Characterizing the mechanical response of materials traditionally requires solving optimization problems in which model parameters are calibrated or trained to minimize the discrepancy between model predictions and experimental data. This process can be computationally expensive and time-consuming. To overcome this limitation, we propose two neural operator architectures that directly map experimentally measured data to the constitutive functions governing the mechanical response of the material: Physics-Augmented Neural Operators (PANO) and Constitutive Artificial Neural Operators (CANO). The proposed neural operators approximate the mapping between the infinite-dimensional input space of full-field displacement measurements and net reaction forces, and the infinite-dimensional output space of hyperelastic strain–energy density functions. The displacement fields are encoded through Laplacian eigenfunctions to obtain discretization-independent and noise-robust predictions. We constrain the output space to physically admissible material models that satisfy fundamental physical requirements by design. Here, we focus on isotropic, incompressible hyperelasticity and assume that the material properties are uniform across the specimen. The neural operators are trained on simulated data tuples of displacement fields and reaction force functions for a range of material models. Once trained, the neural operators enable near-instantaneous material characterization and require only a single forward pass to infer the strain–energy density function from a given experimental dataset. We test the predictive power of the neural operators for unseen data, noisy data, data with missing information, data from different spatial discretizations, and data from geometries of different sizes. We finally discuss the ill-posedness of the inverse material characterization problem and show that constraining the output function space of our neural operator framework sufficiently regularizes the problem. The trained neural operators enable rapid and robust discovery of polyconvex strain–energy density functions while avoiding the need to solve computationally expensive inverse problems.
With the growth in the use of digital payment systems, restaurants have the opportunity to alter how information is visually presented to customers during the payment process. One such option is the directional presentation of tip options included during the transaction (i.e., ascending vs. descending percentage values). In a field study conducted in partnership with an independently owned coffee shop, this research examines how the direction of the tip display impacts the amount customers leave as gratuities. The results support the impact of left-to-right processing, in which customers tip more when tip percentage options are presented in descending order (vs. ascending). These findings provide actionable insights for restaurant operators and demonstrate how subtle changes in digital payment system design can benefit service employees.
Rheumatic diseases constitute a major cause of chronic pain, functional impairment, and long-term disability worldwide. In rheumatic diseases, reliable assessment of structural damage and inflammatory activity is essential for diagnosis, disease monitoring, and treatment response evaluation, yet remains largely dependent on time-intensive expert image interpretation. In this study, we develop and evaluate a pipeline for automatic landmark detection and subsequent pathology scoring in hand magnetic resonance images for three main pathologies in rheumatic diseases, namely erosions, osteitis, and synovitis. We explicitly exploit two orthogonal acquisitions (coronal and transversal) by integrating multi-view information at different stages of the pipeline to assess their impact on automatic scoring performance. We train and compare two landmark detection models that utilize all three magnetic resonance imaging sequences to predict predefined landmarks annotated by experts. The YOLO model achieves better landmark predictions for both metrics and across all distances, with a successful detection rate of 94% for a clinically relevant distance of 6 mm and an overall mean Euclidean distance of 3 mm from ground truth landmarks to predicted landmarks. By using a super-resolution approach to fuse coronal and transversal images for the automatic scoring, we achieve an improved performance for synovitis detection. This paves the way for more fully automated precision medicine in magnetic resonance imaging, reducing the workload for physicians while enabling faster, more standardized results to support the decision-making process.
The increasing operational variability of power systems motivates data-driven methods for fault analysis in protection applications. This paper presents a systematic evaluation of deep learning models for fault detection, classification, line identification, and localization. Models are trained on EMT-simulated voltage and current waveforms from the public PROTECT-90 dataset, a 90 kV double-line benchmark. Recurrent, convolutional, hybrid, and transformer-based architectures are compared under shared data splits, temporal decision horizons, and leakage-aware validation protocols to assess predictive performance, inference latency, and model complexity. The results reveal distinct task characteristics. Detection, classification, and line identification reach near-saturated performance once short post-fault transients are available, whereas localization depends strongly on temporal context and improves with longer observation windows. Additional sensitivity analyses contextualize localization under reduced relay observability, test-time communication perturbations, and impedance-based reference methods. Across tasks, compact recurrent and hybrid architectures nearly match the predictive performance of larger architectures while maintaining shorter inference times, indicating diminishing returns from increased capacity under centralized EMT-based measurements. The findings provide guidance for architectural selection in learning-based protection and post-fault analysis under controlled, reproducible EMT sensing assumptions.
To holistically understand the biology of animals, we must unravel the complexities and specificities of host-microbe interactions across animal taxa. Birds represent enigmatic and scientifically compelling hosts in which to understand these interactions. Here, we present a brief summary of a series of conversations among avian microbiome researchers regarding methodological challenges facing the avian microbiome field, where most research to date has focused on bacterial communities of the gut. Collectively, we acknowledged a commonly shared but underreported issue facing the avian microbiome field: that of difficulty in obtaining high-quality and high-yield microbial DNA from avian fecal samples. We discuss some of the potential reasons underlying low DNA yields, such as inhibitory compounds and rapid DNA degradation, and provide recommendations for how researchers in the avian microbiome field might cope with these methodological challenges. Collective and dedicated efforts to address these challenges will be required for a robust understanding of host-microbe interactions in avian systems.
