萨瓦大学

Earthworms are key drivers of soil function, influencing organic matter turnover, nutrient cycling, and soil structure. Understanding the environmental controls on their distribution is essential for predicting the impacts of land use and climate change on soil ecosystems. While local studies have identified abiotic drivers of earthworm communities, broad-scale spatial patterns remain underexplored. We developed a multi-species, multi-task deep learning model to jointly predict the distribution of 77 earthworm species across metropolitan France, using historical (1960-1970) and contemporary (1990-2020) records. The model integrates climate, soil, and land cover variables to estimate habitat suitability. We applied SHapley Additive exPlanations (SHAP) to identify key environmental drivers and used species clustering to reveal ecological response groups. The joint model achieved high predictive performance (TSS >= 0.7) and improved predictions for rare species compared to traditional species distribution models. Shared feature extraction across species allowed for more robust identification of common and contrasting environmental responses. Precipitation variability, temperature seasonality, and land cover emerged as dominant predictors of earthworm distribution. Species clustering revealed distinct ecological strategies tied to climatic and land use gradients. Our study advances both the methodological and ecological understanding of soil biodiversity. We demonstrate the utility of interpretable deep learning approaches for large-scale soil fauna modeling and provide new insights into earthworm habitat specialization. These findings support improved soil biodiversity monitoring and conservation planning in the face of global environmental change.

Additive manufacturing (3D printing) has emerged as a transformative technology for producing complex metallic components with high precision and design flexibility. Aluminum-based alloys, in particular, are widely used in aerospace, automotive, and biomedical applications, where accurate prediction of their mechanical properties is critical for material selection and process optimization. This study proposes a hybrid predictive framework that combines ab initio simulations using CASTEP with data-driven machine learning models. Four alloy systems were analyzed: Al, Ti, Ni, and their combinations. Two machine learning approaches were evaluated: multiple linear regression (MLR) and deep neural networks (DNN). While MLR provides fast and interpretable results, its performance deteriorates for alloys exhibiting nonlinear elastic behavior, such as Al-Ni (R & sup2; = 0.6463). Conversely, the DNN model, implemented with three hidden layers and L2 regularization, successfully captured the nonlinear correlations between elastic stiffness constants and mechanical moduli, achieving R & sup2; > 0.99 across all alloys. Once trained on DFT-generated data, the DNN enables rapid prediction of mechanical properties at a negligible computational cost compared to performing new full DFT-based simulations. The synergy between deep learning and ab initio simulations offers a promising route for accelerating the design and optimization of aluminum alloys tailored for additive manufacturing. This approach enables precise prediction of bulk, shear, and Young's moduli while reducing the need for repeated computationally expensive quantum-level calculations, thus bridging the gap between first-principles simulations and practical materials engineering.

This work presents an experimental study of ionization wave propagation in a helium plasma jet ignited with a 500 ns pulsed high voltage operating with and without a secondary grounded electrode. Fast camera imaging is coupled with non-perturbative electric field measurements using a Pockels effect-based probe to analyse the ionization front (IF) dynamics. By subtracting the Laplacian electric field components, both radial and axial electric field components generated by the plasma jet are extracted and their temporal evolution is determined. The presence of a grounded electrode is shown to double the IF propagation length and induce restrikes in the inter-electrode region, as evidenced by characteristic drops in the radial electric field component. The axial evolution of the radial electric field and IF velocity is reported, together with their correlation as a function of applied voltage.

OBJECTIVE:Attention impairments are common in children with epilepsy and widely impact their quality of life. Interictal epileptiform discharges (IEDs) may induce subtle dysfunctions of various cognitive processes, but data regarding the impact of IEDs on sustained attention remain limited. The objective of the present study was to evaluate the impact of IEDs on continuous undivided attention in children with epilepsy, controlling for the number of treatments, type of epilepsy, frequency of seizures in the past year, age at onset, and comorbid attention disorder. METHODS:Using a computerized sustained attention test synchronized with the electroencephalogram in 61 children with diverse epilepsy syndromes, reaction time (RT), errors, attention stability over time, and the event-related potential (ERP) in Pz (related to attentional engagement) were collected. The cumulative impact of IEDs was evaluated using multivariate models controlling for epilepsy-related factors. The transient impact of IEDs was assessed by comparing responses in trials with and without IEDs. RESULTS:IEDs were associated with attention fluctuations independently from other epilepsy-related factors. In terms of cumulative impact, a higher IED rate was associated with a poorer sustained attention performance over the entire task. In terms of transient impact, trials disrupted by IEDs were characterized by longer RT and a lower amplitude of the ERP extending over a long time window that included attentional processing (P300). SIGNIFICANCE:These results suggest that IEDs may negatively impact sustained attention, independently of other epilepsy-related features. These findings support the hypothesis that IEDs could contribute to subtle attentional deficits and serve as potential biomarkers of abnormal brain function.

Primordial black holes (PBHs) provide a unique probe of the early Universe and may have an enhanced abundance in bouncing cosmologies, where a long contracting phase can amplify perturbations. We develop a unified framework to study PBH formation in dust-radiation bouncing cosmologies, focusing on the classical contracting phase so that the results are insensitive to bounce details. We compute the curvature power spectrum for an extremely small dust equation of state using a stable semi-analytical (adiabatic) method, derive the Jeans length of the two-fluid system using dynamical-system analysis and the WKB approximation, and extend the three-zone model from the single- to the two-fluid case to model local collapse. We implement two collapse criteria to obtain the curvature perturbation threshold for PBH formation and estimate PBH mass fractions for benchmark masses spanning low-mass (10^-17 M_⊙) to supermassive (10^13 M_⊙) scales. The critical curvature threshold is extremely small and nearly mass-independent over a broad range (ζ_c ∼ 10^-21 for 10^-14 to 10^13 M_⊙), with deviations only near dust-radiation equality. Nevertheless, the square root of the curvature power spectrum at the relevant formation times is many orders of magnitude smaller, yielding vanishingly small PBH mass fractions across the benchmark masses. Compared with the pure-dust case, radiation pressure and the two-fluid collapse conditions significantly suppress PBH production, implying that substantial PBH formation in dust-radiation bouncing cosmologies would require additional mechanisms to amplify curvature perturbations.