This paper introduces a novel surrogate modeling framework for aerodynamic applications based on Neural Fields. The proposed approach, MARIO (Modulated Aerodynamic Resolution Invariant Operator), addresses non parametric geometric variability through an efficient shape encoding mechanism and exploits the discretizationinvariant nature of Neural Fields. It enables training on significantly downsampled meshes, while maintaining consistent accuracy during full-resolution inference. These properties allow for efficient modeling of diverse flow conditions, while reducing computational cost and memory requirements compared to traditional CFD solvers and existing surrogate methods. The framework is validated on two complementary datasets that reflect industrial constraints. First, the AirfRANS dataset consists of a two-dimensional airfoil benchmark with non-parametric shape variations. Performance evaluation of MARIO on this case demonstrates an order of magnitude improvement in prediction accuracy over existing methods across velocity, pressure, and turbulent viscosity fields, while accurately capturing boundary layer phenomena and aerodynamic coefficients. Second, the NASA Common Research Model features three-dimensional pressure distributions on a full aircraft surface mesh, with parametric control surface deflections. This configuration confirms MARIO's accuracy and scalability. Benchmarking against state-of-the-art methods demonstrates that Neural Field surrogates can provide rapid and accurate aerodynamic predictions under the computational and data limitations characteristic of industrial applications.
On the path toward truly autonomous robots, embodied agents will require to be adaptable to unforeseen circumstances. Yet, most robotic agents still suffer from significant performance degradation when scenarios change slightly, with many even failing their tasks entirely. In contrast, organisms in nature exhibit strong adaptability, largely due to bio-diversity, which has prevented the extinction of life throughout severe environmental changes. The concept of quality-diversity aims to emulate this natural resilience, yielding robust results through diversification of embodied agents in the behavior space. However, in nature, diversity occurs simultaneously at multiple levels: body, brain, and behavior. This study on the body-brain optimization of virtual embodied agents spans two brain representations-an artificial neural network (ANN) and a graph-and investigates these levels to determine the most critical scope for diversity in fostering performance, generality, and robustness. We start by optimizing for a simple locomotion task, and then evaluate generality through transfer to a diverse set of tasks, including locomotion in new environments and interaction with objects. Our findings confirm the importance of simultaneously considering multiple axes of diversity for achieving good performance and adaptability-demonstrating zero-shot transfer on 18 new tasks. Moreover, we observe that the graph controller performs on par with the ANN, offering greater interpretability.
The positive rates conjecture states that a one-dimensional probabilistic cellular automaton (PCA) with strictly positive transition rates must be ergodic. The conjecture has been refuted by Gács, whose counterexample is a cellular automaton that is non-ergodic under uniform random noise with sufficiently small rate. For all known counterexamples, non-ergodicity has been proved under small enough rates. Conversely, all cellular automata are ergodic with sufficiently high-rate noise. No other types of phase transitions of ergodicity are known, and the behavior of known counterexamples under intermediate noise rates is unknown. We present an example of a cellular automaton with two phase transitions. Using Gács's result as a black box, we construct a cellular automaton that is ergodic under small noise rates, non-ergodic for slightly higher rates, and again ergodic for rates close to 1.
Graph Neural Networks have been applied to learn the flight and structural dynamics of a High-Altitude Long-Endurance aircraft in response to discrete gusts. The graph network methodology enables the development of a model for structural displacements, loads and aircraft flight dynamics leveraging on the inductive bias provided by the physical connections. Neural Ordinary Differential Equations have been integrated with Graph Neural Network in a novel architecture using exogenous inputs. The results demonstrate promising capabilities in model approximation improving on traditional graph networks, in particular for long term predictions in time by reducing integration drift errors. Even without targeted software optimization, the surrogate model provides an approximately 200-fold increase in computational speed compared to the original simulation environment.
Abstract Computational visual intelligence has been shown to be able to comprehend the content of images, which has been widely used to foster a digitized society, but is often underutilized in applications related to the green transition and climate change mitigation. Here, we evaluate the capacity of convolutional neural networks (CNN) to interpret spatial semantic patterns in optical RGB images to directly estimate forest biomass, an essential climate parameter previously assessed from structural measures of trees. Trained with forest inventory plots, the CNN model demonstrates its learning via interpreting the composition of biomass at tree level, differing from traditional approaches reliant on conversions of aggregated parameters without explanatory rationale. The CNN approach yields consistently low bias across wide biomass ranges, whereas traditional models show insufficiency without information on tree height. Visually interpretable models link advanced computational tools with the power of data, facilitating the sustainable management of resources for a carbon-neutral society.
