on Data Mining Rimsha Hameed Syeda Dustin Kempton Viacheslav Sadykov Irina Kitiashvili Rafal Angryk 10.1109/ICDMW69685.2025.00228 http://arxiv.org/abs/2607.07654v1 2026-07-08T17:09:45Z

The paper establishes extra invariant in magnetohydrodynamics (MHD) of rotating and stratified fluid layer. This invariant is conserved adiabatically, i.e. approximately over long time. The existence of the invariant is interesting by itself, as such invariants are extremely rare. However, in addition, this invariant appears to be connected to the famous dynamo phenomenon.

2026-07-08T17:09:45Z 5 pages, 1 figure Alexander M. Balk http://arxiv.org/abs/2603.17057v2 2026-07-08T16:25:43Z

Active multi-fidelity surrogate modeling is developed for multi-condition airfoil shape optimization to reduce high-fidelity CFD cost while retaining RANS-consistent aerodynamic metrics. The framework couples a low-fidelity-informed Gaussian process regression transfer model with uncertainty-triggered sampling and a synchronized elitism rule embedded in a hybrid genetic algorithm. Low-fidelity XFOIL evaluations provide inexpensive features, while sparse RANS simulations are adaptively allocated when predictive uncertainty exceeds a threshold; elite candidates are mandatorily validated at high fidelity, and the population is re-evaluated to prevent evolutionary selection based on outdated fitness values produced by earlier surrogate states. The method is demonstrated for a two-point problem at $Re=6\times10^6$ with cruise at $α=2^\circ$ (maximize $E=L/D$) and take-off at $α=10^\circ$ (maximize $C_L$) using a 12-parameter CST representation. Independent multi-fidelity surrogates per flight condition enable decoupled refinement. The optimized design improves cruise efficiency by 41.05% and take-off lift by 20.75% relative to the best first-generation individual. Over the full campaign, RANS evaluations were required for only 14.78% and 9.5% of the condition-specific candidate evaluations at cruise and take-off, respectively. These percentages quantify the reduction in high-fidelity usage relative to the fixed automated RANS workflow adopted as the high-fidelity reference in this study.

2026-03-17T18:43:11Z 24 pages, 18 figures Isaac Robledo Alberto Vilariño Arnau Miró Oriol Lehmkuhl Carlos Sanmiguel Vila Rodrigo Castellanos http://arxiv.org/abs/2606.25075v3 2026-07-08T16:11:03Z

Computing the viscous Stokes flow around a shape requires solving a boundary-integral equation, and for a new shape the solve must begin from scratch. Learned operators promise to spread this cost across shapes, but it is not clear what such an operator retains of the solver it replaces, or what determines whether it transfers to shapes it was not trained on. We make both questions answerable by choosing a problem which is exactly solvable except for a single term: a second-kind boundary-integral problem is solved exactly using a kernel-independent fast summation, while the boundary correction, which has no closed form, is learned. The solver's guarantees carry over unchanged: exactness on the closed-form part, $O(N)$ scaling, $SO(3)$-equivariance to machine precision, and an $O(N)$ differentiable adjoint. We then make precise what the learning contributes. It does not contribute to accuracy, differentiability, or scaling $O(N)$, all of which are provided by the solver. Learning contributes a one-time cost in that the forward map is trained once and then evaluated on a new shape in a single pass rather than resolved. Measured against baselines, the learned map is $5$ to $16\times$ more data-efficient than a black-box DeepONet and maintains a $\sim\!2.5\times$ lower in-distribution error than a geometry-aware operator, although that operator is stronger in the low-data limit and the learned map is less reliable out of distribution. We trace this fragility to the global parameterization and reduce it with a local equivariant kernel. The exactly-solvable setting yields clarity about the mechanism: geometric generalization is governed by invariance and coverage, not by conditioning or by capacity.

2026-06-23T18:32:42Z 24 pages, 6 figures, 10 tables Denis Gueyffier Direction Scientifique Générale, ONERA, Institut Polytechnique de Paris, Palaiseau, France http://arxiv.org/abs/2607.07535v1 2026-07-08T15:33:50Z

Wave turbulence is a regime of interacting nonlinear waves occurring in most physical systems. Kelvin waves are helical distortions that propagate along vortex filaments and are believed to play a central role in quantum turbulence up to atmospheric vortices. Yet, Kelvin wave turbulence has remained inaccessible to direct experimental observation. Here, we report the first direct experimental observation of Kelvin-wave turbulence along a single vortex filament in a classical fluid under controlled conditions. Using high-resolution spatiotemporal measurements, we resolve Kelvin-wave dynamics over a broad range of scales and obtain wave-amplitude spectra consistent with the predicted weak-turbulence cascade. We identify six-wave resonant interactions as the mechanism driving this energy transfer, providing direct experimental support for a long-standing prediction of weak-turbulence theory. These results establish an experimental platform for investigating energy transport along vortex filaments, with broader implications for both classical and quantum turbulent systems.

2026-07-08T15:33:50Z in press in Physical Review Letters Jason Barckicke Christophe Gissinger Eric Falcon 10.1103/t3bt-m431 http://arxiv.org/abs/2607.07443v1 2026-07-08T14:14:10Z

Diffuse-interface immersed boundary methods (DIBMs) provide a simple and robust approach for simulating flows involving complex geometries. However, their inherent diffusion effect can contaminate the near-wall flow field and significantly degrade wall-shear-stress prediction in wall-modeled large-eddy simulation (WMLES). To address this limitation, we develop a WMLES approach based on a fully one-sided diffuse-interface immersed boundary method (FODIBM). By performing interpolation and spreading exclusively inside the immersed body, the proposed method removes the cross-boundary diffusion effect that adversely affects wall modeling in conventional DIBMs. A wall-shear-stress enforcement strategy is developed by coupling the wall-parallel immersed-boundary forcing with the wall shear stress predicted by an explicit wall model. In addition, a tau-model based on the modeled turbulent shear-stress tensor is introduced to preserve the total shear-stress balance below the reference height. The method is first validated in high-Reynolds-number turbulent channel flows, showing good agreement with DNS data for the mean velocity, Reynolds shear stress, and skin-friction coefficient. Sensitivity studies with respect to grid resolution, reference height, wall inclination angle, and Reynolds number demonstrate the robustness of the method. Compared with the conventional DIBM, the proposed method substantially improves the overall prediction accuracy, particularly at low reference heights. The approach is further assessed for turbulent flow over a NACA23012 airfoil, where the predicted pressure distribution and lift coefficient agree well with experimental data.

2026-07-08T14:14:10Z Qian Mao Yoshiharu Tamaki Song Zhao Jérôme Jacob Pierre Boivin Julien Favier http://arxiv.org/abs/2607.07432v1 2026-07-08T14:01:32Z

Although one-sided spreading has been shown to improve the near-wall accuracy of diffuse-interface immersed boundary methods (DIBMs), the effect of its asymmetric kernel support on the effective boundary location remains insufficiently understood. In this work, a detailed analysis of the one-sided spreading operator reveals an inward displacement of the effective boundary relative to the geometric boundary. To compensate for this displacement, a target-value reconstruction strategy is developed to ensure consistency between the values imposed at the effective boundary and the prescribed conditions at the geometric boundary. The strategy is incorporated into the fully one-sided diffuse-interface immersed boundary method (FODIBM) and applies to both Dirichlet and Neumann boundary conditions. Although confined to the target-value evaluation step, the modification substantially improves boundary-condition enforcement with negligible additional computational cost. Coupled with a hybrid lattice Boltzmann solver, the improved method consistently reduces L_2 and L_{\infty} error norms across different grid resolutions while retaining approximately second-order grid convergence. The no-slip and isothermal boundary-condition errors are reduced by 77% and 85%, respectively. Simulations involving various two- and three-dimensional geometries further show improved predictions relative to both the conventional DIBM and the original FODIBM. The results agree well with body-fitted reference solutions and experimental data, demonstrating accurate and computationally efficient simulations of compressible flows around complex geometries.

2026-07-08T14:01:32Z Qian Mao Song Zhao Pierre Boivin Julien Favier http://arxiv.org/abs/2603.26169v3 2026-07-08T13:34:29Z

We derive a nonlinear diffusion model for field scale CO$_2$ plume spreading from a Global Buckley--Leverett component balance. The reduced variable $u$ is the vertically averaged mobile gas phase CO$_2$ content normalized by its maximum column value; under vertical segregation, $u=h/H$, where $h$ is plume thickness and $H$ is aquifer thickness. The resulting equation is a nonhomogeneous porous medium type equation in which nonlinear lateral spreading is coupled to source/sink terms for injection, dissolution, mineral fixation, and retention. Using the nonlinear diffusivity $D_u(u)\simeq D_0u^{1-q}$, we analyze Barenblatt-type profiles with prescribed mobile mass and a capped plume constrained by $0\le u\le1$. The capped solution contains a ful-thickness core of radius $a(t)$ and a compact plume edge $R(t)$. Constant net mobile injection can sustain the core and gives square-root growth of $R(t)$, whereas shut-in or weak mobile addition causes the core to shrink and disappear. We compare these regimes with equivalent radii from time lapse seismic plume maps at Sleipner, Aquistore, and Weyburn--Midale. The data distinguish injection controlled growth, delayed layer filling, and tail dominated redistribution, but do not determine a unique nonlinear exponent. The model provides an analytical reference for interpreting plume footprint evolution while separating cumulative injected CO$_2$ from mobile gas phase CO$_2$.

2026-03-27T08:37:12Z Fernando Alonso-Marroquin Christian Tantardini http://arxiv.org/abs/2607.07338v1 2026-07-08T12:27:12Z

Nonlinear dynamics is ubiquitous in nature, ranging from chemical pattern formation to ocean circulation, yet its simulation on quantum computers is fundamentally limited by the unitary nature of quantum evolution. We propose the quantum Koopman method, a data-driven framework that embeds nonlinear dynamics into a learned linear representation and implements the resulting evolution using shallow quantum circuits. This method learns Koopman observables from trajectory data, projects the lifted dynamics onto a finite-dimensional subspace, and decomposes the corresponding non-unitary propagator into parallel spectral channels. We utilize the Koopman method on a superconducting processor to simulate three distinct nonlinear systems, comprising reaction-diffusion dynamics, fluid motion on a sphere, and satellite-derived observations of Gulf Stream currents, employing up to 32 parallel circuits of 10 qubits. These quantum simulations capture the dominant multiscale patterns and statistical signatures of the underlying dynamics, and reveal a transition from performance limited by hardware noise in weakly nonlinear systems to performance limited by finite-dimensional Koopman representations as nonlinear scale interactions increase. This transition identifies a practical boundary for quantum-amenable nonlinear dynamics, establishing a hardware-validated route for simulating moderately nonlinear dynamics on near-term quantum hardware.

2026-07-08T12:27:12Z Baoyang Zhang Dong An Zhaoyuan Meng Yefei Yu Xiaoxiao Xiao Zhen Lu Yue Yang http://arxiv.org/abs/2607.07325v1 2026-07-08T12:14:01Z

We study a bulk-driven nonlinear variant of the Kipnis-Marchioro-Presutti model of stochastic energy diffusion in which local collisions are biased to induce a net energy flow, resembling the effect of an external field. Starting from the microscopic master equation, we derive the hydrodynamic description of the driven system via a local equilibrium approximation, obtaining explicit expressions for the energy current and the associated diffusivity and mobility transport coefficients, which are nonlinear functions of the local energy density. We test our findings in kinetic Monte Carlo simulations of the model and, as a proof of concept, we demonstrate the versatility of this driving mechanism to control nonlinear energy transport by inducing time-crystalline phases. In particular, we show that appropriately designed packing fields induce the spontaneous formation of traveling energy condensates, exhibiting robust long-range temporal order reminiscent of continuous time crystals. Our results provide a simple yet powerful framework to study bulk-driven nonlinear energy diffusion in stochastic many-body systems, offering a bridge between microscopic dynamics, macroscopic transport, and controlled spatiotemporal order.

2026-07-08T12:14:01Z 12 pages, 4 figures P. I. Hurtado G. Cortés-Guillén http://arxiv.org/abs/2607.07300v1 2026-07-08T11:42:41Z

A three dimensional global stability analysis is performed to investigate the problem of screeching jets under turbulent conditions. The study employs an Unsteady Reynolds-Averaged Navier-Stokes (URANS) framework, in which the compressible flow equations are discretised using the high-fidelity solver dNami, the linearised discrete system is obtained through the automatic differentiation tool Tapenade, and the global stability problem is solved in a time-stepping framework. The fixed-point solutions of the URANS equations are first validated against experimental and numerical data, then a three dimensional global stability analysis is performed around fixed points solutions at different levels of under-expanded regimes. The extracted modes are spatially analysed and examined in terms of acustic radiation and validated against experimental data. Comparison with experimental POD data shows that the linear modes reproduce the main wavenumber content and spatial organisation of the screech resonance loop, even at high levels of under-expansion. The staging behaviour is also recovered from the interaction between the Kelvin--Helmholtz wave and the dominant wavenumbers of the shock-cell structure. Finally, a Helmholtz decomposition is applied to the velocity perturbation in order to separate the vortical and irrotational parts of the modes. An energy budget of the wave components is then used to quantify the repartition of the relative feedback-loop energy perturbation. Notably, the Mach-number effects on energy partition vary depending on the type of staged mode. This insight could prove valuable for interpreting receptivity mechanisms at nozzle lips and shocks in future research.

2026-07-08T11:42:41Z Alessandro Franchini Nicolas Alferez Jean-Christophe Robinet http://arxiv.org/abs/2601.11946v3 2026-07-08T11:31:01Z

Modelling fluid dynamics with machine learning (ML) has advanced rapidly, yet most data driven approaches remain opaque because they rely on complex architectures to capture nonlinear flow behaviour. This lack of interpretability limits the reliability and hinders the understanding of when and why some models succeed or fail. To address this, we present a transparent approach that provides insights into how data-driven fluids dynamics and ML work. This is achieved by training a convolutional neural network (CNN), on data from a simple laminar fluid flow, to behave as an operator that exactly matches the finite-difference numerics, providing a direct link between well established theory and this new world of ML models. Importantly, the model demonstrates strong generalisation capability by reproducing the dynamics for a wide range of distinct and unseen flow conditions within the same flow category. The CNN learns the forward Euler three-point stencil weights, capturing physical principles such as consistency and symmetry despite having only three tuneable weights. This interpretable ML model goes beyond pure numerical training (numCNN),the approach is shown to work when trained on analytical (anCNN) and even molecular dynam ics (mdCNN) data. In some cases, the physics is not captured, and thanks to the simple and interpretable form, these CNNs provide insight into the limits, pitfalls and best practice of data-driven fluid models. Because the approach is based on finite-difference operators, it naturally extends to many structured-grid computational fluid dynamics problems, including turbulent, multiphase and multiscale flows as well as systems beyond the continuum such as molecular dynamics.

2026-01-17T07:42:59Z 44 pages, 15 figures. This upload contains the final revised version of the article, incorporating all changes made during peer review and matching the version published in the journal Machine Learning: Science and Technology, Volume 7, Number 4, Year 2026 Kwame Agyei-Baah Muhammad Rizwanur Rahman E. R. Smith 10.1088/2632-2153/ae8072 http://arxiv.org/abs/2604.16449v2 2026-07-08T11:15:30Z

Representing turbulent flow fields in a compact yet physically faithful form remains a central challenge in computational fluid dynamics. We propose a continuous parametric representation based on localized Gaussian primitives, in which the velocity field is modeled as a superposition of kernels with learnable positions, amplitudes, and scales. This formulation yields a compact, grid-independent encoding while enabling evaluation of derived quantities such as vorticity and enstrophy. The approach is assessed on three-dimensional Taylor-Green vortex fields spanning stages from smooth flow to fully developed turbulence. We quantify the compression-accuracy trade-off using both primary variables and derivative-sensitive diagnostics. The baseline isotropic formulation achieves high velocity accuracy at compression ratios exceeding 1e3-1e4, but exhibits substantial enstrophy degradation due to loss of small-scale structure. To address this limitation, we investigate structure-aware extensions including adaptive placement, multi-resolution kernels, and anisotropic Gaussians. The anisotropic formulation provides the most consistent improvement, better aligning with elongated vortical structures and recovering intermediate- and high-wavenumber content, while other strategies yield modest gains. A compact-support Beta basis improves enstrophy in some cases but introduces localized artifacts. Overall, the results indicate that the main limitation of baseline Gaussian representations lies in geometric expressiveness rather than parameter count. The proposed framework provides a compact, interpretable, and continuous representation of turbulent flows, and establishes a foundation for structure-aware and physics-informed flow compression.

2026-04-07T18:40:57Z Published in Computers & Fluids Computers & Fluids, Volume 318, 107202, 2026 Dhanush Vittal Shenoy Steven H. Frankel 10.1016/j.compfluid.2026.107202 http://arxiv.org/abs/2607.07275v1 2026-07-08T11:06:29Z

Bubbly flows exhibit complex multiscale dynamics, with deformable bubbles interacting through the surrounding liquid and giving rise to strongly coupled kinematic and morphological behavior. We present BubbleSH, a bubbly flows dataset consisting of transient, three-dimensional bubble-swarm dynamics obtained from high-fidelity direct numerical simulations of bubbles rising in a periodic domain. The dataset provides time-resolved bubble trajectories, velocities, and shape evolution, with bubble morphology compactly represented using spherical harmonics. Designed to be lightweight yet physically expressive, the dataset enables data-driven modeling of bubbly flow simulators where shape deformation and bubble-bubble interactions play a central role. We characterize the dataset with bubble kinematics, morphology, and interaction patterns, and introduce evaluation metrics for both trajectory and shape prediction. The sensitivity of bubble-swarm dynamics to local perturbations makes BubbleSH particularly well suited to generative models that learn distributions over possible future trajectories. We evaluate a permutationally and translationally equivariant probabilistic emulator on BubbleSH given the proposed metrics. Therefore, we establish a compact, high-fidelity dataset and a benchmark for developing and evaluating data-driven models of deformable, chaotic multiphase systems.

2026-07-08T11:06:29Z 17 pages, 7 figures, dataset available here: see https://doi.org/10.5281/zenodo.21229301 Rachna Ramesh Kiet Bennema ten Brinke Douwe Orij Ivo Roghair Vlado Menkovski http://arxiv.org/abs/2607.07246v1 2026-07-08T10:27:06Z

Boundary layer theory and its analytical methods for skin friction coefficients provide an important basis for aerodynamic analysis. However, classical analytical formulas are mostly limited to flat-plate flows. High-fidelity numerical simulations are not only computationally expensive but also yield predictions that are highly sensitive to physical models, numerical schemes, and grid resolution. To overcome these limitations, symbolic AI opens a new pathway to discover novel laws of complex physical systems from data. Using limited data from surface solutions of the Euler equations and the skin friction coefficient from viscous flows over airfoils, we employ symbolic regression to progressively discover a generalizable, interpretable analytical formula chain for fast skin friction prediction in subsonic and supersonic attached flows. From the perspective of physical mechanisms, the discovered analytical expression chain reveals scaling laws for skin friction at different Mach numbers: the basic form captures the logarithmic decay of skin friction along the streamwise direction in the turbulent boundary layer; the inclusion of a pressure coefficient correction term quantifies the effect of surface pressure variation; and the Mach number correction term evolves with flow regimes, transitioning from the compressibility correction term in subsonic regimes to the thermodynamic effects term in supersonic and hypersonic regimes. This knowledge chain exhibits a unified structure across different Mach numbers, and omitting the correction terms under certain conditions recovers classical theoretical forms, further demonstrating its physical consistency. Validation against typical geometries shows that this analytical formula chain achieves a low average integrated skin friction drag prediction error, with good generalization capability across different freestream conditions and geometric shapes.

2026-07-08T10:27:06Z Mingkun Xia Shule Zhao Weiwei Zhang