activity
20242026
collaborators

6 papers

physics.flu-dyn2026

A Hybrid Generative Reduced-Order Model for the Minimal Flow Unit

Niccolò Tonioni, Lionel Agostini, Marcial Sanchis-Agudo +4

A data-driven reduced-order modelling framework is proposed for wall-bounded turbulent flows to forecast the intermittent near-wall dynamics over extended time horizons from sparse…

physics.flu-dyn2026

X-CAL: Explaining latent causality in physical space for fluid mechanics

Marcial Sanchis-Agudo, Andrés Cremades, Alvaro Martinez-Sanchez +2

We present X-CAL, a pipeline that combines a -variational autoencoder (-VAE) with the synergistic-unique-redundant decomposition (SURD)~\cite{surd} approach for causality a…

physics.flu-dyn2026

A Mixed-Metric Two-Field Framework for Turbulence: Emergent Stress Anisotropy and Wall Asymptotics from a Single Scalar

Marcial Sanchis-Agudo, Ricardo Vinuesa

In our previous work~\cite{SanchisAgudoVinuesa2025PRL}, we argued that viscous dissipation in turbulence can be understood as the macroscopic imprint of microscopic path uncertaint…

physics.flu-dyn2025

A Geometric Foundation for the Universal Laws of Turbulence

Marcial Sanchis-Agudo, Ricardo Vinuesa

We propose a theoretical framework where the dissipative structures of turbulence emerge from microscopic path uncertainty. By modeling fluid parcels as stochastic tracers governed…

cs.LG2025

Easy attention: A simple attention mechanism for temporal predictions with transformers

Marcial Sanchis-Agudo, Yuning Wang, Roger Arnau +4

To improve the robustness of transformer neural networks used for temporal-dynamics prediction of chaotic systems, we propose a novel attention mechanism called easy attention whic…

physics.flu-dyn2024

On Deep-Learning-Based Closures for Algebraic Surrogate Models of Turbulent Flows

Benet Eiximeno, Marcial Sanchís-Agudo, Arnau Miró +3

A deep-learning-based closure model to address energy loss in low-dimensional surrogate models based on proper-orthogonal-decomposition (POD) modes is introduced. Using a transform…